Development of a safe and practical N-oxidation procedure using m-CPBA in acetic acid

Article abstract of DOI:10.1021/op100280c

A safe and practical procedure for the N-oxidation of pyrazolopyridine 4 using meta-chloroperbenzoic acid (m-CPBA) in acetic acid is described. Key safety experiments are outlined which have led to the development and implementation of this reaction on 28-kg scale in 98% isolated yield.

Full text of DOI:10.1021/op100280c

COMMUNICATION
pubs.acs.org/OPRD
Development of a Safe and Practical N-Oxidation Procedure Using
m-CPBA in Acetic Acid
Mahbub Alam,*,† Adrian Goodyear,† Jeremy P. Scott,† and Thomas P. Vickery‡
†Global Process Chemistry, Merck Sharp & Dohme Research Laboratories, Hertford Road, Hoddesdon EN11 9BU, Hertfordshire, U.K.
‡Chemical and Engineering Research and Development, Merck Sharp & Dohme Corporation, Rahway, New Jersey 07065, United States
rate of reaction leading to incomplete conversion of starting
ABSTRACT: A safe and practical procedure for the N-
oxidation of pyrazolopyridine 4 using meta-chloroperben-
zoic acid (m-CPBA) in acetic acid is described. Key safety
experiments are outlined which have led to the development
and implementation of this reaction on 28-kg scale in 98%
isolated yield.
material which, when coupled with the inherently poor physical
properties of 3 and 4 leading to very thick slurries, made the
reaction a concern for scale-up. An urgent program need for
multikilogram quantities of 1, combined with the challenging
conversion of 4 to 3 and the safety concerns of using m-CPBA on
scale, consequently necessitated the rapid development of a safe
and scaleable protocol to prepare the key N-oxide intermediate 3
(Scheme 1).
’ INTRODUCTION
’ RESULTS AND DISCUSSION
During the course of a program to develop orally active non-
nucleoside reverse transcriptase inhibitors (NNRTIs) against
HIV reverse-transciptase mutation, compound 1 was brought
forward for clinical development.1 Amongst the key structural
features of 1 is the pyrazolo-[3,4-β]-pyridine fragment. In particular,
the 2-amino substituent on the pyrazolo-[3,4-β]-pyridine pre-
sented a considerable synthetic challenge that was not initially
obvious from its apparently simple structure. The synthesis of 1
had evolved from its predecessor in clinical development, which
did not contain the 2-amino substituent (Scheme 1).2 Thus, the
installation of the amino group was developed from a recent
report which demonstrated that the reaction of pyridine
N-oxides with tosic anhydride (Ts2O) in the presence of tert-
butylamine, followed by cleavage of the tert-butyl group with
trifluoroacetic acid as an attractive method for the preparation of
2-aminopyridines from the parent unsubstituted pyridine.3 In
this case, the pyrazole nitrogen atom also required protection due
to competitive tosylation; thus, 2 was the target from N-oxide 3.
In turn, the preparation of the key N-oxide intermediate 3 was
required from pyrazolopyridine 4 using m-CPBA, a common
reagent in organic synthesis used for heteroatom oxidations and
epoxidations, but with safety concerns when used on scale.
m-CPBA has been used frequently over the years with many
examples of its use on pilot scale and pharmaceutical
manufacturing.4 However, the safety concerns with its use on
scale are well-known, with the pure solid being shock-sensitive
and potentially explosive in the condensed phase.5,6 The com-
mercial grade (70-77 wt %) although somewhat stabilised with
chlorobenzoic acid and water still represents a significant con-
cern when used on scale. The use of this reagent has been
reported in many solvents, but Bretherick’s Handbook of Reactive
Chemical Hazards notes some incompatibility with certain solvents
such as DMF and DMSO, with storage also recommended at
4 °C. The reagent’s stability in solution has also been observed to
be well below that of the solid form.7
The initial conditions used for the preparation of N-oxide 3
employed m-CPBA in ethyl acetate (EtOAc) at 55 °C. The
protocol consisted of the addition of 1.7 equiv of solid
m-CPBA in multiple portions to a thick and inhomogeneous
slurry of 4 in EtOAc at 55 °C. The limited solubility of starting
material 4 and product 3 in EtOAc led to a poorly controlled
slurry to slurry conversion, culminating in incomplete conversion
(∼ 90%). This was reasoned to be due to entrainment of the
starting material in the N-oxide product. In addition, the thick
product slurry had very poor filtration properties, which sign-
icantly slowed the filtration and would have been impractical on
scale. Also, the excess m-CPBA reagent was not quenched at the
end of the reaction and was of concern.
Reaction Development. The first objective was to increase
the conversion of 4 to 3 above 90% to positively impact the
productivity through to 1 and simultaneously improve the safety
profile of the reaction. As already noted, charging extra m-CPBA
reagent did not increase conversion beyond 90%. This was also
undesirable from a safety standpoint due to the inherent lack
of control when adding an unstable solid oxidant to a thick,
inhomogeneous slurry, thus giving the potential for accumula-
tion of oxidant in the reaction vessel.
Initial attempts to increase the conversion started with adding
m-CPBA slowly as a solution in EtOAc to the pyrazolopyridine
substrate 4 in EtOAc, however, these were not successful. Similarly,
increasing the dilution of the reaction mixture up to 35 mL/g in
EtOAc to aid mobility in the reaction system did not prove fruitful.
Addition of cosolvents (ethanol (EtOH) and H2O) also failed to
facilitate an improvement in the conversion beyond 90%.
Next, a solubility screen was undertaken with both the starting
material 4 and N-oxide 3 (Table 1) to assess the potential for
better physical properties in the reaction and thus, push the
reaction to completion with the added benefit of improving the
safety profile. Most of the solvents in the screen unfortunately did
Initial experimentation showed that the oxidation of pyrazo-
lopyridine 4 to N-oxide 3 (Figure 1) was difficult due to a slow
Received: October 19, 2010
Published: January 07, 2011
r
2011 American Chemical Society
443
dx.doi.org/10.1021/op100280c Org. Process Res. Dev. 2011, 15, 443–448
|

Organic Process Research & Development
COMMUNICATION
not offer better solubility of either 4 or N-oxide 3. Dimethyla-
cetamide (DMAc) was also explored as a solvent (despite the
known concerns over the compatibility of m-CPBA and di-
methylformamide (DMF)5) as 4 and 3 were found to be fully
soluble at 10 mL/g. However, the reaction still disappointingly
only proceeded to 80% conversion after 5 h at 55 °C and resulted
in a significantly worse purity profile.
As part of the safety testing, the mother liquors from the
crystallization were shown to contain residual m-CPBA. As a
result, 13 mL of liquors from a laboratory experiment were
shown an 18 mL Z-3 cell to pressurise to 1 psi over 23 h at 35 °C
(vide infra). Therefore, to avoid pressurisation of waste drums
of liquors, a sodium bisulfite quench was rapidly developed.
Accordingly, 0.7 equiv of sodium bisulfite dissolved in H2O
(2 mL/g) was added directly to the reaction mixture at the end of
reaction, after which testing with starch strips showed peroxide
levels were below detectable limits. After addition of the sodium
bisulfite solution, a further addition of H2O (19.5 mL/g) was
made, and this reliably resulted in agglomerated crystals of
N-oxide 3 with an enhanced filtration rate.
Safety Testing. Given the known safety issues with using m-
CPBA and the introduction of an uncommon solvent (AcOH)
for the reaction, an extensive range of safety experiments were
conducted to aid the development of a safe process amenable for
a multikilogram scale reaction. Differential scanning calorimetry
(DSC) was used to evaluate exothermic decompositions, advanced
reaction system screening tool (ARSST) to characterize exother-
mic decomposition rates as a function of temperature, and accel-
erating rate calorimetry (ARC) to evaluate reaction mixture sample
Whilst the solubility of 4 and 3 was poor in most common
solvents, solubility measurements in acetic acid (AcOH) indicated
its potential use as a solvent, and pleasingly, with 1.6 equiv of
m-CPBA at 55 °C in AcOH, > 99% conversion to N-oxide 3 was
achieved. This was the first time this difficult reaction had been
observed to reach full conversion, and with the starting material
and product being more soluble at the start and end of the
reaction, it gave much better potential for the development of
a safer protocol to be used on scale. Further optimisation of
this procedure resulted in the addition of 1.6 equiv of m-CPBA
as a solution in AcOH (7 mL/g) to a thin slurry of 4 at 55 °C
over 45 min to 1 h to maintain the reaction temperature.
Typically, conversion at the end of the addition of the
m-CPBA solution was 50-60%. By the end of the addition,
the reaction mixture was completely homogeneous, which was
a major improvement for the physical properties of the
reaction, and as a result, after 3 h at 55 °C, the reaction
reached completion (>99%). This protocol also offered good
control of the reaction through limiting the potential for
accumulation of oxidant.
Quench/Crystallization. With the reaction in satisfactory
condition, isolation of the N-oxide product 3 now required some
development. Variation was observed at the end of the reaction
where the N-oxide product crystallized from some reaction
mixtures after overnight ageing at 55 °C, whereas from other
reactions, the N-oxide did not crystallize even upon cooling to
ambient temperature and addition of seed. In addition, the
spontaneous crystallization of 3 afforded a thick slurry with poor
filtration properties and long time cycles due to the crystal
morphology (long, hair-like needles). Where no crystallization
was observed, up to 17-18 mL/g of H2O was required as an
antisolvent which gave better crystal morphology and a more
filterable slurry, and this became the preferred isolation protocol.
Further addition of H2O to reduce mother liquor losses resulted in
entrainment of 3-chlorobenzoic acid which could not be removed
by the usual washing of the wetcake with EtOH/H2O.
Figure 1. N-oxidation of pyrazolopyridine 1.
Table 1. Solubility data for 3 and 4 at 20 °C
solubility of SM,
solubility of N-oxide,
3 (mg/mL)
solvent
4 (mg/mL)
ethyl acetate
ethanol
10.7
2.5
<0.1
<0.1
<0.1
<0.1
<0.1
7.5
methanol
3.9
dichloromethane
acetonitrile
5.4
2.1
acetic acid
19.4
>100
dimethylacetamide
>100
Scheme 1
444
dx.doi.org/10.1021/op100280c |Org. Process Res. Dev. 2011, 15, 443–448

Organic Process Research & Development
COMMUNICATION
Figure 2. DSC of m-CPBA/AcOH solution.
Figure 3. ARSST data: self-heat rate of m-CPBA/AcOH solution.
kinetics. Also, the chemical reaction calorimeter (CRC) was used to
determine reaction stability and exothermic activity.
m-CPBA Solution Preparation. As the oxidant was the main
source of concern for the process, a significant number of
experiments were performed to determine the stability of the
m-CPBA in AcOH.
showed the total pressurization of the solution to be well within
the venting capacity of the reaction vessel (150 L/min). Similarly, an
ARSST experiment of the m-CPBA in AcOH showed the total
pressurization from heating to decomposition to be within the
venting capability of the reaction vessel (maximum pressure
increase of 6.6 psi) (Figures 3 and 4).
DSC analysis of m-CPBA in acetic acid showed a moderate
exothermic decomposition of 14.5 cal/g with an onset at 105 °C
(Figure 2). As this was well above the operating temperature
(ambient) for handling of this solution, preparation of this
solution was not a major concern; however, care was still required
when warming up the solution for dissolution (see below).
Furthermore, an isothermal age experiment in the CRC of this
m-CPBA solution at 55 °C (the reaction operating temperature)
With these observations, it was found that the dissolution of
m-CPBA in AcOH was endothermic, with the temperature dropping
between 9 and 11 °C. As the m-CPBA in AcOH was found to be
thermally unstable, albeit at elevated temperature, care was required
in warming it up to complete the dissolution. Overheating the
solution risked lowering the assay of m-CPBA or triggering an exo-
thermic decomposition; therefore, ∼25 °C was maintained in the
preparation vessel to prevent overheating of the m-CPBA solution.
445
dx.doi.org/10.1021/op100280c |Org. Process Res. Dev. 2011, 15, 443–448

Organic Process Research & Development
COMMUNICATION
Figure 4. ARSST data: pressurisation rate of m-CPBA/AcOH solution.
Figure 5. CRC of mother liquors without quenching at end of reaction.
In addition, from further testing it was calculated that the
vessel vent was required to handle a gas flow of 0.47 L/h/kg of
solution to provide emergency relief for the feed vessel in the case
of a runaway decomposition. This was based on an EAC
experiment which showed that the solution could run away if
cooling was lost.8 In this experiment, 4.24 g of m-CPBA in AcOH
solution (prepared by dissolving 5 g of m-CPBA in 41 mL of
AcOH) was loaded into a SS ARC cell (∼10 mL volume; 12.7 g)
and heated isothermally at 25 °C for ∼15 h. The pressure slowly
increased from 16.5 to 25.2 psi, which indicated the gas genera-
tion was from the decomposition of the m-CPBA. Next, the EAC
was placed into adiabatic mode and a 5 °C step heat-wait-
search experiment was run.9 The sample began to exotherm
at a self-heat rate of ∼0.02 °C/min at 40 °C which continued to
∼53 °C or a temperature rise of 13 °C. Correcting for a φ of 1.6,
the adiabatic temperature rise was 20 °C. The maximum self-heat
and pressure rise rates were 0.07 °C/min and 0.09 psi/min,
respectively. Upon further heating to 78 °C the pressure was
55 psi, and no additional exothermic activity due to reaction
was observed. Upon cool down to 20 °C, the pressure was
38 psi. A sample of this headspace gas was analysed using FT-
IR and indicated the presence of CO2 (O2 not detected by this
method).
For a controlled procedure, the m-CPBA solution had to be
charged at a rate to maintain 55 °C ( 5 °C, and the venting
capacity of the reaction vessel was required to be at least
0.30 L/h/kg of starting material.
Oxidation Reaction. CRC analysis of the reaction showed
that gas was slowly evolved during the reaction (rate of 0.30 L/h/
kg substrate). On the basis of the TG/MS data, this was expected
to be CO2 with some O2; therefore, adequate venting was
required during the reaction. The total pressurization was
measured at 8.6 L/h over 3.5 h = 30.1 L, which was well within
venting capability of the vessel (150 L/min). As a precaution,
446
dx.doi.org/10.1021/op100280c |Org. Process Res. Dev. 2011, 15, 443–448

Organic Process Research & Development
COMMUNICATION
blanket sweep control was used in the vessel as the gas was most
likely to contain oxygen from decomposition of m-CPBA.
Although the reaction was moderately exothermic (-20.8 kcal/
mol substrate), part of the exotherm was masked by the addition
of the cooler feed solution (m-CPBA in AcOH) to the warm
batch. The overall heat of reaction was found to be endothermic,
but the temperature was calculated to rise above the jacket during
the age. The reaction had a pseudo-first-order half-life of 25.8 min at
52 °C, and was expected to take ∼3 h to reach 99% conversion
at 52 °C.
Quench. A reaction was performed where, at the end of the
reaction, product 3 was crystallized without quenching of the
excess m-CPBA and the slurry filtered. CRC analysis showed in
an 18 mL Hastelloy vessel, that these mother liquors generated
1.077 psi/day of pressure at 35 °C and 72% fill (3.77 psi/day at
90% fill). There was no evidence of the rate decreasing with time
over the 24 h of the test (Figure 5). The quench of the excess
oxidant using sodium bisulfite was therefore necessary to address
this pressurization in the waste liquors. The quench was also
characterized and calculated to be exothermic (dHr = -67.6 kcal/
mol, lit., modelled as quenching perbenzoic acid). In the CRC, a
total heat of 2.73 kcal/L of batch was measured during the
quench (ΔTad=4 K). Assuming all of the heat was from quenching
unreacted m-CPBA, the concentration was ∼0.04 mol/L m-CPBA
at the time of the quench (∼66% of the theoretical excess). Even
if the full amount of excess m-CPBA was quenched, the ΔTad
would only be 7 K.
N-Oxide, 3. m-CPBA (27.1 kg) was charged to a 400-L vessel
followed by AcOH (198 L; 208 kg) and stirred for 20 min to fully
dissolve the solid m-CPBA. (Jacket used to counter dissolution
endotherm.)
Pyrazolopyridine 4 (28.3 kg) was then charged to a 1000-L
vessel followed by AcOH (425 L; 445 kg), and the slurry was
heated to ∼55 °C to obtain a thin slurry.
The m-CPBA in AcOH was then transferred to the 1000-L
vessel over 45 min whilst maintaining the temperature in 1000-L
vessel as close to 55 °C as possible using jacket heating and rate of
addition. The line was rinsed with AcOH (5 kg) into the 1000-L
vessel. The reaction mixture was aged at 55 °C, and after 3 h,
HPLC analysis confirmed complete reaction (>99% conversion).
The batch was cooled to 30 °C, and a sodium bisulfite solution
(5.01 kg in 8.0 kg water) was charged over a period of ∼20-40
min whilst maintaining a batch temperature of 30 ( 5 °C using
jacket control. A sample of the reaction mixture was tested for
oxidant using starch test-strips and found to be below the
detectable limit.
Water (550 kg), (19.4 mL/g wrt SM) was then charged over 1 h,
maintaining a batch temperature of ∼20-25 °C, and aged for
30 min. The slurry was filtered and washed with ethanol (140 kg)/
water (90 kg), and dried in a vacuum oven at 40 °C with a
nitrogen bleed for 18 h to afford 32.2 kg (97.9% yield, 97.8 LCAP,
90.0 LCWP, 8.1 wt % H2O) of N-oxide 3 as an off-white solid:
1
mp 197-198 °C; H NMR (DMSO-d6, 400 MHz) δ 5.49 (s,
2H), 7.09 (dd, 1H, J = 8.8, 2.8 Hz), 7.12 (d, 1H, J = 2.7 Hz), 7.19
(dd, 1H, J = 8.0, 6.0 Hz), 7.35 (t, 1H, J = 2.2 Hz), 7.45 (dd, 1H, J =
2.2, 1.4 Hz), 7.55 (d, 1H, J = 8.8 Hz), 7.78 (t, 1H, J = 1.4 Hz), 7.89
(d, 1H, J = 8.0 Hz), 8.40 (d, 1H, J = 6.0 Hz); 13C NMR (DMSO-
d6, 100 MHz) δ 64.1, 110.0, 114.5, 114.6, 117.4, 117.5, 118.5,
118.6, 119.6, 120.1, 122.2, 127.1, 131.8, 135.8, 136.0, 150.5,
152.2, 158.5. Anal. Calcd for C20H12Cl2N4O3: C, 56.22; H, 2.83;
N, 13.11. Found: C, 56.14; H, 2.79; N, 12.99.
The quench was found not to be instantaneous but took ∼90 min
to reach completion. Testing for completion of the quench was
therefore essential, especially in view of the unquenched mother
liquors generating gas in the CRC experiment.
After the addition of sodium bisulfite, small quantities of gas
were formed after the NaHSO3 addition. If this gas was assumed
to be SO2 then 0.3 g of SO2/ L of solution were formed at a slow
rate (0.2 g/L/h), and this rate of gas release was also not a
concern.
’ AUTHOR INFORMATION
Thermal Stability of Product and Intermediates. The
N-oxide product was thermally characterized and found to
decompose exothermically above 200 °C (well above the oper-
ating temperature of drying), and was also not shock sensitive.
None of the intermediate streams tested possessed hazardous
exotherms
Corresponding Author
*E-mail: mahbub_alam@merck.com
’ ACKNOWLEDGMENT
We thank John Edwards, the EPSE group, and George Zhou of
Merck and Co., Inc. for their valuable contributions to this work.
Thanks also go to Ed Cleator and Carl Baxter for proofreading
this manuscript.
’ CONCLUSION
In summary, a safe and practical procedure for the N-oxidation
of pyrazolopyridine 4 was developed and successfully implemented
on 28-kg scale using m-CPBA in AcOH to achieve >99% conversion
and 98% yield of N-oxide 3 from a simple, direct, and controlled
crystallization procedure. Extensive safety testing and analysis
were conducted to determine safe operating parameters for the
use of m-CPBA in acetic acid at elevated reaction temperature.
’ REFERENCES
(1) (a) Tucker, T. J.; Saggar, S.; Sisko, J. T.; Tynebor, R. M;
Williams, T. M.; Felock, P. J.; Flynn, J. A.; Lai, M.-T.; Liang, Y.;
McGaughey, G.; Liu, M.; Miller, M.; Moyer, G.; Munshi, V.; Perlow-
Poehnelt, R.; Prasad, S.; Sanchez, R.; Torrent, M.; Vacca, J. P.; Wan,
B. L.; Yan, Y. Bioorg. Med. Chem. Lett. 2008, 18, 2959–2966. (b) Tucker,
T. J.; Sisko, J. T.; Tynebor, R. M.; Williams, T. M.; Felock, P. J.; Flynn,
J. A.; Lai, M.-T.; Liang, Y.; McGaughey, G.; Liu, M.; Miller, M.; Moyer,
G.; Munshi, V.; Perlow-Poehnelt, R.; Prasad, S.; Reid, J. C.; Sanchez, R.;
Torrent, M.; Vacca, J. P.; Wan, B. L.; Yan, W. J. Med. Chem. 2008, 51,
6503–6511.
(2) Kuethe, J. T.; Zhong, Y-Li.; Alam, M.; Alorati, A. D.; Beutner,
G. L.; Cai, D.; Fleitz, F.; Gibb, A. D.; Kassim, A.; Linn, K.; Mancheno, D.;
Marcune, B.; Pye, P. J.; Scott, J. P.; Tellers, D. M.; Xiang, B.; Yasuda, N.;
Yin, J.; Davies, I. W. Tetrahedron 2009, 65, 5013–5023.
(3) Yin, J.; Xiang, B.; Huffman, M. A.; Raab, C.; Davies, I. W. J. Org.
Chem. 2007, 72, 4554–4557.
’ EXPERIMENTAL SECTION
General. Melting points were determined by closed cell DSC.
HPLC assays were carried using a C-18 reversed-phase column
eluted with 0.1% H3PO4 (aq) and acetonitrile. Assay yields were
obtained by HPLC using pure compounds as standards. Isolated
yields refer to yields corrected for purity based on HPLC assay
using purified standards. All reagents and solvents were used as
received without further purification.
447
dx.doi.org/10.1021/op100280c |Org. Process Res. Dev. 2011, 15, 443–448

Organic Process Research & Development
COMMUNICATION
(4) Caron, S.; Dugger, R. W.; Ruggeri, S. G.; Ragan, J. A.; Ripin,
D. H. B. Chem. Rev. 2006, 106 (7), 2943–2989.
(5) Sigma-Aldrich MSDS for m-CPBA, available on http://
www.sigmaaldrich.com.
(6) Kubota, A.; Takeuchi, H. Org. Process Res. Dev. 2004, 8, 1076–
1078.
(7) Urben, P. G., Ed. Bretherick’s Handbook of Reactive Chemical
Hazards, 7th ed.; Elsevier: Oxford, 2007; Vol. 1; p 944.
(8) EAC stands for “enhanced adiabatic calorimeter” and is a larger
version of the ARC.
(9) The “heat-wait-search” experiment takes into account the
thermal mass of the reaction cell.
448
dx.doi.org/10.1021/op100280c |Org. Process Res. Dev. 2011, 15, 443–448