Development of a selective Friedel-Crafts alkylation surrogate: Safe operating conditions through mechanistic understanding

Article abstract of DOI:10.1021/op300257z

This article describes a selective one-pot, Friedel-Crafts acylation/ketone reduction protocol, effectively a surrogate for the Friedel-Crafts alkylation reaction with a primary alkyl halide. A potentially dangerous failure mode was identified, resulting in the uncontrolled evolution of hydrogen. A series of mechanistic experiments, including analysis by ²⁷Al NMR, was undertaken, and the reaction mechanism elucidated. Finally, the use of React IR to ensure real-time reaction safety was demonstrated.

Full text of DOI:10.1021/op300257z

Article
pubs.acs.org/OPRD
Development of a Selective Friedel−Crafts Alkylation Surrogate: Safe
Operating Conditions through Mechanistic Understanding
Robert N. Bream,* David G. Hulcoop, Steve J. Gooding, Simon A. Watson, and Charlotte Blore
GlaxoSmithKline Medicines Research Centre, Gunnels Wood Road, Stevenage, Hertfordshire SG1 2NY, United Kingdom
S
* Supporting Information
ABSTRACT: This article describes a selective one-pot, Friedel−Crafts acylation/ketone reduction protocol, effectively a
surrogate for the Friedel−Crafts alkylation reaction with a primary alkyl halide. A potentially dangerous failure mode was
identified, resulting in the uncontrolled evolution of hydrogen. A series of mechanistic experiments, including analysis by 27Al
NMR, was undertaken, and the reaction mechanism elucidated. Finally, the use of React IR to ensure real-time reaction safety
was demonstrated.
most direct route would be through a selective Friedel−Crafts
alkylation of phenyl ether 2, but the likelihood of multiple
alkylation products, intramolecular cyclization, and dimeriza-
tion led us to discount this approach. A Friedel−Crafts
acylation/ketone reduction protocol seemed an attractive
alternative (Scheme 1).
INTRODUCTION
■
The Friedel−Crafts alkylation reaction allows the functionaliza-
tion of aromatic systems with carbon electrophiles to form new
carbon−carbon bonds.1 The reaction has broad scope and has
been exploited in numerous industrial applications such as the
production of high octane gasoline,2 synthetic rubber,3 and bulk
chemicals such as ethylbenzene and styrene.4 Nonetheless,
there are significant synthetic limitations. The product is more
reactive than the starting material, and so polyalkylation is a
frequently encountered problem.1 Furthermore, a mixture of
ortho- and para-substituted products is observed with unbiased
substrates.1b Finally, the reaction proceeds via a cationic
intermediate which imposes restrictions on the electrophile:
primary cations rearrange rapidly by Wagner−Meerwein
hydride shift to afford more stable secondary cations, and
thus, it is not possible to introduce primary alkyl groups other
than benzyl, allyl, methyl, and ethyl.1 For example, attempted
alkylation of benzene with n-propyl bromide afforded
predominantly the isopropyl product.5
In contrast, Friedel−Crafts acylation is a highly selective
process favoring monosubstituted products with almost
exclusive para selectivity, controlled by the electronic bias of
the aromatic ring and steric bulk of the metal-coordinated
electrophile.6 Consequently, n-alkyl chains are generally
introduced to aromatic systems by a two-step process starting
with Friedel−Crafts acylation. The resultant ketone moiety can
be exhaustively reduced by a plethora of methods including
Clemmensen reduction,7 Wolff−Kishner reduction,8 hydro-
genation,9 or a combination of silane and either protic10 or
Lewis acid.11 At the inception of this project, a single report of a
one-pot Friedel−Crafts acylation/reduction protocol to gen-
erate an n-alkyl-substituted benzene existed,12 although a
second appeared during the course of our studies.13
The route started with a straightforward alkylation of
hexamethyleneimine with 3-phenoxypropyl bromide 3. The
product 2 is isolated in excellent yield as its HCl salt. The
reaction is much faster under phase transfer conditions than
using a heterogeneous base in organic solvent, and gives very
little quaternary ammonium salt byproduct. With this material
in hand, we set about examining a series of Lewis acids and
solvents for the Friedel−Crafts acylation, using both relevant
literature precedent14 and a PCA-guided approach.15 The
tertiary amine present in the substrate complicates matters by
coordinating an equivalent of Lewis acid, even when its HCl
salt is used. The use of catalytic methods is therefore precluded,
and at least two equivalents of Lewis acid are required. The
results of our preliminary screen are shown (Table 1).
By far the best results were achieved using AlCl3 in apolar
aprotic solvents, although reasonable conversion was observed
using iron and indium chlorides. DCM is the best solvent since
it fully solubilises the reaction mixture and we were unable to
detect the ortho isomer by HPLC or NMR. Keto-chloride 4 is
an oil and thus difficult to isolate and purify on large scale. We
circumvented this issue by formation and crystallization of the
hydrochloride salt (Scheme 2). However, both the salt and free
base were unstable to extended processing times during
aqueous workup, and on storage under inert conditions at 6
°C. The major degradants are keto-alcohol 5 and dihydrofuran
6, formed presumably via cyclic oxonium species 7. This also
accounts for the observation that keto-chloride 4 is a potent
alkylating agent and is genotoxic as indicated by positive AMES
test, whereas 1 is AMES negative.16
As part of our histamine antagonist programme we sought a
short, high-yielding manufacturing route to alkyl chloride 1.
Key attributes were scalability of processes, stability of
intermediates and products to extended processing times, and
purification and isolation by crystallization and filtration. As
such, intermediates and products were isolated as their
crystalline hydrochloride salts as the free bases are oils. The
Special Issue: Safety of Chemical Processes 12
Received: September 13, 2012
© XXXX American Chemical Society
A
dx.doi.org/10.1021/op300257z | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 1. Planned route to alkyl chloride 1
a
Table 1. Lewis acid screen for a Friedel−Crafts acylation reaction
b
solvent
Lewis acid
% conversion to product
solvent
Lewis acid
% conversion to product
DCM
AlCl3
AlCl3
AlCl3
AlCl3
AlCl3
AlCl3
AlCl3
AlCl3
AlCl3
AlCl3
AlCl3
AlCl3
AlCl3
AlCl3
>99
0
DCM
toluene
DCM
FeCl3
80
38
7
THF
FeCl3
toluene
acetone
83
0
ZnCl2
a
toluene
DCM
MeNO2
none
Zn
0
acetophenone
chloroform
p-xylene
0
Zn(OTf)2
Zn(OTf)2
Bis(OTf)2
Bis(OTf)2
Sc(OTf)2
BF3·OEt2
InCl3
13
0
70
76
0
7
TBME
DCM
DCM
DCM
DCM
DCM
0
heptane
71
0
11
0
EtOAc
trifluorotoluene
cyclopentyl methyl ether
chlorobenzene
MeNO2
10
0
67
c
BF3·OEt2
19
40
29
a
Conditions: 4-chlorobutyryl chloride (0.11 mL, 1.1 equiv, 0.94 mmol) was added to a mixture of Lewis acid (1.88 mmol, 2.2 equiv) and substrate
b
(0.20 g, 0.86 mmol) in solvent (2 mL) at 0 °C and allowed to warm slowly to ambient temperature. a) Reaction carried out at 110 °C. Conversion
measured by HPLC analysis of the reaction mixture. Using mixed anhydride rather than acid chloride.
c
Scheme 2. Stability of 4
Scheme 3. Reduction of 4
Nonetheless, we were able to isolate material of sufficient
purity to investigate the subsequent reduction. This process was
found to occur readily, either by hydrogenation over a
palladium catalyst or by triethylsilane reduction in neat
trifluoroacetic acid.9,10 Unfortunately, reduction using triethyl-
silane and stoichiometric TFA in organic solvents was not
successful. Interestingly, the intermediate alcohol is slow to
reduce under hydrogenation conditions, but not observed
B
dx.doi.org/10.1021/op300257z | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 4. Telescoped FC acylation/reduction
Figure 1. Heat flow, gas flow, and gas composition of telescoped reactions.
under the silane conditions. The hydrogenation reaction gave
the product in good yield and purity, but some material was lost
to hydrolysis and protodechlorination (Scheme 3). We were
unable to limit protodechlorination by using a range of
additives that have been reported to suppress Pd insertion into
the C−Cl bond.17 While this two-step protocol is suitable for
laboratory-scale synthesis, several factors led us to question its
suitability as a manufacturing process: (1) Any processing
delays during workup of the Friedel−Crafts acylation would
result in increased hydrolysis, lower yield, and a reduction in
purity as the keto-alcohol is difficult to purge by crystallisation.
(2) The rate of hydrolysis and dechlorination is increased in the
absence of hydrogen in the reduction. Again, processing delays
would lead to a reduction in yield and purity. (3) The use of
trifluoroacetic acid as a solvent on multikilogram scale is
precluded due to its toxicity, volatility, corrosivity, and difficulty
in handling subsequent waste streams. (4) Chloroketone 4,
unlike alkyl chloride 1, is genotoxic and unstable with respect to
hydrolysis, and thus difficulties would be encountered handling
and storing it on scale.
acylation reaction mixture quickly afforded the reduced product
1. We next began the process of optimizing individual reaction
parameters resulting in the following ideal conditions: 2 is
treated with 2.2 equiv of AlCl3 to ensure consistently complete
conversion, followed by 1.1 equiv of 4-chlorobutyryl chloride at
0 °C. Once reaction is complete, 4 equiv of triethylsilane is
added, and the reduction is complete in less than 1 h. When the
amount of acid chloride is reduced to 1 equiv, the reaction
follows a different course: tetrahydronaphthalene 8 is the major
product, formed by intramolecular Friedel−Crafts alkylation of
the now more active product (Scheme 4). The silane addition is
exothermic and accompanied by significant and rapid evolution
of gas, initially thought to be HCl by simple pH paper test.
Concerned by the scalability of this process, we carried out a
series of control experiments, varying individual parameters and
measured reaction calorimetry, gas flow, and off-gas composi-
tion.
To understand the cause of the cyclization event, we treated
chloride 1 with various acidic species in DCM at 0 °C. In the
presence of HCl, there is no reaction. In the presence of AlCl3,
1 cyclizes rapidly to afford tetrahydronaphthalene 8. Treatment
with a combination of HCl and AlCl3 (1:1 ratio) returns 1
uncyclized. Next, we treated keto-chloride 4 with the same
combinations of acids and added triethylsilane to each. In the
presence of HCl, no reaction occurs. In the presence of AlCl3,
slow reduction is observed, followed by cyclization to 8. When
We therefore sought alternatives to this procedure. We
speculated that judicious choice of Lewis acid would allow us to
carry out a Friedel−Crafts acylation followed by silane-
mediated ketone reduction in one pot. Indeed, precedent for
such a process exists.12,13 We were therefore delighted to find
that simply adding triethylsilane to the finished Friedel−Crafts
C
dx.doi.org/10.1021/op300257z | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 2. IR stretching frequencies for proposed reaction intermediates.
Figure 3. IR profiles of the two possible failure modes.
AlCl3 and HCl are combined, the reduction is rapid, but no
cyclization occurs. This suggests that HAlCl4 is the species
responsible for mediating the reduction, while free AlCl3 is
responsible for the subsequent cyclization. The role of excess
acid chloride is therefore to coordinate any free AlCl3. To
further test this hypothesis we added tetra-nbutylammonium
chloride to a Friedel−Crafts acylation carried out with 1.0 equiv
of acid chloride. Addition of triethylsilane then led to smooth
reduction of the ketone moiety without further cyclisation. We
next monitored the heat output, gas flow, and off-gas
composition of reactions carried out with an undercharge and
an excess of 4-chlorobutyryl chloride (Figure 1).
As can be seen from both heat plots, the addition of 2 to a
suspension of AlCl3 in DCM is exothermic. Subsequent
additions of 4-chlorobutyryl chloride and triethylsilane are
also exothermic, but heat output is addition rate-controlled and
so not considered an issue for scale-up. In the case where 4-
chlorobutyryl chloride is undercharged (Figure 1a), each
exotherm is accompanied by a gas evolution. Correlating this
with the off-gas composition (Figure 1b), it can be seen that
HCl is evolved on addition of acid chloride. However, when
triethylsilane is added, the evolved gas is hydrogen. This
evolution can be reproduced by simply combining triethylsilane
and AlCl3 in DCM. In fact, only sub-stoichiometric AlCl3 is
required to consume the silane.18 When an excess of acid
chloride is added (Figure 1c), HCl is still evolved. However,
now when triethylsilane is added, the gas flow stops, (if
anything, gas seems to be absorbed by the reaction), and no
hydrogen is detected in the off-gases (Figure 1d).
To safely run this process on multikilogram scale, we need to
be able to control both the reaction outcome and any evolution
of gas. The potential generation of a mixture of HCl and H2
gases with entrained DCM vapour presents a significant
engineering challenge from an environmental emissions
perspective as the individual components must be treated
very differently: HCl is trapped using a caustic scrubber, DCM
D
dx.doi.org/10.1021/op300257z | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 4. IR profile of successful reaction.
Scheme 5. Mechanistic proposal
is collected using a low-temperature secondary condenser and
collection vessel, whereas H2 must be vented to the atmosphere
as quickly as possible to prevent the build-up of an explosive
atmosphere. Rather than attempt this complex and potentially
hazardous task, we felt the best solution was to ensure the
correct ratio of AlCl3 to 4-chlorobutyryl chloride, thus ensuring
that the reaction was not going to produce hydrogen before
addition of the silane by a noninvasive real-time analytical
technique. The IR stretching frequencies of different carbonyl
compounds are diagnostic of structure.19 We therefore
investigated ReactIR as an analytical tool to probe various
reaction components. We began our studies by determining
diagnostic IR stretching frequencies for each of our presumed
reaction components. We were pleased to find distinct and
separable carbonyl signals for both free and aluminium-bound
acid chloride and chloroketone species (Figure 2).20
Coordination to AlCl3 lowers the stretching frequency of the
carbonyls in both cases by around 150 cm−1. Monitoring the
changes in these signals over time allowed us to follow the
reaction quite clearly and potentially adjust the charge of acid
chloride to ensure an excess is present before silane is added.
Experiments were carried out by adding 4-chlorobutyryl
chloride, AlCl3, and ether 2 successively to DCM so that a
background signal could be measured and the signals for bound
and unbound acid chloride detected. This change in addition
order does not affect the outcome of the reaction. We next
monitored the reaction carried out in two possible failure
modes: with an undercharge of acid chloride (1.0 equiv, Figure
3A) and with an overcharge of AlCl3 (3.2 equiv, Figure 3B). On
addition of acid chloride, the peak at 1800 cm−1 is observed,
but disappears on addition of AlCl3 to give the signal for
aluminium-bound acid chloride at 1653 cm−1. When ether 2 is
added, signals for acid chloride disappear, being replaced by the
aluminium-bound chloroketone signal at 1530 cm−1, thus
indicating a rapid acylation reaction. In neither case is unbound
acid chloride detected. Both reactions would fail on addition of
silane; hydrogen would be evolved rapidly, and the product
would cyclize to afford tetrahydronaphthalene 8.
We next demonstrated that reaction B could be recovered by
charging more acid chloride. A further 0.4 equiv was added,
resulting in reappearance of the signal for aluminium-bound
acid chloride (1653 cm−1). On addition of a further 0.8 equiv,
the signal for free acid chloride (1800 cm−1) reappears.
Addition of triethylsilane then gave clean reduction of all the
carbonyl groups to afford 1 with no H2 evolution and no
destructive cyclisation.
Finally, we monitored a reaction carried out under optimized
processing conditions using an excess of 4-chlorobutyryl
chloride. Gratifyingly, the signal for free acid chloride (1800
cm−1) is observed prior to silane addition which then gives
clean conversion to product 1 (Figure 4).
We propose the following mechanism for the reaction: the
first equivalent of aluminium is sequestered by the protonated
nitrogen as an [AlCl4]− counterion. The second equivalent then
mediates the Friedel−Crafts acylation generating a further
equivalent of HCl, forming additional [AlCl4]−. The excess of
AlCl3 (0.2 equiv) can then cause cyclisation and hydrogen
evolution unless coordinated with excess acid chloride (Scheme
5).
To test this hypothesis, we applied 27Al NMR to investigate
the nature of the aluminium species present at the failure
modes before and after acylation.21 We first acquired reference
spectra of AlCl3 alone and with addition of tetrabutylammo-
nium chloride (TBAC) in DCM-d2. All species present in both
spectra fall in the expected chemical shift region for tetrahedral
halide complexes.22 A single broad resonance at 92.9 ppm (line
E
dx.doi.org/10.1021/op300257z | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 5. 27Al NMR of (a) [AlCl4]− complex and reaction mixtures (b) prior to acid chloride addition, (c) with undercharged acid chloride, (d) with
excess acid chloride.
width 183 Hz) is observed with AlCl3 which may correspond to
the Al2Cl6 dimer, and with the addition of TBAC a single
narrow resonance at 103.5 ppm (line width 3 Hz) is observed
which corresponds to the [AlCl4]− complex (see Figure 5a). A
solution to represent the reaction prior to acylation was next
prepared with AlCl3 and ether 2 in a 1:2.2 molar ratio in DCM-
d2 and a spectrum acquired (see Figure 5b).
support our mechanistic interpretation that an excess of acid
chloride is required to completely sequester AlCl3 and prevent
the exothermic cyclisation and hydrogen evolution events.
In summary, we have developed a Friedel−Crafts acylation/
reduction protocol, effectively a Friedel−Crafts alkylation
reaction, to make a key intermediate in the synthesis of a
potential histamine antagonist drug. Concerns over the
robustness of the reaction led us to a series of experiments to
determine the reaction mechanism. Furthermore, we have
demonstrated ReactIR as a real-time, nonintrusive, analytical
technique to provide a basis of safety for large-scale reactions.
The resulting data show a main resonance at 97.8 ppm (line
width 60 Hz) and a series of lesser, broad resonances at 103.1,
98.5, and 90.1 ppm. It is most likely that the resonance at 103.1
ppm is from the [AlCl4]− complex with a broader line width of
127 Hz as a consequence of being in equilibrium with other
aluminium species in solution. The individual identities of these
other species are not known, but they also lie within the
expected region for tetrahedral halide complexes and may
represent different arrangements of AlCl3 coordinated to ether
2. Finally we acquired spectra of reactions performed with an
undercharge (0.7 eq.) and an excess (1.5 eq.) of acid chloride,
with samples being taken following the acid chloride addition
but before silane addition (Figures 5c and d, respectively). In
both spectra the predominant resonance is at 103.1 ppm,
corresponding to the [AlCl4]− complex, and they also share a
broad resonance at 92.5 ppm which, from the IR data, may
possibly represent AlCl3 coordinated to the carbonyl produced
by the acylation. There is an additional resonance observed in
the undercharged reaction (Figure 5c) at 98.0 ppm which was
also present in the solution prior to acylation and thus is
potentially from a coordination of AlCl3 with part of the
product module originating from ether 2.
EXPERIMENTAL SECTION
■
General Methods. All reactions were carried out in water-,
methanol-, and acetone-washed glassware under a nitrogen
atmosphere. Solvents and reagents were used without any
1
purification or drying. H and 13C NMR spectra were acquired
on a Bruker spectrometer at frequencies of 400 and 100 MHz,
respectively. High-resolution mass spectra were recorded on a
linear ion trap combined with a Fourier transform ion cyclotron
resonance mass spectrometer using an electrospray ionisation
source operated in positive ion mode. IR spectra were recorded
as solids. HPLC chromatograms were recorded on a C18(2)
column at 40 °C; eluent gradient: 100% 0.05% v/v TFA in
water to 95% 0.05% v/v TFA in MeCN over 8 min. Heat
output data was measured in a Mettler Toledo MidTemp RC1e
vessel. Gas flow and composition was recorded using a Hiden
Analytical HPR20 gas analyzer. Melting points were recorded
using an automated melting point apparatus.
These spectra confirm the presence of [AlCl4]− in the
reaction mixture along with other aluminium species and
1-(3-Phenoxypropyl)azepane Hydrochloride 2. Method A.
3-Phenoxypropyl bromide (752 g, 3.5 mol) is added to stirred
F
dx.doi.org/10.1021/op300257z | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
suspension of potassium carbonate (1.06 kg, 7.7 mol) in TBME
(3.75 L), followed by hexamethylene imine (692 g. 7.0 mol)
and then the mixture heated to reflux for 18 h. Once complete
the reaction mixture is cooled to 40 °C, diluted with TBME
(3.75 L) before succinic anhydride (384 g, 3.25 mol) is added
portionwise and stirred for 30 min. The resulting mixture is
washed with water (2 × 7.5 L) and the organic layer dried by
azeotropic distillation to 4.5 L total volume under atmospheric
pressure. The resulting solution is cooled to 40 °C and diluted
with TBME (7.5 L), and the HCl salt of the product is
precipitated by addition of HCl in 2-propanol (772 mL of 5−6
M solution) over 40 min. The resulting slurry is cooled to 20
°C over 1 h, aged for 2 h, and filtered. The wet cake is washed
with TBME (2 × 1.5 L) and dried in vacuo to give 1-(3-
phenoxypropyl)azepane hydrochloride as a white solid (848 g,
3.14 mol, 90%); δH (400 MHz, CDCl3) 12.40−12.15 (1H, brs),
7.32−7.28 (2H, m), 6.98 (1H, t, J = 7.4 Hz), 6.87 (2H, d, J =
8.6 Hz), 4.08 (2H, t, J = 5.5 Hz), 3.63−3.55 (2H, m), 3.24 (2H,
brt, J = 7.5 Hz), 3.03−2.94 (2H, m), 2.50−2..32 (2H, m),
2.26−2.15 (2H, m), 1.93−1.83 (4H, m), 1.75−1.65 (3H, m);
δC (100 MHz, CDCl3) 158.2, 129.6, 121.3, 114.4, 64.8, 55.3,
54.6, 27.0, 24.5, 23.4; νmax cm−1 (solid) 1472, 1493, 1587, 1600,
2512, 2935; HRMS: m/z calcd for C15H24NO 234.1852: found
234.1853; mp 138−140 °C.
Method B. 3-Phenoxypropyl bromide (100 g, 0.47 mol) is
added to a stirred mixture of hexamethylene imine (61.7 mL,
0.54 mol) and tetrabutylammonium bromide (15.0 g, 46.5
mmol) in MiBK (300 mL) and sodium hydroxide solution
(74.4 mL of 10 M solution made up with water to 300 mL total
volume). The reaction mixture is then stirred at reflux for 45
min. Once complete, the reaction is cooled to 40 °C before
acetyl chloride (8.26 mL, 0.12 mol) is added and the reaction
stirred for 1 h. The layers are separated, and the aqueous layer
is washed with MiBK (400 mL); the combined organic layers
are washed with water (600 mL) and then dried by azeotropic
distillation to 400 mL total volume at atmospheric pressure.
TBME (1.5 L) is added followed by HCl in 2-propanol (102
mL of 5−6 M solution), the resulting slurry is stirred for 30
min, then cooled to room temperature, and then aged for 1 h.
The slurry is filtered, washed with TBME (2 × 200 mL), and
dried in vacuo to give 1-(3-phenoxypropyl)azepane hydro-
chloride as a white solid (118.5 g, 0.44 mol, 94% theory);
analytical data as above.
1-(3-(4-(4-Chlorobutyl)phenoxy)propyl)azepane Hydro-
chloride 1. 1-(3-Phenoxypropyl)azepane hydrochloride 2
(750 g, 2.78 mol) is added portionwise to a stirred suspension
of aluminium chloride (818 g, 6.12 mol) in dichloromethane
(7.5 L) at 0 °C. 4-Chlorobutyryl chloride (374 mL, 3.34 mol) is
then added over 20 min, maintaining the internal temperature
below 5 °C. The reaction mixture is then stirred for at least 5
min until complete conversion as judged by HPLC analysis.
Triethylsilane (1.78 L, 11.1 mol) is added over 1 h, maintaining
the internal temperature below 5 °C; then the reaction mixture
is stirred for 30 min. Once reaction is complete, the reaction
mixture is added portionwise to an aqueous HCl solution (7.5
L of 2 M solution), maintaining the internal temperature below
10 °C; the biphasic mixture is then stirred for 10 min. Once
settled the lower organic layer is separated, and the aqueous
layer is washed with DCM (2.5 L); the combined organic
phases are dried by azeotropic distillation at ambient pressure
to 4.5 L total volume. TBME (13.5 L) is added over 45 min at
40 °C and the resulting slurry stirred for at least 1 h. The slurry
is then cooled to 20 °C over 1 h, aged for a further 1 h, filtered,
washed with TBME (2 × 2.3 L) and dried in vacuo to afford 1-
(3-(4-(4-chlorobutyl)phenoxy)propyl)azepane hydrochloride
as a white solid (962 g, 2.70 mol, 96% theory); δH (400
MHz, CDCl3) 12.30−12.10 (1H, brs), 7.08 (2H, d, J = 8.7 Hz),
6.78 (2H, d, J = 8.7 Hz), 4.04 (2H, t, J = 5.6 Hz), 3.61−3.51
(4H, m), 3.25−3.20 (2H, m), 3.02−2.94 (2H, m), 2.58 (2H, t, J
= 7.1 Hz), 2.47−2.39 (2H, m), 2.25−2.15 (2H, m), 1.92−1.60
(10H, m); δC (100 MHz, CDCl3) 156.5, 134.7, 129.4, 114.3,
64.9, 55.3, 54.6, 44.9, 34.1, 32.0, 28.7, 26.9, 24.4, 23.4; νmax
cm−1 (solid) 1509, 1584, 1610, 2514, 2594, 2864, 2602, 2940;
HMRS m/z calcd for C19H31ClNO 324.2094: found 324.2082;
mp 120−122 °C.
1-(4-(3-(Azepan-1-yl)propoxy)phenyl)-4-chlorobutan-1-
one Hydrochloride 4. 1-(3-Phenoxypropyl)azepane hydro-
chloride 2 (300 g, 1.11 mol) is added portionwise to a stirred
suspension of aluminium(III) chloride (327 g, 2.45 mol) in
DCM (3.4 L) at 0 °C. Chlorobutyryl chloride (170 mL, 1.33
mol) is then added over 20 min, maintaining the internal
temperature below 5 °C. The reaction mixture is then stirred
for 5 min. Once complete, ethanol (600 mL) is added over 10
min, maintaining the internal temperature below 10 °C.
Aqueous HCl solution (3 L of a 5 M solution) is then added
over 30 min, maintaining the internal temperature below 20 °C
and the mixture stirred for 10 min. Once settled the lower
organic layer is separated and the aqueous layer is extracted
with DCM (1.8 L) and the combined organic phases dried by
azeotropic distillation at ambient pressure to 1.2 L total volume.
IPA (1.2 L) is added and the reaction mixture is again distilled
under atmospheric conditions to 1.2 L total volume. TBME
(4.8 L) is added over 45 min at 40 °C and the resulting slurry
stirred for at least 1 h. The slurry is then cooled to 10 °C over 1
h, aged for a further hour, filtered, washed with TBME (2 × 1
L) and dried in vacuo to afford 1-(4-(3-(azepan-1-yl)propoxy)-
phenyl)-4-chlorobutan-1-one hydrochloride 4 as a white solid
(369 g, 0.99 mol, 94% theory); δH (400 MHz, CDCl3) 12.25−
12.10 (1H, brs), 7.95 (2H, d, J = 8.6 Hz), 6.92 (2H, d, J = 8.6
Hz), 4.7 (2H, t, J = 5.4 Hz), 3.67 (2H, t, J = 6.4 Hz), 3.63−3.55
(2H, m), 3.29−3.23 (2H, m), 3.12 (2H, t, J = 7.1 Hz), 3.04−
2.97 (2H, m), 2.54−2.44 (2H, m), 2.26−2.14 (4H, m), 1.93−
1.81 (4H, m), 1.72−1.81 (2H, m); δC (100 MHz, CDCl3)
197.4, 162.0, 130.3, 130.2, 114.1, 65.1, 55.0, 54.6, 44.6, 34.9,
26.8, 26.8, 24.2, 23.3; νmax cm−1 (solid) 1509, 1596, 1673s,
2499, 2592, 2935; HMRS m/z calcd for C19H29ClNO2
338.1881: found 338.1878; mp 110−114 °C.
1-(3-((5,6,7,8-Tetrahydronaphthalen-2-yl)oxy)propyl)-
azepane Hydrochloride 8. 1-(3-(4-(4-Chlorobutyl)phenoxy)-
propyl)azepane hydrochloride 1 (1.43 g, 3.97 mmol) was taken
up in DCM (15 mL) under N2. AlCl3 (1.52 g, 11.4 mmol) was
added in three portions over 2 h and the resultant solution aged
at ambient temperature for 18 h, at which point all the input
material was consumed as indicated by HPLC. The mixture was
cooled to 0 °C and quenched by slowly adding 2 M HCl (30
mL) dropwise to afford a cloudy emulsion. The mixture was
poured into a separating funnel and IMS (5 mL) and DCM (15
mL) were added resulting in a clear organic phase. The organic
fraction was decanted and the aqueous re-extracted with DCM
(20 mL). The combined organic fractions were dried over
Na2SO4, filtered, and evaporated down to 15 mL. TBME (50
mL) was added and the mixture evaporated down to 20 mL,
resulting in precipitation of an off-white solid. The slurry was
filtered and washed with TBME, then dried over a weekend at
30 °C under vacuum to afford 1-(3-((5,6,7,8-tetrahydronaph-
thalen-2-yl)oxy)propyl)azepane hydrochloride 8 (960 mg, 2.96
G
dx.doi.org/10.1021/op300257z | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
(10) (a) West, C. T.; Donnelly, S. J.; Kooistra, D. A.; Doyle, M. P. J.
Org. Chem. 1973, 38, 2675−2681. (b) Olah, G. A.; Arvanaghi, M.;
Ohannesian, L. Synthesis 1986, 770−772.
mmol, 75%) as a white solid; δH (400 MHz, CDCl3) 12.20−
12.06 (1H, brs), 6.96 (1H, d, J = 8.2 Hz), 6.61 (1H, d, J = 8.2
Hz), 6.5 (1H, s), 4.02 (2H, t, J = 5.6 z), 3.60−3.53 (2H, m),
3.25−3.19 (2H, m), 3.01−2.94 (2H, m), 2.75−2.67 (4H, m),
2.45−2.38 (2H, m), 2.22−2.14 (2H, m), 1.94−1.82 (4H, m),
1.80−1.72 (4H, m), 1.70−1.60 (2H, m); δC (100 MHz,
CDCl3) 156.0, 138.4, 130.0, 130.0, 114.3, 112.1, 64.8, 55.3,
54.5, 29.6, 28.5, 26.9, 24.5, 23.3, 23.3, 23.0; νmax cm−1 (solid)
1500, 1583, 1605, 2510, 2591, 2927; HMRS m/z calcd for
C19H30NO 288.2322: found 288.2320; mp 152−156 °C.
Chlorotriethylsilane. δH (400 MHz, CD2Cl2) 1.02 (9H, t, J
= 7.8 Hz), 0.82 (6H, q, J = 7.8 Hz).
(11) Fry, J. L.; Orfanopoulos, M.; Adlington, M. G.; Dittman, W. R.,
Jr.; Silverman, S. B. J. Org. Chem. 1978, 43, 374−375.
(12) Jaxa-Chamiec, A.; Shah, V. P.; Kruse, L. I. J. Chem. Soc., Perkin
Trans. 1 1989, 9, 1705−1706.
(13) During the course of this work, a similar strategy was reported
using tetramethyldisiloxane as reductant: Nadkarni, D. V.; Hallissey, J.
F. Org. Process Res. Dev. 2008, 12, 1142−1145. Tetramethyldisiloxane
also works in our process.
(14) (a) Meshram, H. M.; Reddy, G. S.; Reddy, M. M.; Yadav, J. S.
Synth. Commun. 1998, 28, 2203−2206. (b) Litt, M. H.; Summers, J.
W.; Shimko, T. M. J. Org. Chem. 1972, 37, 1045−1047. (c) Miles, W.
H.; Nutaitis, C. F.; Anderton, C. A. J. Chem. Educ. 1996, 73, 272.
(d) He, F.; Wu, H.; Chen, J.; Su, W. Synth. Commun. 2008, 38, 255−
1,1,1,2,2,2-Hexaethyldisiloxane. δH (400 MHz, CD2Cl2)
0.94 (18H, t, J = 7.9 Hz), 0.54 (9H, q, J = 7.9 Hz).
́
264. (e) Repichet, S.; Le Roux, C.; Dubac, J.; Desmurs, J.-R. Eur. J.
ASSOCIATED CONTENT
* Supporting Information
NMR spectra for key compounds. This material is available free
of charge via the Internet at http://pubs.acs.org.
■
Org. Chem. 1998, 12, 2743−2746. (f) Fillion, E.; Dumas, A. M. J. Org.
Chem. 2008, 73, 2920−2923. (g) Komoto, I.; Matsuo, J.-I.; Kobayashi,
S. Top. Catal. 2002, 19, 43−47.
S
(15) (a) Pearson, K. Philos. Mag. 1901, 2, 559−572. (b) Jolliffe, I. T.
Principal Component Analysis; Springer Series in Statistics; Springer:
New York, 2002.
AUTHOR INFORMATION
Corresponding Author
*Robert.N.Bream@GSK.com
■
(16) Mortelmans, K.; Zeiger, E. Mutat. Res. 2000, 455, 29−60.
(17) (a) Wu, G.; Huang, M.; Richards, M.; Poirier, M.; Wen, X.;
Draper, R. W. Synthesis 2003, 1657−1660. (b) Li, J.; Wang, S.;
Crispino, G. A.; Tenhuisen, K.; Singh, A.; Grosso, J. A. Tetrahedron
Lett. 2003, 44, 4041−4043.
(18) One equivalent of AlCl3 consumes 3 equiv of Et3SiH within 1 h
and 8 equiv within 4.5 h to afford a mixture of chlorotriethylsilane and
1,1,1,2,2,2-hexaethyldisiloxane, by comparison of NMR spectra with
authentic commercial samples.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We acknowledge Jan Fiedler and Andy Payne for help in setting
up and interpreting gas measurement experiments, and Mark
Scott for practical assistance with ReactIR experiments.
(19) Williams, D. H.; Fleming, I. Infrared spectra. In Spectroscopic
Methods in Organic Chemistry; McGraw-Hill: London, 2007.
(20) 4-Chlorobutyryl chloride is depicted as being coordinated to
AlCl3 through the oxygen for simplicity of illustration, although
Friedel−Crafts acylation is known to proceed through an acylium
intermediate.
REFERENCES
■
(1) (a) Olah, G. A.; Krishnamurti, R.; Surya Prakash, G. K. Friedel-
Crafts Alkylations. In Comprehensive Organic Synthesis; Trost, B. M.,
Fleming, I., Eds.; Pergamon Press: Oxford, 1991; Vol. 3, pp 293−339.
(b) Smith, M. B.; March, J. Advanced Organic Chemistry: Reactions,
Mechanism and Structure; John Wiley & Sons Inc.: Hoboken, 2007; pp
705−712. (c) Rueping, M; Nachtsheim, B. J. Beilstein J. Org. Chem.
2012, DOI: 10.3762/bjoc.6.6.
(21) The isotope Al27 is 100% naturally abundant, has a nuclear spin
of 5/2 and a relative sensitivity of approximately 0.2 compared to 1H as
1
unity. Nuclei with spin > /2 and are known as quadrupolar. Such
nuclei have nonspherical charge distribution which couples with the
electric field gradient at the nucleus to give a resultant quadrupolar
coupling. The magnitude of the electric field gradient represents the
symmetry of the electron distribution around the nucleus; hence, in
the example discussed in this article, aluminium complexes with less
symmetry experience a greater quadrupolar coupling effect than those
with greater symmetry. The quadrupolar effect provides an efficient
pathway for spin−lattice relaxation, leading to potentially much
shorter T1 values and so greater line widths. These can be many
hundreds of Hz in the case of highly asymmetric species. The spectra
shown here were acquired on a Bruker AV-II 500 MHz spectrometer
at a frequency of 130 MHz using a single-pulse experiment and 64
transients acquired with 16k data points each and a recycle delay of
between 1 and 10 s as appropriate for the T1 of the species in each
sample. All spectra were referenced externally to [Al(H2O)6]3+ at 0.00
ppm, a solution of which was formed by dissolving AlCl3 in water.
(22) (a) Haraguchi, H.; Fujiwara, S. J. Magn. Reson. 1969, 73, 3467−
3473. (b) Akitt, J. W. Prog. Nucl. Mag. Reson. Spectrosc. 1989, 21, 1−
149.
(2) (a) Meisinger, E. E. Process for the alkylation of saturated
hydrocarbons. US Patent 3,113,988, December 10, 1963. (b) Pollitzer,
E. L. Production of high octane unleaded motor fuel. US Patent
3,787,313, January 22, 1974.
(3) (a) (Esso Research and Engineering Company). Polymerisation
process and catalyst thereof. GB Patent Specification 801409,
September 10, 1958. (b) Nishihara, E.; Hiraga, K. Bridge-bond
formation between molecular chains of rubber. Japan Patent
40000537, January 13 1965. (c) Esso Research and Engineering
Company. Verfahren zur Misch polymerisation einer grosseren Menge
eines Isoolefins mit einer leineren Menge eines Diolefins. German
patent 1030029, 14 May 1958.
(4) (a) Welch, V. A., Fallon, K. J. Gelbke, H.-P. Ethylbenzene. In
Ullmann’s Encyclopedia of Industrial Chemistry; Wiley-VCH: Weinheim,
Germany, published online 15 October 2005. (b) James, D. H. Castor,
W. M. Styrene. In Ullmann’s Encyclopedia of Industrial Chemistry;
Wiley-VCH: Weinheim, Germany, published online 15 October 2011.
(5) Olah, G. A; Meidar, D. Nouv. J. Chim. 1979, 3, 269−273.
(6) (a) Heaney, H. The bimetallic aromatic Friedel-Crafts acylation.
In Comprehensive Organic Synthesis; Trost, B. M.; Fleming, I., Eds.;
Pergamon Press: Oxford, 1991; Vol. 2, pp 733−752. (b) Smith, M. B.;
March, J. Advanced Organic Chemistry: Reactions, Mechanism and
Structure; John Wiley & Sons Inc.: Hoboken, NJ, 2007; pp 719−724.
(7) Vedejs, E. Org. React. 1975, 22, 401−422.
(8) Todd, D. Org. React. 1948, 4, 378−422.
(9) Ram, S.; Spicer, L. D. Tetrahedron Lett. 1988, 29, 3741−3744.
H
dx.doi.org/10.1021/op300257z | Org. Process Res. Dev. XXXX, XXX, XXX−XXX