Safe and scaleable oxidation of benzaldoximes to benzohydroximinoyl chlorides

Article abstract of DOI:10.1021/op100013m

Benzohydroximinoyl chlorides are useful precursors to nitrile oxides used in the preparation of various heterocycles via 1,3-dipolar cycloadditions. These intermediates are typically accessed by oxidation of aldoximes using N-chlorosuccinimide. This simpl

Full text of DOI:10.1021/op100013m

Organic Process Research & Development 2010, 14, 574–578
Safe and Scaleable Oxidation of Benzaldoximes to Benzohydroximinoyl Chlorides†
Eric C. Hansen,* Mahmut Levent, and Terrence J. Connolly
Chemical DeVelopment, Wyeth Research, 401 North Middletown Road, Pearl RiVer, New York 10965, U.S.A.
Scheme 1. Synthesis of isoxazoles
Abstract:
Benzohydroximinoyl chlorides are useful precursors to nitrile
oxides used in the preparation of various heterocycles via 1,3-
dipolar cycloadditions. These intermediates are typically accessed
by oxidation of aldoximes using N-chlorosuccinimide. This simple
and efficient reaction is highly exothermic and a significant
induction period can be observed. The potential for a sudden and
significant heat release makes large-scale reactions difficult to
control and potentially hazardous. Herein we describe a thermal
analysis of this reaction to determine the heat flow in the presence
of common additives to determine the most effective means of safe
scale-up. Of the additives screened, aqueous HCl consistently
avoided an induction period and enabled a dose-controlled process
to be developed.
the nitrile oxide and subsequent cycloaddition is achieved by
addition of an aqueous solution of sodium carbonate. Ethyl
acetate was found to be a suitable solvent for the oxime
oxidation, the cycloaddition, and the subsequent aqueous
workup and isolation which allows for a simple, streamlined
procedure.
Introduction
In a recent program we required kilogram quantities of the
known compound 5-bromomethyl-3-(4-fluorophenyl)isoxazole
1a.1 Substituted isoxazoles of this type are well documented in
the literature and are typically derived from the 1,3-dipolar
cycloaddition of nitrile oxides with alkynes.2 The highly reactive
nitrile oxides are typically generated in situ from the hydrox-
iminoyl chlorides which are readily available by oxidation of
the corresponding aldoximes with N-chlorosuccinimide (NCS)
(Scheme 1).3,4
First, we adapted the known procedures for isoxazole
synthesis4,5 to the scale-up of 1a, deciding on a one pot, two-
step synthesis that allows the reaction to be done in a single
solvent. The benzonitrile oxide 4 is only generated in the
presence of alkyne, thereby minimizing dimerization of this
unstable intermediate. The procedure starts with the addition
of a solution of 2a in ethyl acetate to a slurry of NCS in the
same solvent. Due to the low solubility of NCS in ethyl acetate,
we opted for this mode of addition to avoid dosing of solid to
the reactor. After the oxidation is complete, the excess NCS is
quenched with aqueous sodium bisulfite. Propargyl bromide is
then added as an 80 wt % solution in toluene.6 Generation of
During the development of this process, a significant
exotherm was observed for the oxidation step that occurred only
after a considerable amount of the benzaldoxime had been added
to the NCS. This poses a significant barrier to scale-up because
of the safety issues involved with releasing a large amount of
heat in an uncontrolled fashion. Consequently, we evaluated
this reaction in the Mettler Toledo RC1 calorimeter to determine
the heat output. 4-Fluorobenzaldehyde oxime 2a was charged
over 30 min as a 35 wt % solution in ethyl acetate using a
dosing pump to the slurry of NCS in ethyl acetate at 25 °C.
Figure 1 shows the dosing curve and the resulting heat output
of the reaction. The integral of the heat curve corresponds to a
heat of reaction of 114 kJ/mol and an adiabatic temperature
rise (∆Tad) of 67 K. It is apparent from the graph that the vast
majority of the reaction takes place after the dosing of the oxime
is complete. The thermal conversion plot indicates that at the
end of dosing, only 8% of the reaction has taken place.
In order to safely scale this reaction, it was apparent that
the proper controls would need to be in place in order to prevent
the sudden release of the reaction heat. Ideally, the induction
period would be eliminated and the amount of heat controlled
by the dosing of substrate. Attempts to achieve this by
performing the reaction at higher temperatures (35 and 50 °C)
were not successful in completely eliminating the induction
period. A survey of the literature indicated that a catalytic
amount of hydrochloric acid4 or pyridine7 are often used in this
reaction; however, no systematic evaluation of the function of
† Wyeth became part of Pfizer on October 16, 2009.
* Address correspondence to this author at the above mailing address:
Attention B240/239. E-mail: hansene4@wyeth.com.
(1) Sen, H. G.; Seth, D.; Joshi, U. N.; Rajagopalan, P. J. Med. Chem.
1966, 9, 431–433.
(2) Grundmann, C.; Grunanger, P. The Nitrile Oxides; Springer-Verlag:
New York, 1971.
(3) Stevens, R. V. Tetrahedron 1976, 32, 1599–1612.
(4) Liu, K.-C.; Shelton, B. R.; Howe, R. K. J. Org. Chem. 1980, 45, 3916–
3918.
(5) Bast, K.; Christ, M.; Huisgen, R.; Mack, W.; Sustmann, R. Chem.
Ber. 1973, 106, 3258–3274.
(6) Propargyl bromide is shock sensitive as a neat liquid. See Bretherick’s
Handbook of ReactiVe Chemical Hazards, 7th ed.; Elsevier: Boston,
2007; Vol. 1.
(7) Larsen, K. E.; Torssell, K. B. G. Tetrahedron 1984, 40, 2985–2988.
10.1021/op100013m  2010 American Chemical Society
574
•
Vol. 14, No. 3, 2010 / Organic Process Research & Development
Published on Web 03/11/2010

Figure 1. RC1 heat flow for oxidation of 2a.
Figure 2. Oxidation of 2a-d with no initiator.
these additives has been reported. In an effort to determine their
effect, we proposed to screen a variety of substrates in this
oxidation in the presence of these initiators and to record the
heat output vs dosing in order to get a qualitative picture of
reaction rate over time. To do this, we utilized the Mettler
Toledo EasyMax reactor. This system consists of two 100-mL
glass reactors fitted in temperature-controlled reactor wells and
equipped with overhead propeller stirring and internal temper-
ature probes. Temperature can be controlled by either a set
jacket temperature (Tj mode) or a set reaction temperature (Tr
mode). While running in Tr mode, any rise in temperature due
to an exothermic reaction is met with a proportional drop in
jacket temperature in an attempt to remove the heat and maintain
the reaction temperature. The resulting difference between
reaction and jacket temperatures (Tr - Tj) is directly proportional
to the heat of reaction and the reaction rate.8,9 The dosing pump
attachment allows substrate dosing to be recorded and correlated
with heat flow similar to the RC1 calorimeter setup. Using this
apparatus, we evaluated the oxidation of a series of substituted
benzaldoximes (2a-d) both with and without additives.
Figure 2 shows the oxidation of 2a-d at 25 °C with no
additives present. In all cases the substrate was charged over
30 min as a solution in ethyl acetate. Several trends are apparent
from the heat flow curves. First, the more electron-rich
substrates showed shorter initiation periods. The benzaldehyde
oxime 2c and p-methoxy substrate 2d had induction times of
16 and 23 min, respectively, while substrates 2a and 2b
containing electron-withdrawing groups had induction times of
around 45 min. Second, following initiation, the reaction rate
was much faster for the substrates with electron-withdrawing
groups. Oxime 2b showed complete reaction in less than 5 min,
while 2d took over 90 min to react completely. The small spike
in heat flow observed in the latter half of the reaction was
determined to be an exothermic crystallization of the succin-
imide byproduct from the reaction mixture.10
Next, we evaluated the oxidation of 2a-d in the presence
of 5 mol % pyridine. To our surprise, the induction periods for
2a, 2c, and 2d actually increased (Figure 3). Furthermore, once
the oxidation initiated, the reaction rates were extremely fast,
creating a large spike in heat flow. Substrate 2d, which required
approximately 90 min for completion in the uncatalyzed
reaction, was complete in less than 30 min in the presence of
pyridine. It quickly became apparent that this was not an
improvement to the reaction conditions and in fact represents
a worst-case scenario for scale-up of the reaction. Substrate 2b
showed very different behavior in the presence of pyridine. The
induction period decreased slightly to 25 min and the reaction
rate decreased significantly, requiring over 14 h to reach
completion (Figure 4). Pyridine is typically used as an additive
with DMF as solvent. When 2a, 2c, and 2d were oxidized in
the presence of pyridine using DMF as solvent, the reactions
(10) Filtration of the reaction mixture slurry identified pure succinimide
as the solid phase. This small thermal event is not observed with
reactions using DMF as solvent where the reaction remains homog-
enous throughout.
(8) Jacobsen, J. P. Thermochim. Acta 1990, 160, 13–23.
(9) Landau, R. N. Thermochim. Acta 1996, 289, 101–126.
Vol. 14, No. 3, 2010 / Organic Process Research & Development
•
575

Figure 3. Oxidation of 2a, c, d with 5 mol % pyridine in ethyl acetate.
Figure 4. Oxidations with 5 mol % pyridine.
Figure 5. Oxidation of 2a-d with 5 mol % HCl.
proceeded similar to the reaction of 2bsa very slow rate with
a small heat output that extended 10-14 h was observed. The
concern over initiation periods was effectively eliminated
because the heat was released over an extended time. However,
given the solvent dependence of this behavior, we determined
this was not the ideal strategy for scaling the reaction.
Finally, the oxidation was performed in the presence of 5
mol % HCl that was added to the reaction mixture as an ether
solution prior to dosing of the oxime. Figure 5 shows that these
reactions had essentially no induction periods and the heat
curves showed the roughly square heat output indicative of an
addition controlled reaction. Oxime 2d still showed a much
slower reaction rate with a heat curve extending beyond the
end of the addition. Accumulation of oxime in this case could
be mitigated by slowing the addition rate. The lack of induction
period in all cases allowed the rate of heat to be controlled by
the rate of addition.
Given these positive results, we reevaluated the oxidation
of 2a in the RC1 calorimeter in the presence of HCl. For large-
scale reactions, concentrated HCl was used with no detrimental
effects. Figure 6 shows the heat output vs dosing. As expected,
no induction period was observed, and the heat curve is typical
of a dose-controlled reaction. The dosing was done in two equal
portions to demonstrate that the dosing can be paused and
restarted without the presence of a second induction period. The
similarity of the thermal conversion and dosing plots further
576
•
Vol. 14, No. 3, 2010 / Organic Process Research & Development

Figure 6. RC1 calorimetry with 5 mol % HCl.
Scheme 2. Proposed mechanism of the NCS oxidation of
oximes
The large exotherm can be controlled by introduction of a
suitable initiator that allows an addition-controlled release of
heat from the reaction. Hydrochloric acid was identified as the
most effective initiator by using the EasyMax reactor to perform
small-scale “pseudo” calorimetry. This method has been used
to synthesize kilogram quantities of isoxazole 1a.
Experimental Section
General. Oxime 2a was purchased from Wychem Ltd. and
2c was purchased from Aldrich. Oximes 2b,d were synthesized
from the corresponding aldehydes using standard methods
(NH2OH ·HCl, pyridine, EtOH). Reaction monitoring was
carried out on a Waters Aquity UPLC with PDA monitoring
(210 nm) equipped with an Aquity UPLC BEH C-18 (1.7 µm)
column (2.1 mm × 100 mm) at 50 °C. Elution was performed
with a flow rate of 0.9 mL/min and a gradient method using
water containing 0.1% H3PO4 (mobile phase A) and acetonitrile
(mobile phase B) (Initial: 90% A, 0.1 min: 90% A, 3.0 min:
10% A, 3.75 min: 10% A). Observed retention times were as
follows: Compound 2a (1.38 min); 3a (1.80 min); 2b (1.37
min); 3b (1.80 min); 2c (1.26 min); 3c (1.75 min); 2d (1.28
min); 3d (1.73); 1a (2.20 min).
Procedure for EasyMax Calorimetry. In the 100-mL
cylindrical EasyMax reaction vessel fitted with propeller stirring,
temperature probe, and nitrogen inlet was charged NCS (5.11
g, 37.13 mmol) and ethyl acetate (20 mL). The temperature
was adjusted to 25 °C (Tr mode), and the initiator (5 mol %)
was added. A stock solution of oxime in ethyl acetate was
prepared (1.83 M). The oxime solution (19 mL) was charged
over 30 min using the dosing pump attachment. The data for
dosing, Tr and Tj measurements, were collected by the internal
computer and exported to Excel.
Scale-Up of 1a. N-Chlorosuccinimide (1.2 kg, 8.37 mol)
and ethyl acetate (4.0 L) at 20 °C were treated with concentrated
HCl (29 mL, 0.35 mol). An oxime solution of 1a (1.0 kg, 7
mol) in ethyl acetate (2.0 L) was then added to the slurry of
NCS over 1.5 h at a constant rate. The jacket temperature was
adjusted between 5-20 °C in order to maintain a reaction
temperature in the range 20-30 °C. The reaction was stirred
30 min to ensure complete conversion to 2a as determined by
HPLC. A solution of sodium sulfite (145 g in 350 mL water)
was then added over 3 min. Propargyl bromide (1.24 kg, 80 wt
confirms that there was no accumulation of oxime in the
reaction. Interestingly, the calorimeter records the small exo-
therm corresponding to the crystallization of succinimide near
the end of the first dosing period.
Using this method we were able to safely synthesize our
target compound 1a on 1 kg scale. The oxime was added over
1.5 h, maintaining a reaction temperature between 20-30 °C
using a jacket temperature between 5-20 °C.
Mechanism. The potency of HCl as an initiator can be
explained by considering the reaction mechanism proposed in
Scheme 2. We believe that the HCl reacts with NCS to form
chlorine which is the reactive oxidant.11 The reaction between
chlorine and the oxime generates one equivalent of HCl which
propagates the catalytic cycle. The presence of chlorine is
consistent with the deep blue-green color observed in several
of the reactions described above. This hypothesis is further
supported by the fact that the addition of a catalytic amount of
chlorine gas to the reaction is equally as effective at initiating
the catalytic cycle.
The presence of pyridine in the reaction is likely to buffer
the HCl and reduce the rate at which the active oxidant is
generated, thereby slowing the overall reaction rate as observed
in Figure 4. The dependence of the induction period and reaction
rate on the reaction solvent, as demonstrated in Figure 3,
highlights that adding pyridine is a risky strategy for scale-up
of this reaction.
Conclusion
A safe and effective means of scaling the oxidation of
benzaldoximes to hydroximinoyl chlorides has been developed.
(11) Nishiguchi, A.; Maeda, K.; Miki, S. Synthesis 2006, 24, 4131–4134.
Vol. 14, No. 3, 2010 / Organic Process Research & Development
•
577

% in toluene Caution: Neat propargyl bromide is shock
sensitive. This reagent must be handled as a stock solution.
See reference 6.) was added over 20 min. The temperature
was adjusted to 20 °C, and a solution of sodium carbonate (750
g in 2.5 L water) was added over 1.5 h. The reaction temperature
was maintained in the range 17-27 °C, using a jacket
temperature of 5-20 °C. The reaction was stirred 5 min at 24
°C, and complete reaction was confirmed by HPLC. Water (1.6
L) was added, and the layers were separated. The organic layer
was washed a second time with water (4 L). Sodium chloride
(20 g) was added to the second wash to aid in layer separation.
The solution was concentrated to a volume of ∼4.0 L, and
heptane (8.0 L) was added over 1 h as the distillation continued.
The resulting slurry was filtered and washed with heptane (500
mL). The solids were dried in a vacuum oven at 50 °C for 24 h.
Isolated 1.08 kg (60%) as an off-white solid (97 area % purity
and 95 wt %). 1H NMR: (400 MHz, CDCl3) δ 7.78 (dd, JH-H
) 8.7 Hz, JH-F ) 5.2 Hz, 2 H); 7.15 (t, JH-H≈ JH-F ) 8.7 Hz,
2 H); 6.60 (s, 1 H); 4.51 (s, 2 H); 13C NMR: (100 MHz,
CDCl3): δ 168.2, 163.9 (d, 1JC-F ) 250 Hz, 1 C), 161.9, 128.8,
124.8, 116.1 (d, 2JC-F ) 22 Hz, 2 C), 101.8, 18.6.
Acknowledgment
We thank Simon Rea of Mettler Toledo for continued
support with EasyMax and RC1 calorimetry.
Received for review January 19, 2010.
OP100013M
578
•
Vol. 14, No. 3, 2010 / Organic Process Research & Development