Scale-up of azide chemistry: A case study

Article abstract of DOI:10.1021/op3002646

We report research and development conducted to enable the safe implementation of a highly enantioselective palladium-catalyzed desymmetrization of a meso-bis-ester using trimethylsilylazide (TMSN₃) as the nucleophile. This work is used as a ca

Full text of DOI:10.1021/op3002646

Communication
pubs.acs.org/OPRD
Scale-up of Azide Chemistry: A Case Study
Francisco Gonzalez-Bobes,*,† Nathaniel Kopp,*,† Li Li,‡ Joerg Deerberg,† Praveen Sharma,†
́
Simon Leung,§ Merrill Davies,† Joseph Bush,§ Jason Hamm,§ and Michael Hrytsak§
†Chemical Development, ‡Analytical and Bioanalytical Development, and §Chemical Development Operations, Bristol-Myers Squibb
Company, One Squibb Drive, New Brunswick, New Jersey 08903, United States
S
* Supporting Information
ABSTRACT: We report research and development con-
ducted to enable the safe implementation of a highly
enantioselective palladium-catalyzed desymmetrization of a
meso−bis-ester using trimethylsilylazide (TMSN3) as the
nucleophile. This work is used as a case example to discuss
safe practices when considering the use of azide reagents
or intermediates, with a focus on the thermodynamic and
quantitative analysis of the hazards associated with hydrazoic
acid (HN3).
transformations: (a) the formation of the allylic azide inter-
mediate 2; (b) the reduction of the azide intermediate to the
INTRODUCTION
amine 3; and (c) the formation of the desired allylic carbamate
4. Herein, we report a detailed safety analysis and process
assessment that enabled the safe scale up of this chemistry
without the need of specialized facilities. Whereas most published
approaches rely on preventing accumulation of HN3 in the
condensate by implementing engineering controls, we consid-
ered the risks associated with this approach not acceptable
for implementation in our multipurpose scale up facilities. We
focused on a quantitative approach in which, even in the event
of enriched condensate accumulation, the lower explosive limit
(LEL)17 would not be reached. To the best of our knowledge,
such an approach has not been previously disclosed in the
context of preparation of pharmaceutical intermediates.
■
Azide chemistry offers an efficient entry to synthesize a range of
nitrogen-containing compounds with a wide variety of functional
groups.1,2 When considering a reaction involving azide reagents
and intermediates, safety aspects, such as material compatibility,
exposure, and explosivity, must be evaluated and addressed
prior to large scale execution. Specifically, the incompatibility of
azide reagents with some metals3 and with certain chlorinated
solvents,4 and the thermal instability of azide intermediates5
constitute major hurdles when developing routes that involve
azide-containing intermediates. A topic of even greater concern
is the potential for generation of hydrazoic acid (HN3).6 HN3
is a relatively weak acid (pKa = 4.7),7 with a low boiling point
(37 °C).8 The recommended NIOSH exposure limit for in-
halation, expressed as the threshold limit value (TLV), is 0.11 ppm
for HN3.9 Exposure above these levels has been shown to lower
blood pressure in animals and humans. Other symptoms often
associated with inhalation of HN3 include the following:
bronchitis, eye, nose, throat, and lung irritation, headache, and
weakness that can lead to collapse. There is no specific antidote
for azide intoxication.9d In addition to exposure, HN3 is an
unstable and shock sensitive compound that can decompose to
generate large amounts of gas and energy.10 Equilibrium con-
centrations of HN3 can be formed under a range of pH
conditions,11 and its volatile nature creates the potential for
explosions.12 Despite these concerns, azide chemistry has been
employed on large scale,13 primarily by the implementation of
process and engineering controls.14 For example, there are several
reports involving the scale-up of azide reactions, in many cases
carrying out critical operations in bunkers.15
RESULTS AND DISCUSSION
■
Initial Assessment: First Generation Process. We
undertook development efforts with two major goals: (a) to
design a safe and scalable process to support downstream route
development and (b) to develop a safe and robust process to
support long-term project needs. Our first step to assess the
viability of using this reaction on scale was to evaluate the thermo-
stability of the allylic azide 2 in THF. The decomposition of 2 was
highly exothermic (−388.4 J/g), yet it rated as low in probability
(ADT 24 was 93 °C by ARC and 94 °C by Setaram C80 and
AKTS Thermokinetic analysis).18 This did not pose a reactivity
problem, since the palladium-catalyzed desymmetrization pro-
ceeds efficiently at 0 °C.16,19 Based on the thermochemistry
data, we also decided that the allylic azide 2 would not be
isolated but used directly in the next step, which would be
carried out in the same solvent system as the desymmetrization
reaction to avoid the need for distillations.
During the development of a synthetic route to a pharma-
ceutical candidate, we became interested in accessing an inter-
mediate via a highly enantioselective palladium-catalyzed de-
symmetrization of a meso-bis-ester 1 using TMSN3 as the
nucleophile (eq 1).16 The process requires three chemical
Special Issue: Safety of Chemical Processes 12
Received: September 21, 2012
© XXXX American Chemical Society
A
dx.doi.org/10.1021/op3002646 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
Scheme 1. Description of the First Generation Process
Our initial process development relied heavily on the literature
conditions.16 We quickly established a basic understanding of
the unit operations required to obtain 4 with the desired quality
attributes (Scheme 1).
Scheme 2. Secondary Reactions for Residual TMSN3 under
Processing Conditions
The desymmetrization performed well in THF as the solvent,16
and the reaction mixture could be easily telescoped into the
Staudinger reaction via addition of water and a dilute solution
of trimethylphosphine in THF.20 The thermochemistry of the
azide reduction was also evaluated. The reaction was highly
exothermic (−161.6 J/g) and produced significant amounts of
nitrogen as a byproduct. Nonetheless, the thermal profile and
gas evolution were addition-controlled and easily managed by
adjusting the rate of addition of the reactants to adapt to the
cooling and venting capacities of the equipment. Extracting the
allylic amine 3 into the aqueous phase by acidification with
diluted aqueous acid was crucial to ensure stability of the
process stream and process robustness. Addition of n-heptane
allowed for an easy phase separation, in which most of the
process- and catalyst-derived impurities remained in the organic
phase and could be removed via a simple liquid−liquid separation.
The aqueous solution was then treated with isopropyl acetate
(IPAc) and an excess of aqueous sodium hydroxide to effi-
ciently partition the allylic amine 3 into the organic phase. After
a phase separation, the IPAc layer was treated with di-tert-butyl-
dicarbonate (Boc2O) to afford the desired allylic carbamate 4.
Finally, solvent exchange from IPAc21 to IPA followed by
addition of water as antisolvent allowed for the isolation of
the product in excellent chemical and optical purity (>99.5%
e.e.; >99% pure).
Through a basic risk analysis we identified two transfor-
mations that could generate HN3 during the process (high-
lighted in red in Scheme 1). The formation of HN3 from hydrolysis
of TMSN3 due to residual water during the desymmetrization
reaction and during the downstream processing from any un-
reacted TMSN3 (Staudinger reaction and acidification) posed a
safety concern. The hydrolysis during the desymmetrization
reaction could be minimized by controlling the water content
of the input reagents and solvent. In addition, we reasoned that
HN3 would be a competent nucleophile in the desymmetriza-
tion reaction. The potential for formation of HN3 from residual
TMSN3 upon addition of water required the development of a
control strategy to establish safe operating limits for HN3 after
the desymmetrization reaction (Scheme 2).22
We then focused on the development of an in-process
control (IPC) for residual azide to guarantee that the solution, gas,
and enriched condensate phases would remain below the LEL.10,23
Development of the Process Control Strategy:
Vapor−Liquid Equilibrium (VLE) Assessment. We con-
ducted a complete assessment of the three phases in two steps:
(a) evaluation of the process vapor−liquid equilibrium with
respect to reaction completion (azide consumption); and (b)
determination of the process IPC limit through interpolation of
literature data for the LEL of the enriched condensate in equilib-
rium with the solution and vapor phase.
Due to the issues associated with the preparation and
handling of HN3, few studies have developed complete vapor−
liquid equilibrium profiles. However, independent reports have
reasonably agreed on the Henry’s law constant (KH) to describe
the equilibrium at concentrations ranging from 11 M to 10 μM.11,24
The Henry’s law constant is expressed as the ratio between the
aqueous phase solution concentration [HN3] and the partial
pressure of HN3 in the gas phase, PHN3 (eq 2):
[HN3]aq
KH =
P
HN3
(2)
Betterton and Robinson11 determined this value at 25 °C
(aq) as 12.0 0.7 M/atm, which may be adjusted based on the
pH and temperature of the system. Our study specifically
examined the effect of pH, as the downstream addition of
aqueous acid significantly lowers the pH and increases the HN3
vapor pressure in the system.
The pH dependence of the Henry’s law constant (KH) may
be described through calculation of the effective Henry’s law
B
dx.doi.org/10.1021/op3002646 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
constant (K′H) with respect to the [H+] concentration and the
acidity constant for HN3 (Ka = 2.24 × 10−5) (eq 3):
aqueous system at 25 °C. The process conditions added a
buffer into the approximation, since the temperature was con-
trolled below 15 °C in a volatile organic solvent. As presented
in Figure 2, the effects of pH dominate the HN3 gas phase
concentration (vol %). To maintain safety during downstream
processing, a high pH and/or near complete reaction conver-
sion must be achieved. Since the process required an acid-
ification for intermediate stabilization (approximate pH 3.5),
our control strategy relied on achieving high reaction con-
versions (low residual free azide). Therefore, the next step
in the quantitative analysis became the determination of
the acceptable residual free azide level after the palladium-
catalyzed desymmetrization reaction. To accomplish this, we
first interpolated equilibrium data from the PUREX process, a
nuclear fuel reprocessing method which forms HN3 as a
byproduct.23a Using this data, we established a gas phase limit
of 0.625 vol % which is based on the enriched condensate
LEL in equilibrium with the solution and gas phases. As an
additional layer of safety, our IPC limit calculation used a
more conservative HN3 gas phase limit of 0.30 vol %, which
translated to a final limit of 200 ppm of free azide after the
desymmetrization reaction.25 By using substoichiometric
amounts of TMSN3 (0.95 equiv), we were able to comfortably
meet the proposed IPC (typical experimental values of <100
ppm). This strategy was employed during scale-up.
+
′
KH = KH(1 + Ka/[H ])
(3)
Figure 1 highlights the relationship between the vapor−
liquid equilibrium and pH derived from eq 3. This plot
Figure 1. Combined effects of solution free azide concentration and
pH on the partial pressure of HN3 in the gas phase over H2O at 25 °C.
Adapted with permission: 1997 Taylor & Francis Group.11
First Generation Process: Process Execution and
Results. Although the focus of our control strategy was limiting
the level of free azide in solution post-desymmetrization, the
process execution included a comprehensive evaluation from
reagent and waste handling to emergency preparedness. Before
processing, an inert transfer of the azide reagent to a com-
patible pressure container was conducted to minimize exposure
to moisture during charging and limit personnel exposure. A
nitrogen sweep was implemented during processing to purge
the process train, and the vent line was sent through a caustic
scrubber. Waste streams were segregated and treated with
excess base prior to disposal by a qualified contractor.
quantitatively shows that a higher solution concentration of
azide and/or a lower pH both increase the gas phase partial
pressure significantly. The TLV at 0.11 ppm9a illustrates that
azide solution concentrations must be kept very low or at high
pH in order to maintain safe exposure levels. Since the chemistry
was conducted in an inert, closed environment, the safe solution
azide level did not depend on the TLV but focused on the LEL.
The Henry’s law approximation afforded a method to
evaluate the vapor−liquid equilibrium relationship with respect
to reaction conversion and pH for our process (Figure 2). It
must be noted that these calculations are based on a purely
Figure 2. The vol % of HN3 in the gas phase corresponding to conversion of starting material or consumption of trimethylsilyl azide reagent with
respect to pH.
C
dx.doi.org/10.1021/op3002646 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
Scheme 3. Description of the Second Generation Process
The process was successfully scaled-up in two batches to
deliver 8.42 kg of 4 (76% yield based on TMSN3) in high purity
(>99%), potency (≥99%), and enantiomeric excess (≥99.5%).
In both cases the reaction IPC for residual azide after the
desymmetrization reaction met the proposed limit.26
exothermic decomposition behavior (−326.5 J/g) upon
evaluation with the ARSST (advance reactive system screening
tool) and DSC (differential scanning calorimetry). This decom-
position was also rated as low in probability (ADT 24 was 71 °C
by the ARSST and 80 °C by DSC and AKTS Thermokinetics
Analysis),18 and it did not pose a problem since the palladium-
catalyzed desymmetrization proceeded efficiently at 0 °C. Upon
reaction completion (30 min), the mixture was treated with
excess aqueous sodium hydroxide (2 N), followed by azide
reduction, which proceeded smoothly under the basic con-
ditions of the reaction media.29 The concentration of azide in
the organic layer was <10 ppm, well below the 200 ppm limit
previously discussed, showcasing that the residual levels of azide
after the desymmetrization were no longer critical (Scheme 3).30
A comparison of both processes against different key
parameters is presented in Table 1. The initial process required
Second Generation Process. Although we successfully
demonstrated our first generation process at kilogram scale and
thus met the first objective of our development work, we
recognized that the process had some key limitations that
would need to be addressed before considering larger scale
operations. Specifically, in order to meet the IPC for residual
azide, a very accurate charge of TMSN3 would be critical.
The use of TMSN3 as the limiting reagent resulted in the
undesirable sacrifice of variable amounts of the more elaborate
meso-bis-benzoate starting material. More importantly, the
control strategy was based on minimizing the generation of
HN3 whereas avoiding its formation altogether would be
intrinsically safer. We decided to pursue the development of an
alternative process in which the stoichiometry of TMSN3 would
not play a critical role.27
Table 1. Comparison of the Process against Key Parameters
2nd
key parameters
1st generation
critical
generation
We were interested in leveraging the chemical properties of
HN3 to aid its efficient removal. We envisioned that, upon
completion of the desymmetrization reaction, the reaction
media could be adjusted to a basic pH via addition of an
aqueous solution of sodium hydroxide (eq 4). This would
stoichiometry of TMSN3
not critical
not critical
azide levels post-desymmetrization
control to ppm
level
potential to form HN3 downstream
HN3 control strategy
yield
yes
no
IPC control
process design
76%
85
88%
52
PMI (kg total process input/kg
product)
tight control of the stoichiometry and a highly sensitive IPC. In
contrast, the second generation process addresses the topic of
HN3 generation through an improved process design in which
neither the stoichiometry nor the levels post-desymmetrization
of azide play a critical role. Importantly, significant yield improve-
ment and waste reduction were realized while maintaining the
desired product quality.
immediately convert any excess of TMSN3 present in the
reaction mixture to aqueous sodium azide, which could be
easily removed via extractive workup, hence eliminating any
concerns about formation of HN3.28
To use this strategy, a series of requirements would need to
be met: (a) the desymmetrization reaction should proceed in a
solvent that allows for an efficient phase separation from the
aqueous phase; (b) the Staudinger reaction should also proceed
efficiently in that same solvent at low temperature;19 (c) the
sodium azide generated as byproduct should efficiently partition
into the aqueous phase to allow for quantitative removal. With
these objectives in mind, we found that toluene could be used
as the solvent in the palladium-catalyzed desymmetrization
reaction without compromising chemical and optical purity.
The allylic azide intermediate 2 in toluene also exhibited highly
CONCLUSION
■
We have described the safe implementation of a highly enantio-
selective palladium-catalyzed desymmetrization using an azide
nucleophile. We conducted a quantitative analysis of the
vapor−liquid-condensate equilibrium using Henry’s law based
on the literature data for HN3, which shows that the safe
operating limits are dictated by the enriched condensate. This is
in contrast with most published reports which either do not
D
dx.doi.org/10.1021/op3002646 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
consider enriched condensate or mitigate the risk by imple-
menting engineering controls. This quantitative analysis allowed
for the establishment of a successful strategy to address explosivity
concerns associated with HN3 generation. We used this approach
to prepare a key intermediate in kilogram amounts with excellent
quality in our multipurpose scale up facilities. In addition, we also
designed and demonstrated a more efficient process based on an
understanding of the secondary reactions involved in the fate of
TMSN3 under the reaction conditions. The learnings from this
case example may be applicable to other reactions involving azides.
the allylic azide had been successfully converted to the allylic
amine intermediate. n-Heptane (15.020 kg), water (11.935 kg),
brine (20 wt %, 11.985 kg), and aqueous citric acid (20 wt %,
41.00 kg) were sequentially added. The reaction mixture was
then allowed to warm up (15−25 °C) and stirred for at least
15 min. The layers were allowed to split. The aqueous layer,
which contained the allylic amine, was pressure-transferred into
a clean 200 L reactor, and was held overnight. The organic layer
was disposed of and treated as azide containing waste. To the
aqueous solution containing the product was added IPAc
(24.086 kg). The mixture was then cooled to 0 °C, followed by
the addition of aqueous sodium hydroxide (25 wt %, 23.207 kg).
The addition was done at such a rate that the temperature
stayed in the desired range (0−10 °C). The mixture was
warmed to room temperature (15−25 °C), and after no less
than 15 min the phases were allowed to split. The lean aqueous
was discharged, and the rich organic layer was washed with
brine (20 wt %, 18.162 kg) and diluted with water (17.948 kg).
After no less than 15 min of contact time, the phases were
allowed to split and the aqueous layer was discharged. A
solution of Boc2O (3.540 kg) in IPAc (9.551 kg) was then
added to the reactor over a period of 1 h. After 90 min the
reaction mixture was sampled for completion. A put and take
distillation was implemented to remove IPAc and replace it in
solution with IPA for crystallization of the allylic carbamate.
Distillation was continued until an IPA/IPAc ratio of >95:5 was
obtained as measured by GC-FID. The concentration of the
product was adjusted with IPA to ∼200 mg/mL, and a seeded/
cooling crystallization using water as antisolvent was conducted.
The slurry was filtered and then washed with IPA/H2O (1/3).
The wet cake was placed into a vacuum oven, and it was dried
at 40−45 °C under vacuum. 3.67 kg of product were obtained.
99% wt; 99.9 area %; 76% corrected yield (based on azide).
b. Second Generation Process. cis-cyclohex-2-ene-1,4-diyl
dibenzoate (1 equiv, 874.8 mol (corrected for purity), 300 g)
(1), [Pd(π-allyl)Cl]2 (0.0015 equiv, 1.31 mmol, 0.48 g, 0.15 mol %),
and (R,R)-DACH-Ph (0.0045 equiv, 3.94 mmol, 2.77 g, 0.45 mol %,
Chirotech) were introduced in a 5 L reactor equipped with a
mechanical stirrer. The reactor was placed under a nitrogen
atmosphere (three vacuum/refilling cycles over 45 min, 100 Torr).
Toluene (<0.003% H2O by KF; 3 L/kg of crude meso-ester,
900 mL) was then added, and the resulting mixture was stirred
for 5−10 min at room temperature. The resulting homoge-
neous solution was then cooled. Once the internal temperature
reached −3 °C to −5 °C, TMSN3 (98% wt, 1.15 equiv, 118 g,
137 mL, Sigma Aldrich) was added at such a rate that the
internal temperature was kept below 0 °C. After the addition
was complete, the jacket temperature was adjusted to −5 °C,
and the mixture was stirred for 1 h. A sample was analyzed by
HPLC (residual starting material <1 area %). 2 N NaOH (1.4 equiv,
703 g, 651 mL, ∼2 L/kg crude meso-ester) and a 1 M solution
of PMe3 in toluene (1.25 equiv, 1165 mol, 988 g, 1165 mL,
Sigma Aldrich) were sequentially added at such a rate that the
internal temperature stayed at 0−10 °C. The reaction mixture
was stirred for 1 h at 0−10 °C, and then an aliquot was taken to
ensure complete conversion of the allylic azide to the allylic
amine had occurred. The reaction mixture was then treated
with NaCl 10% (2 L/kg of crude meso-ester, 600 mL), warmed
to 18−22 °C, and held for 5 min with stirring, and then the
phases were allowed to split. The aqueous layer was collected
and was segregated to be disposed as azide containing waste.
TBME was added (2 L/kg of crude meso-ester, 600 mL), and
the organic layer was then cooled to ∼ +15 °C. A sample of the
EXPERIMENTAL SECTION
■
General. TMSN3 was transferred from the commercial
container to a pressure bomb excluding air and moisture. In
laboratory scale, TMSN3 was always handled in a ventilated
enclosure (fume hood) to prevent exposure to HN3 vapors. On
kilogram scale, handling of TMSN3 was done using double
gloves (inner-nitrile surgical style, outer-silvershield), a silver-
shield apron, and a supplied air respirator.
The potency of cis-cyclohex-2-ene-1,4-diyl dibenzoate was
determined by HPLC, and the stoichiometry of TMSN3 was
corrected accordingly. For details on the analytical method, see
the Supporting Information.
Waste treatments to eliminate azide were not conducted on
site. The waste streams containing azide were adjusted to pH ≥12
with aqueous NaOH and handled by a qualified contractor.
The palladium-catalyzed reaction is oxygen sensitive, and
inefficient inertion could cause catalyst deactivation and stalling,
thus compromising the ability to meet the IPC. The following
actions were taken to ensure adequate inertion: (a) extensive
evacuation/refilling cycles with nitrogen and (b) circulation of
the nitrogen gas through a gas purifier to adsorb residual oxygen
(Oxyclear).
A scrubber with 5 wt % sodium hypochlorite was used to
collect the PMe3 vapors from the vent line.
The Staudinger reaction and the addition of Boc2O produced
gaseous byproducts. The rate of addition was adjusted to the
cooling and venting capacities of the equipment.
Preparation of (1R,4S)-4-(tert-Butoxycarbonylamino)-
cyclohex-2-enyl Benzoate (4). a. First Generation Process
(TMSN3 as the Limiting Reagent). A 200 L lined-glass reactor
was inerted via three consecutive vacuum/nitrogen refilling
cycles (vacuum pressure: −0.8 bar). cis-Cyclohex-2-ene-1,4-diyl
dibenzoate (5.465 kg, 1.0 equiv) (1), [Pd(π-allyl)Cl]2 (9.44 g,
0.0015 equiv), and (R,R)-DACH-Ph (54.12 g, 0.0045 equiv,
Chirotech) were charged to the reactor, followed by three
additional inertion cycles (vacuum pressure: −0.8 bar). THF
(12.200 kg, 13.7 L) was added, and the mixture was first stirred
until a homogeneous solution was obtained (<10 min, endo-
thermic process) and then cooled to −5 °C. TMSN3 (1.785 kg,
0.95 equiv, Sigma Aldrich) was then added to the reaction
mixture over approximately 20−25 min in order to maintain
the temperature at 0
5 °C. An additional THF charge
(1.900 kg) was conducted to remove any residual TMSN3 from
the transfer line. The reaction was stirred at 0 5 °C for an
additional 2 h and then sampled for residual azide in solution
(<70 ppm of azide versus <200 ppm target). Water (11.030 kg)
and 1 M trimethylphosphine in THF (17.075 kg, Sigma
Aldrich) were then sequentially added to the reactor at such a
rate that the internal temperature remained 0−15 °C. Once the
phosphine had been charged, the transfer line was rinsed with
THF (1.15 kg) and the reaction mixture was stirred at 0−5 °C
for 1 h. An aliquot was then analyzed by HPLC to verify that
E
dx.doi.org/10.1021/op3002646 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
organic layer was analyzed for azide concentration (<10 ppm
azide observed with a target of <200 ppm). Aqueous citric acid
20% (2 equiv, 1.86 mol, 1.79 kg) was added with vigorous
stirring, maintaining the internal temperature within the range
10−20 °C. The mixture was stirred for 15 min, and then the
phases were allowed to split. A sample of the aqueous phase
was taken to check the pH (pH 3−4 was targeted). The reaction
mixture was held overnight. The rich aqueous layer was placed
into a clean 5 L reactor and treated with IPAc (4 L/kg of
meso-bis-ester, 1.2 L) and the resulting mixture was cooled to
0−5 °C. 10 N NaOH (5.5 equiv, 5.1 mol, 773 g, 512 mL) was
then slowly added at such a rate that the internal temperature
did not exceed 10 °C. The reaction mixture was warmed to
15−25 °C and held for at least 15 min with stirring, and then
the phases were allowed to split. The aqueous layer was
collected and was disposed as waste. The organic layer was then
washed with 10 wt % brine (4 L/kg of meso-bis-ester, 1.2 L).
The mixture was vigorously stirred for no less than 15 min, and
then the phases were allowed to split. The lean aqueous layer
was collected and was disposed of as waste. The rich IPAc
phase was then cooled to 0−5 °C. A solution of Boc2O (1 equiv,
930.6 mmol, 204 g) in IPAc (1−1.5 L/kg of Boc2O) was
prepared (endothermic). This solution was added to the reaction
mixture over approximately 1 h to control the carbon dioxide
evolution. The reaction mixture was stirred at 10 °C for 90 min,
and then a sample was collected and was analyzed by HPLC to
ensure the allylic amine had been efficiently converted to the
allylic carbamate. The reaction mixture was then warmed to
room temperature and was held overnight. A put and take
distillation strategy was implemented to remove IPAc and replace
it in solution with IPA for crystallization of the allylic carbamte.
Distillation was continued until an IPA/IPAc ratio of >95:5 was
obtained as measured by GC-FID. The concentration of the
product was adjusted with IPA to ∼200 mg/mL, and a seeded/
cooling crystallization using water as antisolvent was conducted.
The slurry was filtered and then washed with an IPA/H2O ratio
of 1/3. The wet cake was dried at 40−45 °C under vacuum.
244.79 g of product were obtained. 98% wt; >99.9 area %; 88%
corrected yield.
REFERENCES
■
(1) Organic Azides: Synthesis and Applications; Brase, S., Banert, K.,
Eds.; Wiley-VCH: Weinheim, 2010.
̈
(2) (a) Brase, S.; Gil, C.; Knepper, K.; Zimmermann, V. Angew.
̈
Chem., Int. Ed. 2005, 44, 5188−5240. (b) Scriven, E. F. V.; Turnbull,
K. Chem. Rev. 1988, 88, 297−368.
(3) For selected examples: (a) Urbanski, T. Chemistry and
Technologies of Explosives; Pergamon Press: Oxford, Great Britain,
1964; Vol. III; p 164. (b) Gray, P.; Waddington, T. C. Proc. R. Soc.,
London A 1957, 241, 110−121. (c) Egghart, H. C. Inorg. Chem. 1968,
7, 1225. (d) Tomlinson, W. R.; Ottoson, K. G.; Audrieth, L. F. J. Am.
Chem. Soc. 1949, 71, 375−376. (e) Talaward, M. B.; Agrawal, A. P.;
Anniyappan, M.; Wani, D. S.; Bansode, M. K.; Gore, G. M. J. Hazard.
Mater. 2006, 137, 1074−1078.
(4) (a) Hassner, A.; Stern, M.; Gottlieb, H. E.; Frolow, F. J. Org.
Chem. 1990, 55, 2304−2306. (b) Churchill, D. G. J. Chem. Educ. 2006,
83, 1798−1803. (c) Conrow, R. E.; Dean, W. D. Org. Process Res. Dev.
2008, 12, 1285−1286.
(5) The heuristic of violent decomposition expected for azido
compounds having a (C + O)/N ratio of <3 is typically used as a
qualitative measure of risk: Smith, P. A. S. The Chemistry of Open-
Chain Organic Nitrogen Compounds; W. A. Benjamin Inc.: New York,
USA, 1966; Vol. II, pp 211−256. Nonetheless, reactions involving
azide intermediates must be first conducted in small laboratory scale
and evaluated with specialized equipment before considering addi-
tional development. For a discussion: Keicher, T.; Lobbecke, S.
Organic Azides: Synthesis and Applications; Brase, S., Banert, K., Eds.;
̈
̈
Wiley-VCH: Weinheim, 2010; pp 3−27. See also: Tsuritani, T.;
Mizuno, H.; Nonoyama, N.; Kii, S.; Akao, A.; Sato, K.; Yasuda, N.;
Mase, T. Org. Process Res. Dev. 2009, 13, 1407−1412.
(6) Curtius, T. Ber. Dtsch. Chem. Ges. 1890, 23, 3023−3033.
(7) Bordwell, F. G. Acc. Chem. Res. 1988, 21, 456−463.
(8) Curtius, T.; Radenhausen, R. J. Prakt. Chem. 1891, 43, 207−208.
(9) (a) NIOSH (National Institute for Occupational Safety and
Health), International Chemical Safety Cards No. 09050, Recom-
mended Exposure Limits (REL), 1997. (b) Alben, J. O.; Fager, L. Y.
Biochemistry 1972, 11, 842−847. (c) Bormett, R. W.; Asher, S. A.;
Larkin, P. J.; Gustafson, W. G.; Ragunathan, N.; Freedman, T. B.;
Nafie, L. A.; Balasubramanian, S.; Boxer, S. G.; Yu, N.-T.; Gersonde,
K.; Noble, R. W.; Springer, B. A.; Sligar, S. G. J. Am. Chem. Soc. 1992,
114, 6864−6887. (d) Chang, S.; Lamm, S. H. Int. J. Toxicol. 2003, 22,
175−186. (e) Bowler, M. W.; Montgomery, M. G.; Leslie, A. G. W.;
Walker, J. E. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 8646−8649.
(10) Urben, P. Bretherick’s Handbook of Reactive Chemical Hazards,
6th ed.; Butterworth-Heinemann: Oxford, 1999; pp 1602.
ASSOCIATED CONTENT
■
(11) Betterton, E. A.; Robinson, J. L J. Air Waste Manage. Assoc. 1997,
47, 1216−1219.
S
* Supporting Information
Analytical methods and compound characterization data. This
material is available free of charge via the Internet at http://
pubs.acs.org.
(12) The use of microreactors can provide a safer alternative to batch
mode in processes where HN3 is a risk. For selected examples:
(a) Kopach, M. E.; Murray, M. M.; Braden, T. M.; Kobierski, M. E.;
Williams, O. L. Org. Process Res. Dev. 2009, 13, 152−160. (b) Gutmann,
B.; Roduit, J.-P.; Roberge, D.; Kappe, O. Angew. Chem., Int. Ed. 2010,
49, 7101−7105. (c) Palde, P. B.; Jamison, T. F. Angew. Chem., Int. Ed.
2011, 50, 3525−3528. (d) Gutmann, B.; Roduit, J.-P.; Roberge, D.;
Kappe, C. O. Chem.Eur. J. 2011, 17, 13146−13150. (e) Gutmann,
B.; Obermayer, D.; Roduit, R.; Roberge, D. M.; Kappe, C. O. J. Flow.
Chem. 2012, 1, 8−19.
AUTHOR INFORMATION
■
Corresponding Author
*E-mail: francisco.gonzalezbobes@bms.com; nathaniel.kopp@
bms.com.
Notes
(13) (a) Haese, J. Organic Azides: Synthesis and Applications; Brase, S.,
̈
The authors declare no competing financial interest.
Banert, K., Eds.; Wiley-VCH: Weinheim, 2010; pp 29−51.
(b) Alimardanov, A.; Nikitenko, A.; Connolly, T. J.; Feigelson, G.;
Chan, A. W.; Ding, Z.; Ghosh, M.; Shi, X.; Ren, J.; Hansen, E.; Farr,
R.; MacEwan, M.; Tadayon, S.; Springer, D. M.; Kreft, A. F.; Ho, D.
M.; Potoski, J. R. Org. Process Res. Dev. 2009, 13, 1161−1168.
(c) Connolly, T. J.; Hansen, E. C.; MacEwan, M. F. Org. Process Res.
Dev. 2010, 14, 466−469.
(14) (a) Wiss, J.; Fleury, C.; Onken, U. Org. Process Res. Dev. 2006,
10, 349−353. (b) Chen, C.-Y.; Frey, L. F.; Shultz, S.; Wallace, D. J.;
Marcantonio, K.; Payack, J. F.; Vazquez, E.; Springfield, S. A.; Zhou,
G.; Liu, P.; Kieczykowski, G. R.; Chen, A. M.; Phenix, B. D.; Singh, U.;
ACKNOWLEDGMENTS
■
The authors thank Professor Barry Trost and Drs. M. Eastgate,
T. La Cruz, C. Sfouggatakis, and B. Zheng for helpful
discussions during the development of this work; Drs. A.
Ramirez, J. Burt, A. Braem, and D. Conlon for helpful
discussions during the preparation of this manuscript; and
Drs R. Parsons, J. Tom, D. Kronenthal, and R. Waltermire for
support of this work.
F
dx.doi.org/10.1021/op3002646 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
Strine, J.; Izzo, B.; Krska, S. W. Org. Process Res. Dev. 2007, 11, 616−
623. (c) Wilson, R. D.; Cleator, E.; Ashwood, M. S.; Bio, M. M.;
Brands, K. M. J.; Davies, A. J.; Dolling, U.-H.; Emerson, K. M.; Gibb,
A. D.; Hands, D.; McKeown, A. E.; Oliver, S. F.; Reamer, R. A.; Sheen,
F. J.; Stewart, G. W.; Zhou, G. X. Org. Process Res. Dev. 2009, 13, 543−
547. (d) Wallace, D. J.; Baxter, C. A.; Brands, K. J. M.; Bremeyer, N.;
(29) The palladium-catalyzed desymmetrization−basification−Stau-
dinger sequence also worked well in 2-MeTHF. Toluene was chosen
due to its lower price.
(30) These results show that the sodium azide partitions exclusively
into the aqueous layer. In the unlikely scenario that higher levels were
observed after the first split, a simple recourse would be to conduct an
additional wash with 2 N NaOH.
́
Desmond, R.; Emerson, K. M.; Foley, J.; Fernandez, P.; Hu, W.; Keen,
S. P.; Mullens, P.; Muzzio, D.; Sajonz, P.; Tan, L.; Wilson, R. D.; Zhou,
G.; Zhou, G. Org. Process Res. Dev. 2011, 15, 831−840. (e) Girardin,
M.; Dolman, S. J.; Lauzon, S.; Oullet, S. G.; Hughes, G.; Fernandez, P.;
́
Zhou, G.; O’Shea, P. D. Org. Process Res. Dev. 2011, 15, 1073−1080.
(f) Scott, J. P.; Alam, M.; Bremeyer, N.; Goodyear, A.; Lam, T.;
Wilson, R. D.; Zhou, G. Org. Process Res. Dev. 2011, 15, 1116−1123.
(g) Zhou, G.; Emerson, K.; Majusiak, E.; Anderson, C.; Sudah, O. Org.
Process Res. Dev. 2012, 16, 204−213.
(15) (a) Peer, M. Specialty Chemicals 1998, 18, 256−263.
(b) Hagenbuch, J.-P. Chimia 2003, 57, 773−776.
(16) Trost, B. M.; Pulley, S. R. Tetrahedron Lett. 1995, 36, 8737−
8740.
(17) LEL is defined as the lowest concentration of a gas or a vapor in
air capable of producing a flash of fire in the presence of an ignition
source.
(18) Stoessel, F. Thermal Safety of Chemical Processes: Risk Assessment
and Process Design; Wiley-VCH: 2008; pp 66−67. ARC: Accelerating
Rate Calorimeter. C80: Setaram C80 Calvet Calorimeter. ADT 24:
Temperature at which time to maximum rate at adiabatic conditions is
24 h. AKTS: Advanced Kinetics and Technology Solutions
Thermokinetics Software.
(19) Low temperatures are also required in order to prevent a [3,3]-
sigmatropic rearrangement that results in formation of the 1,2-
disubstituted adduct. While this is more pronounced for the five-
membered ring series, Trost, B. M.; Stenkamp, D.; Pulley, S. R.
Chem.Eur. J. 1995, 1, 568−562 , we also observed it for the six-
membered ring upon holding the reaction mixture at room
temperature for extended periods of time.
(20) Alternative conditions for the reduction of the allylic azide were
evaluated including hydrogenation and thiol-based methods. Never-
theless, the Staudinger reaction proved superior in terms of rate,
overall impurity profile, and process control. Different phosphines can
be used in the Staudinger reaction: Gololobov, Y. G.; Zhmurova, I. N.;
Kasukiiin, I. F. Tetrahedron 1980, 37, 437−472. Attempts to replace
PMe3 have not been explored at this stage.
(21) IPAc was chosen as the solvent for the Boc-protection to
facilitate the solvent exchanged to IPA required for the crystallization
and to minimize unit operations and cycle time. Chen, C.-K.; Singh, A.
K. Org. Process Res. Dev. 2001, 5, 508−513.
(22) Hydrazoic acid has been reported to undergo Staudinger
reaction with phosphines and phosphites: (a) Staudinger, H.; Hauser,
E. Helv. Chim. Acta 1921, 4, 861. (b) Kabachnik, M. I.; Gilyarov, V. A.
Russ. Chem. Bull. 1961, 10, 816−818.
(23) The Lower Explosive Limit (LEL) is often cited as >17 wt %
HN3 for aqueous solutions and >8−15 vol % HN3 for the vapor phase.
(a) Ertel, D.; Schmieder, A.; Stollenwerk, H. Nukl. Ensorgung 1989, 4,
107−119. (b) Ullmann’s Encyclopedia of Industrial Chemistry; VCH
Verlag: Weinheim, 1989; Vol. A13, pp 193−197. (c) Rozenberg, A. S.;
Arsen’ev, Y.; Voronkov, V. G. Combust. Explos. Shock Waves 1973,
271−277. (d) Wiss, J.; Fleury, C.; Heuberger, C.; Onken, U.; Glor, M.
Org. Process Res. Dev. 2007, 11, 1096−1103.
(24) (a) Templeton, J. C.; King, E. L. J. Am. Chem. Soc. 1971, 93,
7160−7166. (b) Wilhelm, E.; Battino, R.; Wilcock, R. J. Chem. Rev.
1977, 77, 219−262.
(25) For details on how we interpolated the data from the PUREX
process, see the Supporting Information.
(26) We developed contingency plans to be used in the case that the
IPC was not met when working on scale. For a discussion, see the
Supporting Information.
(27) The use of nonazide nucleophiles was also considered. A
discussion on this topic is beyond the scope of this manuscript.
(28) A target pH of ≥12 was set for the aqueous layer.
G
dx.doi.org/10.1021/op3002646 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX