Enhanced O-dealkylation activity of SiCl4/LiI with catalytic amount of BF3

Article abstract of DOI:10.1016/j.tetlet.2004.03.095

Dibenzyloxydihydrobenzoxathiin 1, which resisted debenzylation with SiCl₄/LiI, was effectively debenzylated with SiCl₄/LiI in the presence of a catalytic amount of BF₃. The HCl salt of the bis-debenzylated product 2 was isolated in 90% yield and 99% purity. This enhanced dealkylation activity has also been observed with other substrates.

Full text of DOI:10.1016/j.tetlet.2004.03.095

Tetrahedron
Letters
Tetrahedron Letters 45 (2004) 3729–3732
Enhanced O-dealkylation activity of SiCl4/LiI with catalytic
amount of BF3
Daniel Zewge,* Anthony King, Steven Weissman and David Tschaen
Department of Process Research, Merck Research Laboratories, PO Box 2000, Rahway, NJ 07065, USA
Received 24 February 2004; accepted 16 March 2004
Abstract—Dibenzyloxydihydrobenzoxathiin 1, which resisted debenzylation with SiCl4/LiI, was effectively debenzylated with SiCl4/
LiI in the presence of a catalytic amount of BF3. The HCl salt of the bis-debenzylated product 2 was isolated in 90% yield and 99%
purity. This enhanced dealkylation activity has also been observed with other substrates.
ꢀ 2004 Published by Elsevier Ltd.
Dihydrobenzoxathiin derivatives are known to function
as selective estrogen receptor modulators (SERMs),
which are estrogen receptor ligands that act like estro-
gen in some tissues while blocking estrogen action in
others. These molecules are of pharmaceutical interest
due to their potential in preventing estrogen related
diseases including osteoporosis, hot flashes, increased
level of LDL cholesterol, breast cancer, and obesity.1
probably generated via intramolecular Friedel–Crafts
alkylation reactions, were detected from 3% to 4%. Due
to their structural similarity to the final product,1 these
impurities were difficult to reject. To minimize these
impurities, the use of additives, such as thiourea and
N-methylimidazole, was explored to scavenge the benzyl
iodide and obviate the generation of the byproducts.
The resulting debenzylation procedure was highly
effective and provided 2 in 81% yield although the cost
of the reagents was unacceptably high for large-scale
production. Another disadvantage with the use of
thiourea was the formation of benzylthiol during the
aqueous workup, which gave an unpleasant odor.
In our recent efforts to synthesize estrogen receptor
modulator 2, a final double debenzylation step of 1 was
required (Scheme 1).1 Due to catalyst poisoning result-
ing from the presence of sulfur in the molecule, hetero-
geneous catalytic hydrogenolysis was unsatisfactory. As
an alternative, trimethylsilyliodide (TMSI) was initially
developed to accomplish the debenzylation. The TMSI
reaction was complete in 8–10 h at rt. However, the
basic workup resulted in a complex mixture of
byproducts including ꢀ40% of the N-benzylated impu-
rity. In addition, three aryl ring-benzylated impurities,
We therefore sought to develop a more efficient and cost
effective debenzylation procedure that would facilitate
the large-scale synthesis of high purity 2. A variety of
reagents have been reported for the cleavage of alkyl
ethers including AlCl3,2 SnCl4,3 lithium naphthalenide,4
Sc(NTf)3 (cat.)/anisole,5 TMSCl/NaI,6 TMSI,7 BCl3/
1. SiCl4/LiI/BF3 (cat.)
ACN/toluene
70 o
O
OH
O
S
C
HO
S
O
O
N
N
HCl
2. HCl/EtOH
O
O
1
2
Scheme 1. Debenzylation of 1.
* Corresponding author. Tel.: +1-732-594-2496; fax: +1-732-594-1499; e-mail: daniel_zewge@merck.com
0040-4039/$ - see front matter ꢀ 2004 Published by Elsevier Ltd.
doi:10.1016/j.tetlet.2004.03.095

3730
D. Zewge et al. / Tetrahedron Letters 45 (2004) 3729–3732
n-Bu4NI,8 BBr3,9 BF3/EtSH,10 and SiCl4/NaI.11 The use
of catalytic scandium(III) triflimide was investigated
but, the overnight reaction at 100 ꢁC resulted in no
debenzylation of 1 and further reaction resulted in
decomposition. The BBr3/LiI/acetic acid reagent system
was more reactive than TMSI, and conversion of 1 to 2
was complete in 2 h at rt giving an improved yield (92%)
and impurity profile. The presence of acetic acid during
the reaction helped to prevent the N-benzylation reac-
tion. Although thiourea could be replaced by the
odorless 2-mercapto-1-methylimidazole or DABCO as
effective benzyl iodide scavengers, the requirement of a
benzyl iodide scavenger for the process remained
undesirable.
results showed that the debenzylation works best in
toluene–acetonitrile (3:1). Using optimized reaction
conditions, product 2 was obtained in 92% isolated yield
and 99% purity. LiI can be replaced by NaI, however,
the reaction takes longer (ꢀ3 h). While the use of
BF3ÆAcOH is preferred and gives a slightly cleaner
reaction profile, BF3 etherate has also been used suc-
cessfully. Interestingly the use of excess BF3/LiI (up to
5 equiv) resulted in decomposition and low yield
(ꢀ20%).
Encouraged by the enhanced reactivity observed with
catalytic amounts of BF3, we investigated the general
application of this reagent system to other substrates. A
systematic comparison of SiCl4/LiI with and without
catalytic amounts of BF3 on different substrates high-
lights the improved reactivity of the former system
(Table 1).
Among the reagents reported for O-dealkylation, the
SiCl4/NaI11 system was very attractive in terms of cost
and availability of reagents. Unfortunately, minimal
debenzylation of 1 was detected after aging the reaction
mixture over 2 days at 70 ꢁC. We were however,
delighted to find that by adding catalytic amounts of
BF3 (ꢀ25 mol % per benzyl site) to the reaction, the
debenzylation activity was significantly enhanced giving
complete conversion in 45 min.12 Furthermore, the
impurities generated were minimized. A scavenger-free
workup procedure was also developed. Experimental
It is clear from this table and examples from the litera-
ture that the SiCl4/LiI system is sufficiently reactive
toward many substrates. Essentially no reaction rate
difference is observed for the dealkylation of 4-benzyl-
oxyphenylacetonitrile (entry 8). On the other hand,
2,3-dihydrobenzofuran, allyl tolyl ether, 3-methoxy-
benzonitrile, and 6-methoxy-2-naphthonitrile all greatly
a
Table 1. SiCl4/LiI dealkylation with and without BF3
Entry
Substrate
Product
# of equiv
SiCl4/LiI/BF3
Time
Conversion with
BF3 (yield)b
Conversion with-
out BF3
I
1
1.5/1.5/0.25
1.2/1.2/0.5
45 min
100% (90%)c
15%
O
OH
OH
O
2
6 h
99.8% (90%)
16%
NC
NC
OMe
NC
NC
OH
3
4
1.5/1.5/0.25
2.0/2.0/0.3
15 h
10 h
99.7% (82%)
99.8% (83%)
25%
30%
OMe
OH
SMe
OMe
SMe
OH
5
6
2.0/2.0/0.4
13 h
15 h
99.8% (89%)
99.6% (82%)
50%
65%
OMe
OH
1.5/1.5/0.25
7
8
1.2/1.2/0.25
2.0/2.0/0.3
10 h
2 h
100% (89%)
99.9% (98%)
65%
90%
OMe
NC
OH
NC
OBn
OH
a Typical reaction conditions: starting material is dissolved in sieve dried toluene/acetonitrile (0.12 M). LiI is added to the solution and system kept
under N2 while SiCl4 and BF3ÆAcOH are consecutively added via syringe. Reaction mixture is then aged at 70 ꢁC and progress of reaction is
followed by HPLC.
b Yields refer to assay yield by HPLC.
c Isolated yield of product (entry 1); characterized by 1H, 13C NMR, HRMS, and elemental analysis data.13

D. Zewge et al. / Tetrahedron Letters 45 (2004) 3729–3732
3731
OBn
OBn
OBn
BnO
BnO
S
S
+
TFA/LiI
toluene/acetonitrile
O
O
O
I
I
I
4
5
3
less than 1%
Scheme 2.
H+ I-
R1
R1
4
H
3
R1
H
I
BnO
BnO
S
BnO
S
R1
R2
BnO
S
S
I-
HI
H
2
1
O
R2
R2
O
O
O
R2
11
10
9
3
I-, HI
I-
3a R1 =
H2O
R1
R2 =
BnO
H
H
S
BnO
SH R1
BnO
SH
+ I2
Br
H
H+
H
R1
+ I2
3b R1, R2 =
R2
O
4
O
12
R2
O
R2
13
Scheme 3.
benefited from the presence of the BF3 (entries 1–4).
With 2,3-dihydrobenzofuran the reaction with BF3 was
complete within 45 min whereas only 15% conversion
was observed without BF3. Deallylation of allyl tolyl
ether required 6 h in the presence of BF3 giving 90%
yield of 4-methylphenol, but gave only 16% conversion
in the absence of BF3 within this time frame. Similarly,
3-methoxybenzonitrile and 6-methoxy-2-naphthonitrile
were dealkylated at a much faster rate with BF3. Several
substrates were found to give marginally accelerated
dealkylation rates with BF3 (entries 5–7). Provided with
longer reaction time, these substrates would have been
effectively dealkylated even in the absence of BF3.
4 in 93% isolated yield. The use of TMSI and H2O also
gave 4 in 86% isolated yield.
Based on these experimental findings, we propose the
following mechanisms for the transformation of 3 to 4
(Scheme 3). Thus, the initial protonation of the double
bond generates an unstable oxonium ion 9, which
facilitates the migration of the sulfur. The resulting new
benzylic carbocation 10 is then trapped by iodide to give
11, which is subsequently reduced by another iodide15 to
give iodine and the observed product 4.16 Alternatively a
nucleophilic addition of iodide at C-2 of 9, followed by
its elimination as iodine could form a carbanion
resulting in benzoxathiin ring opening and formation of
a reactive thiophenol intermediate 12. Subsequent acid
catalyzed ring closure of 13 gives 1,3-benzoxathiol
product 4. This procedure has also been applied to
substrates 3a17 and 3b18 with equal success. In each case
the sulfur migrates selectively to give only one product.
The application of the SiCl4/LiI/BF3 reaction conditions
to 3, an intermediate in the total synthesis of 2, resulted
in no debenzylation. Instead, an interesting reductive
rearrangement product 4, formed via the selective 1,2-
migration of the sulfur, was recovered along with a
small amount of the ketone 5. The rearranged product 4
was also obtained when 3 was treated with 4 equiv of
SiCl4/LiI. The addition of BF3 to this reaction only
resulted in the accelerated formation of 4. Use of HCl
and HBr did not give 4.
In conclusion, catalytic amounts of BF3 can tremen-
dously enhance the dealkylation power of SiCl4/LiI, a
reagent system that is inexpensive and readily available.
An interesting novel reductive rearrangement reaction
was also observed for the transformation of benzo-
xathiin 3 to 1,3-bezoxathiol 4.
A similar intramolecular rearrangement under oxidizing
conditions has been reported.14 To rationalize this novel
reductive rearrangement, we carried out several inves-
tigative NMR experiments. When the reaction was done
in the presence of 2–3 equiv of D2O, ꢀ80% deuterium
incorporation was detected at the methylene carbon of
4, suggesting the reductive rearrangement reaction could
be due to the unintentionally generated HI. Further
proof was obtained when the reaction was run using in
situ generated HI from trifluoroacetic acid and LiI
(Scheme 2). The reaction was instantaneous at rt to give
Acknowledgements
We thank Dr. P. G. Dormer and R. A. Reamer for
NMR support, Dr. T. A. Blizzard for helpful sugges-
tions, Marjorie Waters for providing substrates 3a and
3b, and Dr. T. J. Novak for HRMS analysis.

3732
D. Zewge et al. / Tetrahedron Letters 45 (2004) 3729–3732
J ¼ 7:7 Hz, 2H). 13C NMR (100 MHz, CDCl3): d 153.5,
References and notes
130.9, 128.4, 127.3, 121.3, 115.7, 35.3, 5.0. HRMS: Calcd
for C8H9IO (MꢁH) ¼ 246.9620, found ¼ 246.9624. Anal.
Calcd for C8H9IO: C, 38.7; H, 3.7. Found: C, 38.9; H, 3.4.
14. (a) Cermola, F.; DeLorenzo, F.; Giordano, F.; Graziano,
M. L.; Iesce, M. R.; Palumbo, G. Org. Lett. 2000, 2, 1205–
1207; (b) Cermola, F.; Iesce, M. R. J. Org. Chem. 2002, 67,
4937–4944.
1. Song, Z. J.; King, A. O.; Waters, M. S.; Lang, F.; Zewge,
D.; Bio, M.; Leazer, J. L., Jr.; Javadi, G.; Kassim, A.;
Tschaen, D. M.; Reamer, R. A.; Rosner, T.; Chilenski,
J. R.; Mathre, D. J.; Volante, R. P.; Tillyer, R. Proceedings
of the National Academy of Sciences, in press.
2. Watanabe, S.; Nakazumi, H.; Akagi, N.; Maeda, K.;
Kitao, T. J. Hetetrocycl. Chem. 1990, 27, 1241–1245.
3. Arnoldi, A.; Bassoli, A.; Caputo, R.; Merlini, L.; Pal-
umbo, G.; Pedatella, S. J. Chem. Soc., Perkin Trans. 1
1994, 1241–1244.
15. Sakai, T.; Miyata, K.; Utaka, M.; Takeda, A. Bull. Chem.
Soc. Jpn. 1987, 60, 1063–1066.
16. Representative procedure for compounds 4, 4a, and 4b. To
a solution of 3 (1 g, 1.6 mmol) in toluene/acetonitrile
(9 mL/3 mL) was added LiI (0.86 g, 6.4 mmol). Reaction
mixture was kept under N2 while TFA (0.49 mL,
6.4 mmol) was added via syringe. Reaction was complete
over 10 min at rt as followed by HPLC (96% assay yield).
Reaction mixture was diluted with EtOAc (20 mL) and
washed with water (2 · 20 mL) followed by 5% Na2S2O3
(10 mL). The organic layer was dried with Na2SO4,
filtered, and concentrated. The light yellow oily material
was treated with methyl tert-butyl ether and aged at rt
overnight affording an off white solid 4 (0.93 g, 93%). Mp
139–141 ꢁC; 1H NMR (400 MHz, CDCl3): d 7.62–7.59 (m,
2H), 7.43–7.29 (m, 10H), 7.12–7.08 (om, 3H), 6.83 (d,
J ¼ 8:7 Hz, 1H), 6.82 (m, 1H), 6.76 (d, J ¼ 2:6 Hz, 1H),
6.62 (dd, J ¼ 8:7 Hz, 2.6, 1H), 6.55–6.59 (om, 2H), 4.97 (s,
2H), 4.94 (ABq, Dm ¼ 5:5, J ¼ 12:0, 2H), 3.64 (d,
J ¼ 13:8 Hz, 1H), 3.52 (d, J ¼ 13:8 Hz, 1H). 13C NMR
(100 MHz, CDCl3) 158.5, 154.8, 149.4, 142.7, 137.2, 137.2,
137.1, 136.4, 129.0, 128.7, 128.1, 128.1, 127.8, 127.6, 127.6,
127.1, 123.6, 117.2, 114.0, 112.1, 111.1, 109.7, 103.0, 94.0,
71.0, 70.1, 49.6. HRMS calcd for C34H27SO3I
(MþH) ¼ 643.0804, found ¼ 643.0809. Anal. Calcd for
C34H27SO3I: C, 63.6; H, 4.2. Found: C, 63.6; H, 4.1.
17. Product 4a crystallized from EtOH–ACN (5:1). Mp 108–
110 ꢁC; 1H NMR (400 MHz, CDCl3) 7.42–7. 32 (m, 7H), d
7.27–7.15 (m, 5H), 7.00–6.95 (m, 2H), 6.87 (d, 1H,
J ¼ 8:7 Hz), 6.77 (d, J ¼ 2:6 Hz, 1H), 6.63 (dd, J ¼ 8:7 Hz,
2.6, 1H), 4.97 (s, 2H), 3.68 (d, J ¼ 13:8 Hz, 1H), 3.56 (d,
J ¼ 13:8 Hz, 1H). 13C NMR (100 MHz, CDCl3): d 154.8,
149.5, 142.0, 137.2, 134.9, 131.2, 130.9, 128.8, 128.2, 128.1,
127:62, 127.58, 127:19, 127:17, 122.2, 112.1, 111.1, 109.7,
103.0, 71.1, 49.6. HRMS calcd for C27H21SO2Br
(MþH) ¼ 489.0524, found ¼ 489.0523. Anal. Calcd for
C27H21SO2Br: C, 66.3; H, 4.3. Found: C, 66.2; H, 4.2.
18. Product 4b, re-crystallized from EtOH. Mp 89–91 ꢁC; 1H
NMR (400 MHz, CDCl3): d 7.38–7.25 (m, 10H), 7.19–7.13
(m, 3H), 7.00–6.93 (m, 2H), 6.86 (d, J ¼ 8:7 Hz, 1H), 6.76 (d,
J ¼ 2:7 Hz, 1H), 6.61 (dd, J ¼ 8:7 Hz, 2.7, 1H), 4.97 (s, 2H),
3.69 (d, J ¼ 13:8 Hz, 1H), 3.57 (d, J ¼ 13:8 Hz, 1H). 13C
NMR (100 MHz, CDCl3): d 154.7, 149.7, 142.8, 137.3, 135.3,
130.9, 128.8, 128.2, 128.1, 127.9, 127.6, 127.5, 127.0, 125.7,
112.0, 111.0, 109.7, 103.5, 71.1, 49.9. HRMS calcd for
C27H22SO2 (MþH) ¼ 411.1419, found ¼ 411.1421. Anal.
Calcd for C27H22SO2: C, 78.9; H, 5.4. Found: C, 79.0; H, 5.4.
4. Liu, H. J.; Yip, J. Tetrahedron Lett. 1997, 38, 2253–2256.
5. Ishihara, K.; Hiraiwa, Y.; Yamamoto, H. Synlett 2000,
80–82.
6. Olah, G. A.; Narang, S. C.; Gupta, B. G. B.; Malhotra, R.
J. Org. Chem. 1979, 44, 1247–1251.
7. (a) Jung, M. E.; Lyster, M. A. J. Org. Chem. 1977, 42,
3761–3764; (b) Vickery, E. H.; Pahler, L. F.; Eisenbraun,
E. J. J. Org. Chem. 1979, 44, 4444–4446.
8. Brooks, P. R.; Wirtz, M. C.; Vetelino, M. G.; Rescek,
D. M.; Woodworth, G. F.; Morgan, B. P.; Coe, J. W.
J. Org. Chem. 1999, 64, 9719–9721.
9. (a) McOmie, J. F. W.; Watts, M. L.; West, D. E.
Tetrahedron 1968, 24, 2289–2292; (b) Niwa, H.; Hida,
T.; Yamada, K. Tetrahedron Lett. 1981, 22, 4239–4240; (c)
Greene, T. W.; Wuts, P. G. M. Protective Groups in
Organic Synthesis. 3rd ed.; John Wiley: New York, 1999;
pp 248–269.
10. Fuji, K.; Ichikawa, K.; Node, M.; Fujita, E. J. Org. Chem.
1979, 44, 1661–1664.
11. (a) Bhatt, M. V.; El-Morey, S. S. Synthesis 1982, 1048–
1050; (b) Bhatt, M. V.; Kulkarni, S. U. Synthesis 1983,
249–282.
12. Debenzylation of 1: To a solution of 1 (2.0 g, 3.2 mmol) in
toluene (20 mL) was added LiI (2.2 g, 16 mmol) followed
by acetonitrile (7 mL) and reaction mixture was kept under
N2. To the heterogeneous mixture was added SiCl4
(1.83 mL, 16 mmol) followed by BF3ÆAcOH (0.22 mL,
1.6 mmol). Reaction mixture was then aged at 70 ꢁC over
45 min. Progress of reaction was followed by HPLC.
When conversion was complete, the reaction mixture was
quenched with 20 mL of EtOH, excess solid NaHCO3 was
added, and aged over 20 min at rt. Mixture was filtered
(quant assay ¼ 98% yield), and concentrated. The resulting
oily material was treated with excess concd HCl and aged
at 60 ꢁC over 2 h. The resulting slurry was cooled to rt,
seeded, and aged at rt overnight. 1.44 g of 2 was collected
as an off white solid after filtration. 90% Isolated yield and
99A% by HPLC.
13. Purification: 1.5 g of crude oily product (entry 1) by silica
gel chromatography (CH2Cl2–MeOH ¼ 80:1) afforded
1.3 g of solid. Mp 45–47 ꢁC; 1H NMR (400 MHz, CDCl3):
d 7.22–7.10 (m, 2H), 6.92 (t, J ¼ 7:4 Hz, 1H), 6.75 (d,
J ¼ 7:9 Hz, 1H), 3.42 (t, J ¼ 7:8 Hz, 2H), 3.23 (t,