Development of an efficient and stereoselective manufacturing route to idoxifene

Article abstract of DOI:10.1021/op0100394

A literature route to 1-(2-{4-[(E)-1-(4-iodophenyl)-2-phenyl-but-1-enyl]phenoxy}ethyl)pyrrolidine (idoxifene) has been modified to tackle various scale-up issues and provide initial supplies. A new highly efficient, robust, and stereoselective manufacturing route is described in detail. This route involves diastereoselective synthesis of tertiary alcohol (1RS,2SR)-1-(4-iodophenyl)-2-phenyl-1-[4-(2-pyrrolidin-1-yl-ethoxy)phenyl]butan- 1-ol by Grignard addition to the ketone 1-(4-iodophenyl)-2-phenyl-1-butanone followed by derivatisation and stereoselective syn elimination to provide idoxifene in excellent yield and geometric purity. Evaluation of a more direct route to idoxifene using a McMurry low-valent titanium coupling reaction is also described.

Full text of DOI:10.1021/op0100394

Organic Process Research & Development 2001, 5, 479−490
Development of an Efficient and Stereoselective Manufacturing Route to
Idoxifene
Karl W. Ace, Mark A. Armitage, Richard K. Bellingham,* Paul D. Blackler, David S. Ennis,† Nigel Hussain,
David C. Lathbury,† David O. Morgan, Noah O’Connor, Graham H. Oakes,‡ Stephen C. Passey,§ and
Laurence C. Powling
Chemical DeVelopment, GlaxoSmithKline Pharmaceuticals, Old Powder Mills, Nr. Leigh,
Tonbridge, Kent TN11 9AN, U.K.
Abstract:
briefly discuss the initial supply route, which was subtley
modified from that previously reported,8 and will then
describe the evaluation of a very direct route using McMurry
chemistry9 and the development of a new, efficient, and
stereoselective route.
A literature route to 1-(2-{4-[(E)-1-(4-iodophenyl)-2-phenyl-but-
1-enyl]phenoxy}ethyl)pyrrolidine (idoxifene) has been modified
to tackle various scale-up issues and provide initial supplies. A
new highly efficient, robust, and stereoselective manufacturing
route is described in detail. This route involves diastereoselective
synthesis of tertiary alcohol (1RS,2SR)-1-(4-iodophenyl)-2-
phenyl-1-[4-(2-pyrrolidin-1-yl-ethoxy)phenyl]butan-1-ol by Grig-
nard addition to the ketone 1-(4-iodophenyl)-2-phenyl-1-
butanone followed by derivatisation and stereoselective syn
elimination to provide idoxifene in excellent yield and geometric
purity. Evaluation of a more direct route to idoxifene using a
McMurry low-valent titanium coupling reaction is also de-
scribed.
Results and Discussion
The route shown (Scheme 1) was used to produce material
on an 18-kg scale. The route differs from that previously
Introduction
Idoxifene 1, was first synthesised in 1986 by the Cancer
Research Campaign1 (CRC) and was licensed from the
British Technology Group. Idoxifene 1 is a selective estrogen
receptor modulator (SERM). It has estrogen agonist effects
in bone tissue and upon lipid metabolism and estrogen
antagonist effects in the uterus and breast tissues.2-7 This
broad profile gave idoxifene 1 the potential to treat or prevent
several diseases associated with female menopause. Os-
teoporosis prevention was the lead indication, but clinical
trials were also being conducted with idoxifene 1 in the
treatment of advanced breast cancer, currently treated with
the closely related compound tamoxifen 2. This paper will
Scheme 1. Route of synthesis used to prepare initial
supplies
* To whom correspondence should be addressed. Chemical Development,
GlaxoSmithKline Pharmaceuticals, Old Powder Mills, Nr. Leigh, Tonbridge, Kent
TN11 9AN, U.K. E-mail: Richard•K•Bellingham@sbphrd.com.
† Present address: Process R&D, AstraZeneca Charnwood, Bakewell Road,
Loughborough, Leicestershire LE11 5HT, U.K.
‡ Present address: University of Liverpool, Department of Chemistry, Crown
Street, Liverpool L69 7ZD, U.K.
§ Present address: School of Chemistry, University of Bristol, Bristol BS8
1TS, U.K.
(1) McCague, R. Br. Patent GB 2174088, 1986.
(2) McCague, R.; Leclercq, G.; Legros, N.; Goodman, J.; Backburn, G. M.;
Jarman, M.; Foster, A. B. J. Med. Chem. 1989, 32, 2527.
(3) Rowlands, M. G.; Parr, I. B.; McCague, R.; Jarman, M.; Goddard, J. M.
Biochem. Pharmacol. 1990, 40, 283.
(4) Chander, S. K.; McCague, R.; Luqmani, Y.; Newton, C.; Dowsett, M.;
Jarman, M.; Coombes, R. C. Cancer Res. 1991, 51, 5851.
(5) Haynes, B. P.; Parr, I. B.; Griggs, L. J.; Jarman, M. Breast Cancer Res.
Treat. 1991, 19, 174.
(6) Hardcastle, I. R.; Rowlands, M. G.; Houghton, J.; Parr, I. R.; Potter, G. A.;
Jarman, M. J. Med. Chem. 1995, 38, 241.
(7) Hardcastle, I. R.; Rowlands, M. G.; Grimshaw, R. M.; Houghton, J.; Jarman,
M.; Sharff, A.; Neidle, S. J. Med. Chem. 1996, 39, 999.
10.1021/op0100394 CCC: $20.00 © 2001 American Chemical Society and The Royal Society of Chemistry
Published on Web 08/22/2001
Vol. 5, No. 5, 2001 / Organic Process Research & Development
•
479

published8 in a number of key areas. The original route
involved isolation of the intermediate chlorosulphite 8,
formed by reaction of alcohol 3 with excess thionyl chloride,
and subsequent reaction with catalytic pyridine to give the
chloroalkane 4.
Scheme 2
The uncontrolled evolution of sulphur dioxide was a major
concern and led to the development of a method that
premixed the alcohol 3 and 1 mol equiv of pyridine at reflux
in toluene before the addition of 1 mol equiv of thionyl
chloride. In this way the evolution of sulphur dioxide could
be controlled by addition rate. Formation of ketone 5 from
2-phenylbutyric acid was initially performed neat, but on
scale-up, addition of a solvent, primarily as a heat sink, was
necessary. Isolation of the ketone 5 was modified by dilution
of the reaction mixture with toluene before aqueous work-
up. Crystallisation of the product 5 was then achieved by
concentration followed by dilution with petroleum ether (80-
100). Another key issue was the instability of 4-iodophen-
yllithium over time. On large scale the rate at which ketone
5 could be safely added severely limited batch size. The
problem was overcome by adding butyllithium to a mixture
of 1,4-diiodobenzene and ketone 5; this approach was the
subject of a previous publication.10 Changing the solvent
from THF to toluene also lowered levels of a dimeric
impurity formed in this reaction.10 Alcohol 6 and alkene 7
were not isolated but taken forward to crude idoxifene 1.
Separation of the geometric isomers was achieved by
crystallisation to afford pure idoxifene 1. Despite these
improvements in the supply route there still remained some
major drawbacks of this procedure. The low-temperature
lithiation, which also leads to the formation of one mole
equivalent of butyl iodide, and a lack of selectivity with
respect to alkene 7 geometry meant that in the long term a
new route was required.
Scheme 3
Propiophenone is commercially available, and the ben-
zophenone 9 was prepared in one step via a Friedel-Crafts
acylation reaction in 62% yield (Scheme 3).
Using the TiCl3-Li protocol11,12 the coupling was per-
formed in DME to afford a 25% yield of the desired E-olefin
7 (Table 1, entry 1). A number of other systems were
investigated (Table 1, entries 2-4) although none offered
any significant advantage in terms of stereoselectivity or
yield. In most cases the E-isomer 7 was the predominant
product. Changing the solvent to THF had a pronounced
effect on the stereoselectivity of the coupling. In both cases
(Table 1, entries 5 and 6) the Z-olefin was observed as the
predominant product. The effect of the solvent in controlling
the stereochemistry of the McMurry reaction has been
reported previously.14
Since the choice of solvent appeared to be important in
controlling stereoselectivity, a number were evaluated. No
reaction was observed in acetonitrile, DME, dioxolane, or
thioxane (Table 1, entries 7-10); however in 1,4-dioxane a
considerable predominance of the desired E-olefin 7 was
observed (Table 1, entry 11). These results suggested that
dioxygenated solvents form stable octahedral titanium com-
plexes.15 It was reasoned that bidentate ligands capable of
complexing the titanium in a similar manner could also
influence the stereoselectivity of the reaction. A number of
reactions were performed in 1,4-dioxane in the presence of
various bidentate phosphine and amine ligands (Table 1,
entries 13-16). Although the E-isomer 7 was the predomi-
nant product in all cases, only the reaction containing 1,2-
bis(diphenylphosphino)ethane (Table 1, entry 13) offered any
advantage in terms of stereoselectivity and yield.
McMurry Approach. As part of the program directed
at identifying alternative routes to idoxifene 1, it was
considered that a low-valent titanium-meditated coupling
(McMurry reaction9) between propiophenone and diaryl
ketone 9 (Scheme 2) may provide a very direct route to the
final product. It has been reported that the reaction works
particularly well if one of the components is a diaryl ketone.11
The related antitumour agent tamoxifen 2 has been success-
fully prepared by this approach.12,13
(8) McCague, R.; Potter, G. A.; Jarman, M. Org. Prep. Proced. Int. 1994, 26(3),
343.
(9) (a) Mukaiyama, T.; Sato, T.; Hanna, J. Chem. Lett. 1973, 1041. (b) McMurry,
J. E.; Fleming, M. P.; J. Am. Chem. Soc. 1974, 96, 4708. (c) McMurry, J.
E.; Fleming, M. P.; Kees, K. L.; Krepski, L. R. J. Org. Chem. 1978, 43,
3255. (d) Robertson, G. In ComprehensiVe Organic Synthesis; Trost, B.
M., Fleming, I., Pattenden, G., Eds.; Pergamon: Oxford; 1991, p 563.
(10) Ennis, D. S.; Lathbury, D. C.; Wanders, A.; Watts, D. Org. Process Res.
DeV. 1998, 2(5), 287.
(13) (a) Shani, J.; Gazit, A.; Livshitz, T.; Biran, S. J. Med. Chem. 1985, 28,
1504. (b) Schwartz, W.; Hartmann, R. W.; Schonenberger, H. Arch.
Pharmacol. 1991, 324, 223. (c) Gauthier, S.; Mailhot, J.; Labrie, F. J. Org.
Chem. 1996, 61, 3890.
(11) (a) McMurry, J. E. Acc. Chem. Res. 1983, 16, 405. (b) McMurry, J. E.
Chem. ReV. 1989, 89, 1513.
(12) Coe, P. L.; Scriven, C. E. J. Chem. Soc., Perkin Trans. 1 1986, 475.
(14) Nayak, S. K.; Banerji, A. J. Org. Chem. 1991, 56, 1940.
(15) Reetz, M. T. In Organotitanium Reagents in Organic Synthesis; Springer
Verlag: Berlin, 1986; p 21.
480
•
Vol. 5, No. 5, 2001 / Organic Process Research & Development

Table 1.
E (%)
isolated
Z (%)
isolated
entry
conditions
additive
solvent
E:Z ratio
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
TiCl3-Li
none
none
none
none
none
none
none
none
none
none
none
none
dppe
dppp
PPh3
TMEDA
DME
DME
DME
DME
THF
THF
MeCN
glyme
1,3-dioxolane
1,4-thioxane
1,4-dioxane
diethoxy methane
1,4-dioxane
1,4-dioxane
1,4-dioxane
1,4-dioxane
1.8:1
1.3:1
1.4:1
1:1
1:1.9
1:11
no reaction
no reaction
no reaction
no reaction
3.9:1
1.4:1
4.1:1
3.6:1
3.5:1
26
20
30
6
-
-
14
15
21
6
-
-
TiCl4-Zn
TiCl3-LAH
TiCl3(dme)1.5-Zn-Cu
TiCl4-Zn
TiCl3-LAH
TiCl4-Zn
-
-
-
-
TiCl4-Zn
TiCl4-Zn
-
-
45
-
51
40
26
-
-
-
11.5
-
12
11
7.5
-
TiCl4-Zn
TiCl4-Zn
TiCl4-Zn
TiCl4-Zn
TiCl4-Zn
TiCl4-Zn
TiCl4-Zn
3.4:1
To demonstrate the utility of the McMurry reaction for
the preparation of idoxifene 1, the reaction was performed
on a 25 mmol scale using the optimum conditions (Table 1,
entry 13). After work-up the crude mixture of E- and
Z-alkenes was reacted with pyrrolidine in refluxing ethanol
to give a mixture of E- and Z-idoxifene. Purification by
chromatography followed by recrystallisation from ethanol
afforded E-idoxifene 1 in 21% yield from 9. Idoxifene 1
could also be prepared directly from 10, but this procedure
was inferior in both stereoselectivity and yield. Although an
extremely direct approach to idoxifene 1, the route suffered
from a lack of E:Z selectivity, and the inevitable chromato-
graphic purification of the final product. Work in this area
was discontinued.
Scheme 4
Syn-Elimination Approach. During development work
on the original synthetic route the intermediate tertiary
alcohol 6 had been isolated.10 Single-crystal X-ray diffraction
experiments had demonstrated that tertiary alcohol 6 was
obtained effectively as the single diastereoisomer10 predicted
by Cram’s rule.16 Much of this stereochemical information
was lost in the acid-catalyzed dehydration reaction. It was
therefore envisaged that much better E:Z ratios of idoxifene
1 could be obtained if a concerted elimination of a diaste-
reomerically enriched substrate 11 could be performed
(Scheme 4). The research pursued a syn-elimination ap-
proach; hence, synthesis of the tertiary alcohol 12 became
the focus of our work.
Scheme 5
Alcohol 12 should be accessible by a Grignard addition
of 14 to ketone 13 with the addition occurring via a Felkin-
Anh transition state17 (Scheme 5).
Scheme 6
Synthesis of Ketone 13. Attempts to make the precursor
ketone 13 initially centred around nucleophilic addition of
4-iodophenyllithium to various 2-phenylbutyric acid deriva-
tives. The reaction of the Weinreb amide18 15 was the only
procedure to give a synthetically useful yield of ketone 13
(48%) that was isolable without chromatography (Scheme
(16) Cram, D. J.; Abd Elhafez, F. A. J. Am. Chem. Soc. 1952, 74, 5828.
(17) (a) Che´rest, M.; Felkin, H.; Prudent, N. Tetrahedron Lett. 1968, 9, 2199.
(b) Anh, N. T. Top. Curr. Chem. 1980, 88, 145.
6). This procedure was used to prepare supplies of ketone
13 for laboratory-scale investigations on downstream chem-
istry.
(18) Einhorn, J.; Einhorn, C.; Luche, J.-L. Synth. Commun. 1990, 20(8), 1105.
Vol. 5, No. 5, 2001 / Organic Process Research & Development
•
481

Scheme 7
Online studies using the React-IR system eventually
demonstrated that decarboxylation started at -25 °C and
accelerated above -10 °C. The carbon dioxide was presum-
ably being trapped in solution as an adduct with diisopro-
pylamine. The results from the React-IR studies also showed
that formation of dianion 16 and the reaction with ester 18
at room temperature took no longer than 30 min. These
results led to the development of an inverse addition protocol
where the dianion 16 solution could be added to a solution
of ester 18 at room temperature. When the reaction was
deemed complete, the solution was added to 5 M hydro-
chloric acid, carbon dioxide evolution being controlled by
addition rate. The product isolation using solvent-exchange
protocol was also replaced by a procedure that concentrated
the toluene extract to two volumes and then diluted it with
2-propanol to precipitate the ketone 13. This process has been
used to synthesise ketone 13 reproducibly on >50 kg scale
with yields of 77-94% and HPLC assay of >98.9%.
Synthesis of Aryl Bromide 23 and Grignard Reagent
14. Initial laboratory supplies of the bromide 23 were made
via the Williamson ether synthesis from commercially
available 4-bromophenol, N-(2-chloroethyl)pyrrolidine hy-
drochloride and sodium ethoxide in ethanol.20 The bromide
23 can be purified by distillation and obtained pure on 100-g
scale. The first batches produced on pilot-plant scale used a
modification of this procedure using potassium carbonate as
a base and 4-methyl-2-pentanone (MIBK) as a solvent
(Scheme 8).
To reduce iodinated waste and cost, approaches via
4-iodobenzoic acid derivatives were evaluated. Direct ap-
proaches via reaction of various benzylic anions with
4-iodobenzoic acid derivatives were unsuccessful. Formation
of the dianion of phenylbutyric acid 16 and subsequent
reaction with 4-iodobenzoyl chloride 17 gave a 70% yield
of the desired ketone 13 on aqueous work-up (Scheme 7)
without the need for a separate decarboxylation reaction. This
result quickly led to the evaluation of ethyl-4-iodobenzoate
18 as a potential electrophile (Scheme 7). This reaction gave
the desired ketone 13 in a comparable yield of 75%. Further
work concentrated on optimisation of the reaction utilising
ethyl-4-iodobenzoate 18 since this was by far the cheapest
reagent. Formation of the dianion 16 was always performed
by addition of 2-phenylbutyric acid solution to the LDA
solution to prevent precipitation of the less soluble lithium
carboxylate.
Before the reaction could be scaled up, a more detailed
appreciation of the mechanism was needed, most importantly,
the point in the reaction sequence at which the intermediate
19 underwent decarboxylation. The conditions established
for synthesis of laboratory samples were the reaction of
2-phenylbutyric acid with LDA at room temperature, fol-
lowed by addition of ester 18 in THF at 0 °C. The reaction
was then quenched with 5 M hydrochloric acid and extracted
into toluene. The toluene solution was then concentrated to
a low volume and diluted with ethanol; this procedure was
repeated until crystallisation of ketone 13 occurred. Initial
work with 4-iodobenzoyl chloride 17 indicated that decar-
boxylation occurred before work-up since small quantities
of the by-product enol ester 20 were isolated, suggesting that
the enolate was present during the addition of 4-iodobenzoyl
chloride. However, in the case of the ester 18 reaction, safety
testing of the procedure showed that carbon dioxide evolution
occurred with acidic work-up. These results led to the
postulation that the intermediate 21 was involved and that
the ethoxide group leaves on exposure to strong acid. Efforts
to trap an intermediate of type 21 with bis electrophiles
(diiodomethane and dimethyldichlorosilane) failed. On scale-
up of the reaction with ester 18 the longer addition times of
ester 18 to the dianion 16 led to the isolation of a new
impurity19 22 that clearly indicated that decarboxylation
occurred prior to acidic work-up since carbon dioxide had
been trapped by the dianion 16.
Scheme 8
The quality of bromide 23 had a direct impact on the
quality of idoxifene 1 produced. One of the major impurities
(19) Fadel, A.; Garcia-Argote, S. Tetrahedron: Asymmetry 1996, 7(4), 1159.
(20) Robertson, D. W.; Katzenellenbogen, J. A. J. Org. Chem. 1982, 47, 2387.
482
•
Vol. 5, No. 5, 2001 / Organic Process Research & Development

seen was the diamine 24 which was easily removed by
employing a dilute aqueous zinc chloride wash to the
procedure.
dibromoethane) in THF. After the reaction had initiated, the
remainder of the bromide was added at reflux. To obtain
complete consumption of the bromide 23 2.3 equiv of
magnesium and 1 equiv of 1,2-dibromoethane had to be
employed, and initiation at reflux was necessary. Further
research showed that 1,2-dibromoethane was not needed if
the THF was anhydrous, with initiation occurring at 62 °C.
Evidence that initiation had occurred was essential for safe
addition of the remainder of the aryl bromide 23. An
exothermic event close to the boiling point of a solvent is
difficult to observe; therefore, the React-IR system was used
in early laboratory studies to monitor the initiation. A
characteristic band at 1036 cm-1 (tentatively assigned as a
magnesium etherate stretch) was a clear indication that
initiation had started. However, there were large differences
in initiation times depending on the source of magnesium
metal. To tackle this issue another method of activating
magnesium24 metal was employed. The addition of a small
quantity (3.1 mol % with respect to (wrt) aryl bromide 23)
of sodium bis(2-methoxyethoxy)aluminum hydride (Red-Al)
(65 wt % in toluene) solution provided more reproducible
results and was used on several 20-kg batches, initiation
occurring smoothly at 64 °C. Although the activation of
magnesium metal with Red-Al had proved reliable, there
were still a number of processing issues associated with its
use:
This method was used to synthesise bromide 23 on 30-
35 kg scale in 65% yield. The mechanism of bromide 23
formation from chloroethyl pyrrolidine was briefly investi-
gated, and the intermediacy of the aziridinium ion 26 was
established (Scheme 9). The deuterated adduct 25 was made
Scheme 9
(1) The viscous toluene solution was difficult to handle.
(2) Initiation occurred near the boiling point of THF
causing the release of large quantities of vapour with the
exothermic initiation.
(3) Only a limited amount of moisture in the THF was
tolerated before the initiation failed.
by using standard literature methods.21-23 The deuterated
adduct 25 was reacted under standard conditions to give a
1:1 ratio of the products 27 and 28.
Laboratory investigations into alternative routes to bro-
mide8 23 led to a route that gave product 23 of extremely
high purity (99.89% PAR HPLC) and in an overall yield of
63% (Scheme 10). However, supply of the aryl bromide 23
(4) employment of Red-Al led to small quantities of
hydrogen in the waste stream gases.
The release of large quantities of vapour on commence-
ment of the reaction meant that very low vessel occupancy
was essential such that initiation of the reaction did not cause
a large increase in headspace pressure. To circumvent this
problem it was desirable to have initiation of the reaction at
low temperature such that the exotherm of reaction did not
cause the solution to boil.
Scheme 10
It was found that magnesium metal activated by the
addition of 2.3 mol % (wrt aryl bromide 23) of TMSCl
reacted smoothly with the 10% charge of bromide 23 at 28
°C in GPR grade THF, causing an adiabatic temperature rise
of 22 °C well below the boiling point of THF. The remainder
of the bromide 23 is added below 50 °C followed by a 1 h
reflux period. The TMSCl activation method of initiation
tolerated up to 500 ppm water and up to 600 ppm of
2-hydroperoxy tetrahydrofuran in separate experiments.
There was some dependence of the initiation time (5 to 40
min) on the source of THF, ironically a batch of anhydrous
THF had the longest initiation time of 40 min, the exact
reason for this variation was never deduced. Batches of THF
that gave long initiation times gave short initiation times (ca.
5 min) after distillation. The only new impurity that arose
was out-sourced before the chemistry in Scheme 10 was
scaled up.
Formation of the Grignard reagent 14 from bromide 23
was originally achieved on laboratory scale by reaction of
ca. 10% of bromide 23 with activated magnesium24 (1,2-
(21) Canonne, P.; Belley, M.; Fytas, G.; Plamondon, J. Can. J. Chem. 1988; 66,
168.
(22) Poindexter, G. S.; Owens, D. A.; Dolan, P. L.; Woo, E. J. Org. Chem.
1992, 57, 7(23), 6257.
(23) Cason J.; Baxter, W. M.; DeAcetis, W. J. Org. Chem. 1959, 24, 247.
(24) Silverman, G. S.; Rakita, P. E. Handbook of Grignard Reagents; Marcel
Dekker: New York, 1996; pp 53-78.
Vol. 5, No. 5, 2001 / Organic Process Research & Development
•
483

Scheme 12
from the use of TMSCl to initiate the Grignard reaction was
the aryl silane 31 and was never seen in the final product.
The TMSCl method to activate magnesium was success-
fully used with seven batches of bromide 23 on 20-45 kg
scale. The concentration of the Grignard reagent 14 solution
was established by titration of a sample against s-BuOH in
p-xylene using 1,10-phenanthroline as indicator.25 Towards
the end of the project an on-line React-IR method had been
successfully demonstrated in the pilot plant.
Throughout this research it was found that a rigorously
anaerobic atmosphere (O2 content <0.2%) was required;
otherwise, two impurities (32 and 33) were observed at
unacceptably high levels. The impurity 33 reacts with ketone
13 and is carried through to produce an idoxifene-related
impurity that is not efficiently removed by recrystallisation.
Table 2
E:Z ratio
at end of
reaction
reaction
time/h
E:Z ratio
after 1 h
R′
11a
11b
11c
11d
11e
11f
11g
11h
11i
11j
11k
11l
11m
11n
11o
11p
Me
Et
MeOCH2
Me2CCH2
Me3C
MeO2C
4.5
3
1
2
7.5
2
1
1
2
1
8.5
1
2
2
1
79:21
71:29
79:21
88:12
88:12
76:24
75:25
77:28
78:22
74:26
88:12
80:20
89:11
81:19
79:21
77:23
85:15
77:23
-
89:11
93:7
-
PhCH2
Ph
-
-
4-(MeO)C6H4
4-(NO2)C6H4
2,4-(MeO)2C6H3
3,4,5-(MeO)3C6H2
2-(MeO)C6H4
2-(Me)C6H4
2-ClC6H4
78:22
-
Synthesis of Tertiary Alcohol 12. Initially a 4-fold excess
of Grignard reagent 14 was used to completely consume
ketone 13 in THF with the alcohol 12 being obtained in a
94:6
-
1
79% yield essentially as a single diastereoisomer by H
90:10
84:16
-
NMR. Conditions were soon optimised to use 1.3 equiv of
the Grignard reagent 14 and consistently obtain a yield of
the alcohol 12 of 85-89% (Scheme 11). The relative
2-furyl
2
77:23
Scheme 11
step. All derivatives of alcohol 12 were formed by reaction
of electrophiles with the potassium alkoxide of 12 in THF
at room temperature formed using either potassium hydride
or potassium hexamethyldisilazide (KHMDS). A large
number of functionalities (xanthate salts,26 xanthate esters,26
isocyanates, thioisocyanates, chlorosulphite, carbodiimide
adducts, diphenyl phosphite, and acetate) were made and
their high-temperature elimination reactions evaluated. Elimi-
nation of acetate 11a yielded the highest E:Z ratios (79:21);
hence, an extensive range of esters (again synthesised via
the potassium alkoxide of alcohol 12) (Scheme 12, Table 2)
were screened (Table 2).
After these results were obtained, work concentrated on
optimisation of the reaction with pivaloyl chloride to give
adduct 11e. Although the E:Z ratio was the same as for the
elimination reaction of the 2,4-dimethoxybenzoyl adduct 11k,
the pivaloyl chloride was a far cheaper reagent.
stereochemistry was determined by single-crystal X-ray
diffraction.
The phenyl ether 34 by-product formed on work-up was
easily removed by slurrying the crude product in petroleum
ether (60-80)/toluene (9:1).
(25) Furniss, B. S.; Hannaford, A. J.; Smith, P. W. G.; Tatchell, A. R. Vogel’s
Textbook of Practical Organic Chemistry, 5th ed.; Longman: Essex, 1989;
pp 443-445.
Selection of Ester 11e and Optimisation of syn Elimi-
nation. In the early stages of the project the alcohol 12 was
isolated as a white solid before being derivatised in the next
(26) McCague, R. J. Chem. Soc., Perkin Trans. 1 1987, 1011.
484
•
Vol. 5, No. 5, 2001 / Organic Process Research & Development

From these results it became apparent that, as the reaction
progressed, the E:Z ratio decreased. Two possible scenarios
were proposed to account for the observation:
(1) The carboxylic acid by-product from the elimination
was catalysing a competing E1 elimination mechanism.
(2) The carboxylic acid by-product was isomerising the
olefin after formation.
The second scenario was eliminated by refluxing geo-
metrically pure idoxifene in xylene with acetic acid (1 equiv),
no isomerisation was seen.
The reaction of the pivalate ester 11e in refluxing xylene
with 10 equiv of sodium carbonate demonstrated that the
E:Z ratio was maintained (ca. 14:1), but 48 h at reflux was
required for complete reaction. Other solvents and bases were
then investigated. Switching solvent to 1,2,4-trimethylben-
zene brought reaction times down to 16 h, although E:Z ratios
were variable when using sodium carbonate. When 1,1,1,3,3,3-
hexamethyldisilazane (HMDS) was employed, the E:Z ratios
were reliably high (91:9 to 93.5:6.5), much more so than
with other amines (Hu¨nig’s base and DBU). HMDS may
well be acting as a silylating agent to form the TMS ester of
pivalic acid rather than simply as a base.
Summary
In conclusion a reliable and efficient synthesis of ketone
13 has been developed and demonstrated on scale. A route
to aryl bromide 23 was developed and used to make initial
multikilogram quantities before the supply was outsourced.
An addition reaction of Grignard reagent 14 onto ketone 13
yielded tertiary alcohol 12 essentially as a single diastere-
oisomer. When alcohol 12 was derivatised as the pivalate
ester 11e, a syn-elimination reaction could be performed,
yielding idoxifene 1 as a 93:7 mixture of the E- and
Z-isomers. The E-idoxifene 1 can be obtained pure after
crystallisation from methanol and recrystallisation from IMS
99% in 70% yield from ketone 13 (Scheme 13).
Scheme 13. Manufacturing route to idoxifene
Streamlining the Process. To streamline the process
further we envisaged that isolation of the tertiary alcohol 12
and the pivalate ester 11e could be avoided if the reaction
solvent was changed to 1,2,4-trimethylbenzene throughout
the process. Hence, the Grignard reagent 14 was formed in
THF and subsequently treated with a solution of the ketone
13 in 1,2,4-trimethylbenzene. After aqueous work-up and
distillation of a small amount of the solvent to remove traces
of water, the solution of the crude tertiary alcohol 12 was
treated with a 1,2,4-trimethylbenzene solution of KHMDS
followed by pivaloyl chloride. The crude solution of ester
11e was then washed with water and again dried by
azeotropic distillation. HMDS was then added and the
solution brought to reflux for 12-14 h to give complete
conversion to idoxifene (E:Z ratio 93:7). Isolation of idox-
ifene 1 was achieved by concentration of the solution to two
volumes followed by dilution with eight volumes of methanol
to crystallise the product. The crude product was isolated
by filtration and then recrystallised from ethanol/methanol
(99:1, IMS 99%) to yield the pure product 1 in 70% yield
from ketone 13.
Direct acylation of the magnesium alkoxide of alcohol
12 with pivaloyl chloride after addition of the Grignard
reagent 14 was attempted, but the reaction failed to go to
completion. Conversion of the crude solution of ester 11e
to idoxifene 1 without aqueous work-up was briefly inves-
tigated, but the procedure gave higher levels of the impurities
35 and 36.
Experimental Section
Melting points were recorded on a Buchi 510 melting
point apparatus and are uncorrected. Infrared spectra were
recorded in ATR mode or as a dispersion in mineral oil at 4
cm-1 resolution on a Nicolet 710 FT-IR spectrometer.
Nuclear magnetic resonance spectra were recorded at 400
MHz (1H) and 100 MHz (13C) on a JEOL GSX400 FT-NMR
spectrometer. Mass spectra were recorded using CI or EI
mode at an ionisation energy of 70 eV on a VG Trio-2 single
quadrupole mass spectrometer; exact mass measurements
were made using EI mode on a JEOL SX102 double focusing
mass spectrometer. Elemental analysis was performed on a
Control Equipment 440 elemental analyser.
Vol. 5, No. 5, 2001 / Organic Process Research & Development
•
485

Initial Supply Route. 2-Phenoxyethyl Chloride (4).
Toluene (395 kg) was charged to a clean, dry, nitrogen-
flushed vessel and the solvent heated with stirring to reflux
under a Dean-Stark water separator. The contents of the
vessel were then cooled to 100-105 °C before the addition
of 2-phenoxyethanol 3 (70 kg, 506.6 mol) followed by
pyridine (43.9 kg, 554.9 mol, 1.1 equiv), washed in with
toluene (21 kg). The reaction mixture was then heated to
reflux before the addition of thionyl chloride (63.5 kg, 533.8
mol, 1.05 equiv) as a solution in toluene (42 kg) over 90
min, washed in with toluene (21 kg). The reaction mixture
was then heated at reflux for 1 h before being cooled to 20-
25 °C. Water (490 L) was then added and the solution stirred
at 20-25 °C for 10 min before settling for 15 min. The
aqueous phase was then separated and discarded. The
washing procedure was repeated one more time before the
vessel was set for vacuum distillation and heated at less than
60 °C until toluene (500 L) had been removed. The residue
was then cooled to 30-35 °C before being transferred to a
smaller vessel, washed in with toluene (3.6 kg). The residue
was then distilled under vacuum (10-15 mbar) at 110-120
°C to give 2-phenoxyethyl chloride27 4 (71.44 kg, 69.7 kg
at 100%, 445 mol, 88%).
1-[4-(2-Chloroethoxy)phenyl]-2-phenyl-1-butanone (5).
2-Phenoxyethyl chloride 4 (71.7 kg, 69.94 kg at 100%, 446.6
mol) was charged to a reactor containing 1,2-dichloroethane
(184.0 kg) and washed in with 1,2-dichloroethane (25.0 kg).
2-Phenylbutyric acid (74.4 kg at 98.7%, 73.4 kg at 100%,
447.0 mol) was added. The reaction mixture was adjusted
to 14 °C, and trifluoroacetic anhydride (161.4 kg, 768.5 mol)
was added over 46 min, maintaining the temperature at 14-
18 °C and washed in with 1,2-dichloroethane (37.3 kg). The
reaction mixture was stirred at 18-21 °C for 32 min and
then slowly heated over 2 h to 33 °C and held at this
temperature for 23 h. The reaction mixture was cooled to
23 °C, and toluene (353.0 kg) was added. A solution of
sodium carbonate (86.0 kg) in demineralised water (507.0
L) was slowly added over 2.5 h, maintaining the temperature
at 18-22 °C. The reaction mixture was then allowed to
separate for 35 min. The lower aqueous layer was separated
and discarded. The organic layer was washed with dem-
ineralised water (1 × 355.0 L) and the mixture allowed to
settle before the aqueous layer was separated and discarded.
Solvent (475.0 L) was distilled off at atmospheric pressure.
The solution was cooled to 20 °C, and petroleum ether 80-
100 (477.0 kg) added. The suspension was cooled to 5 °C
and stirred at 0-5 °C for 1 h. The product was isolated by
filtration and washed with cold petroleum ether (80-100)
(135.0 kg). The wet product was dried in the filter drier at
15-25 °C for 30 h to give ketone27 5 (105.5 kg, 348.4 mol,
78%).
over 2 h, maintaining the temperature between -67 and -54
°C. The mixture was stirred at this temperature for 53 min.
Further butyllithium in hexane (2.0 kg at 15.0%, 0.3 kg at
100%, 4.68 mol) was added over 2 min and washed in with
hexane (2.7 kg). The reaction mixture was transferred to a
second vessel containing a solution of ammonium chloride
(50.0 kg) in water (150.0 L) and washed in with toluene
(52.0 kg), and the reaction mixture temperature was adjusted
to 20 °C. The reaction mixture was stirred for 15 min and
allowed to separate for 30 min. The lower aqueous phase
was separated and discarded. Solvent (475 L) was distilled
off at atmospheric pressure to leave an approximate volume
of 120.0 L. The solution was transferred to a third vessel
and washed in with toluene (20.0 kg). Solvent (75 L) was
distilled off in vacuo to leave an approximate volume of 40.0
L. Ethanol (30 L) was then added and a “put and take”
distillation carried out, maintaining the volume in the vessel,
by the addition of ethanol (8 × 30.0 L), until an internal
temperature of 80.5 °C was obtained. The ethanolic solution
of tertiary alcohol 6 was then treated with ethanol (94.6 kg)
and concentrated hydrochloric acid (2.5 kg). The reaction
mixture was heated to reflux (78 °C) and held at reflux for
30 min. The reaction mixture was cooled to 40 °C over 30
min, and pyrrolidine (43.0 kg at 99.2%, 42.7 kg at 100%,
600.4 mol) was added. The reaction mixture was heated to
reflux (83 °C) and held at reflux for 16 h. The solution of
crude idoxifene 1 was filtered hot, at 70 °C, into another
reactor. The transfer line was washed through with hot
ethanol (4.0 kg). Idoxifene 1 seed crystals (1.0 g) were added
at 51 °C. The product started to crystallise after 35 min at
28 °C. The reaction mixture was cooled to 1 °C over 30
min and stirred at 0-5 °C for 2 h. The product was isolated
by centrifugation and washed with cold (5 °C) filtered ethanol
(72.0 kg). The wet product 1 (24.4 kg) was added to filtered
absolute ethanol (100.0 kg). The mixture was heated to reflux
(78 °C) and held at reflux for 10 min to confirm that
dissolution had occurred. The solution was cooled and the
product crystallised at 52 °C after 13 min. The reaction
mixture was cooled to 5 °C over 72 min and held at this
temperature for 3 h. The product was isolated by centrifuga-
tion and washed with cold, filtered ethanol (48.0 kg). The
wet product 1 (19.6 kg) was dried in a vacuum pan drier at
45-50 °C for 24 h to give, after sieving, idoxifene 1 (18.0
kg at 99.9%, 17.98 kg at 100%, 34.3 mol, 35%) as an off-
white crystalline solid; mp ) 106-109 °C; IR (Nujol, cm-1)
3050 (m), 1604 (m), 1507 (s), 1247 (m), 1044 (m), 831 (s);
1H NMR (CDCl3, 400 MHz) 7.66 (2H, d, J ) 8.4 Hz), 7.19-
7.05 (5H, m), 6.98 (2H, d, J ) 8.4 Hz), 6.73 (2H, d, J ) 8.8
Hz), 6.55 (2H, d, J ) 8.8 Hz), 3.96 (2H, t, J ) 6.0 Hz),
2.81 (2H, t, J ) 6.0 Hz), 2.62-2.57 (4H, m), 2.44 (2H, q,
J ) 7.5 Hz), 1.82-1.77 (4H, m), 0.91 (3H, t, J ) 7.5 Hz);
13C NMR (CDCl3, 100 MHz) 156.9, 143.3, 142.0, 141.8,
137.2, 137.1, 134.9, 131.8, 131.5, 129.6, 127.9, 126.2, 113.4,
92.1, 66.7, 55.0, 54.7, 29.0, 23.4, 13.6; MS m/z 523 (M)+
(30); HRMS calcd for C28H30INO m/z 523.1372 found
523.1368; Anal. calcd for C28H30INO: C 64.25, H 5.78, N
2.68, I 24.24; found: C 64.21, H 5.76, N 2.72, I 24.27.
1-(2-{4-[(E)-1-(4-Iodophenyl)-2-phenyl-but-1-enyl]-
phenoxy}ethyl)pyrrolidine, Idoxifene (1). Dry toluene
(340.0 kg), 1,4-diiodobenzene (36.28 kg, 36.1 kg at 100%,
109.4 mol), and ketone 5 (30.0 kg, 99.1 mol) were charged
to a vessel and cooled to -60 °C. n-Butyllithium in hexane
(48.7 kg at 15.0%, 7.31 kg at 100%, 114.1 mol) was added
(27) Bell, R. A.; Dickson, K. C.; Valliant, J. F. Can. J. Chem. 1999, 77, 146.
486
•
Vol. 5, No. 5, 2001 / Organic Process Research & Development

McMurry Approach. 1-[4-(2-Chloroethoxy)phenyl]-1-
(4-iodophenyl)methanone (9). To a solution of 2-phenoxy-
ethyl chloride 4 (14.7 g, 93.9 mmol) and 4-iodobenzoyl
chloride (25.0 g, 93.9 mmol) in carbon disulfide (65 mL)
magnetically stirred at 25 °C under nitrogen was added
aluminum chloride (10.65 g, 79.8 mmol) over 10 min; an
exothermic reaction resulted in a deep red mixture. The
reaction mixture was stirred at room temperature for 2 h
before being poured onto ice/water/concentrated hydrochloric
acid (1:1:2) (300 mL). The mixture was warmed to room
temperature and the product extracted into DCM (1 × 300
mL and 1 × 125 mL). The combined organic extract was
washed with water (150 mL) before being dried (MgSO4),
filtered, and concentrated in vacuo, and the residue was
recrystallised from ethyl acetate/hexane (1:2) (300 mL),
affording benzophenone 9 (22.6 g, 58.2 mmol, 62%) as a
pale pink crystalline solid; mp ) 149 °C; IR (ATR mode,
cm-1) 1637 (m), 1602 (m), 1577 (m), 1253 (m), 1031 (s),
solution of n-butyllithium (12.6 mL of 1.6 M in hexanes,
20.22 mmol) and the resulting green suspension stirred for
30 min. In a separate flask amide 15 (3.99 g, 19.26 mmol)
in THF (30 mL) was cooled to -78 °C and treated with the
solution of 4-iodophenyllithium prepared above in a single
portion. The reaction mixture was then stirred at -78 °C
for 30 min before being warmed to room temperature and
quenched by the addition of 1 M ammonium chloride
solution (80 mL). The product was extracted into ether (2
× 40 mL), and the combined ethereal extracts were washed
with 2 M hydrochloric acid (50 mL) followed by saturated
sodium bicarbonate solution (50 mL) before being dried
(MgSO4), filtered, and concentrated in vacuo to give the
crude product 13 (6.63 g). The crude material was then
recrystallised from petroleum ether (80-100) giving pure
ketone 13 (3.24 g, 9.23 mmol, 48%) as a white crystalline
solid; mp ) 88-89 °C; IR (ATR mode, cm-1) 2969 (w),
2925 (w), 1676 (s), 1577 (s), 1000 (s); 1HNMR (CDCl3, 400
MHz) 7.73 (2H, dt, J ) 8.6, 1.9 Hz), 7.65 (2H, dt, J ) 8.6,
1.9 Hz), 7.32-7.16 (5H, m), 4.34 (1H, t, J ) 7.2 Hz), 2.25-
2.13 (1H, m), 1.91-1.78 (1H, m), 0.88 (3H, t, J ) 7.3 Hz);
13CNMR (CDCl3, 100 MHz) 199.7, 139.7, 138.2, 136.6,
130.5, 129.4, 128.6, 127.5, 101.2, 55.8, 27.4, 12.7; HRMS
calcd for C16H15IO m/z 350.0168 found 350.0155. Anal. calcd
for C16H15IO: C 54.88, H 4.32; found: C 54.80, H 4.34.
Synthesis of Ketone 13 from 4-Iodobenzoyl Chloride.
1-(4-Iodophenyl)-2-phenyl-1-butanone (13). To a stirred
solution of diisopropylamine (56.4 mL, 401.9 mmol) in THF
(100 mL) at 0 °C under argon was added dropwise n-
butyllithium (240 mL of 1.6 M in hexanes, 383.7 mmol).
The yellow solution was then stirred at 0 °C for 1 h before
the dropwise addition of a solution of 2-phenylbutyric acid
(30.0 g, 182.7 mmol) in THF (130 mL). The solution was
then stirred at 0 °C for 5 min before being warmed to room
temperature and stirred for a further 2 h. The solution was
then cooled to -10 °C and treated dropwise with a solution
of 4-iodobenzoyl chloride (48.6 g, 182.7 mmol) in THF (140
mL). The solution was then stirred at -10 °C for a further
1 h before being allowed to warm to room temperature. The
reaction mixture was then quenched with 1 M ammonium
chloride solution (800 mL). The product was then extracted
into ether (2 × 750 mL and 1 × 300 mL). The combined
organic phase was then washed with 2 M hydrochloric acid
(400 mL) followed by saturated sodium bicarbonate solution
(400 mL). The ethereal extract was then dried (MgSO4),
filtered, and concentrated in vacuo to give the crude product
13 (60.9 g). The crude solid was then recrystallised from
ethanol (150 mL) to give pure ketone 13 (44.76 g, 127.8
mmol, 70%) as a white crystalline solid.
1
752 (s); H NMR (CDCl3, 400 MHz) 7.84 (2H, d, J ) 6.4
Hz), 7.80 (2H, d, J ) 6.8 Hz), 7.47 (2H, d, J ) 6.4 Hz),
6.98 (2H, d, J ) 6.4 Hz), 4.32 (3H, t, J ) 5.6 Hz), 3.86
(3H, t, J ) 5.6 Hz); 13C NMR (CDCl3, 100 MHz) 194.5,
161.9, 137.5, 132.4, 131.2, 130.3, 114.2, 99.5, 68.1, 41.6;
MS m/z 386 (35ClM)+ (100), 388 (37ClM + 1)+ (32); HRMS
calcd for C15H1235ClIO2 m/z 385.9571 found 385.9569.
1-(2-{4-[(E)-1-(4-Iodophenyl)-2-phenylbut-1-enyl]-
phenoxy}ethyl)pyrrolidine, Idoxifene (1). To a magneti-
cally stirred mixture of zinc dust (10.67 g, 163.2 mmol),
1,2-bis(diphenylphosphino)ethane (32.0 g, 80.3 mmol), and
dry 1,4-dioxane (260 mL) at room temperature under
nitrogen was added titanium(IV)chloride (15.2 g, 8.80 mL,
80.3 mmol) dropwise over 15 min. The reaction mixture was
then heated to reflux for 2 h before being cooled to room
temperature and treated with a mixture of benzophenone 9
(10.0 g, 25.9 mmol), propiophenone (3.47 g, 3.44 mL, 25.9
mmol) in dry 1,4-dioxane (130 mL). The reaction mixture
was then heated to reflux for 2.5 h before being cooled to
room temperature and partitioned between DCM (500 mL)
and water (200 mL). The organic layer was separated and
washed with 2 M hydrochloric acid (200 mL) and then dried
(MgSO4), filtered, and concentrated in vacuo to give a yellow
oil. The residue was extracted into boiling hexane (2 × 200
mL) and the combined organic solution filtered before being
concentrated in vacuo to give crude alkene 7 (E:Z 4.1:1) as
a yellow oil. The crude product was dissolved in ethanol
(60 mL) and treated with pyrrolidine (11.1 g, 13 mL, 155
mmol), and the reaction mixture was heated at reflux under
nitrogen for 18 h. The solution was then cooled to room
temperature and concentrated in vacuo. The residue was
chromatographed on silica eluting with ethyl acetate/
methanol (9:1) to give crude idoxifene 1 as a yellow oil (5.5
g). The crude idoxifene 1 was then crystallised from ethanol
(15 mL) to give pure idoxifene 1 (2.80 g, 5.34 mmol, 21%)
as an off-white crystalline solid.
Laboratory Preparation of Aryl Bromide 23. 1-[2-(4-
Bromophenoxy)ethyl]pyrrolidine (23). Sodium ethoxide
(70.88 g of 96%, 1.00 mol) was dissolved in ethanol (1 L)
with stirring under nitrogen. A solution of 4-bromophenol
(86.5 g, 0.50 mol) in ethanol (500 mL) was added in one
portion, and the reaction mixture was then treated with a
solution of N-(2-chloroethyl)pyrrolidine hydrochloride (85.04
g, 0.50 mol) in ethanol (500 mL). A precipitate formed
immediately. The reaction mixture was then heated at reflux
Synthesis of Ketone 13 from Weinreb Amide 15. 1-(4-
Iodophenyl)-2-phenyl-1-butanone (13). 1,4-Diiodobenzene
(6.99 g, 21.18 mmol) in THF (80 mL) at -78 °C, magneti-
cally stirred under argon, was treated dropwise with a
Vol. 5, No. 5, 2001 / Organic Process Research & Development
•
487

for 60 h. Approximately half of the solvent was removed
by distillation and the remainder of the reaction mixture
cooled to room temperature and diluted with water (3 L).
The product was extracted into ether (4 × 750 mL) and the
combined ethereal extract washed with 1 M sodium hydrox-
ide solution (2 × 750 mL) followed by brine (1 L). The
organic solution was then dried (MgSO4), filtered, and
concentrated in vacuo to give a brown liquid (118.4 g). The
liquid was distilled and a fraction (104.42 g) (bp 122-125
°C, 0.4 mmHg) collected. This fraction was distilled again
to give aryl bromide28 23 (99.43 g, 368 mmol, 74%) (bp
139-141 °C, 1.5 mmHg) as a pale yellow oil. IR (ATR
mode, cm-1) 2962 (w), 2782 (w), 1487 (s), 1241 (s), 819
(m); 1H NMR (CDCl3, 400 MHz) 7.40-7.32 (2H, m), 6.82-
6.75 (2H, m), 4.07 (2H, t, J ) 5.9 Hz), 2.89 (2H, t, J ) 5.9
Hz), 2.68-2.57 (4H, m), 1.88-1.72 (4H, m); 13C NMR
(CDCl3, 100 MHz) 158.0, 132.2, 116.4, 112.8, 67.3, 545.0,
54.7, 23.5; MS m/z 270 (79BrM + 1)+ (98), 272 (81BrM +
1)+ (100).
Pilot-Plant Preparation of Aryl Bromide 23. 1-[2-(4-
Bromophenoxy)ethyl]pyrrolidine (23). 4-Methyl-2-pen-
tanone (360.0 kg), 325 mesh potassium carbonate powder
(120.0 kg), 4-bromophenol (30.0 kg at 99.6%, 29.9 kg at
100%, 172.8 mol) and N-(2-chloroethyl)pyrrolidine hydro-
chloride (29.5 kg, 173.2 mol) were charged to a vessel. The
reaction mixture was heated to 87 °C over 33 min and held
at 80-87 °C for 8 h. The reaction was then cooled to 20-
25 °C over 46 min. Water (300.0 L) and n-hexane (140.0
kg) were charged, and the reaction mixture was stirred for
10 min and allowed to separate for 15 min. The aqueous
layer was separated and discarded. The product was extracted
from the organic phase into 2 M hydrochloric acid solution
(135.0 kg). The organic phase was separated and discarded.
Sodium hydroxide solution (2 M, 169.6 kg), was slowly
added to the solution of product in 2 M hydrochloric acid,
maintaining the temperature at 20-25 °C. n-Hexane (140.0
kg) was added and the mixture stirred for 5 min and allowed
to separate for 15 min. The aqueous phase was separated
and discarded. Zinc chloride solution (1.2 kg in water 100.0
kg) was added to the organic phase and the mixture stirred
for 5 min and then filtered to remove a gelatinous precipitate.
The mixture was allowed to separate, and the aqueous layer
was separated and discarded. The organic phase was then
washed with saturated sodium chloride solution (135.0 kg).
The aqueous layer was separated and discarded. The organic
phase was concentrated in vacuo and n-hexane (175.0 L)
distilled off to leave approximately 30.0 L of oil. The oil
was cooled to 20 °C over 65 min and transferred to a new,
clean polydrum to give the crude product, 33.68 kg at 96.9%
(GC PAR). Three batches were combined for purification,
thus providing crude aryl bromide 23 (88.97 kg at 96.8%,
86.15 kg at 100%, 319mol) which was distilled at 139-143
°C, at 0.25-0.3 mbar to give the purified aryl bromide28 23
(82.33 kg at 98.9%, 81.43 kg at 100%, 301.6 mol, 65%) as
a pale yellow oil.
mL) and pyridine (4.00 g, 4.1 mL, 50.6 mmol) were brought
to reflux under a Dean and Stark water separator, and reflux
continued for 1 h. The solution was then cooled to room
temperature and treated with 2-(4-bromophenoxy)ethanol 29
(10.0 g, 46.1 mmol), and the solution was heated to reflux
before the dropwise addition of thionyl chloride (5.79 g, 3.55
mL, 48.7 mmol) over 30 min. The reaction mixture was
heated at reflux for 2 h before being cooled to room
temperature and quenched by the addition of water (50 mL).
The organic layer was separated, dried (MgSO4), filtered,
and concentrated in vacuo, affording a solid that was
recrystallised from n-hexane to give the alkyl chloride28 30
(7.71 g, 32.7 mmol, 71%) as a white crystalline solid; mp
55-57 °C; IR (ATR mode, cm-1) 2945 (w), 1485 (m), 1233
(m), 1033 (m), 817 (s); 1H NMR (CDCl3, 400 MHz) 7.42-
7.35 (2H, m), 6.83-6.75 (2H, m), 4.20 (2H, t, J ) 5.9 Hz),
3.80 (2H, t, J ) 5.9 Hz); 13C NMR (CDCl3, 100 MHz) 157.7,
132.8, 117.0, 114.1, 68.7, 42.2.
1-[2-(4-Bromophenoxy)ethyl]pyrrolidine (23). 2-(4-
Bromophenoxy)ethyl chloride 30 (5.0 g, 21.2 mmol), pyr-
rolidine (7.55 g, 8.9 mL, 106.2 mmol) and ethanol (30 mL)
were heated at reflux under nitrogen for 18 h. The reaction
mixture was then cooled to room temperature and concen-
trated in vacuo. The residue was dissolved in water (50 mL)
and the pH adjusted to 12 by the addition of 40% w/v sodium
hydroxide solution. The product was then extracted into
hexane (2 × 50 mL). The combined organic extract was dried
by filtration through phase-separation paper and concentrated
in vacuo to give a red oil (5.36 g) that was Kugelrohr distilled
(ot 175 °C, 0.1 mbar) to give aryl bromide28 23 (5.10 g,
18.9 mmol, 89%) as a clear colourless oil.
Laboratory Preparation of Tertiary Alcohol 12.
(1RS,2SR)-1-(4-Iodophenyl)-2-phenyl-1-[4-(2-pyrrolidin-
1-ylethoxy)phenyl]butan-1-ol (12). Approximately 20 mL
of a solution of aryl bromide 23 (40.0 g, 0.148 mol) and
1,2-dibromoethane (6.4 mL, 74 mmol) in dry THF (300 mL)
was added to a mixture of magnesium raspings (5.94 g, 0.244
mol) in THF (50 mL) and magnetically stirred at room
temperature under nitrogen. The reaction mixture was then
heated to 50 °C at which point a visible reaction took place.
Once the reaction had subsided, the reaction mixture was
heated to reflux, the remainder of the aryl bromide 23
solution was added over 2 h, and the reaction mixture was
heated at reflux for a further 30 min. The reaction mixture
was then cooled to room temperature, and the concentration
of the Grignard reagent 14 solution was ascertained by
titration with 2-butanol in p-xylene using 1,10-phenanthroline
as an indicator. Ketone 13 (40.0 g, 0.114 mol) in THF (100
mL) was then added dropwise over 30 min, maintaining the
reaction temperature at ca. 25 °C with a cold water bath.
After the addition the reaction mixture was stirred for a
further 30 min before being poured onto a mixture of crushed
ice (600 g) and 1 M aqueous ammonium chloride solution
(1.2 L). The crude product was extracted into ether (3 ×
800 mL), and the combined organic phases were washed
with brine (1 L), dried (MgSO4), filtered, and concentrated
in vacuo to give the crude tertiary alcohol 12 (71.38 g). The
crude product was slurried in petroleum ether (60-80)/
Alternative Laboratory Preparation of Aryl Bromide
23. 2-(4-Bromophenoxy)ethyl Chloride (30). Toluene (50
(28) Reagent available from Sigma-Aldrich, Ltd.
488
•
Vol. 5, No. 5, 2001 / Organic Process Research & Development

toluene (9:1) (250 mL) for 1 h before isolation by filtration
and drying in a desiccator at 1 mmHg to give the tertiary
alcohol 12 (54.44 g, 0.101 mol, 88%) as an off-white solid;
mp ) 120-123 °C; IR (ATR mode, cm-1) 3102 (w), 2964
min. A separate nitrogen-purged vessel was charged with
aryl bromide 23 (45 kg, 166.6 mol) and THF (100 kg) and
stirred at 20-25 °C to ensure homogeneity; 15 L of the aryl
bromide 23 solution was then added to the suspension of
activated magesium raspings in THF. The reaction mixture
was then stirred until initiation occurred, apparent by an
internal temperature rise to ca. 50 °C (time taken to initiate
was between 5 and 40 min, depending on the quality of the
THF). The remainder of the aryl bromide 23 solution was
then added over 45 min, maintaining the temperature between
38 and 51 °C. The aryl bromide 23 solution was then washed
in with THF (25 kg). The solution of Grignard reagent 14
was then stirred at 45-50 °C for 30 min before being cooled
to 20-25 °C. The concentration of the Grignard reagent 14
solution was then determined by on-line IR spectroscopy
using the React-IR system or by titration of a sample from
the vessel with s-BuOH in p-xylene using 1,10-phenanthro-
line as an indicator.25 The solution of Grignard reagent 14
was then filtered into a clean, dry, nitrogen-purged vessel
and washed in with THF (25 kg). To the Grignard reagent
14 at 20-25 °C was added a solution of ketone 13 (45 kg,
128.5 mol) in 1,2,4-trimethylbenzene (234 kg) over 1 h,
maintaining the temperature at 20-25 °C, and washed in
with 1,2,4-trimethylbenzene (25 kg); the reaction mixture
stirred for a further 1 h 15 min at 20-25 °C. The solution
was then added to a stirred solution of trisodium citrate
trihydrate (74 kg) in water (340 L), maintaining the tem-
perature at 20-25 °C, and washed in with 1,2,4-trimethyl-
benzene (25 kg). The biphasic mixture was stirred for 30
min and allowed to settle for 15 min before the aqueous
phase was separated and discarded. The organic phase was
then washed twice with water (280 L) each with a 5-min
stir period and a 15-min settling period. The remaining
organic phase was then concentrated in vacuo to remove
residual water and THF (160 L). The solution of crude
tertiary alcohol 12 was then cooled to 15-20 °C and the
resulting suspension treated with a solution of KHMDS in
1,2,4-trimethylbenzene (196 kg of 17% w/w, 33.3 kg of
KHMDS, 167.0 mol, 1.3 equiv) over 1 h, maintaining a
temperature of 20-25 °C and washed in with 1,2,4-
trimethylbenzene (25 kg). The reaction mixture was then
stirred for 1 h at 20-25 °C before the addition of pivaloyl
chloride solution (21.7 kg, 180.0 mol, 1.4 equiv, in 1,2,4-
trimethylbenzene (40 kg)) over 1.5 h, maintaining the
temperature below 30 °C, and washed in with 1,2,4-
trimethylbenzene (9 kg). The solution was then stirred at
20-25 °C for 1.5 h before being washed with water (230
L) three times. Each wash was stirred for 5 min and allowed
to settle for 15 min before separation. The vessel was then
set for distillation, and the reaction mixture was concentrated
in vacuo by distilling out 200 L of solvent. The solution of
crude ester 11e was then cooled to 80-90 °C and treated
with HMDS (31.2 kg, 192.2 mol); the reaction mixture was
then brought to reflux (165 °C) for 12 h. The solution was
then cooled to 20-25 °C and treated with a solution of
sodium carbonate (29.8 kg, 278.5 mol) in water (280 L) and
the biphasic mixture stirred for 5 min and allowed to settle
for 15 min before the aqueous layer was separated and
1
(w), 2827 (w), 1507 (m), 1480 (m), 1248 (s), 1178 (m); -
HNMR (CDCl3, 400 MHz) 7.45-7.40 (2H,m), 7.40-7.35
(2H, m), 7.17-7.06 (5H, m), 6.94-6.86 (4H, m), 4.09 (2H,
t, J ) 6.1 Hz), 3.48 (1H,. dd, J ) 6.1, 4.0 Hz), 2.89 (2H, t,
J ) 6.1 Hz), 2.67-2.58 (4H, m), 2.37 (1H, br s), 1.88-
1.72 (6H,m), 0.75 (3H, t, J ) 7.3 Hz); 13CNMR (CDCl3,
100 MHz) 155.8, 144.8, 137.7, 136.1, 134.6, 128.2, 126.0,
125.9, 125.4, 125.6, 112.4, 89.8, 78.6, 65.1, 54.2, 53.2, 52.8,
21.6, 21.5, 10.7; MS m/z 542 (M + 1)+ (100); HRMS calcd
for C28H32NIO2 m/z 541.1478, found 541.1467.
Manufacturing Route. 1-(4-Iodophenyl)-2-phenyl-1-
butanone (13). Dry THF (11.0 kg) and lithium diisopropy-
lamide in heptane/ethylbenzene/THF (183.5 kg at 25.2%,
46.2 kg at 100%, 431.3 mol) were charged to a vessel and
washed in with dry THF (4.0 kg). A solution of 2-phenyl-
butyric acid (35.9 kg, 218.7 mol) in dry THF (65 kg) was
added over 85 min, maintaining the temperature at 23-24
°C and washed in with dry THF (30 kg). The reaction
mixture was stirred at 10-20 °C for 1.5 h. The solution of
dianion 16 was added over 30 min to a solution of ethyl
4-iodobenzoate 18 (48.4 kg, 175.3 mol) in toluene (156.4
kg) and washed in with dry THF (30 kg). The reaction
mixture was stirred at 20-23 °C for 1 h. The reaction
mixture was quenched into concentrated hydrochloric acid
(110 kg) in water (140.0 L) over 1 h, maintaining the
temperature between 17 and 27 °C, and washed in with
toluene (30 kg). The reaction mixture was then stirred for 5
min before the phases were allowed to separate for 45 min.
The lower aqueous phase was separated and discarded. A
solution of sodium carbonate (24.5 kg) in water (195 L) was
charged, the reaction mixture was stirred for 5 min, and the
phases were allowed to separate for 25 min. The lower
aqueous phase was separated and discarded. Water (170 L)
was charged, the reaction mixture was stirred for 5 min, and
the phases were allowed to separate for 15 min. The lower
aqueous phase was separated and discarded. The organic
layer was transferred to another vessel and washed in with
toluene (10 kg). Solvent (450 L) was distilled off in vacuo
to leave ca. 70 L of solution. iso-Propanol (192 kg) was
added, the solution was heated to 59 °C, and complete
dissolution was confirmed. The reaction mixture was then
cooled to 0 °C over 3.5 h to cause crystallisation and then
stirred at 0 °C for 1 h. The product 13 was then isolated by
filtration and washed with cold 2-propanol (77 kg). The wet
product was then dried in vacuo at 25-30 °C for 24 h to
give the pure ketone 13 (57.5 kg, 164.2 mol, 94%) as a white
crystalline solid.
1-(2-{4-[(E)-1-(4-Iodophenyl)-2-phenylbut-1-enyl]-
phenoxy}ethyl)pyrrolidine, Idoxifene (1). To a thoroughly
nitrogen-purged vessel was added THF (100 kg) and
magnesium raspings (4.8 kg, 197.5 mol), and the mixture
was stirred at 25-28 °C before the addition of a 1.0 M
solution of TMSCl in THF (3.6 kg, 4.1 L, 4.1 mol of TMSCl)
washed in with THF (2 kg). The mixture was stirred for 10
Vol. 5, No. 5, 2001 / Organic Process Research & Development
•
489

discarded. The organic phase was then washed twice with
water (230 L), each wash being stirred for 5 min and allowed
to settle for 15 min before being separated and discarded.
The organic solution of idoxifene 1 was then concentrated
in vacuo to ca. 90 L in volume and cooled to 55-60 °C
before being treated with methanol (292 kg). The solution
was then heated to 60-65 °C and filtered into another vessel
and washed in with methanol (64 kg) where the reaction
mixture was cooled to 25 °C to crystallise and then cooled
further to -10 to -5 °C and the slurry stirred for 1 h. The
product was isolated by filtration at -10 to -5 °C and
washed with cold methanol (71 kg) and blown dry at 1.5
bar. The solid idoxifene 1 was then partially dried at 45-50
°C in vacuo for 1.5 h. The partially dried idoxifene 1 (67.8
kg) in a nitrogen-purged vessel was then treated with IMS
99% (235 kg) and heated to 65-70 °C until dissolution. The
solution was then filtered into another vessel and washed in
with hot IMS 99% (39 kg). The solution was then cooled to
crystallise and then further to 0 °C and stirred for 1 h. The
idoxifene 1 was then isolated by filtration and washed with
cold IMS 99% (28.5 kg) and blown dry at 1.5 bar and then
dried in vacuo at 45-50 °C to give pure idoxifene 1 (47.1
kg, 90.0 mol, 70%) as an off-white crystalline solid.
ester 11e in 1,2,4-trimethylbenzene (500 mL) obtained from
the pilot plant after aqueous work-up and azeotropic drying
was concentrated in vacuo (ca. 1 mmHg at 60 °C) to afford
crude ester 11e as a thick orange/brown oil (140 g). The
crude material was chromatographed on silica (870 g) eluting
with toluene/triethylamine (30:1) to give ester 11e (21.2 g)
as a pale yellow foam. This material was rechromatographed
on silica (860 g) eluting with toluene/triethylamine (25:1)
to afford pivalate ester 11e (16.9 g, 27.0 mmol) as an off-
white granular solid; mp ) 69-70 °C; IR (ATR mode, cm-1)
2965 (w), 2777 (w), 1727 (s), 1281 (s), 1146 (s), 815 (s);
1H NMR (CDCl3, 400 MHz) 7.65-7.6 (2H, m), 7.22-7.10
(5H, m), 6.93-6.87 (2H, m), 6.82-6.72 (4H, m), 4.44 (1H,
dd, J ) 12.1, 1.9 Hz), 4.12 (2H, t, J ) 6.7 Hz), 2.92 (2H, t,
J ) 6.7 Hz), 2.70-2.58 (4H, m), 2.02-1.90 (1H, m), 1.88-
1.76 (4H, m), 1.32-1.08 (1H, m), 0.88 (9H, s), 0.72 (3H, t,
J ) 7.3 Hz); 13C NMR (CDCl3, 100 MHz) 176.3, 158.5,
144.2, 139.3, 137.0, 133.7, 131.1, 131.0, 130.6, 127.7, 127.2,
113.0, 93.9, 88.5, 67.5, 55.6, 55.2, 51.3, 39.6, 27.1, 26.2,
23.9, 12.7; MS m/z 626 (M + 1)+ (100), 524 (5).
Acknowledgment
We thank John Lewis for the pilot plant support and Jeff
Richards, Kevin Sutcliffe, and Mike Kingswood for the
analytical support they provided during this project.
Isolation of a Pure Sample of Pivalate Ester 11e. 2,2-
Dimethylpropionic Acid (1RS,2SR)-1-(4-Iodophenyl)-2-
phenyl-1-[4-(2-pyrrolidin-1-ylethoxy)phenyl]butyl Ester
(11e). Ester 11e could be isolated from large-scale pilot plant
work in the following manner. A solution of the pivalate
Received for review May 11, 2001.
OP0100394
490
•
Vol. 5, No. 5, 2001 / Organic Process Research & Development