Asymmetric synthesis of (S)-mirtazapine: Unexpected racemization through an aromatic ipso-attack mechanism

Article abstract of DOI:10.1002/ejoc.200701224

An asymmetric synthesis of (S)-mirtazapine has been achieved from the synthesis of the racemate by using (S)-1-methyl-3-phenylpiperazine as the starting material. Unfortunately, significant racemization was encountered in the final step, which involved an electrophilic aromatic ring closure of a alcohol by concentrated sulfuric acid. A significantly higher ee was observed when polyphosphoric acid (PPA) was used instead. A remarkable correlation between the amount of PPA used and the ee of the product was revealed, namely, an increase in the ee upon decreasing the amount of PPA. This trend was paralleled by the formation of an increasing amount of a side-product upon lowering the amount of PPA. The racemization and formation of a side-product can be explained by an ipso-attack mechanism during the electrophilic aromatic ring-closure reaction. This mechanism was supported by a mechanistic study using a deuterium-labeled substrate. Wiley-VCH Verlag GmbH & Co. KGaA, 2008.

Full text of DOI:10.1002/ejoc.200701224

FULL PAPER
DOI: 10.1002/ejoc.200701224
Asymmetric Synthesis of (S)-Mirtazapine: Unexpected Racemization through
an Aromatic ipso-Attack Mechanism
Marco van der Linden,[a] Judith Borsboom,[a] Frans Kaspersen,[a] and Gerjan Kemperman*[a]
Keywords: Asymmetric synthesis / Reaction mechanisms / (S)-Mirtazapine / Isotopic labeling
An asymmetric synthesis of (S)-mirtazapine has been
achieved from the synthesis of the racemate by using (S)-1-
methyl-3-phenylpiperazine as the starting material. Unfortu-
nately, significant racemization was encountered in the final
step, which involved an electrophilic aromatic ring closure of
a alcohol by concentrated sulfuric acid. A significantly higher
ee was observed when polyphosphoric acid (PPA) was used
instead. A remarkable correlation between the amount of
PPA used and the ee of the product was revealed, namely,
an increase in the ee upon decreasing the amount of PPA.
This trend was paralleled by the formation of an increasing
amount of a side-product upon lowering the amount of PPA.
The racemization and formation of a side-product can be ex-
plained by an ipso-attack mechanism during the electrophilic
aromatic ring-closure reaction. This mechanism was sup-
ported by a mechanistic study using a deuterium-labeled
substrate.
(© Wiley-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim,
Germany, 2008)
pine. It would be particularly interesting to see if an asym-
metric synthesis could be deduced from the synthesis of ra-
cemic mirtazapine, which starts with 2-chloro-3-cyanopyr-
idine (1) and 1-methyl-3-phenyl-piperazine (2), as shown in
Scheme 1.[1] It was anticipated that the chiral centre of (S)-
mirtazapine could be introduced via the 1-methyl-3-phenyl-
piperazine building block and consequently the use of
enantiopure (S)-1-methyl-3-phenylpiperazine in the synthe-
sis of mirtazapine should in principle lead to a viable stereo-
convergent synthesis of (S)-mirtazapine.
Introduction
Mirtazapine is a tetracyclic compound that as a racemate
finds widespread use in the treatment of depression. The
synthesis of mirtazapine has mainly been covered in
patents.[1–3] In addition, the synthesis of tritium-, carbon-
13-, and carbon-14-radiolabelled mirtazapine has been de-
scribed in literature.[4] The biological effects of mirtazapine
have been comprehensively reviewed over the past two dec-
ades.[5,6] Investigations into the biological effects of the
enantiomers of mirtazapine revealed interesting properties
of the compound in its pure enantiomeric form,[7] for exam-
ple, the (S) enantiomer has demonstrated potential in the
treatments of insomnia and the climacteric symptoms asso-
ciated with the menopause. In order to obtain (S)-mirt-
azapine in sufficient quantities to start pharmaceutical and
clinical development, a classical process for the resolution
of the enantiomers of the readily available mirtazapine was
developed. This process, employing dibenzoyl--tartaric
acid as the resolving agent, has been used to manufacture
over 1300 kg of (S)-mirtazapine maleate in a batch scale up
to 250 kg.
Scheme 1. Retrosynthesis of mirtazapine.
This paper describes the efforts to develop an asymmetric
synthesis of (S)-mirtazapine starting from 2-chloro-3-cya-
nopyridine (1) and (S)-1-methyl-3-phenylpiperazine [(S)-2].
In addition, the mechanistic aspects of an unexpected race-
mization are revealed.
The resolution process is, however, inherently associated
with a 65% loss of mirtazapine and with the formation of
large amounts of waste. For the commercial manufacture
of (S)-mirtazapine maleate this process is thus not an eco-
nomically or environmentally benign option. This has trig-
gered the search for an asymmetric synthesis of (S)-mirtaza-
Results and Discussion
[a] Organon N. V., part of Schering-Plough, Department of Pro-
cess Chemistry,
In order to obtain significant quantities of (S)-1-methyl-
3-phenylpiperazine [(S)-2] for use in the synthesis of (S)-
mirtazapine, a classical resolution using (S)-(+)-Anicyphos
P. O. Box 20, 5340 BH Oss, The Netherlands
E-mail: gerjan.kemperman@organon.com
Eur. J. Org. Chem. 2008, 2989–2997
© 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2989

M. van der Linden, J. Borsboom, F. Kaspersen, G. Kemperman
FULL PAPER
was exploited (see Figure 1).[8] From a series of resolving 4 was obtained in a yield of 57% and with an ee Ͼ 99%.
agents, this compound appeared the most effective. First Reaction of carboxylic acid 4 with lithium aluminium hy-
the salt of (S)-2 with (S)-(+)-Anicyphos was crystallized dride furnished cleanly and in excellent yield alcohol 5. In
from water. The free base of (S)-2 was liberated by treat- the last step, alcohol 5 was treated with concentrated sulfu-
ment with sodium hydroxide and extraction with ethyl ace- ric acid and the thus formed cation underwent electrophilic
tate. In this way (S)-2 was obtained in an overall yield of aromatic substitution at the phenyl ring. Disappointingly, a
between 35–40% and with an ee Ͼ98%. This process has considerable loss in enantiomeric excess was observed in
been applied to the preparation of more than a kg of this this last step as the ee of (S)-mirtazapine (compound 6) was
optically active building block. Although by this process found to be only 36%.
65% of 1-methyl-3-phenylpiperazine 2 was lost, from an
economic point of view, this is preferred over resolution of
mirtazapine as the latter is a compound with significantly
more added value. Moreover, several alternative routes to
(S)-1-methyl-3-phenylpiperazine have been discovered. One
of these proceeds by enzymatic resolution of the oxalamate
derivative of the racemic mixture.[9] In addition, a number
of stereoconvergent synthetic routes starting from (S)-phen-
ylglycine have been elaborated of which one is shown in this
paper (see Scheme 4).[10]
It was demonstrated that (S)-mirtazapine is enantiostable
in concentrated sulfuric acid and thus the observed racemi-
zation had to be related to the mechanism of the electro-
philic ring-closing reaction. However, the mechanism of the
racemization was totally unclear and it was presumed that
the cation was not entirely enantiostable. In case the racemi-
zation takes place in the cation, its possible stabilization
by using coordinative solvents could result in an improved
enantiopurity of the product. It can be seen from Table 1
that when a noncoordinating solvent like dichloromethane
was used (entry 2), the ee of the process did not improve.
On the other hand when coordinating solvents like alcohols
or carboxylic acids were used (entries 3–7) there was a
strong positive effect on the enantiopurity, leading to values
twice as high, that is, 60–70%. Nevertheless, an ee of 60–
70% is still insufficient for a viable asymmetric synthesis.
A series of different conditions, known for their ability to
accomplish dehydration reactions, were investigated
(Table 1 entries 8–17). Surprisingly, a considerably higher ee
Figure 1. Structure of (S)-(+)-Anicyphos.
The optically active (S)-1-methyl-3-phenylpiperazine (S)- was obtained when phosphorus pentoxide was used even
2 was used in the synthesis of (S)-mirtazapine, which is de- with the noncoordinating solvent xylene. More remarkable
picted in Scheme 2. The first step involves coupling of (S)- was the finding that when phosphorus pentoxide was used
2 and 2-chloro-3-cyanopyridine (1) in boiling DMF using in combination with the polar solvents DMF or N-methyl-
potassium fluoride to facilitate the nucleophilic aromatic pyrrolidinone (NMP) no racemization was observed at all.
substitution. During this reaction no loss of optical purity As the reaction mixtures with phosphorus pentoxide were
was found and the nitrile 3 was directly converted into car- very sticky and therefore difficult to stir, a liquid analogue
boxylic acid 4 by treatment with sodium hydroxide in a mix- of phosphorus pentoxide, namely polyphosphoric acid, was
ture of methanol and water. After crystallization, the acid used instead. The use of only polyphosphoric acid (PPA)[11]
Scheme 2. Synthesis of (S)-mirtazapine.
2990
www.eurjoc.org
© 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2008, 2989–2997

Asymmetric Synthesis of (S)-Mirtazapine
(entry 13) also resulted in a higher ee than was obtained Table 2. Effect of the amount of PPA on the ee of the product.
with sulfuric acid. When PPA was used with the polar coor-
dinating solvents DMF or NMP (entries 14 and 15) no loss
of ee was observed, as was the case for phosphorus pentox-
ide.
Entry
Equiv. PPA (m/m)
% ee
Impurity (a/a%)[a]
1
2
3
4
5
6
7
30
19
5
4
3
2.5
2
73
76
81
82
91
95
99
n.d.
n.d.
n.d.
0.3
1.8
2.1
Table 1. Effect of various reaction conditions on the ee of the elec-
trophilic ring-closure step.
6–8
Entry
Acid (anhydride) Solvent
% ee
4[b]
2.3[b]
2.1[b]
2[b]
83
99
99
99
99
n.d.
6.6
7.7
8.4
10
1
2
3
4
5
6
7
H2SO4
H2SO4
H2SO4
H2SO4
H2SO4
H2SO4
H2SO4
–
36
36
67
57
57
57
70
8
9
10
11
12
CH2Cl2
EtOH
MeOH
ethylene glycol
acetic acid
acetic anhydride
1.6[b]
[a] Determined by GC relative to the peak of (S)-mirtazapine. [b]
The oxalate salt of the alcohol 5 was used.
8
9
MeSO3H
MeSO3H
–
55
75
dibutyl ether
place leading to intermediate 8. After breaking of the C–C
bond with the piperazine ring, the stabilized achiral piper-
azine cation 9 is formed. Cation 9 could attack the phenyl
ring, which, after loss of a proton, would lead to racemic
mirtazapine. Alternatively, cation 9 can be intercepted by a
nucleophile, for example, a phosphate group or a nucleo-
philic solvent.[12] This would lead to cyclic N,O-acetal-like
compound 10. Because N,O-acetals are generally not very
stable, under the rather harsh reaction conditions com-
pound 10 decomposes in a number of steps to the side-
product shown in Figure 2. Quantitatively, the amount of
side-product observed under conditions in which racemiza-
tion is prevented (Table 2, entries 7, 9, 10, 11, and 12; 8–
10% of side-product) is not the same as the amount of race-
mization at higher dilutions (Table 2, entry 1; 27% racemi-
zation). This can be explained by other nonidentified side-
products that are formed in smaller amounts by the inter-
ception of the same piperazine cation intermediate 9. The
compound shown in Figure 2 and referred to in Table 2 as
the side-product predominates, and was formed in such
amounts that it could be isolated. The effect of the amount
of PPA relative to the alcohol on the enantiomeric excess
of the product and on the amount of side-product forma-
tion can also be explained by this mechanism. In highly
diluted reaction mixtures, that is, with a large excess of
PPA, the medium is very acidic and thus not very nucleo-
philic. Presumably, under these conditions, the longevity of
the piperazine cation is sufficient to re-attack the phenyl
ring thereby forming racemic mirtazapine. Conversely, with
smaller amounts of PPA, due to the basic nature of the
starting material, the medium is more nucleophilic such that
the piperazine cation is intercepted completely and the path
towards racemization is blocked.
10
11
12
P2O5
P2O5
P2O5
xylene
DMF
NMP
70
Ͼ99
Ͼ99
13
14
15
16
17
PPA
PPA
PPA
PPA
PPA
–
73
Ͼ99
Ͼ99
77
DMF[a]
NMP[b]
dibutyl ether[c]
[d]
H3PO4
80
[a] 6–60 volumes of DMF per gram of starting material. [b] 1–
60 volumes of NMP per gram of starting material. [c] 2 volumes
of dibutyl ether per gram of starting material. [d] 3–5 volumes of
phosphoric acid (85%) per gram of starting material.
The ring-closure reaction with PPA was studied in more
detail. When the amount of PPA relative to alcohol 5 was
varied, a remarkable trend was revealed. The data in Table 2
show that the enantiomeric excess of the product increased
upon decreasing the amount of PPA (entries 1–7). Whereas
with 30 wt.-equiv. of PPA the enantiomeric excess of the
product was only 73%, with 2 wt.-equiv. complete retention
of the ee was observed. Unfortunately there was a flip side
to this discovery. Upon decreasing the amount of PPA, a
side-product was formed in increasing amounts and in up
to 8% yield when 2 wt.-equiv. of PPA were used. A similar
series of experiments was carried out with the oxalate salt
of alcohol 5 (entries 8–12) that showed the same trends, that
is, an increase in the ee of the product paralleled by an
increasing amount of an impurity. It should be noted that
a number of other impurities were formed as well but the
impurity referred to in Table 2 predominates.
The data in Table 2 suggest a relationship between the
mechanism of the racemization and the mechanism in-
volved in the formation of the impurity. More precisely, it
seems that the path leading to racemization at higher di-
lutions in PPA, that is, with a large excess of PPA, is being
blocked by a pathway leading to the predominant impurity.
The structure of the impurity was elucidated and is shown
in Figure 2. Both the racemization and the side-product for-
mation can be explained by the mechanism shown in
Scheme 3. It is conceivable that besides ortho attack an ipso
Figure 2. Structure of the side-product formed in the synthesis of
attack of the cation 7 at the phenyl ring could also take (S)-mirtazapine.
Eur. J. Org. Chem. 2008, 2989–2997
© 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
2991

M. van der Linden, J. Borsboom, F. Kaspersen, G. Kemperman
FULL PAPER
Scheme 3. Mechanism proposed for the racemization and side-product formation.
It was found that by using 2 wt.-equiv. of PPA for the label at the para position of the phenyl ring according to
ring-closing reaction, an asymmetric synthesis of (S)-mirta- the synthesis shown in Scheme 4. To this end, p-bro-
zapine could be achieved. By treating the alcohol or its oxa- mobenzaldehyde (12) was protected as a diethyl acetal, lithi-
late salt with 2 wt.-equiv. of PPA at 130 °C, (S)-mirtazapine ated at the position of the bromide and quenched in deute-
was obtained in excellent enantiopurity.[13] The impurities rium oxide. Hydrolysis of the ethyl acetal 14 furnished p-
formed were effectively removed by crystallization of the deuteriobenzaldehyde (15), which, by a Strecker synthesis,
maleate salt. The maleate salt was obtained in a yield be- was converted into 16. Compound 16 was converted into 1-
tween 60 and 70%.
methyl-3-(p-deuteriophenyl)piperazine (20) according to a
In order to substantiate the mechanism shown in procedure developed by Medichem.[14] From the deuter-
Scheme 3, the alcohol 21 was prepared with a deuterium iated piperazine 20, alcohol 21 was prepared by the route
Scheme 4. Synthesis of p-deuteriophenyl alcohol 21.
2992
www.eurjoc.org
© 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2008, 2989–2997

Asymmetric Synthesis of (S)-Mirtazapine
Scheme 5. Anticipated position of the deuterium atom in mirtazapine after ortho and ipso attack.
described in Scheme 2. The NMR spectrum of 21 con-
The deuteriated alcohol 21 was subjected to both race-
firmed the presence of the deuterium atom exclusively at mizing and nonracemizing ring-closing conditions. Follow-
the para position of the phenyl substituent.
ing the mechanism in Scheme 5, it can be deduced that after
1
Figure 3. H NMR spectra (600 MHz in acetone) obtained after the ring-closure step under different reaction conditions.
Eur. J. Org. Chem. 2008, 2989–2997
© 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
2993

M. van der Linden, J. Borsboom, F. Kaspersen, G. Kemperman
FULL PAPER
Table 3. Effect of different reaction conditions on the ring-closure step.
Exp.
Conditions
Yield (%)
Purity by GC (a/a%)
I
II
III
PPA (6 equiv., m/m) + NMP (28 vol.); 100 °C
PPA (2.5 equiv., m/m); 100 °C
PPA (7 equiv., m/m); 100 °C
86
31
51
96[a]
92[a]
99
[a] After silica gel chromatography.
ortho attack the deuterium atom will eventually be present pered by racemization in the final step, this problem could
in mirtazapine at the 12-position. On the other hand, if an be solved by using 2 wt.-equiv. of PPA instead of sulfuric
ipso attack is involved as well, this would lead to mirtazap- acid for the electrophilic aromatic ring closure. A corre-
ine with the deuterium atom at the 13-position. It was dem- lation between the extent of racemization and the amount
onstrated that no scrambling of deuterium can take place of PPA used in this ring-closure reaction was observed. An
under the reaction conditions using PPA. However, when ipso-attack mechanism was postulated to explain the race-
the alcohol was subjected to reaction with deuteriated sul- mization under more dilute conditions of PPA and the re-
furic acid scrambling of the hydrogen atoms of the phenyl tention of ee concurrent with side-product formation at
ring was observed. Therefore, for the study with deuteriated higher concentrations. This ipso attack is likely driven by
alcohol 21, only PPA was used.
the formation of a stabilized piperazine cation. The ipso-
The deuteriated alcohol was subjected to three different attack mechanism was supported by a mechanistic study
sets of reaction conditions, which resulted in no racemiza- using deuterium-labeled starting material for the final elec-
tion (Experiment I), little racemization (Experiment II), trophilic ring-closing reaction.
and significant racemization (Experiment III). The NMR
spectra of the products resulting from these experiments are
shown in Figure 3[15] and the yields and purities of the
products are compiled in Table 3. From Table 3 it can be
seen that the reaction in diluted PPA (Experiment III) gave
a relatively clean reaction product but that those with less
Experimental Section
General Information: NMR spectra were recorded with a Bruker
DPX 400 spectrometer. Chemical shifts are reported in parts per
1
million (ppm). H NMR chemical shifts are referenced to TMS as
equivalents of PPA and with PPA in combination with internal standard (abbreviations: s = singlet, d = doublet, t = trip-
let, m = multiplet, dd = double doublet, dt = double triplet, ax =
axial, eq = equatorial). Mass spectra were recorded with a PE
SCIEX API 165 spectrometer. Gas chromatography was performed
with a HP6890N instrument obtained from Agilent with a Restek
RTX-column (5 mϫ0.25 mm inner diameterϫ0.5 µm diameter
film; injection temperature 250 °C with a FID detector; tempera-
ture program: starting at 60 °C, 30 °Cmin–1 up to 300 °C, held at
300 °C for 2 min.
NMP required chromatographic purification. The purities
of the products, as assessed by GC, were over 90% for all
samples. From the NMR spectrum of the product formed
under nonracemizing conditions, that is, Experiment I, it
can be seen that the signal of 12-H is missing. Clearly, this
position is completely blocked by the deuterium atom
pointing to the formation of mirtazapine by normal ortho
attack only. The experiment employing slightly racemizing
conditions, Experiment II,[16] showed a signal arising from
a hydrogen atom at the 12-position, whereas the signal of
13-H is smaller relative to the other signals. This indicates
that part of the mirtazapine is formed by ortho attack, leav-
ing the deuterium atom at the 12-position, and part of the
mirtazapine is formed by ipso attack resulting in a shift of
the deuterium atom to the 13-position. The NMR spectrum
from Experiment III, in which significant racemization oc-
curred, showed that the signal of 13-H had decreased fur-
ther, whereas that of 12-H was higher. This is in line with
the expectations in more diluted PPA without a coordinat-
ing solvent such as NMP, as in this medium racemization
was more pronounced. In conclusion, the spectra in Fig-
ure 3 nicely support the mechanism proposed for racemiza-
tion shown in Scheme 3.
The polyphosphoric acid was prepared by adding 1 weight equiva-
lent of phosphorus pentoxide in portions to 85% phosphoric acid
while maintaining the temperature below 140 °C. For example,
100 g P2O5 was added in portions to 100 g phosphoric acid.
(S)-1-Methyl-3-phenylpiperazine·(+)-Anicyphos
Salt:
(R,S)-1-
Methyl-3-phenylpiperazine (1) (100 g, 567 mmol) and (+)-An-
icyphos (154.5 g, 571 mmol) were dissolved in water (250 mL) by
heating the mixture at reflux. After cooling down to room tempera-
ture a seed crystal was added. After 2 h, the white crystals formed
were collected by filtration and dried in a vacuum oven at 40 °C
for 21 h. This provided 121 g of the salt (48%) with an ee of 85.5%.
The crystals were dissolved in water (119 mL) at reflux tempera-
ture. After cooling down crystallization started. After 1 h the crys-
tals were collected by filtration and dried in a vacuum oven at
40 °C. The yield of the crystals of the (S)-1-methyl-3-
phenylpiperazine·Anicyphos salt was 105.8 g (42%, ee 99.0%).
The ee was determined by HPLC analysis: Chiralcel OD
250ϫ4.6 mm i.d. (Daicel), 5% isopropyl alcohol in hexane, flow
rate 1.0 mLmin–1, UV detector, column temperature 40 °C, reten-
tion times 5.6 and 6.3 min.
Conclusions
An asymmetric synthesis of (S)-mirtazapine has success-
fully been derived from the synthesis of racemic mirtazap-
ine starting from enantiomerically pure (S)-1-methyl-3-
(S)-1-Methyl-3-phenylpiperazine [(S)-2]: The (S)-1-methyl-3-phenyl-
piperazine·(+)Anicyphos salt (100 g) was suspended in dichloro-
methane (378 mL). A 25% ammonium hydroxide (63.2 mL) aque-
phenylpiperazine. Although this synthesis was initially ham- ous solution (315 mL) was added to this stirred suspension. The
2994 Eur. J. Org. Chem. 2008, 2989–2997
www.eurjoc.org
© 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Asymmetric Synthesis of (S)-Mirtazapine
layers were separated. The dichloromethane layer was washed three
times with a saturated aqueous solution of sodium chloride. The
dichloromethane layer was dried with magnesium sulfate and the
solvents evaporated to dryness. This yielded 30.3 g (75%, ee 99.0%)
of (S)-2 as a colorless oil that crystallized upon standing. 1H NMR
(CDCl3): δ = 1.90 (s, 1 H, NH), 2.0 (t, 1 H, CHax), 2.16 (dt, 1 H,
CHax), 2.32 (s, 3 H, CH3), 2.85 (m, 1 H, CHeq), 2.89 (m, 1 H,
THF (4.6 mL), water (2.8 mL), and 33% sodium hydroxide solu-
tion (5.1 mL) were added. The mixture was heated at reflux for
30 min. The solution was cooled, magnesium sulfate was added,
the salts were filtered, and the solvent was evaporated. The title
compound was obtained as a colorless oil (15.9 g, 98%, purity by
1
GC 99.7%, ee Ͼ99%). H NMR (CDCl3): δ = 2.31 (t, 1 H), 2.38
(s, 3 H, CH3), 2.48 (dt, 1 H, CH), 2.98 (m, 2 H), 3.17 (m, 2 H),
CHeq), 3.12 (dt, 1 H, CHax), 3.15 (m, 1 H, CHeq), 3.88 (dd, 1 H, 4.61 (m, 1 H), 4.5 (dd, 1 H), 4.87 (d, 1 H, CHbenzyl), 5.30 (s, 1 H,
CHbenzyl), 7.23–7.41 (m, 5 H, Ar-H) ppm.
OH), 6.88 (dd, 1 H, Ar-H), 7.08–7.19 (m, 2 H, Ar-H), 7.27–7.36
(m, 2 H, Ar-H), 8.16 (dd, 1 H, Ar-H) ppm.
The ee was determined by HPLC analysis: Chiralcel OD
250ϫ4.6 mm i.d., 10 µm (Daicel), 5% isopropyl alcohol in hexane,
flow rate 1.0 mLmin–1, column temperature 40 °C, UV detector
(210 nm), retention times 5.8 and 6.7 min.
The ee was determined by HPLC analysis: Chiralcel OJ
250ϫ4.6 mm i.d., 10 µm (Daicel), 8% isopropyl alcohol in heptane
+ 0.1% diethylamine, flow rate 1.0 mLmin–1, column temperature
40 °C, UV detector (295 and 300 nm), retention times 7.0 and
12.5 min.
2-[(2S)-4-Methyl-2-phenyl-1-piperazinyl]-3-pyridinecarbonitrile (3):
A
mixture of (S)-1-methyl-3-phenylpiperazine (S)-2 (21.7 g,
123 mmol), 2-chloronicotinitrile (1) (21.7 g, 157 mmol), KF (21.7 g,
373 mmol), and DMF (65 mL) was heated at reflux for 23.5 h. The
DMF was evaporated and ethyl acetate (45 mL) and water (64 mL)
at 60 °C were added to the crude product. The mixture was refluxed
for 10 min. The aqueous layer was separated from the organic layer
and was extracted with ethyl acetate (45 mL). The collected ethyl
acetate layers were evaporated to dryness. This furnished 3 (53.5 g,
quantitative, ee 97.6%) as a brown oil that was directly used in the
(14bS)-1,2,3,4,10,14b-Hexahydro-2-methylpyrazino[2,1-a]pyrido-
[2,3-c][2]benzazepine [(S)-Mirtazapine]. Procedure A: A solution of
5 (7.02 g, 24.7 mmol) in N-methylpyrrolidinone (10 mL) was added
to a mixture of polyphosphoric acid (41.8 g) and N-methylpyrolid-
ine (10.5 mL). The reaction mixture was heated at 130 °C for 1 h.
Water (152 mL), Dicalite (8.8 g), toluene (76 mL), and 33% sodium
hydroxide solution (128 mL) were added to the reaction mixture.
The aqueous layer was separated and extracted twice with toluene
(76 mL). The combined toluene layers were washed three times
with water (76 mL), dried with MgSO4, and evaporated, to yield
1
subsequent step. H NMR (CDCl3): δ = 2.38 (s, 3 H, CH3), 2.57
(m, 1 H, CH), 2.78 (m, 2 H), 2.95 (dd, 1 H, CH), 3.6 (m, 1 H,
CH), 5.43 (t, 1 H, CHbenzyl), 6.79 (dd, 1 H, Ar-H), 7.18 (m, 1 H, 4.63 g (S)-Mirtazapine (70%, ee 99%). 1H NMR (MeOD): δ = 2.98
Ar-H), 7.26 (m, 2 H, Ar-H), 7.37 (m, 1 H, Ar-H), 7.78 (dd, 1 H, (s, 3 H, CH3), 3.30 (dt, 1 H, CH), 2.86 (dt, 1 H, CH), 2.95 (dd, 1
Ar-H), 8.26 (dd, 1 H, Ar-H) ppm.
H, CH), 3.42 (dd, 1 H, CH), 3.50 (dt, 1 H, CH), 3.69 (dt, 1 H,
CH), 4.35 (dd, 1 H, CH), 4.52 (d, 1 H, CHbenzyl), 6.72 (dd, 1 H,
Ar-H), 7.12–7.21 (m, 3 H, Ar-H), 7.25 (t, 1 H, Ar-H), 7.32 (dd, 1
H, Ar-H) ppm.
The ee was determined by HPLC analysis: Chiralcel OD-H
250ϫ4.6 mm i.d., 10 µm (Daicel), 3% isopropyl alcohol in hexane
+ 0.1% diethylamine, flow rate 1.0 mLmin–1, column temperature
40 °C, UV detector (295 nm), retention times 6.6 and 7.7 min.
The ee was determined by HPLC analysis: Chiralcel OD-H
250ϫ4.6 mm i.d. (Daicel), 10% isopropyl alcohol in heptane, flow
rate 1.0 mLmin–1, column temperature 40 °C, UV detector, reten-
tion times 6.2 and 7.1 min.
2-[(2S)-4-Methyl-2-phenyl-1-piperazinyl]-3-pyridinecarboxylic Acid
(4): Nitrile (3) (53.5 g) was dissolved in methanol (86.8 mL) and the
solution was heated to 50 °C. A 33% sodium hydroxide solution
(127.4 mL) was added and the reaction mixture was refluxed for
3 days. The mixture was diluted with water (22 mL) and the meth-
anol was evaporated. The thus formed oil was separated from the
salty aqueous phase and dissolved in water (96 mL) by increasing
the temperature to 80 °C. The pH was adjusted to 5.5Ϯ0.5 by add-
ing sulfuric acid and the solution was stirred for 15 min. Water was
removed by co-evaporation with toluene (320 mL). The solution
was filtered to remove the salts. The filtrate was evaporated to dry-
ness and the crude product was co-evaporated twice with a mixture
of methanol (87 mL) and water (2 mL). The residue was dissolved
in acetone (109 mL) at 50 °C after which the solution was cooled
to ambient temperature at which it was stirred overnight. The mix-
ture was stirred for another hour at –10 °C after which the crystals
of 4 were collected by filtration and dried in a vacuum oven. The
yield of the crystals was 20.9 g (57%, ee Ͼ 99%). 1H NMR
(CDCl3): δ = 2.43 (s, 3 H, CH3), 2.61 (t, 2 H, CH2), 3.12 (m, 3 H,
CH2, CH), 3.42 (dt, 1 H, CH), 4.79 (dd, 1 H, CHbenzyl), 7.11–7.28
(m, 4 H, Ar-H), 8.25 (dd, 1 H, Ar-H), 8.54 (dd, 1 H, Ar-H) ppm.
Procedure B: A mixture of 5 (1.0 g, 3.53 mmol) and polyphosphoric
acid (2 g) was stirred and heated at 130 °C for 18 h. The reaction
mixture was diluted with water (6.5 mL) and the pH was brought
to 8 by adding a 4  sodium hydroxide solution. The water layer
was extracted with ethyl acetate. The organic layer was washed with
water, dried with magnesium sulfate, and the solvents evaporated.
This provided the title compound (0.71 g, 76%, ee 98%).
Procedure C: A mixture of polyphosphoric acid (20 gram) and
5·oxalic acid (13.2 g, 35.3 mmol) was stirred and heated at 130 °C
for 18 h. Water (220 mL) was added followed by the addition of
ethyl acetate (220 mL) and 33% sodium hydroxide solution
(65 mL). The aqueous layer was separated and extracted twice with
ethyl acetate (220 mL). The combined organic fractions were
washed three times with water (220 mL) and the solvents evapo-
rated. This gave 7.9 g (S)-mirtazapine (84%, ee 99.2%).
Isolation of the Side-Product 3-Benzyl-2-(2-methylaminoethyl)ami-
nopyridine Trifluoroacetic Acid Salt (Figure 2): From the reaction
mixture obtained by the above-described ring-closure reaction of 5
with 2 wt.equiv. of PPA, the title compound was isolated by pre-
parative HPLC using a Gilson-semiPrep system with 25 wti pumps.
An isocratic eluent with 90% acetonitrile and 10% water contain-
ing 0.1% formic acid was used. Flow-rate 20 mLmin–1. The col-
The ee was determined by HPLC analysis: Chiralcel OD-H
250ϫ4.6 mm i.d., 10 µm (Daicel), 10% isopropyl alcohol in hep-
tane, flow rate 1.0 mLmin–1, UV detector, column temperature
40 °C.
2-[(2S)-4-Methyl-2-phenyl-1-piperazinyl]-3-pyridinemethanol
(5): umn was obtained from Waters: Sunfire prep C18, OBP 10 µm,
Carboxylic acid 4 (15 g, 50 mmol) was dissolved in THF (80 mL)
and LiAlH4 (44 mL, 1  in THF) was added. After stirring the
reaction mixture for one night, the solution was cooled 10 °C and
19ϫ150 mm. The injection volume was 750 µm. By using multiple
injections 205 mg of 3-benzyl-2-[(2-methylaminoethyl)amino]pyr-
idine was isolated.
Eur. J. Org. Chem. 2008, 2989–2997
© 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
2995

M. van der Linden, J. Borsboom, F. Kaspersen, G. Kemperman
FULL PAPER
MS: m/z = 242 [M + H]+. MS/MS of m/z = 242: m/z = 211, 167,
CHNH2), 7.24–7.32 (m, 4 H, ArH) ppm. Note: The 1H NMR spec-
133, 92. 13C ATP (CDCl3): δ = 33.4 (CH3), 37.0 (CH2), 39.6 (CH2), trum was recorded for the potassium salt.
114.0 (CH), 122 (C), 127.3 (CH), 129.3 (4 CH), 137.7 (C), 139.0
(CH), 143.7 (CH), 155.79 (CH) ppm. H NMR (CDCl3): δ = 2.63
(s, 3 H, CH3), 3.18 (s, 2 H, CH2), 3.72 (s, 2 H, CH2), 3.83 (s, 2 H,
CH2), 6.63 (m, 1 H, Ar-H), 7.16 (d, 2 H, Ar-H), 7.19 (d, 1 H, Ar-
H), 7.25 (m, 1 H, Ar-H), 7.30 (m, 2 H, Ar-H), 7.95 (d, 1 H, Ar-
H), 8.38 (s, 1 H, NH) ppm.
Ethyl (rac)-Amino(4-deuteriophenyl)acetate (17): Concentrated sul-
1
furic acid (1013 mmol, 54 mL, 99 g) was added to a solution of
(rac)-4-deuteriophenylglycine (16) (417 mmol, 63.45 g) in ethanol
(650 mL) over 30 min at room temperature. After stirring over-
night, the reaction mixture was poured into a saturated Na2CO3
solution and extracted twice with ethyl acetate. The organic layer
was separated and washed with a saturated NaCl solution, dried
with MgSO4, and concentrated in vacuo to yield 16.4 g of the title
2-(4-Bromophenyl)-1,3-dioxolane (13): Ethylene glycol (1627 mmol,
91 mL, 101 g) and p-toluenesulfonic acid monohydrate
(10.51 mmol, 2.0 g) were added to a solution of 4-bromobenzalde-
hyde (811 mmol, 150 g) in toluene (3.15 L). The mixture was heated
at reflux (110 °C) and by using a Dean–Stark apparatus water was
removed. After stirring overnight, the mixture was cooled to room
temperature and extracted with water/ethyl acetate, washed with
saturated aqueous NaCl, dried with MgSO4, and concentrated in
vacuo to yield the title compound in 181.5 g (792 mmol, 98%, pu-
rity according to GC 97% a/a) as a colorless oil. 1H NMR
(CDCl3): δ = 4.01–4.15 (m, 4 H,CH2O), 5.76 (s, 1 H, CH) 7.35 (d,
2 H, ArH, meta), 7.52 (d, 2 H, ArH, ortho) ppm.
1
compound (22%) as a liquid. H NMR (CDCl3): δ = 1.22 (t, 3 H,
CH3), 3.69 (q, 2 H, CH2), 4.55 [s, 1 H, CH(CO)NH2], 7.31–7.43
(m, 4 H, ArH) ppm.
rac-2-Amino-2-(p-deuteriophenyl)-N-methylacetamide (18): Ethyl
(rac)-amino(p-deuteriophenyl)acetate (17) (139 mmol, 25 g) was
dissolved in an ethanolic solution of methylamine (2196 mmol,
264 mL, 227 g). The reaction mixture was stirred overnight at 20 °C
and concentrated in vacuo to yield 16.3 g of 18 (84%) as a colorless
oil. 1H NMR (CDCl3): δ = 2.81 (s, 3 H, CH3NH), 4.52 [s, 1 H,
CH(CO)NH2], 7.31–7.43 (m, 4 H, ArH) ppm.
2-(4-Deuteriophenyl)-1,3-dioxalane (14): A 1.6  solution of n-bu-
tyllithium (1200 mmol, 750 mL, 510 g) in hexane was slowly added
to a solution of 2-(4-bromophenyl)-1,3-dioxolane (13) (792 mmol,
181.5 g) in THF (2.5 L) at –65 °C. After stirring for 30 min at
–65 °C, deuterium oxide (5493 mmol, 100 mL, 110 g) was added
carefully at a maximum temperature of –60 °C. The mixture was
warmed to room temperature and was extracted with toluene
(2ϫ2 L). The combined organic layers were washed with water
(2ϫ1 L), dried with MgSO4, and the solvents evaporated. This
yields 14 in quantitative yield (purity according to GC 95% a/a) as
rac-2-Chloroacetamido-2-(p-deuteriophenyl)-N-methylacetamide (19):
Sodium carbonate (59.2 mmol, 6.27 g) was added to a solution
of (rac)-2-amino-2-(p-deuteriophenyl)-N-methylacetamide (18)
(98 mmol, 16.2 g) in water (53 mL) and acetone (44 mL) at 20 °C.
The mixture was cooled to 0 °C and stirred for 1.5 h. A solution
of chloroacetyl chloride (104 mmol, 8.24 mL, 11.70 g) in acetone
(30 mL) was slowly added. The resulting milky suspension was
stirred at 0 °C for 1.5 h which was followed by the addition of
Na2CO3 (1.25 g, 11.85 mmol) and a solution of chloroacetyl chlo-
ride (1.57 mL, 2.23 g, 19.7 mmol) in acetone (6 mL). After stirring
for an additional 1.5 h, approximately 50 mL of solvent was re-
moved at 44 °C under vacuum. The obtained thick suspension was
cooled to 20 °C and filtered. The residue was washed with water
(30 mL). The obtained material was dried overnight by passing ni-
trogen through the filter to yield 16.1 g of 19 (67.8%) as off-white
crystals.
1
a colorless oil. H NMR (CDCl3): δ = 4.01–4.18 (m, 4 H,CH2O),
5.82 (s, 1 H, CH), 7.38 (d, 2 H, ArH, meta), 7.50 (d, 2 H, ArH,
ortho) ppm.
4-Deuteriobenzaldehyde (15): Water (100 mL) and p-toluenesulfonic
acid monohydrate (10.51 mmol, 2 g) were added to a solution of 2-
(4-deuteriophenyl)-1,3-dioxalane (14) (840 mmol, 127 g) in acetone
(1 L). The reaction mixture was stirred overnight at room tempera-
ture. In order to achieve a complete conversion, four additional
portions of p-toluenesulfonic acid monohydrate (10.51 mmol, 2 g)
we added. The reaction mixture was diluted with toluene (2 L) and
saturated sodium hydrogen carbonate (2ϫ1 L). The toluene phase
was separated and washed with saturated NaCl (2ϫ1 L), dried
with MgSO4, and concentrated in vacuo to yield 67.5 g of 15 (75%,
rac-3-(p-Deuteriophenyl)-1-methylpiperazine (20): A solution of bo-
rane–tetrahydrofuran (300 mmol, 300 mL, 330 g) was added drop-
wise to
a suspension of (rac)-2-chloroacetamido-2-(p-deuter-
iophenyl)-N-methylacetamide (19) (63.2 mmol, 15.28 g) in THF
(160 mL). The reaction mixture was stirred overnight. Concen-
trated hydrochloric acid (336 mmol, 56 mL, 58.8 g) was added
slowly at room temperature (a lot of hydrogen gas was formed).
The mixture was distilled at reduced pressure to remove the THF.
Ethyl acetate (150 mL) was added and the mixture was extracted
with 2  hydrochloric acid (3ϫ100 mL). The pH of the combined
acidic phases was adjusted to 12–13 by addition of an aqueous
solution of sodium hydroxide (33%). The aqueous phase was ex-
tracted with ethyl acetate (3ϫ100 mL). The combined organic
phases were dried on MgSO4, and concentrated in vacuo to give
1
purity according to GC 95% a/a) as a liquid. H NMR (CDCl3):
δ = 7.55 (d, 2 H, ArH, meta), 8.11 (d, 2 H, ArH, ortho), 10.2 (s, 1
H, CH) ppm.
rac-4-Deuteriophenylglycine (16): Ammonium chloride (654 mmol,
35 g) was added to a stirred solution of sodium cyanide (653 mmol,
32 g) in water (128 mL). This was followed by the addition of a
solution of 4-deuteriobenzaldehyde (15) (627 mmol, 67.15 g) in
MeOH (128 mL). During the addition the temperature rose from
10 to 30 °C. The reaction mixture was stirred for 2 h, water was
added (300 mL), and the mixture was extracted twice with toluene
(180 and 120 mL). The toluene phases were washed with water
(2ϫ100 mL) followed by extraction with a 5  HCl solution
(2ϫ100 mL). The acidic extract was heated for 22 h at reflux. After
cooling, the tarry black material was removed by filtration through
a plug of cotton wool. The pH of the filtrate was adjusted to 6.2
by the addition of an ammonia solution (25 wt.-%) under cooling.
The resulting precipitate was collected by filtration, washed with
cold water (300 mL), and dried overnight to yield 63.4 g of 16
(66.5%) as off-white crystals. 1H NMR (D2O): δ = 4.23 (s, 1 H,
1
7.7 g of 20 as a yellowish liquid (68.7%). H NMR (CDCl3): δ =
1.70 (s, 1 H, NH), 2.0 (t, 1 H, CHax), 2.16 (dt, 1 H, CHax), 2.33 (s,
3 H, CH3), 2.85 (m, 1 H, CHeq), 2.89 (m, 1 H, CHeq), 3.02–3.16
(m, 2 H, CHax, CHeq), 3.88 (dd, 1 H, CHbenzyl), 7.35 (m, 2 H, Ar-
H), 7.38 (m, 2 H, Ar-H) ppm.
2-[(2S)-2-(4-Deuteriophenyl)-4-methyl-1-piperazinyl]-3-pyridine-
methanol (21): The title compound was prepared from (rac)-3-(p-
deuteriophenyl)-1-methylpiperazine (20) and 2-chloronicotinitrile
according to the procedures described above for compounds 3–5.
1H NMR (CDCl3): δ = 2.29 (t, 1 H), 2.38 (s, 3 H, CH3), 2.50 (dt,
1 H, CH), 2.96 (m, 2 H), 3.20 (m, 2 H), 4.66 (m, 1 H, CH), 4.76
(d, 1 H, CH) 4.81 (d, 1 H, CHbenzyl), 6.99 (dd, 1 H, Ar-H), 7.08
2996
www.eurjoc.org
© 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2008, 2989–2997

Asymmetric Synthesis of (S)-Mirtazapine
[8]
(m, 2 H, Ar-H), 7.28 (m, 2 H, Ar-H), 7.77 (dd, 1 H, Ar-H), 8.03
(dd, 1 H, Ar-H) ppm.
The IUPAC name of (S)-(+)-Anicyphos is (S)-(+)-2-hydroxy-
4-(2-methoxyphenyl)-5,5-dimethyl-1,3,2-dioxaphosphorinan-2-
one.
[9]
M. C. A. van Vliet, G. J. Kemperman, M. Schreuder, M.
Goedheijt (Organon N.V.), WO 2007144409, 2007.
[1] W. J. van der Burg, US patent 4062848, 1977.
[2] Y. Kang, F. Qu, K. Liu (Beijing D-Venturepharm. T. Corp.),
CN 1939918, 2007; C. Arnalot Aguilar (Medichem, S.A.), WO
2006008302, 2006; Y. Yang, B. Guo, K. Chen, R. Ji (Shanghai
Institute of Pharmacy), CN 1429819, 2003; S. Claude, A. Lib-
erman, N. Finkelstein (Teva Pharmaceutical Industries, Ltd.),
US 2003069417, 2003; L. Metzger, S. Wizel (Teva Pharmaceuti-
cal Industries Ltd.), WO 2002070513, 2002; S. Sebastian, H. V.
Patel, R. Thennati (Sun Pharmaceutical Industries Ltd.), WO
2002038552, 2002; S. Claude, A. Liberman, N. Finkelstein
(Teva Pharmaceutical Industries Ltd.), WO 2000062782, 2000.
[3] T. Zhang, F-h. Wu, Huadong Ligong Daxue Xuebao 2006, 32,
318–320.
[4] F. M. Kaspersen, F. A. M. van Rooij, E. G. M. Sperling, J. H.
Wieringa, J. Labelled Compd. Radiopharm. 1989, 27, 1055–
1068.
[5] K. J. Holm, A. Markham, Drugs 1999, 57, 607–631; T. de Boer,
J. Clin. Psychiatry 1996, 57, 19–25; T. de Boer, F. Nefkens, A.
van Helvoirt, A. M. L. van Delft, J. Pharmacol. Exp. Ther.
1996, 277, 852–860; C. de Montigny, N. Haddjeri, R. Mon-
geau, P. Blier, CNS Drugs 1995, 4, 13–17; J. M. A. Sitsen, M.
Zivkov, CNS Drugs 1995, 4, 39–48.
[10]
Economically more viable syntheses for (S)-1-methyl-3-phenyl-
piperazine have also been identified, which will be reported in
a separate paper.
The polyphosphoric acid was prepared by adding 1 weight
equivalent of phosphorus pentoxide in portions to phosphoric
acid (85%) while maintaining the temperature below 140 °C.
For example, 100 g P2O5 was added in portions to 100 g phos-
phoric acid.
It should be noted that in reactions with PPA and DMF (en-
try 14 in Table 1) a side-product was observed in amounts of
up to 26% relative to the product. This side-product was char-
acterized by MS as having a MIM of 338, corresponding to a
DMF adduct of a cationic intermediate. The side-product was
not isolated or further characterized. However, it is a plausible
hypothesis that the piperazine cation was attacked by DMF
thereby preventing re-attack of the piperazine cation and thus
avoiding formation of racemic mirtazapine.
J. H. Wieringa, A. A. M. van De Ven, G. J. Kemperman (Akzo
Nobel N.V.), WO 2005/005410, 2005.
J. Bosch i Llado, P. Camps Garcia, J. Contreras Lascorz, M.
Onrubia Miguel (Medichem S.A.), WO 2003/024918, 2003.
The NMR spectra were recorded in deuteriated acetone in or-
der to achieve sufficient resolution of the signals of the relevant
protons 12 and 13.
Because of the small scale of the reaction a little more than the
intended 2.5 wt.-equiv. of PPA were used.
Received: December 25, 2007
[11]
[12]
[13]
[14]
[15]
[6] Over 200 papers on the biological properties of mirtazapine
have appeared over the past two decades. Only a selection of
the earliest publications are collected in ref.[5]
.
[7] W. T. O’Connor, B. E. Leonard, Neuropharmacology 1986, 25,
267–270; A. R. Kooyman, R. Zwart, P. M. L. Vanderheijden,
J. A. van Hooft, H. P. M. Vijverberg, Neuropharmacology 1994,
33, 501–507; T. De Boer, G. Maura, M. Raiteri, C. J. De Vos,
J. Wieringa, R. M. Pinder, Neuropharmacology 1988, 27, 399–
408.
[16]
Published Online: April 29, 2008
Eur. J. Org. Chem. 2008, 2989–2997
© 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
2997