Safety assessment for the scale-up of an oxime reduction with melted sodium in standard pilot-plant equipment

Article abstract of DOI:10.1021/op300101d

A pilot-plant process is described for the reduction of 2-allyl cyclohexanone oxime with melted sodium in xylenes, toluene, and 4-methyl-2-pentanol. The trans/cis ratio was 3-4:1. Safety data are presented from a range of thermokinetic experiments (heat flow calorimetry, differential scanning calorimetry, and accelerating rate calorimetry). The process has been designed and developed to enable an expedient and safe scale-up in a standard enameled 100-L steel reactor and has been reproduced six times on 209 mol scale (each 4.8 kg sodium). Crystallization of the product 2-allyl cyclohexylamine with oxalic acid from the reaction mixture in tert-butyl methyl ether successfully avoided the yield losses associated with the isolation of the volatile free 2-allyl cyclohexylamine.

Full text of DOI:10.1021/op300101d

Article
pubs.acs.org/OPRD
Safety Assessment for the Scale-up of an Oxime Reduction with
Melted Sodium in Standard Pilot-Plant Equipment
Roland A. Breitenmoser,*,† Thomas Fink,† and Stefan Abele†,‡
†CARBOGEN AMCIS AG, Hauptstrasse 171, CH-4416 Bubendorf, Switzerland
S
* Supporting Information
ABSTRACT: A pilot-plant process is described for the reduction of 2-allyl cyclohexanone oxime with melted sodium in xylenes,
toluene, and 4-methyl-2-pentanol. The trans/cis ratio was 3−4:1. Safety data are presented from a range of thermokinetic
experiments (heat flow calorimetry, differential scanning calorimetry, and accelerating rate calorimetry). The process has been
designed and developed to enable an expedient and safe scale-up in a standard enameled 100-L steel reactor and has been
reproduced six times on 209 mol scale (each 4.8 kg sodium). Crystallization of the product 2-allyl cyclohexylamine with oxalic
acid from the reaction mixture in tert-butyl methyl ether successfully avoided the yield losses associated with the isolation of the
volatile free 2-allyl cyclohexylamine.
For the oxime reduction discussed in this paper, the initial
procedure as given in Scheme 1 used 13.5 equiv sodium that
was added to a refluxing solution of the oxime 2 in 64 equiv
ethanol. As also indicated above, a number of critical issues
were eminent: (i) solid sodium addition to refluxing ethanol,
(ii) large excess of sodium and ethanol and the asso-
ciated large amounts of hydrogen gas released, (iii) control of
diastereoselectivity (cis/trans), and (iv) volatile and unstable
2-allyl cyclohexyl amine 3. We envisioned a procedure focusing
on the following goals. First, sodium should be charged into the
empty vessel (or to a nonreactive solvent such as xylenes), and
the excess should be minimized. Second, the three substeps, i.e.
generation of granulated or melted sodium, the reduction of the
oxime, and the quench of excess sodium, should be fully
controlled by dosage. Both the generation of heat and hydrogen
were to be controlled by dosage.
INTRODUCTION
■
trans-2-Allyl cyclohexylamine, trans-3, was an intermediate for
the synthesis of an active pharmaceutical ingredient (API) at
Pfizer (Scheme 1). Multikilogram quantities were required. The
initial synthesis relying on Medicinal Chemistry protocols
started with the oxime formation with 2-allyl cyclohexanone 1.
Reduction of the oxime 2 was performed by treatment with
sodium in ethanol, affording 2-allyl cyclohexyl amine 3 that was
isolated by stripping to dryness prior further derivatization.1a
The use of elemental sodium on pilot-plant scale mandates
stringent safety control. Herein, the design and safe scale-up of
a sodium reduction of the 2-allyl cyclohexanone oxime 2 on
multikilogram scale is described.
We first tried to circumvent the use of elemental sodium for
the reduction of this α-substituted cyclohexanone oxime. Accord-
ing to literature−and underlined by experimental data gathered at
Pfizer−generation of equatorial (trans) alcohols from hindered
cyclohexanones often required the use of sodium2−4 or other alkali
metals,5,6 while metal hydrides7,8 or hydrogenation9 gave mainly
the axial (cis) product. Hutchins et al. showed10 that the intro-
duction of a phosphinyl imine can greatly improve the stereo-
selectivity in favor of the trans product, using tri-sec-butylborohy-
dride. More recently, DeNinno et al.1b used lithium in anhydrous
ammonia at −78 °C to afford the amine 3 with a trans/cis ratio
of 10:1 in 67% yield. In addition, SiGNa Chemistry, Inc.
disclosed the use of encapsulated sodium/potassium alloys in
silica gel (Na−SG); these are free-flowing solids without the
dangers associated with the handling of alkali metals.11 To cope
with stringent timelines for the delivery of trans-3, considering
the lack of efficient alternatives, we aimed at studying the
sodium reduction in more detail to render it safe for the
intended pilot-plant scale.
RESULTS AND DISCUSSION
■
Initial Model Runs with Cyclohexanone Oxime.
Scheme 2 outlines a simplified reaction mechanism of the
sodium-mediated oxime reduction.13 Potential radical inter-
mediates are not shown. Mechanistically, a minimum of 4 equiv
of sodium and 3 equiv of protons delivered from an alcohol are
necessary for complete conversion.
We wanted first to optimize the reproducible production of
a sodium sand (also named melted or granulated sodium,
molecular sodium, or powdered sodium) in xylenes13,14 and to
find a safe way for the reduction of the oxime and the alco-
holysis of the sodium. Dosage of solid sodium into a reacting
alcohol atmosphere was excluded due to equipment limita-
tions. We envisioned mass transfer issues associated with the
stirring of the granulated or melted sodium, such as sedi-
mentation, stirring, and gas evolution. To study both the
Banziger et al. described a Birch/Bouveault Blanc reduction
̈
using kg amounts of sodium. Precut pieces (2 cm cubes) of
sodium (5.8 kg) were added to a solution of the substrate in
n-butanol over 7 h.12 As a measure to avoid the open exposure
to a hydrogen emitting flammable fluid, the addition took place
through a nitrogen-inertized double air-lock system.
Special Issue: Safety of Chemical Processes 12
Received: April 19, 2012
Published: June 19, 2012
© 2012 American Chemical Society
2008
dx.doi.org/10.1021/op300101d | Org. Process Res. Dev. 2012, 16, 2008−2014

Organic Process Research & Development
Article
Scheme 1. Protocol as used by Medicinal Chemistry
upon restart of stirring. Therefore, all further experiments were
carried out above 100 °C (sodium mp 98 °C) and at
approximately 250−500 rpm, avoiding such problems by
handling only liquid sodium. No stirring was applied during
the melting of sodium. A related behavior was reported for
sodium in n-butanol: at temperatures below 100 °C, sodium
butoxide formation outweighed the desired reduction.12
Scheme 2. Simplified mechanism of oxime reduction by
sodium in alcohols13
For familiarization with oxime reduction, we used cyclo-
hexanone oxime (4) as model substrate (Table 2). Entry 1 is
based on our envisioned ideal conditions, i.e. melting of sodium
in inert xylenes and dosage of the oxime 4 in equimolar 4M2P.
However, to our surprise, this did not look promising: 0%
conversion was observed which was likely due to poor solubility
in the solvent mixture. Therefore, the following two experi-
ments used protocols more closely related to the original
procedure, i.e. addition of solid sodium to the reaction mixture.
Except for entry 2 all runs were not worked up after cooling to
room temperature as only the conversion was studied. Entry 2
basically reproduced the original procedure using the model
substance cyclohexanone oxime 4. The low yield of 35% was
probably caused by the high volatility of cyclohexyl amine 5 (bp
134 °C). In entry 3, ethanol was replaced by the higher boiling
4M2P to check its performance giving a conversion >90%. With
entry 4 we simulated a safe charge of sodium to the empty
reactor by using a solution of sodium in 4M2P prior to the
inverse addition of a solution of the oxime 4 in 4M2P: 91%
conversion was obtained. The main drawback was the
dissolution of sodium in a large excess of 4M2P (64 equiv, as
in the original procedure) which would likely lead to almost
quantitative formation of the corresponding alkoxide on larger
scale due to the associated longer stirring times. Hence, in entry
5, sodium was melted in xylenes without stirring followed by
the addition (with stirring) of the oxime 4 in 64 equiv of 4M2P
(main difference as compared to entry 1). This led to 98%
conversion. Additional ethanol was added to ensure that excess
stability of the thus formed melted sodium and the hydrogen
evolution, various alcohols have been added at reflux (Table 1).
Table 1. Alcohols as additives for the reduction with melted
Na
alcohol
bp (°C)
ethanol
78
82
isopropanol
cyclohexanol
160−161
163−167
132
2-methylcyclohexanol cis/trans mixture
4-methyl-2-pentanol
Solvents with a higher boiling point than that of ethanol were
desirable in order to be above the melting point of sodium.
Boiling point and cost triggered the choice of 4-methyl-2-
pentanol (4M2P) from the selection of potential alcohols. The
stability of sodium in various alcohols was also briefly discussed
by Sugden and Patel,15 and further details on the stability in
n-butanol were given by Banziger et al.12
̈
A first familiarization experiment to generate granulated
sodium by cooling melted sodium in xylenes below its melt-
ing point while stirring ended up with a broken glass impeller
stirrer at 750 rpm. A second experimentwithout stirring
during coolingled to sodium plates, which tended to aggregate
a
Table 2. Familiarization runs using cyclohexanone oxime 4 as model substrate
b
conversion
(%)
entry
alcohol
xylenes
procedure
c
1
4M2P 6 equiv
EtOH 64 equiv
4M2P 64 equiv
4M2P 64 equiv
10.5 vol
none
dose 4 in 4 equiv 4M2P to Na in xylenes at 105 °C; dose 2 equiv 4M2P
0
99
92
91
98
d
2
add Na to 4 in 37 equiv EtOH at 80 °C; dose 27 equiv EtOH; heating to 110 °C
add Na to 4 in 37 equiv 4M2P at 110 °C; dose 27 equiv 4M2P; heating to 156 °C
dose 4 in 27 equiv 4M2P to Na in 37 equiv 4M2P at 115 °C; heat to 156 °C; hold for 1 h 45 min
dose 4 in 64 equiv 4M2P to Na in xylenes at 121 °C; dose 27 equiv EtOH; hold for 1 h; dose 16 equiv
EtOH; hold for 39 min
3
4
5
none
none
4M2P 64 equiv
EtOH 43 equiv
9 vol
a
All runs used 13.5 equiv Na, except for entry 1 which was performed with 6 equiv Na. Scale: 16 (entries 4, 5) and 32 mmol 4 (entries 2, 3).
Conversion as determined by GC. RC1 experiment, 167 mmol 4. Isolated yield 35%.
b
c
d
2009
dx.doi.org/10.1021/op300101d | Org. Process Res. Dev. 2012, 16, 2008−2014

Organic Process Research & Development
Article
a
Table 3. Initial process development with 2-allyl cyclohexanone oxime 2
IPC: purity (%)
yield (%)
c
b
entry
1
alcohol
xylenes
procedure
trans/cis
trans/cis
as entry 5,
Table 2
9 vol dose 2 in 64 equiv 4M2P to Na in xylenes at 121 °C; dose 27 equiv EtOH; dose 16
>90 3.0:1.0
62 4.3:1.0
equiv EtOH
2
3
EtOH 128 equiv
EtOH 64 equiv
EtOH 43 equiv
−
based on Medicinal Chemistry protocol (Scheme 1)
>45 2.9:1.0
>90 2.9:1.0
72 5.9:1
9 vol dose 2 in xylenes to Na in xylenes at 121 °C; dose 64 equiv EtOH
54 4.5:1.0
dose 43 equiv EtOH
a
b
All runs used 16.1 mmol 2 and 13.5 equiv Na. Conversion and trans/cis ratio in IPC sample after approximately 1 h as determined by GC.
Conversions of 87−99% were detected. Sample preparation: 0.5 mL of the reaction mixture (waxy solid after cooling) was quenched with 0.5 mL
c
water and extracted with 1 mL MTBE (methyl tert-butyl ether). The organic phase was analyzed by GC (% a/a). Isolated yield of crude product;
trans/cis ratio as determined by GC.
a
Table 4. Process development with reaction calorimetry (RC1)
b
c
entry Na (equiv)
alcohol
xylenes
9 vol
procedure
conv. (%)
(Qr/mr)max
1
13.5
4M2P 63.8
equiv
dose 2 in 64 equiv 4M2P to Na in xylenes at 121 °C over 120 min;
72
−233 during dosage: after
hold 40 min
13 equiv 4M2P dosed
2
13.5
4M2P 10 equiv 9 vol
dose 2 in 10 equiv 4M2P to Na in xylenes at 120 °C over 47 min
99
−206 30 min after end of
dosage
EtOH 20 equiv
hold 14 h; dose 20 equiv EtOH over 77 min
3
6.0
4M2P 5 equiv 11 vol
dose 2 in 5 equiv 4M2P to Na in xylenes at 120 °C in 3 portions
(each over 10 min and 1 h waiting time)
68
−150 at the end of the
EtOH quench
EtOH 5 equiv
dose 5 equiv EtOH over 30 min
4
5
6
6.0
10.0
10.0
4M2P 5 equiv 9 vol
dose 2 in 5 equiv 4M2P to Na in xylenes at 120 °C over 30 min
hold 14 h
96
94
98
−123 4 h after dosage
−247 45 min after dosage
−204 during dosage of 2
4M2P 9 equiv 9 vol
dose 2 in 9 equiv 4M2P to Na in xylenes at 120 °C over 31 min
hold 27 h; add water at 90 °C
dose 2 equiv 4M2P to Na in xylenes at 120 °C hold 36 min
dose 2 in 7 equiv 4M2P at 120 °C over 31 min
hold 21 h, dose 2 equiv 4M2P, add water at 90 °C
4M2P 2 equiv 9 vol
4M2P 9 equiv
d
7
10.0
4M2P 2 equiv 4.5 vol + 4.5 vol dose 2 equiv 4M2P to Na in xylenes/toluene at 120 °C
70
reflux mode without
calibration
toluene
4M2P 9 equiv
dose 2 in 7 equiv 4M2P over 62 min
hold 1.25 h dose 2 equiv 4M2P add water at 90 °C
a
b
Entries 1−6 used 64.5 mmol 2. Crude 3 was isolated in >90% a/a purity by GC. Conversion as determined by GC. Sample preparation: 0.5 mL of
the reaction mixture (waxy solid after cooling) was quenched with 0.5 mL water and extracted with 1 mL MTBE. The organic phase was analyzed by
c
d
GC (% a/a). Maximum heat flow (W/kg mr) without Qdos term. 193.5 mmol 2.
sodium was destroyed. With this procedure, a safe handling of
sodium on scale seemed possible. However, the high dilution
would result in a very low volume throughput. As this last
familiarization run (entry 5) gave promising results, further
experiments were continued with the 2-allyl cyclohexanone
oxime 2 (Table 3).
Process Development of the Oxime Reduction. The
first experiment with 2-allyl cyclohexanone oxime 2 (Table 3,
entry 1) was run under the conditions previously defined in
Table 2, entry 5. Conversion and purity levels greater than 90%
were obtained. However, the trans/cis ratio was only 3:1 in the
In-Process-Control (IPC) by GC compared to the 10:1 value
reported in the Medicinal Chemistry protocol. Repetition of the
original Medicinal Chemistry procedure in ethanol (entry 2) with
2 revealed that in our hands a ratio of only 2.9:1 was obtained
in the reaction mixture. However, it was shown for all three
runs that the trans isomer was enriched up to a ratio of 5.9:1
upon workup according to the original procedure.16 Run 3 was
carried out to check if the potential dimerization (via ketyl
radicals)a conceivable side reaction prior to reprotonation
took place, by allowing a 75 min aging period after addition of
the oxime 2 in xylenes and before the addition of ethanol. No
dimerization was detected by GC.
Further Development and Hazard Evaluation Using
Reaction Calorimetry. For practical reasons, further experi-
ments were carried out in the RC1 with heat flow and gas
generation rate in our focus (Table 4). It is noteworthy that the
heat flow trace is, in general, calculated without the Qdos term
which is quite large due to a dosage of a solution of 20 °C to a
reaction mixture at 120 °C. For this evaluation we did not
distinguish between the desired reaction (the reduction of the
oxime 2 with sodium and alcohol) and the inevitable
alcoholysis of sodium. Interestingly, a large heat flow is always
accompanied by a large hydrogen flow despite the fact that only
the alcoholysis of the sodium should lead to hydrogen generation
based on the assumed reaction mechanism.
2010
dx.doi.org/10.1021/op300101d | Org. Process Res. Dev. 2012, 16, 2008−2014

Organic Process Research & Development
Article
In a separate experiment, the stability of sodium in 4M2P was
studied: 0.033 mol of 4M2P was added during 10 min to 1 mol
of sodium in 200 mL xylenes at 105 °C, whereby a reaction power
of −40 W/kg reaction mixture (mr) and a dosage-controlled gas
evolution were measured, corresponding to the expected amount
of 0.39 L H2. However, this side reaction may be influenced by the
solvent composition (polarity) and temperature.12
Upon scaling up by a factor of 4 (Table 4, entry 1), the
longer dosage time (120 min) led to a lower conversion due to
the expected consumption of sodium by alkoxide formation.
This effect was also promoted by the large excess of alcohol
(64 equiv), thus reducing volume productivity substantially. It is
also assumed that more product was lost during workup due to
the high dilution (isolated yield 62%). The mass of the dosed
2-allyl cyclohexanone oxime 2 and 4M2P solution was very
high as compared to the initial mass (xylenes and sodium).
Dosage started when the sodium was present as large droplets
in xylenes. At this point, the sodium droplet size changed
frequently. Small-sized sodium droplets seemed to result in
higher exothermic heat flow than those of larger size (low
surface area). At the end of the dosage, the sodium was
completely consumed, and gas evolution ceased. No additional
alcohol was necessary to destroy excess sodium.
dissolved by adding a mixture of ethanol and 4M2P. This
approach was not further evaluated due to the low conversion.
In entry 4, in contrast to the procedure in entry 3, the solution
of 2 in 4M2P and xylenes was added in one portion within
31 min (equivalents were kept unchanged). As observed in
previous experiments (e.g., entry 2), a delayed exothermic effect
was measured. In entry 5, the equivalents of sodium were
lowered to 10 equiv as compared to entry 2 while keeping the
equivalents of 4M2P at 9 equiv. Heat release was again delayed
and reached a maximum heat flow of −247 W/kg, 43 min after
the end of dosage. The accumulation of the process had to be
further minimized for a scale-up into the pilot plant.
In entry 6, 2 equiv of 4M2P was added prior to the addition of
the oxime 2 with the aim of mitigating the delayed exotherm. This
was based on the hint from entry 1 that large excess of alcohol led
to less accumulation (exotherm during dosage) and the experience
that very low amounts of alcohol led to delayed exotherms. The
increase of polarity of the reaction mixture was beneficial. The
thermal safety of this process was improved (Chart 2).
Chart 2. RC1 plot of entry 6, Table 4
In entry 2, the excess of 4M2P was reduced from 64 to 10 equiv.
The goal was to implement a procedure with stoichiometric
amounts of sodium and 4M2P. The RC1 data for this run are
depicted in Chart 1. During the tremendous heat flow after
Chart 1. RC1 plot of entry 2, Table 4
The following enthalpies consider the Qdos term. However,
the accuracy is reduced by the small starting volume, the still
ongoing alcoholysis after the dosage of 4M2P, and the ongoing
alcoholysis even 10 h after the dosage of the oxime/4M2P
solution. For the temperature range of 20−120 °C an average
Cp value of 3200 J/(kg*K) for 4M2P has been extrapolated
from the value of 2670 J/(kg*K) at 20 °C. On the basis of the
hydrogen release 0.75 L/0.13 mol 4M2P after 10 min of dosage
and a waiting time of 36 min, only about 50% of the alcohol did
react with the sodium. During the 31 min dosage of the oxime
solution 62% of the heat is released (−463 kJ/kg total) with a
maximum heat flow of −285 W/kg (−204 W/kg without taking
Qdos into account). This value is too high for heat removal by
jacket cooling and therefore calls for a heat removal by reflux
condenser. Hydrogen release again parallels the reaction enthalpy.
Two hours after the end of dosage about one-third of the 10 equiv
sodium is converted into hydrogen. In case of a power failure,
taking into account an enthalpy of evaporation for xylene of −354
kJ/kg, the 38% heat accumulated would lead to an evaporation of
about 40% of the xylene without recondensation.
completion of the dosage of the solution of the starting
material, small sodium droplets were observed. When the heat
flow dropped down, again large sodium droplets were observed.
The process is highly controlled by the sodium surface area.
The surface area is a function of stirrer speed, baffles, stirrer
type, solvent density, polarity, and surface tension (it is easier to
generate fine droplets in apolar xylenes than in more polar
mixtures of 4M2P, sodium, and 2).
The following experiments focused on the accumulation as
the remaining safety issue. First, the solution of the oxime 2 in
4M2P and xylenes was added in three portions, each with at
least 1 h reaction time after the end of dosage (entry 3). In
addition, the equivalents of sodium and 4M2P were lowered to
6 equiv and 5 equiv, respectively, as mechanistically a minimum
of 4 equiv of sodium and 3 equiv of alcohol are necessary (see
Scheme 2). After completion of the reaction, most of the
sodium should be consumed. The remaining sodium was
To reduce the boiling point of the reaction mixture for a
process under reflux conditions, a 1:1 w/w mixture of toluene/
xylenes was proposed in order to have a solvent mixture with a
boiling point of approximately 120 °C (entry 7). A further
driving force to reduce the process temperature was the fact
that the boiling point of pure xylenes is too close to the onset of 2
2011
dx.doi.org/10.1021/op300101d | Org. Process Res. Dev. 2012, 16, 2008−2014

Organic Process Research & Development
Article
according to DSC and ARC. The size of the sodium droplets in
xylenes was very small (approximately 1 mm). The addition of
4M2P resulted in an agglomeration of sodium consisting of one
single large drop. During dosage of the oxime 2 solution, smaller
drops were formed, and their size continuously dec-
reased and finally reached 2−3 mm in diameter at the end of
the dosage. An additional 2 equiv 4M2P was necessary to dissolve
the sodium completely in a reasonable time.
Finally, the aim of run in entry 7 was to reproduce entry 6 on
larger scale (scale-up factor 3) including the use of toluene/xylenes
as solvents. The heat balance coefficient of the reflux condenser
was not calibrated. Therefore, the thermal data of the reduction
(Chart 3) were only evaluated qualitatively. It is evident that the
4M2P was present in up to 20% w/w in the product solution
when the original procedure was used. Any residual alcohol
impeded the follow-on steps that did not tolerate protic
solvents. Therefore, isolation of the 2-allyl cyclohexyl amine 3
as a salt was deemed attractive (Scheme 3). A first experiment
with addition of 1.0 equiv oxalic acid in MTBE to precipitate 3
from the organic phase as a salt gave 82% recovery on a 0.68 g
scale. Although the oxalate precipitated readily, it was shown
that the recovery after filtration was only 60% after stirring for
2 h; stirring for an additional 5 h gave another 24% and yet
another 6% yield resulted after two more hours. Therefore, the
following experiments always included stirring the suspension
for >10 h, usually overnight. On scale, the procedure was
changed to inverse addition of the amine 3 solution to a
solution of oxalic acid in MTBE for practical reasons. Titration
of the resulting oxalate with 0.1 N NaOH indicated that the
depicted 1:1 salt 6 (Scheme 3) was present.
Chart 3. RC1 plot of entry 7, Table 4
DSC Data. Due to the low onset of 2-allyl cyclohexanone
oxime 2 (177 °C, see Supporting Information) the temper-
atures given in the experimental procedure should not be ex-
ceeded, and solutions of 2-allyl cyclohexanone oxime 2 should
not be distilled to dryness. Handling its solution in MTBE was
deemed safe due to the presence of a boiling barrier. Isolated 2-
allyl cyclohexyl amine oxalate 6 can be considered stable up to
60 °C. However, the free amine 3 is not stable according to DSC.
Still, prolonged handling under the reaction conditions at 120 °C
or subsequent distillation at 80 °C had not affected the quality.
Accelerating Rate Calorimetry (ARC) Data. Self-heating
was detected starting at 140 °C, and the pressure vs
temperature indicated the generation of noncondensable
gases (Chart 4). It is recommended to use a boiling barrier
heat flow and hydrogen evolution mostly ceased after the end of
the dosage. The hydrogen flow during dosage is constant.
Chart 4. ARC data of deprotonated 2-allyl cyclohexanone
oxime 2
Likely, it seems possible to further reduce the equiv of
sodium towards 6 equiv. However, this was not studied further.
For all runs (entries 1−7, Table 4) the trans/cis ratio varied
between 2.8:1 and 3.9:1 in the IPC and was only slightly
increased to 2.9:1 or 4.0:1 after workup. For entries 2−7 in
Table 4, the new workup described below was carried out.
Isolation and Workup. In our hands, workup following the
original procedure, i.e. quench with water, acidification, and
extraction with MTBE followed by adjusting the pH to 14 and
extraction with MTBE, gave an organic phase containing the
2-allyl cyclohexylamine 3. Due to the high volatility described
for 2-allyl cyclohexyl amine 3 (boiling point 65 °C at 16 mbar,
estimated boiling point at normal pressure approximately
170 °C) and the use of 4M2P (bp 130 °C) isolation by evapora-
tion to an oil was not feasible. The experiments showed that
a
Scheme 3. Telescoped oxime formation−oxime reduction sequence as used for pilot-plant batches
a
Reagents and conditions. (a) NH2OH·HCl (1.3 equiv), pyridine, MeOH, MTBE, solvent exchange to 4-methyl-pentanol, concentration: 2 in 7.1
vol 4-methyl-pentanol. (b) (i) Na (10 equiv), xylenes, toluene, 110 °C, (ii) 4M2P (1 equiv), 110 °C, (iii) 2 in 4-methyl-2-pentanol (7 equiv),
120 °C, (iv) 4-methyl-2-pentanol (2 equiv), 120 °C, (v) water, 32% HCl, 100 °C. (c) Oxalic acid (1.0 equiv), MTBE, 4-methyl-2-pentanol, 20 °C
(97% over 3 steps).
2012
dx.doi.org/10.1021/op300101d | Org. Process Res. Dev. 2012, 16, 2008−2014

Organic Process Research & Development
Article
(xylenes/toluene 1:1 w/w, bp 120 °C) for the reduction of
2-allyl cyclohexanone oxime 2 with sodium in order to avoid
undesired byproduct resulting from the thermal decomposition
of 2-allyl cyclohexanone oxime 2 deprotonated by sodium
alcoholate (for sample preparation see Supporting Information,
ARC data). It is noteworthy that in the final process the 2-allyl
cyclohexanone oxime 2 is added to sodium in the solvent mixture.
So, 2 is not present as full batch like in the ARC. This allowed the
reaction to run only 20 °C below the onset.
Scale-up in Pilot Plant. To ensure a safe scale-up of the
procedure described above, an intermediate scale-up with 410 g
sodium (5 L reaction volume) was performed in a 100-L vessel
(scale-up factor 90) to test the different reactor geometry
equivalents of alcohol was used. This measure minimized the
degradation of sodium to the alkoxide upon scale-up. (4)
Xylenes/toluene were used as cosolvents to melt sodium. This
solvent system generated a stirrable form of sodium during the
reaction and secured a boiling barrier at 120 °C. (5) The
isolation was streamlined by precipitating 2-allyl cyclohexyl
amine 3 as oxalate salt 6 instead of evaporation of the volatile 3.
In summary, a safe and practical procedure for the reduction
of 2-allyl cyclohexanone oxime 2 was developed and success-
fully implemented on kilogram scale, using 4.8 kg sodium, to
yield the corresponding amine 3 as oxalate salt 6. The pro-
cedure was safely reproduced 6 times with 4.8 kg sodium in a
100-L reactor.
(cylindrical Buchi steel-enamel reactor with impeller stirrer and
̈
EXPERIMENTAL SECTION
General. Reported temperatures are internal temperature
unless otherwise stated.
baffle located at 11.5 L, minimal stirring volume 1 L). One
change with respect to the run in entry 7 (Table 4) was carried
out: only 1 equiv 4M2P instead of 2 equiv 4M2P was added
prior to the addition of the oxime 2 to increase the excess of
sodium by 1 equiv. Possibly, this amount could be further
reduced, but this was not studied anymore. A lower than
expected conversion of 64% was measured after 5 h at 115 °C.
The low volumes led to rather inefficient stirring (baffle with
different plates not immersed), meaning that the sodium was
not finely dispersed by all baffles. Upon further scale-up, a faster
conversion was expected. The product was isolated in 57%
yield with a purity of 67/22% a/a GC (trans/cis). Indeed, in the
next intermediate scale-up run with 2.0 kg sodium (23 L reaction
volume), a conversion of 91% after only 2 h at 120 °C was
measured. The trans/cis ratio in IPC was 2.8:1.0. During the
quench with water a thick suspension was formed. The product
was isolated in 95% yield with a purity of 71/24% a/a GC
(trans/cis). On the basis of these results, the remaining material
was converted in six runs, each using 4.8 kg sodium (56 L
reaction volume) based on the elaborated procedure (entry 7,
Table 4). For details see also the Experimental Section. During
the addition of water the suspension, as formed during the
quench, became partly unstirrable (stirring continued at the
bottom, and the upper volume remained an unstirred, gel-like
mixture) during a short time of the addition, partially due to
the stirrer and reactor design (impeller and large height-to-diameter
ratio). This could likely be improved by quenching with a larger
excess of the polar 4-methyl-2-pentanol but was not tested.
After removal of 50% of the original amount of MTBE, the
solution was diluted with additional MTBE and the solution
stored after each run. The final oxalate salt formation was
carried out with combined runs 1−3 and runs 4−6 (Scheme 3)
by an inverse addition of the solution in MTBE to 1.0 equiv
oxalic acid in MTBE at 20 °C. Both combined workup runs led
to the expected white solid rac-6 in quantitative yield and 71/
24% a/a GC (trans/cis).
■
Safety Measures with Elemental Sodium. Generally, the
process should only be carried out in a reactor equipped with
an oil-cooled reflux condenser. Also, due to the ambiguous
results of the reaction exotherm, a mixture of xylenes or toluene
was placed in a feeding tank as a contingency for quick
dilution of the reaction mixture in case of a sudden exotherm
that would lead to uncontrolled boiling. Use a constant nitrogen
purge during operations generating hydrogen gas to ensure
dilution in exhaust gas. Sodium reacts with many substances
very exothermically, in some cases under explosion, the given
list not being complete:17 all halogenated solvents, water, DMF,
hydrazine, charcoal, sulfides, acetylenes, metal halogenides,
ammonium salts, oxides, oxidants, acids, alcohols, oxygen,
carbon dioxide, and chlorinated organic compounds. It is recom-
mended to test seals and gaskets under friction if they are un-
avoidable, e.g. charging via a butterfly valve (in analogy to Teflon
that is not stable under friction with lithium). For decontamination
and fires: use only extinguisher for metals, powder extinguisher,
dry sand, or in case of a spillage also sodium chloride or soda. Do
not use water or carbon dioxide extinguisher.
2-Allyl Cyclohexanone Oxime (2). A dry 100-L reactor
was purged with nitrogen, and the jacket temperature was set to
20 °C. NH2OH·HCl (10.1 kg, 1.1 equiv) was loaded, followed
by MeOH (95 L). Note: a solution was obtained. Pyridine
(19 L, 1.7 equiv) was added at 17−18 °C over 8 min. Allyl
cyclohexanone 1 (18.2 kg, 1 equiv, 131.7 mol) in MeOH (18 L)
was added at 18−21 °C over 31 min. Stirring was continued at
20 °C for 1 h 17 min. IPC by GC: 100% conversion. Sat.
aqueous NaCl solution (60 L) was added at 20 °C over 10 min.
Note: a suspension was obtained. Aqueous 0.5 N HCl (132 L,
0.5 equiv) was dosed at 20−22 °C during 18 min. MTBE (152 L)
was added, and stirring was continued at 20 °C for 5 min. After
phase separation, the inorganic phase was washed with MTBE
(105 L). The organic phase was washed twice with sat. NH4Cl
solution (2 × 45 L). The organic phase was washed with water
(49 L). 131 L of a total of 256 L solvent was removed at jacket
temperature 60 °C to afford 2 as a yellow solution in MTBE
(106 L). Yield: 97%, corrected for LOD (loss on drying).
Purity: 94.7% a/a (sum of isomers), LOD 81.5% w/w, off-white
solid after concentration to dryness of an aliquot. NMR
corresponds to reference 8.
CONCLUSION
■
On the basis of extensive thermokinetic experiments, our goal
to develop a safe procedure of the oxime reduction by melted
sodium was achieved. The following features have been key in
optimizing the initial protocol. (1) It was shown that the
presence of 4M2P instead of ethanol was essential to avoid
accumulation. (2) An inverse addition was applied, i.e. addition
of 2-allyl cyclohexanone oxime 2 in 4M2P to melted sodium in
xylenes/toluene instead of solid addition of sodium to refluxing
ethanol. This prevented the open handling of sodium in
ethanol vapors. (3) The equivalents of sodium were reduced
from 13.5 equiv to 10.0 equiv, and only 10.0 instead of 64.5
2-Allyl Cyclohexyl Amine Oxalate (6). A dry 100-L reactor
was purged with nitrogen, and the jacket temperature was set
to 20 °C. 4-Allyl cyclohexanone oxime 2 (19.6 kg) in MTBE
(106 L) was transferred into the reactor. Solvent (67 L) was
removed at a jacket temperature of 60 °C. 4-Methyl-2-propanol
2013
dx.doi.org/10.1021/op300101d | Org. Process Res. Dev. 2012, 16, 2008−2014

Organic Process Research & Development
Article
Notes
(114 L) was added to the solution. Solvent was removed at a
jacket temperature of 60 °C and p = 80 mbar until the
distillation stopped. After cooling to 20 °C, the solution was
stored in a barrel for the subsequent six analogous runs with
sodium followed by two isolations as oxalate salt.
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
We thank Bharat Shah, Michael DeNinno, and Step
■
́
hane Caron
The reactor was rinsed with xylenes. The reactor was loaded
with xylenes (12.8 L). Sodium (4.8 kg, 10.0 equiv) was loaded
into the reactor. Note: sodium was kept under kerosene or
xylenes. The reactor was purged with nitrogen, and toluene
(13.0 L) was added. The temperature was set to 110 °C. Note:
sodium is melted at jacket temperature (max) = 130 °C, no
stirring was applied. The temperature was set to 120 °C.
Stirring was started. 4-Methyl-2-propanol (2.6 L, 1.0 equiv) was
added at >112 °C during at least 15 min. Stirring was continued
for 30 min at 115−120 °C. A sixth part of the above solution of
2-allyl cyclohexanone oxime 2 (22.7 L, 1 equiv) in 4M2P
(7 equiv) was added during at least 60 min at >115 °C. Note:
reflux during part of the addition. Stirring was continued for 2−
14 h at 120 °C. IPC by GC: 90% conversion.
4-Methyl-2-propanol (5.2 L, 2.0 equiv) was added at 120 °C
during at least 4 min, and stirring continued for at least 1 h at
120 °C. The temperature was set to 90 °C. Water (23.5 L) was
added at 120−100 °C during at least 30 min for first 8 L. Note:
in the beginning the addition was very exothermic, and the
suspension, as formed during the quench, became partly un-
stirrable. The two-phase mixture was cooled to 15 °C, and 32%
HCl (12.5 L, 6.0 equiv) was added at 15−25 °C; pH = 11.
30% aqueous NaOH was added to readjust the pH to >13. The
phases were separated. The organic phase was washed with
0.1 N NaOH (15.0 L). MTBE (45 L) was added to the organic
phase. Solvent (50% of the MTBE amount added) was
removed at jacket temperature 80 °C. MTBE (23 L) was
added, and the solution OP 1 (approximately 70 L each) was
stored after each of the six runs without further analysis.
A dry 630-L reactor was loaded with oxalic acid (5.60 kg,
1.0 equiv), and MTBE (60 L) was added. Upon stirring, a clear
solution was obtained. At 20 °C OP 1 (210 L combined for
three runs with sodium) was added during at least 15 min.
Note: a white suspension was formed. Stirring was continued
for at least 12 h at 20 °C. The suspension was filtered over a
filter nutsch. The filter cake was washed with MTBE (39 L).
The white solid was dried on the nutsch by a flow of N2. Lot 1:
23.44 kg, lot 2: 17.65 kg. Yield: quant., corrected for loss on
drying (LOD). Purity: 71/24% a/a (trans/cis) (identical result
for both lots), LOD lot 1: 19% w/w, lot 2: 8% w/w. NMR and
GC correspond to reference from Pfizer.
(Pfizer) for their professional collaboration and encouragement
to publish this work. We are grateful for Cornelia Struppe’s
analytical support and Bernd Englert for running the numerous
thermokinetic experiments. Florian Zemp and Samuel Vogel at
CARBOGEN AMCIS AG are thanked for operating the
chemistry in the laboratory and the pilot plant.
REFERENCES
■
(1) (a) Initial Medicinal Chemistry procedure given by Pfizer.
(b) DeNinno, M.; Andrews, M.; Bell, A. S.; Chen, Y.; Eller-Zarbo, C.;
Eshelby, N.; Etienne, J. B.; Moore, D. E.; Palmer, M. J.; Visser, M. S.;
Yu, L. J. Bioorg. Med. Chem. Lett. 2009, 19, 2537−2541.
(2) Booth, H.; King, F. E. J. Chem. Soc. 1958, 2688−2693.
(3) Rausser, R.; Weber, L.; Hershberg, E. B.; Oliveto, E. P. J. Org.
Chem. 1966, 31, 1342−1347.
(4) Masamune, T.; Ohno, M.; Koshi, M.; Ohuchi, S.; Iwadare, T.
J. Org. Chem. 1963, 29, 1419−1424.
(5) Solodar, J. J. Org. Chem. 1976, 41, 3461−3465.
(6) Huffmann, J. W.; Charles, J. T. J. Am. Chem. Soc. 1968, 90, 6486−
6492.
(7) Hutschins, R. O.; Wei-Yang, S.; Sivakumar, R.; Cistone, F.;
Stercho, J. P. J. Org. Chem. 1983, 48, 3412−3422.
(8) Kawada, K.; Tsushuima, T. Heterocycles 1989, 28, 573−578.
(9) Kennewell, P. D.; Matharu, S. S.; Taylor, J. B.; Westwood, R.
J. Chem. Soc., Perkin Trans. I 1982, 2553−2562.
(10) Hutchins, R.; Adams, J.; Rutledge, M. C. J. Org. Chem. 1995, 60,
7396−7405.
(11) Silica gel stabilized Na for the Bouveault−Blanc reduction of
esters, see: Bodnar, B. S.; Vogt, P. F. J. Org. Chem. 2009, 74, 2598−
2600.
(12) Banziger, M.; Kusters, E.; La Vecchia, L.; Marterer, W.; Nozulak,
̈
̈
J. Org. Process Res. Dev. 2003, 7, 904−912.
(13) Becker, H. G. O. Organikum, 20th ed.; Wiley-VCH: Weinheim,
1999; pp 474−477.
(14) Reduction of oxime with 10 equiv Na in 2 vol of ethanol, see:
Organic Synthesis; Wiley & Sons: New York, 1943; Collect. Vol. II, pp
318−320.
(15) Sugden, J. S.; Patel, J. J. B. Chem. Ind. 1972, 683.
(16) The cooled reaction mixture was quenched with water/ice and
acidified with concentrated aqueous HCl. Ethanol was removed by
distillation, and the residue was taken up in water and extracted with
diethyl ether. The aqueous layer was basified with 15% aqueous NaOH
and extracted with diethyl ether. The organic layers were washed with
brine, dried over magnesium sulfate, filtered, and concentrated. This
workup procedure was used for entries 1−3, Table 3, and entry 1,
Table 4.
ASSOCIATED CONTENT
■
(17) Urben, G. Bretherick’s Handbook of Reactive Chemical Hazards,
6th ed.; Butterworth-Heinemann: Oxford, UK, 1999; pp 1815−1823.
S
* Supporting Information
Details on DSC and ARC data; GC method including a GC-
trace of 3; copies of H NMR of compounds 2 and 6. This
material is available free of charge via the Internet at http://
pubs.acs.org.
NOTE ADDED AFTER ASAP PUBLICATION
1
■
This paper was published on the Web on June 19, 2012, with
incorrect legends in Charts 1 and 2. The corrected version was
reposted on June 21, 2012.
AUTHOR INFORMATION
■
Corresponding Author
*E-mail: roland.breitenmoser@carbogen-amcis.com. Telephone:
+41 61 935 53 91.
Present Address
‡Process Research Chemistry, Actelion Pharmaceuticals Ltd.,
Gewerbestrasse 16, CH-4123 Allschwil, Switzerland.
2014
dx.doi.org/10.1021/op300101d | Org. Process Res. Dev. 2012, 16, 2008−2014