Execution of a performic acid oxidation on multikilogram scale

Article abstract of DOI:10.1021/op700039r

In order to run a peracid oxidation on multikilogram scale in a pilot plant facility with an acceptable level of risk, an extensive study of the thermal characteristics of the reaction, reactants, and quench was undertaken. A number of features were engineered into the process in order to confidently run the process without the potential for a thermal or pressure excursion on scale.

Full text of DOI:10.1021/op700039r

Organic Process Research & Development 2007, 11, 762−765
Communications to the Editor
Execution of a Performic Acid Oxidation on Multikilogram Scale
David H. Brown Ripin,* Gerald A. Weisenburger,* David J. am Ende, David R. Bill, Pamela J. Clifford,
Clifford N. Meltz, and James E. Phillips
Chemical Research and DeVelopment, Pfizer Global Research DiVision, Pfizer Inc., Eastern Point Road, Groton,
Connecticut 06340, U.S.A.
Abstract:
2). Either product could then be oxidized under a variety of
conditions, including some metal-ion-free conditions such
as performic or peracetic acid in formic or acetic acid.2 Under
the acidic conditions, no N-oxide formation was observed.
The crystallinity of the thioacetate derivative in conjunction
with the higher atom efficiency and lower cost as compared
with the trityl protecting group led us to focus our attention
on the thioacetate route.
In order to run a peracid oxidation on multikilogram scale in
a pilot plant facility with an acceptable level of risk, an extensive
study of the thermal characteristics of the reaction, reactants,
and quench was undertaken. A number of features were
engineered into the process in order to confidently run the
process without the potential for a thermal or pressure excur-
sion on scale.
The success of the new route hinged upon the ability to
cleanly oxidize thioacetate 9 and quench the reaction in the
absence of any metal-ion-containing salts. This could be
accomplished using a performic acid or peracetic acid
oxidation followed by quenching of the excess peroxide with
activated carbon (Darco G-60).3 A number of issues were
identified that needed to be addressed in order to confidently
run this process on scale. First and foremost, the thermal
stability of the peracid reagent and the heat output from the
reaction itself required study. Further, the quench of the
reaction generates heat and oxygen gas, a situation requiring
control in a bulk manufacturing facility. Finally, in the case
of the performic acid oxidation, the stability of formic acid
during the subsequent distillation required investigation.
Thermal Potential of the Oxidants. Performic acid and
peracetic acid are generated in situ by the addition of
hydrogen peroxide to the corresponding carboxylic acid.4 A
solution of performic acid was made by adding 1.0 volume
of 30 wt % hydrogen peroxide to 2.5 volumes of 96 wt %
formic acid. The performic acid solution and a commercial
solution of 32 wt % peracetic acid in acetic acid were tested
by differential scanning calorimetry (DSC) which indicated
that these solutions decompose at relatively low temperatures
with the generation of a large amount of heat. DSC screening
of 8, 9, and 96 wt % formic acid did not reveal any
significant thermal concerns for this process.
Introduction
CP-865,569 (8) is a CCR1 antagonist. The candidate
contains a tertiary amine and the unusual benzylic sulfonic
acid functionalities and exists as a zwitterion as its neutral
species. The synthesis of this candidate involved a number
of synthetic and processing challenges due to the interesting
functionality present and the physical properties thereof.
Herein is described execution of the peracid oxidation of
thioacetate (9) to CP-865,569 on multikilogram scale.
Caution: The processes described herein haVe the
potential to release large amounts of energy and gas in an
uncontrolled fashion. InVestigators considering running a
large-scale oxidation reaction should consult the literature,
run appropriate safety tests, and take proper precautions
when running the reaction.
Why Use a Peracid Oxidation?
The original synthesis of the candidate installed the
sulfonic acid functionality utilizing a sulfite displacement1
(Scheme 1). While the installation of the sulfonic acid
proceeded in reasonable selectivity relative to sulfite ester
formation (S vs O alkylation ratio 9:1), the primary problem
with this route was the extreme difficulty encountered when
trying to remove sodium from the material prior to forming
the desired salt. For this reason, a method which introduced
the sulfonic acid at the end of the synthesis in the absence
of metal counterions was sought.
Accelerating rate calorimetry (ARC) was conducted to
accurately determine the onset temperature and to investigate
off-gassing. Analysis of the performic acid solution by ARC,
Identification of a New Method for the Introduction
of the Sulfonic Acid. The chloride could be very cleanly
displaced by tritylmercaptan or potassium thioacetate (Scheme
(2) Oxidation of thioacetates to sulfonic acids: Koenig, N. H.; Swern, D. J.
Am. Chem. Soc. 1957, 79, 4235.
(3) Residual metals in the activated carbon are postulated to be the active catalyst
in this reaction. No lot-to-lot variability was observed in the effectiveness
of the quench, either between different lots of the same grade of carbon, or
other types of carbon (KBB was also tried). Each lot was use-tested prior
to use on scale.
* Authors to whom correspondence should be addressed. G.A.W.: E-mail:
gerald.a.weisenburger@pfizer.com; D.B.R.: E-mail: david.b.ripin@pfizer.com.
(1) For a full synthetic report, see: Belecki, K.; Berliner, M.; Bibart, R. T.:
Meltz, C.; Ng, K.; Phillips, J.; Brown Ripin D. H. Org. Process Res. DeV.
2007, 11, 754-761.
(4) For a review of organic peracids see: Swern, D. Chem. ReV. 1949, 45, 1.
762
•
Vol. 11, No. 4, 2007 / Organic Process Research & Development
10.1021/op700039r CCC: $37.00 © 2007 American Chemical Society
Published on Web 06/15/2007

Scheme 1. Original synthesis of CP-865,569 using a sulfite displacementa
a Reagents and conditions: a) Na2SO3, MeCN, b) HCl, MeCN, i-PrOH.
Scheme 2. Installation of the sulfonic acid functionality of CP-865,569 via oxidation of thioacetate 9a
a Reagents and conditions: a) KSAc, PhMe, b) HCO2H, 30% H2O2 in H2O or 30% AcOOH in AcOH.
is clear from these results that accumulation of performic
acid could lead to severe consequences.
Investigation of a commercial solution of 32 wt %
peracetic acid in acetic acid by ARC revealed two exotherms.
The first exotherm occurred at a φ-corrected onset temper-
ature of 30 °C with an ATR of 358 °C, a maximum
temperature rate of 2000 °C/min, and a 3.5 h TMR at 30
°C.6 A maximum pressure rate of 1479 psi/min and a residual
pressure of 400 psia upon cool down to 20.8 °C were
observed. The experiment terminated during the second
exotherm because the 300 °C limit was reached.
Oxidation with peracetic acid would have been the
preferred method for a number of reasons: peracetic acid is
more stable than performic acid,7 the oxidation of 9 proceeds
at a lower temperature and is therefore dose controlled at a
lower temperature, and the isolation does not involve the
distillation of formic acid, which poses a possible hazard if
purified to a high grade during distillation. Unfortunately,
peracetic acid does not form as readily as performic acid on
addition of hydrogen peroxide to acetic acid.8 As a result,
in order to run a peracetic acid oxidation, a preformed
solution of peracetic acid would have to be purchased and
stored in bulk quantities. Alternatively, addition of acetic
anhydride to an acetic acid solution of 9 followed by addition
of hydrogen peroxide did give facile oxidation at 0 °C.
However, peracetic acid generated under anhydrous condi-
tions has been reported to explode.9 Therefore, we chose to
oxidize multikilogram quantities of 9 with performic acid
generated in situ.
Figure 1. Accelerating rate calorimetry (ARC) of 1.0 vol. 30
wt % H2O2 in 2.5 vol. formic acid (14.6 wt % performic acid
assuming 100% conversion); 5 °C , O-corrected onset temper-
ature, 179 °C adiabatic temperature rise (ATR), 252 °C/min
maximum temperature rate, 1972 psi/min maximum pressure
rate.
Figure 1, revealed a φ-corrected onset temperature of 5 °C.5
The exotherm was severe and rapid, resulting in a φ-corrected
adiabatic temperature rise (ATR) of 179 °C and a maximum
self-heat rate of 252 °C/min. The time to maximum rate
(TMR) was 30 min at 25 °C under these experimental
conditions. The TMR is the time required to reach the
maximum rate from the initial time of onset. As a general
rule, a TMR of greater than 24 h at the processing
temperature is considered to be acceptable in a manufacturing
facility, providing sufficient time for corrective action if
cooling, stirring, or other critical utilities are unexpectedly
interrupted. In addition to the thermal event, a maximum
pressure rise of 1972 psi/min was observed. The decomposi-
tion generated a non-condensable gas as indicated by a
residual pressure of 861 psia upon cool down to 22.7 °C. It
RC1 Experiments. RC1 calorimetry was performed to
determine: (1) The heat of the overall oxidation process,
which includes (a) the heat of reaction of formic acid with
H2O2 to give performic acid and (b) the oxidation of 9 by
performic acid to give 8. (2) The rate at which the heat is
(6) Experiment was carried out using a titanium bomb and a φ-factor of 2.42.
A corrected onset of 41 °C was observed when the experiment was repeated
in a glass bomb with a metal stem and a φ-factor of 4.43. The MSDS from
FMC reports: SADT > 55 °C for a 55-gal drum (35 wt % peracetic acid
solution in acetic acid).
(7) This is not a direct comparison since we did not test a 32 wt % solution of
performic acid.
(8) When the process was repeated using acetic acid instead of formic acid, no
product was observed at the end of the H2O2 addition. These relative rates
of peracid formation have been noted in the literature, see ref 4.
(9) Urben, P.G., Ed. Brethericks Handbook of ReactiVe Chemical Hazards, 6th
ed.; Butterworth-Heinemann: Woburn, MA, 1999; p 1627.
(5) Experiment was carried out using a titanium bomb and a φ-factor of 1.69.
A corrected onset of -1 °C was observed when the experiment was repeated
in a glass bomb with a metal stem and a φ-factor of 3.62.
Vol. 11, No. 4, 2007 / Organic Process Research & Development
•
763

testing of the vessel, agitator, baffle, etc. was performed to
ensure the integrity of the glass lining. In addition, careful
consideration was given to all parts that were to contact the
reaction mixture to ensure compatibility with peroxide
species. Passivation of the equipment with 35 wt % HNO3
was considered but deemed unnecessary after discussions
with the H2O2 manufacturer. Careful cleaning of the reaction
vessel was also performed to reduce the risk of rapid
decomposition caused by contaminants.
H2O2 Charge. Given the low onset temperature and high
decomposition energy of performic acid, engineering controls
were implemented to prevent accumulation. Vent-sizing
studies showed that a rapid addition of the 30 wt % H2O2
would result in a runaway reaction producing a large amount
of heat and pressure, highlighting the importance of a
controlled addition. In order to eliminate the potential for a
rapid addition, the 30 wt % H2O2 charge was subdivided
into nine portions. Given the 172 °C ATR, subdivision of
the H2O2 charge into nine portions should provide an
acceptable safety margin, since in the worst-case, a 44 °C
reaction temperature would result from rapid addition of one
subdivision. In order to confirm the validity of this assump-
tion, the vent-sizing experiment was repeated using one-ninth
of the H2O2 charge, resulting in small increases in both
temperature and pressure.
During the scale-up runs, the reaction was monitored by
in situ IR and grab sample HPLC assay to confirm progres-
sion of the reaction and consumption of the peroxide. ARC
analysis of the crude reaction mixture prior to quench
revealed a maximum self-heating rate of about 0.1 °C/min,
a minimal thermal hazard. Stirring and cooling was continued
after reaction completion to remove any heat resulting from
decomposition of the crude reaction mixture.
Quench. The oxygen off-gassing posed both a foaming
and explosion potential.12 A rapid nitrogen purge was
maintained throughout the quench in order to prevent the
oxygen level from rising above the LOC of formic acid. G-60
Darco was used at 5 wt % relative to 9. An increase in the
amount of carbon used in the quench led to more vigorous
gas generation and a significant decrease in isolated yield
of product. Quench completion was monitored with semi-
quantitative peroxide test strips, which indicated a final
peroxide concentration of less than 1 ppm. Prior to proceed-
ing in our pilot plant, an ARC test was run on the quenched
solution to confirm no exotherms would be observed during
concentration. We considered using small tubing to transfer
the crude reaction mixture to the Darco slurry to prevent
rapid addition, but felt that it was not necessary due to careful
operator training, the relatively low heat flow, and the
relatively low rate of oxygen generation.
Figure 2. RC1 data for a 3-h addition of 100 g of 30 wt %
H2O2 to a solution of 99.7 g of 9 in 305 g (250 mL) of 96 wt %
formic acid. ∆Hrxn ) -975 kJ/mol of 9, ATR ) 172 °C, 9%
thermal accumulation.
released throughout the process. (3) The amount of thermal
accumulation.
The oxidation was carried out by charging 4.2 equiv of
30 wt % H2O2 over 3 h to a solution of 9 in 2.5 volumes of
formic acid maintained at 25 °C. The heat flow data is shown
in Figure 2. The heat generated by the overall process, -975
kJ/mol of 9, was determined by integrating the area under
the heat-flow curve.10 The ATR was calculated to be 172
°C using the integrated heat, final reaction mass, and final
heat capacity, 2.33 J/g °C, of the reaction mass. The
maximum heat flow, -44 W/kg of final reaction mixture,
would challenge the cooling capacity of the scale-up equip-
ment since we planned to maintain the batch temperature
throughout the addition. Since the reaction heat was nearly
dose controlled, with 9% thermal accumulation for the 3 h
addition, the maximum heat flow could be easily reduced
by slowing the addition rate.11
The reaction quench was also investigated using the RC1.
The crude reaction mixture was charged over 1 h to a slurry
of G-60 Darco maintained at 25 °C. The heat for this process
was determined to be -135 kJ/mol of 9, resulting in a 17
°C ATR. The maximum heat flow was -37 W/kg of reaction
mixture with 40% thermal accumulation for the 1 h addition.
Off-gassing was observed at a maximum rate of 40 L h-1
kg-1 of H2O2. This gas could easily be vented without
pressurization of the scale-up equipment. With this data in
hand, engineering controls for the scale-up were proposed
and implemented.
Isolation of the Product. Azeotropic drying with acetic
acid under reduced pressure, followed by addition of
methanol and precipitation with isopropyl ether provided
Pilot-Plant Preparation
Safety Considerations and Engineering Controls. Tank
Preparation. Trace metal impurties can catalyze the decom-
position of H2O2 into O2 and H2O and potentially lead to a
foaming or over-pressurization incident. Therefore, spark
(11) Note that the percent accumulation was calculated based on the heat evolved
after 4.2 equiv of H2O2 were added. Theoretically, only 3 equiv of H2O2
are needed for conversion of the thioacetate to the sulfonic acid. An
accumulation of 34% was calculated after 3 equiv of H2O2 were added in
the RC1 experiment.
(12) The limiting oxygen concentration of 96 wt % formic acid was determined
to be 7.26-7.72% by Chilworth Technology.
(10) Integration was performed on a curve plotted as W versus seconds.
764
•
Vol. 11, No. 4, 2007 / Organic Process Research & Development

solid zwitterion isolated by filtration. High purity formic acid
is known to liberate carbon monoxide upon heating.13
Therefore, acetic acid was added to the quenched reaction
mixture prior to distillation. DSC and ARC analysis of the
pot residue from a laboratory pilot at the start, midpoint,
and end of the distillation did not reveal any significant
exotherms or pressure events.
Summary of Pilot Plant Runs and Results. In execution,
five runs of this process were performed in our pilot plant
on 30-35-kg scale without incident, and produced product
in high purity and >90% yield.14 The oxidation reaction was
run in a 300-gal tank with a maximum operating volume of
100 gal to provide a large headspace as an extra layer of
protection. Likewise, the quench was executed in a 500-gal
tank to provide a large headspace.
1-((2R,5S)-(4-(4-fluorobenzyl)-2,5-dimethylpiperazin-1-yl)-
2-(4-chloro-2-(acetoxythiomethyl)phenoxy)ethanone (9) (34.2
kg,71.3 mol) and 96% formic acid (7.20 L/kg ethanethioate).
To the formic acid solution at 20 °C is added one subdivision
of 30 wt % hydrogen peroxide (3.40 kg, 3.77 L, 0.42 equiv).
The exotherm is noted, and the temperature rise is recorded.
The reaction is stirred at 20 °C for 2 h followed by the
addition of a second aliquot of 30 wt % hydrogen peroxide
(3.40 kg, 3.77 L, 0.42 equiv). Again, the exotherm is noted,
and the temperature rise is recorded. The reaction is stirred
at 20 °C for 2 h. The reaction mixture is sampled and
analyzed by HPLC to monitor starting material consumption
and sulfonic acid formation. (A lack of starting material
consumption and sulfonic acid formation implies a buildup
of hydrogen peroxide, and subsequent additions should be
discontinued.) Upon determination of safe hydrogen peroxide
consumption, the reaction mixture is treated with seven
further subdivisions of 30 wt % hydrogen peroxide (each
subdivision: 3.40 kg, 3.77 L, 0.42 equiv) with 2-h reaction
times at 20 °C in between each subdivision. Samples for
HPLC analysis are taken following subdivision 5 and 8 to
confirm peroxide consumption. Upon complete addition of
hydrogen peroxide, the reaction mixture is added to a
suspension of 96% formic acid (39 L) and Darco G-60 (1.70
kg) at 20 °C under a high-volume nitrogen sweep over 2 h.
A gentle gas evolution is noted and confirms a controlled
consumption of excess hydrogen peroxide. The resulting
suspension is stirred at 20 °C for 12 h, and a reaction sample
is analyzed by accelerating rate calorimetry (ARC) to ensure
complete peroxide consumption. The quenched reaction is
filtered through Celite, and the filtrate is diluted with glacial
acetic acid (316 L). The resulting solution is concentrated
under reduced pressure to a volume of approximately 30 L
with a maximum pot temperature of 50 °C. The desired 8 is
diluted with methanol (140 L) and added to isopropyl ether
(1400 L) at 20 °C over 1 h. The resulting slurry is stirred at
20 °C for 2 h followed by filtration, washed with isopropyl
ether (35 L), and vacuum dried at 50 °C to afford 8 in 91%
yield.
Experimental Section
DSC experiments were performed using a Mettler Toledo
822e instrument with a 20-min isothermal hold at 30 °C
followed by a 4 °C/min scan rate to 300 °C in 40-µL, high-
pressure (150 bar max), sealed, gold plated steel pans under
an air atmosphere. ARC experiments were performed under
an atmosphere of air using a clamshell style ARC from TIAX
(Acorn Park, Cambridge, MA) with either a titanium or glass
bomb with a 1/8 in. neck. Heat-wait-search experiments
were perfomed starting at 30 °C in 10 °C steps with a
sensitivity of 0.02 °C/min. Reaction heats were measured
using a Mettler Toledo Mid-Temp RC1e with a 1-L MP06
reactor, Hastelloy C head, Hastelloy C anchor stirrer under
a small positive pressure of nitrogen. The reactor was
passivated with 35 wt % HNO3 at 50 °C for 1 h followed
by 3 × 1 L water rinses. The reactor was rinsed with acetone
and blown dry with nitrogen overnight. Linear doses were
performed using an integrated prominent pump and balance.
The reaction temperature was maintained at 25 °C by
allowing the jacket temperature to fluctuate. QuickCal
calibrations were performed before additions and after
reaction completion to obtain heat capacities and heat transfer
coefficients. The heat flow curve was analyzed using WinRC
NT version 7.11 (SR-6) with the following equation:
Acknowledgment
Q ) Qaccum + Qflow + Qdose
We thank Professor Steven Ley and Drs. Robert Dugger,
Tamim Braish, Stephane Caron, and John Ragan for helpful
discussions. We also thank Kyle Leeman and Natalie Harner
for analytical assistance. Jeff Adler, Paul Allard, Joseph
Bellavance, Todd Carden, Wayne Crandall, Douglas Davis,
Thomas Limani, Larry Lumbert, Brian Morgan, Alden
Peckham, Steven Stott, Ralph Scheck, III, Scott Schelkly,
and Walden Smolen are thanked for operating the chemistry
in pilot-plant equipment.
Heat transfer coefficients were interpolated proportional to
virtual volume, heat capacities were interpolated proportional
to the reaction mass, and the baseline was interpolated
proportional to conversion.
(2-(2-((2R,5S)-4-(4-Fluorobenzyl)-2,5-dimethylpiperazin-
1-yl)-2-oxoethoxy)-5-chlorophenyl)methanesulfonic Acid
(8). To a nitrogen-purged reactor at 20 °C is added solid
(13) (a) For details on formic acid, see: http://www.basf.com. (b) Reference 9,
p 163.
(14) The process as performed on scale used 7.2 volumes of 96 wt % formic
acid instead of 2.5 volumes as in the RC1 run.
Received for review February 14, 2007.
OP700039R
Vol. 11, No. 4, 2007 / Organic Process Research & Development
•
765