Hydrolysis of phosphoryl trichloride (POCl3): Characterization, in situ detection, and safe quenching of energetic metastable intermediates

Article abstract of DOI:10.1021/op1001484

The accumulation of metastable intermediates resulting from the incomplete hydrolysis of phosphoryl trichloride-containing mixtures carries the risk of latent exothermic events. Significant accumulation of two P-Cl species containing reactive phosphorus-c

Full text of DOI:10.1021/op1001484

Organic Process Research & Development 2010, 14, 1490–1500
Hydrolysis of Phosphoryl Trichloride (POCl3): Characterization, in Situ Detection, and
Safe Quenching of Energetic Metastable Intermediates
Michał M. Achmatowicz,*,† Oliver R. Thiel,† John T. Colyer,† Jack Hu,‡ Maria Victoria Silva Elipe,‡ Joe Tomaskevitch,†
Jason S. Tedrow,† and Robert D. Larsen†
Department of Chemical Process Research and DeVelopment, Department of Analytical Research and DeVelopment, Amgen
Inc., One Amgen Center DriVe, Thousand Oaks, California 91320-1799, U.S.A.
Scheme 1. Hydrolysis of phosphoryl trichloride (1)
Abstract:
The accumulation of metastable intermediates resulting from the
incomplete hydrolysis of phosphoryl trichloride-containing mix-
tures carries the risk of latent exothermic events. Significant
accumulation of two P-Cl species containing reactive phospho-
rus-chlorine bonds was detected spectroscopically (31P NMR)
during inverse quench of POCl3-MeCN mixtures under typical
literature conditions. The dominant reactive intermediate was
unequivocally assigned as phosphorodichloridic acid (X-ray, 31P
NMR). Quantitative 31P NMR time-course experiments allowed
for the determination of kinetic parameters of POCl3 hydrolysis
under synthetically relevant concentration and temperature condi-
tions in batch settings. Development of an in situ Raman method
allowed to further expand these studies to semibatch conditions
under different pH regimes. Furthermore, we hereby describe an
in situ Raman method suitable to ascertain completeness of the
quench for large-scale preparations involving POCl3. These
analytical techniques can be supported by differential scanning
calorimetry (DSC) and accelerated rate calorimetry (ARC) in
order to confirm absence of reactive species.
deserve care during hydrolytic processing in the workup as they
contain significant additional energy. The thermodynamic values
for the hydrolysis of POCl3 have been reported in the literature
and, depending on the amount of water that is present during
the hydrolysis, either aqueous HCl or gaseous HCl is formed
along with phosphoric acid.4
POCl3 + 13.5H2O f H3PO4 + 3HCl(aq) +
10.5H2O + ∆H
∆H ) -286.19 kJ/mol POCl3
(1)
POCl3 + 3H2O f H3PO4 + 3HCl(g) + ∆H
∆H ) -89.54 kJ/mol POCl3
(2)
The hydrolysis of POCl3 predominantly follows the sequence
shown in Scheme 1. A detailed mechanistic discussion of the
hydrolysis of a range of phosphorus compounds including
phosphoryl trichloride was published in a series of papers by
Hudson et al.5 Pseudofirst-order rate constants (k1 and k2) were
determined in dilute aqueous dioxane solutions (10-5-10-2 M)
under mildly acidic to neutral conditions (pH 4-7) at 25 °C.
The initial fast reaction (k1 ≈ 66 s-1, t1/2 ≈ 0.01 s) led to the
formation of phosphorodichloridic acid (2) followed by a slower
process (k2 ≈ 3 × 10-3 s-1, t1/2 ≈ 250 s) leading directly to
phosphoric acid (4). Phosphorochloridic acid (3) was not
observed suggesting that k3.k2.6 The rate of hydrolysis of 2
(k2) was similar regardless of pH, showing that the main reaction
pathway does not involve the undissociated acid. Moreover, a
large rate increase by nucleophilic catalysis suggested that the
hydrolysis of 2 proceeds through a bimolecular reaction between
a water molecule and a phosphorodichloridate anion. Only
recently7 the intermediacy of the previously elusive 3 was
unambiguously confirmed by Mitchell’s 18O-labeling experi-
ments (Scheme 2). Phosphorus pentachloride was partially
Introduction
Phosphoryl trichloride1 is a powerful dehydroxychlorinating
agent commonly used to convert hydroxyheteroaromatics into
the corresponding chloroheteroaromatics2 or for in situ genera-
tion of the Vilsmeier-Haack reagent from DMF.3 Most
frequently at least one equivalent of phosphoryl trichloride is
used in these transformations, thereby at the most employing
only one P-Cl bond per POCl3 molecule in a productive
manner. Following the first substitution of chloride in POCl3
the remaining two P-Cl bonds are kinetically less reactive and
rendered sufficiently thermodynamically strong to be of lesser
synthetic use. Nonetheless, all the remaining P-Cl bonds
* To whom correspondence should be addressed. Telephone: 805-447-5774.
Fax: 805-480-1346. E-mail: michala@amgen.com.
† Department of Chemical Process Research and Development.
‡ Department of Analytical Research and Development.
(1) Common synonyms: phosphorus oxychloride, POCl3.
(2) (a) Albert, A.; Clark, J. J. Chem. Soc. 1964, 1666. (b) Golovchinskaya,
E. S. Russ. Chem. ReV. (Engl. Transl.) 1974, 43, 1089. (c) Robins,
R. K.; Revankar, G. R.; O’Brien, D. E.; Springer, R. H.; Novinson,
T.; Albert, A.; Senga, K.; Miller, J. P.; Streeter, D. G. J. Heterocycl.
Chem. 1985, 22, 601. (d) Andersen, K.; Begtrup, M. Acta Chem.
Scand. 1992, 46, 1130.
(4) Kaspias, T.; Griffiths, R. F. J. Hazard. Mater. 2001, A81, 223.
(5) Hudson, R. F.; Moss, G. J. Chem. Soc. 1962, 3599.
(6) The rate in the phosphorochloridic acid series (k1, k3 . k2) is in contrast
to that found for the corresponding phosphorofluoridic acid series
where k1 > k2 > k3: Lange, W. Ber. 1929, 62, 786. Ber. 62 793; Ber.
62, 1084; Lange, W.; Livingstone, R. J. Am. Chem. Soc. 1950, 72,
1280.
(3) (a) Dyer, U. C.; Henderson, D. A.; Mitchell, M. B.; Tiffin, P. D. Org.
Process Res. DeV. 2002, 6, 311. (b) Miyake, A.; Suzuki, M.; Sumino,
M.; Iizuka, Y.; Ogawa, T. Org. Process Res. DeV. 2002, 6, 922. (c)
Bollyn, M. Org. Process Res. DeV. 2005, 9, 982.
(7) Mitchell, R. A. J. Chem. Soc., Dalton Trans. 1997, 1069.
1490
•
Vol. 14, No. 6, 2010 / Organic Process Research & Development
10.1021/op1001484  2010 American Chemical Society
Published on Web 08/23/2010

Scheme 2. Demonstration of intermediacy of 3 in the hydrolysis of PCl5 (POCl3)
Scheme 3. Vilsmeier-Haack reaction, workup and isolation
hydrolyzed using H218O as a limiting reagent (0.28-3.72 equiv)
and the hydrolysis was completed using a large excess of H216
O
followed by exhaustive methylation using diazomethane. The
resulting mixture of trimethyl phosphate isotopomers (m/z )
140, 142, 144, 146, 148) was analyzed by GC/MS. A significant
amount (e30%) of m/z ) 146 corresponding to Me3P18O316
O
was observed confirming the intermediacy of 18O-3 in the initial
stage of hydrolysis.
Examination of the recent literature related to the use of
phosphoryl trichloride8 showed that two general methods were
primarily used to quench postreaction mixtures on large scale:
(Method A) inverse quench into ice-water or (Method B)
inverse quench into an aqueous base. Selection of the particular
workup protocol (if discussed at all) was invariably dictated
by product quality considerations (e.g., stability to aqueous base,
ease of isolation). The composition of aqueous waste-streams
resulting from workups was rarely discussed; leaving the
impression that approaches Method A and Method B are
equivalent in the sense of completeness of the hydrolysis of
any residual byproduct containing P-Cl bonds. An additional
complicating factor in the workup of some of these reactions
is the use of a biphasic workup protocol due to the presence of
a water-immiscible reaction solvent, leading to the potential for
delayed quench due to mass transfer limited hydrolysis rate.
significantly decreased yields. The resulting viscous postreaction
mixture containing a large excess of unreacted POCl3 was
diluted with acetonitrile (5.0 vol) and following Method A10
was inversely quenched into 50:50 v/v acetonitrile/water (20
vol).11 The rate of the addition was adjusted to control a
significant exotherm and maintain the internal temperature
below 5 °C. The desired II gradually crystallized out from the
quench solution and was subsequently separated from
the resulting highly acidic medium by filtration. The sequence
described above was carried out reproducibly on multiple
occasions on a bench scale (e0.6 kg of POCl3). Nevertheless,
it was noted that an exothermic reaction occurred in the aqueous
filtrates. The temperature profile of filtrates after product
isolation is shown in Figure 1. Independent of the quench
temperature (5 and 25 °C), a latent exotherm leading to a
temperature increase of up to 50 °C was observed. Formation
of a colorless crystalline material was also observed upon
prolonged storage of these filtrates and analysis of these solids
identified them as ammonium chloride. The observation of a
latent and potentially uncontrollable exotherm in the filtrates
prompted further investigation into the kinetic parameters of
Rationale for Study
A significant quantity of a building block 2-chloroquinoline-
3-carbaldehyde II was required for an ongoing program. The
desired quinoline II could be synthesized from an acetanilide I
using a classical Vilsmeier-Haack type process as described
previously by Meth-Cohn (Scheme 3).9 The reaction is note-
worthy as three carbon-carbon bonds and one carbon-chlorine
bond are formed in a single transformation.
An addition-controlled exothermic reaction of DMF with
neat POCl3 (6.0 equiv) at e5 °C afforded the Vilsmeier reagent.
No reaction between I and POCl3 was observed in the absence
of DMF at this temperature. The formylation reaction between
the Vilsmeier reagent and I was carried out at 70 °C. Any
attempts to reduce the amount of POCl3 in this reaction led to
(8) Experimental procedures published in Org. Process Res. DeV., J. Org.
Chem., Tetrahedron or Tetrahedron Lett. from 1998 to 2009. A total
of 81 examples were found using from 10 g to 90 kg of POCl3 per
run.
(9) (a) Meth-Cohn, O.; Narine, B.; Tarnowski, B. Tetrahedron Lett. 1979,
20, 3111. (b) Meth-Cohn, O. Heterocycles 1993, 35, 539. (c) Ali,
M. M.; Tasneem, K. C.; Rajanna, K. C.; Sai Prakash, P. K. Synlett
2001, 251.
Figure 1. Temperature profile of reaction filtrates (eq 1, 20 g
of POCl3). Monitoring began immediately following filtra-
tion.
(10) Quench into an aqueous base (Method B) failed to provide product II
with a sufficient purity.
Vol. 14, No. 6, 2010 / Organic Process Research & Development
•
1491

Scheme 4. Hydrolysis of acetonitrile
50 v/v acetonitrile/water (3.2 vol, 12 equiv of water per POCl3)
at less than 5 °C (see Scheme 5). The resulting solution was
incubated at room temperature and analyzed periodically by
31P NMR. The characteristic feature of these spectra was the
presence of three 31P resonances (singlets) at approximately -2,
0, and +3 ppm (Figure 3). Their intensities as a function of
incubation period are shown in Chart 1.
the hydrolysis of POCl3, ultimately resulting in the design of a
safe quench.
Results and Discussion
The peak at -2 ppm (initially >97% by area) was determined
to gradually convert into the peak at 0 ppm (H3PO4). The
intensity of the minor peak at +3 ppm remained relatively
constant at approximately 2-5% of the intensity of the -2 ppm
peak throughout the study. The peak at 0 ppm was assigned as
H3PO4 by spiking experiments. The peak at -2 ppm was
consistent with phosphorodichloridic acid, which was previously
studied,13 and recently indirectly demonstrated,7 but never fully
structurally characterized. The postulated 3,5,6-triphenyl-2,3,5,6-
tetraza-bicyclo[2.1.1]-hex-1-ene (Nitron; C20H16N4) complex of
phosphorodichloridic acid (2·C20H16N4), previously prepared
by Grunze, was synthesized and recrystallized from hot aceto-
nitrile (Scheme 6). The 31P NMR spectra of both nonrecrys-
tallized and recrystallized 2·C20H16N4 were identical with that
of the material prepared by the quenching of POCl3 in aq
MeCN. All the spectra showed a single resonance at -2 ppm.
The structure of phosphorodichloridic acid was elucidated by
single crystal X-ray diffraction of the Nitron complex (Figure
4). To the best of our knowledge, this is the first X-ray crystal
structure of a phosphorodichloridic acid complex.
The hydrolysis of phosphoryl trichloride (Scheme 1) to
phosphorodichloridic acid (k1) is much faster than the following
step (k2) as determined by Hudson (k1 ≈ 66 s-1, k2 ≈ 0.0031
s-1)13g and confirmed recently by Mitchell.7 Therefore, the initial
formation of 2 can be regarded as essentially instantaneous. As
will be evident from the data presented below, under highly
acidic conditions the rate of hydrolysis of phosphorodichloridic
acid (2) is significantly retarded relative to the dilute and mildly
acidic systems studied by Hudson.13g The decay of phospho-
rodichloridic acid (2) (initially 1.22 M in 66:33 v/v acetonitrile/
water) under controlled temperature conditions (25 °C) was
quantified using 31P NMR. A plot of ln([2]t)0/[2]t) versus time
(Chart 2) showed good linearity (R ) 0.997), confirming that
the reaction is first order with respect to 2. However, the k2obs
determined by measuring the slope is only 5.15 × 10-5 s-1
(t1/2 ≈ 3.5 h) as opposed to the corresponding value of 3.06 ×
10-3 s-1 (t1/2 ≈ 4 min) reported by Hudson, giving the observed
rate difference of 59 times.14 The difference of the inherent
Three potential sources were identified for the latent exo-
therm observed in the filtrates: (i) exothermic degradation of
acetonitrile in the strongly acidic medium; (ii) delayed hydroly-
sis of phosphoryl trichloride (1) or its downstream byproduct
(2 and 3); (iii) combination of the two events (first event
triggered by second event).
Stability of Acetonitrile. The hydrolysis of acetonitrile in
the presence of hydrogen chloride leads to the formation of
acetic acid and ammonium chloride, with acetamide as an
intermediate (Scheme 4).
The decomposition of acetonitrile in the presence of HCl
was monitored by 1H NMR spectroscopy.12 Only trace amounts
of decomposition (<1%) were observed over 24 h using 1 or 5
M HCl at room temperature or 60 °C. In the presence of
concentrated HCl, however, acetonitrile readily decomposed at
both room temperature and at 60 °C. Similarly decomposition
was observed in a mixture that was obtained by adding POCl3
to a mixture of acetonitrile and water followed by heating to
60 °C. In contrast, negligible decomposition was observed in
this system at ambient temperature (Figure 2).
The conclusion that can be drawn from these data points is
that the decomposition of acetonitrile was not the primary factor
in the exothermicity of the filtrates. Decomposition of aceto-
nitrile under strongly acidic conditions, however, can occur at
elevated temperatures, and a suitable hazard evaluation is
warranted when operating under these conditions.
Hydrolysis of Phosphoryl Trichloride (POCl3). The
composition of the aqueous filtrates was initially studied using
31P NMR on a model system consisting of aqueous acetonitrile
and POCl3. It was reasoned that organic entities involved in
the Vilsmeier-Haack reaction were unlikely to be of impor-
tance, since most of the mass balance resulting from acetanilide
I was removed as 2-chloroquinoline-3-carbaldehyde II upon
filtration of the aqueous solution prior to the exotherm. The
postquench filtrates were modeled by diluting neat POCl3 with
acetonitrile (1.6 vol), followed by an inverse quench into 50:
(11) Fourteen equivalents of water per POCl3.
(12) Integration of the protons on the methyl-group (acetonitrile [2.07 ppm],
acetic acid [1.90 ppm], and acetamide [1.76 ppm]).
(13) (a) Meerwein, H.; Bodendorf, K. Ber. 1952, 62, 1952. (b) Van Wazer,
J. R.; Fluck, E. J. Am. Chem. Soc. 1959, 81, 6360. (c) Goubeau, J.;
Schulz, P. Z. Anorg. Allg. Chem. 1958, 294, 224. (d) Goubeau, J.;
Schulz, P. Z. Physik. Chem. 1958, 14, 49. (e) Grunze, H. Acta Chim.
Acad. Sci. Hung. 1959, 18, 303. (f) Grunze, H. Z. Anorg. Allg. Chem.
1961, 313, 316. (g) Hudson, R. F.; Moss, G. J. Chem. Soc. 1962,
3599.
(14) Difference in the observed rates of hydrolysis: 59 times (3.06 × 10-3
s-1/5.15 × 10-5 s-1).
Figure 2. Acetonitrile decomposition under acidic condi-
tions.
1492
•
Vol. 14, No. 6, 2010 / Organic Process Research & Development

Figure 3. Hydrolysis of POCl3 in aq acetonitrile at room temperature. Overlay of 31P NMR spectra at selected time points.
Chart 1. Hydrolysis of POCl3; major species quantified by
31P NMR
Figure 4. X-ray structure of phosphorodichloridic Nitron
complex (2 ·C20H16N4).
reaction rate obtained by taking water concentration into account
was 18 times.15
assumption that the pH dependence is negligible. A likely
candidate for the +3 ppm peak is phosphorochloridic acid (3)
(Scheme 5). Its progenitor (2) was present in high concentration
while the intensity of the peak at +3 ppm remained ap-
proximately constant relative to 2 (3-5 mol % of 2). Further-
more, a highly acidic medium favors undissociated forms of
acids 2 and 3 for which the relative rate constants k2 and k3
were shown to be comparable (k3/k2 ≈ 4),7 thereby supporting
the accumulation of phosphorochloridic acid (3) under these
conditions. The identity of the +3 ppm peak, however, has not
been unequivocally confirmed. A rate of decay of phospho-
rodichloridic acid (2) to phosphoric acid measured at 5, 25, and
35 °C by 31P NMR followed Arrhenius behavior with the
approximate half-lives of 26 h, 4 h, and 1 h, respectively (Chart
3, Table 1).
The reason for this disparity is likely two-fold. Sufficiently
low pH keeps the phosphorodichloridate anion protonated (in
the form of 2), thus inhibiting the hydrolysis pathway via the
phosphorodichloridic anion. This pathway is presumably re-
placed by a slower reaction via the undissociated phosphorod-
ichloridic acid (2). Simultaneously, the nucleophilicity of the
already sparse water molecules is greatly diminished under the
strongly acidic conditions. Therefore, the kinetic parameters
established by Hudson are no longer valid under the regime of
high concentration and low pH since they are based on the
(15) Rateobs ) k2obs × [HPO2Cl2] ≈ k2 × [HPO2Cl2] × [H2O]. Difference
in the inherent rate of hydrolysis (k2) obtained by taking concentration
of water into account: 18 times (3.06 × 10-3 s-1 × 16.5 M)/(5.3 ×
10-5 s-1 × 55.6 M).
Vol. 14, No. 6, 2010 / Organic Process Research & Development
•
1493

Table 1. Values of k2obs and t1/2 determined from the plot of
ln([2]t)0/[2]t) against time (Chart 3)
Chart 3. Hydrolysis rate of 2 as a function of temperature
temp
composition
(°C) k2obs [s-1
]
t1/2
water/MeCN (33:66 v/v)
water/MeCN (33:66 v/v),
1 equiv of NaOH
water/MeCN (33:66 v/v)
water/MeCN (33:66 v/v)
5
5
7.1 × 10-6 9.5 × 104 s 26 h
2.4 × 10-5 2.9 × 104 s 7.9 h
25 5.2 × 10-5 1.3 × 104 s 3.7 h
35 1.7 × 10-4 4.1 × 103 s 1.1 h
Additional 31P NMR experiments were performed to un-
derstand the effect of base (aq NaOH) and water concentration
on the kinetics of hydrolysis of phosphorodichloridic acid.
Solutions of phosphorodichloridic acid used in these experi-
ments were prepared by adding solutions of POCl3 in aceto-
nitrile to a 50:50 (v/v) acetonitrile-water mixture at e5 °C to
ensure fast and quantitative conversion of POCl3 to phospho-
rodichloridic acid. The resulting solutions ([H2O] ) 16.5 M)
were diluted with water or aqueous NaOH to an approximately
equal water concentration ([H2O] ≈ 24 M). The decay of
phosphorodichloridic acid in the presence of varying amount
of NaOH (0, 1, 2 equiv) was quantified by 31P NMR at 25 °C.
The plot of ln([2]t)0/[2]t) versus time (Chart 4) was used to
determine k2obs for each of the reaction composition. The
inherent reaction rate constants (k2) were derived directly from
k2obs and are compared in Chart 5.16,17
It is evident from Chart 5 that significant rate increase was
observed only after addition 2 equiv of NaOH. The observed
reaction rate increase (k2obs) in the presence of additional water
with or without 1 equiv of NaOH was similar (proportional to
[H2O]) (Table 2). The peak at +3 ppm attributed to phospho-
rochloridic acid (3) was only observed in the absence of base.
All the plots show a slight downward trendsa deviation from
first-order kinetics at high conversions. This is consistent with
an inhibitory effect caused by HCl byproduct which lowers the
fraction of ionized 2 (phosphorodichloridate). The nonionized
phosphorodichloridic acid undergoes hydrolysis more slowly
than phosphorodichloridate anion.13g
Several conclusions can be drawn from the mechanistic
studies that were performed on phosphorodichloridic acid. First,
the hydrolysis of phosphorodichloridic acid (2) is first order in
2 and first order in H2O, with Arrhenius behavior observed
between 5 and 35 °C. Second, the hydrolysis of phosphorod-
ichloridic acid (2) is the rate-limiting step of hydrolytic
conversion of POCl3 to H3PO4. However, the undesired
accumulation of 2 can be minimized by either increasing the
water concentration and/or by increasing pH of the solution.
Third, the beneficial effect of increasing the pH with base has
been demonstrated with 2 equiv of aq NaOH per 2, whereas
only a modest rate increase with 1 equiv of aq NaOH was
observed relative to pure water.
Raman Spectroscopy: In Situ Monitoring of HPO2Cl2
(2). Due to the multistep nature of the hydrolysis of POCl3 a
specific, quantitative, and sensitive analytical method was
desired to monitor the conversion of the intermediate species
2. After a preliminary evaluation which considered the solvent
Chart 2. Solvolysis of 2 at room temp; plot of ln([2]t)0/[2]t) ) k2obs × t provided k2obs ) 5.15 × 10-5 s-1
1494
•
Vol. 14, No. 6, 2010 / Organic Process Research & Development

Chart 4. Effect of base and water concentration on k2obs
Table 2. Values of k2obs and k2 determined from the plots
shown in Charts 4 and 5
Scheme 5. Hydrolysis of POCl3 in aq MeCN at room
temperature
conditions
k2obs [s-1
]
k2 [s-1
]
[H2O] ) 16.5 M, MeCN, 25 °C
[H2O] ) 24.4 M, MeCN, 25 °C
5.2 × 10-5 3.1 × 10-6
[H2O] ) 24.8 M, 1.0 equiv NaOH, 1.5 × 10-4 5.9 × 10-6
MeCN, 25 °C
1.1 × 10-4 4.6 × 10-6
[H2O] ) 24.8 M, 2.0 equiv NaOH, 6.8 × 10-4 2.7 × 10-5
MeCN, 25 °C
Chart 5. Effect of base and water concentration on k2
increase in temperature from 5 to 15 to 25 °C. Autoinhibition
previously noted by 31P NMR was evident at high conversions
(Figure 7). The limit of detection of phosphorodichloridic acid
(2) by the in situ Raman was determined by spiking an aqueous
acetonitrile medium with known amounts of POCl3 leading to
immediate formation of HPO2Cl2 (2). A clear response at 204
cm-1 was observed at concentrations of 2 as low as 0.015 M
(Figure 8). More than 99% conversion of HPO2Cl2 (2) to H3PO4
(4) could be ascertained under most of the realistic applications,
including the previously studied quench conditions.
In summary, the in situ Raman method was developed as a
complementary method for monitoring the POCl3 hydrolysis
reaction. It allows for monitoring of formation and disappear-
ance of phosphorodichloridic acid (2) in aqueous acetonitrile
mixtures at concentrations as low as 0.015 M. While no attempts
have been made to use the current Raman method for quantitation,
it was capable of semiquantitative monitoring of processes involv-
ing POCl3. In particular, its usefulness for a real-time monitoring
of accumulation of reactive intermediates containing P-Cl bonds
during aqueous quenches of POCl3-containing mixtures has been
demonstrated. Additionally, 31P NMR spectroscopy allows for a
straightforward quantitation of the reactive species and determining
the kinetic parameters in batch settings.
Hazard Evaluation of POCl3 Hydrolysis Using Calorim-
etry. In order to understand the latent exotherm in the reaction
filtrates, the quench experiments were conducted on a simplified
model system. The reaction was performed in the reaction
calorimeter (RC-1) by adding a solution of POCl3 in acetonitrile
into a mixture of acetonitrile and water over 2 h, maintaining
an internal temperature between 0 and 5 °C. The amount of
water (43 equiv) was sufficient to completely dissolve the
compatibility (aqueous acetonitrile) and method specificity,
Raman spectroscopy was determined to be a more suitable
method than FTIR. Subsequently, all the Raman peaks were
assigned to the known POCl3 hydrolysis intermediates/products
and confirmed with the actual reaction mixture. The first step
evaluated the specificity and peak assignments. Several key
Raman shifts had been identified: (i) 204 cm-1 corresponding
to the PCl2 deformation, (ii) 546 cm-1 corresponding to the
P-Cl stretch, and (iii) 889 cm-1 corresponding to the P-O
stretch (Figure 5).18,19
Consequently, simultaneous appearance and increase in
intensity of 204 and 546 cm-1 Raman shifts correlated with a
formation and accumulation of phosphorodichloridic acid (2)
during the first step of POCl3 hydrolysis (Figure 6). A
simultaneous and gradual disappearance of these bands with
concomitant gradual appearance of 889 cm-1 Raman shift was
indicative of the rate-limiting hydrolysis of 2 resulting in
formation of phosphoric acid (4). Increasing rate of conversion
of 2 to 4 consistent with the Arrhenius model was observed by
Vol. 14, No. 6, 2010 / Organic Process Research & Development
•
1495

Scheme 6. Preparation of a crystalline phosphorodichloridic Nitron complex (2 ·C20H16N4)
generated HCl. A sample of the reaction mixture was investi-
gated by accelerated rate calorimetry (ARC) showing three
exothermic events at ∼25 °C, 80 and 180 °C (Figure 9). These
exothermic events could also be detected using DSC as
analytical technique. The temperatures of the observed exo-
thermic events coincide with the observations from the ARC
(Figure 10). Estimation of the reaction mixture composition by
31P NMR showed 25% phosphoric acid and 75% phosphorod-
ichloridic acid upon the completion of the POCl3 addition. The
reaction mixture was warmed to 20 °C, and aged for 18 h at
which point analysis followed by 31P NMR showed 98%
phosphoric acid and 2% phosphorodichloridic acid. Analysis
of a sample of the reaction mixture by ARC showed that the
low-temperature exotherm was no longer present (Figure 11),
and instead the first exotherm was observed around 90 °C. This
exotherm is most likely related to decomposition of acetonitrile
under the strongly acidic conditions, as it was absent from the
sample that was investigated after complete neutralization of
the reaction mixture with NaOH.
Figure 5. Raman bands: PCl2 deformation (204 cm-1), P-Cl stretch (546 cm-1), P-O stretch (889 cm-1).
Figure 6. Phosphorodichloridic acid (2) monitored by Raman at 204 and 546 cm-1
.
Figure 7. Effect of temperature on hydrolysis of HPO2Cl2 monitored by Raman.
1496
•
Vol. 14, No. 6, 2010 / Organic Process Research & Development

Figure 8. Limit of detection of phosphorodichloridic acid by in situ Raman (e0.015 M at 204 cm-1).
Figure 9. ARC after POCl3 addition to mixture of acetonitrile/water at 0-5 °C.
Figure 10. DSC after POCl3 addition to mixture of acetonitrile/water at 0-5 °C.
The hazard studies on the model system highlight the
potential for significant latent exothermic events in the hydroly-
sis of POCl3, especially if the quench is conducted at low
temperatures and in the absence of base. These findings led to
the requirement to design a safe quench for the heterocycle
synthesis described above.
Development of the Improved Quench Protocol. An
improved quench protocol was devised on the basis of
Vol. 14, No. 6, 2010 / Organic Process Research & Development
•
1497

Figure 11. ARC after stir-out for 18 h at 20 °C.
mechanistic studies. It was decided to continue to use acetonitrile
as a cosolvent for the quench since it has significant miscibility
with the aqueous system, thereby making it inherently safer in
the quench as compared to water immiscible solvents. The
generated data also supported that acetonitrile showed suitable
stability under the quench conditions if a partial neutralization
with NaOH was performed.
reaction mixture before filtration, thereby avoiding the problems
caused by salt formation and phase separation. Between 1 and
3 equiv of 5 N NaOH (based on the initial amount of POCl3)
were added simultaneously during the quench, keeping the pH
< 7 and the temperature below 25 °C. Phase separation of the
supernatant was observed when g2 equiv NaOH were em-
ployed. The pH of the filtrate was determined in all cases, and
with the exception of the addition of 3 equiv of NaOH (pH )
1.6) the resulting filtrates where strongly acidic (pH < 1). The
latent exothermic events previously observed in the filtrates were
eliminated by aging the quenched mixture at 20 ( 5 °C until
>97% of active P-Cl bonds had been quenched as monitored
by 31P NMR. It was concluded that the optimal conditions
involved using 1.5 equiv of NaOH, since this was the largest
amount of base that could be added without an impact on
product quality. The new conditions were successfully dem-
onstrated on 50-g scale (98.3% LCAP, 58.1% yield). Analysis
by 31P NMR immediately before filtration showed the active
species was >97% quenched.
A safety evaluation was performed in the RC-1. The quench
was performed over 2 h, and the reaction mixture was allowed
to age for over an additional hour before filtration. The
quenching of the P-Cl bonds was monitored by 31P NMR and
was shown to be g98% complete before filtration. An ARC of
the final filtrates taken immediately after the quench was
complete (Figure 12) and 2 h later after filtration (Figure 13)
showed no exothermic events, which led to the conclusion that
all the P-Cl bonds were sufficiently hydrolyzed and that the
product filtrates from this process are sufficiently neutralized
to avoid the exothermic decomposition of acetonitrile.
Different approaches in employing a partial neutralization
during the quench were investigated. Complete neutralization
of the supernatant was attempted first. Two methods were
examined: (A) a reverse quench into a precalculated amount
of NaOH so that the final pH would be 7 and (B) a simultaneous
addition of aq NaOH during the normal reverse quench which
would keep the pH 6-7.20 While both of these methods
eliminated the latent exotherm and fully quenched all P-Cl
bonds,21 the quality and yield of the isolated product was
unacceptable. The increased amounts of salts and water
introduced into the crystallization caused the acetonitrile and
water to phase split, lowering the isolated yields and resulting
in the product oiling out of solution. The majority of the mass
recovered was actually determined to be sodium phosphate salts.
After the difficulties of achieving a neutral pH in the
supernatant before filtration, a study was conducted to examine
the feasibility of using aq NaOH to accelerate the hydrolysis
of reactive P-Cl bonds while only partially neutralizing the
(16) A correct rate law should use activities instead of concentrations. For
simplicity it was assumed that activities of components are equal to
their concentrations.
(17) The hydrolysis of phosphorodichloridic acid is first order in 2 according
to the approximate rate equation: Rateobs ) k2obs × [2]. However, it is
also first order in water: Rateobs ) k2 × [2] × [H2O]. Therefore, the
inherent rate constant (k2) can be calculated from: k2 ) kobs/[H2O],
assuming [H2O] ≈ const.
Conclusions
(18) Infrared and Raman spectra, conformational stability, and normal
coordinate analysis of ethyldichlorophosphine-borane: Odom, J. D.;
Hizer, T. J.; Stanley, A. E.; Tonker, T. L.; Durig, J. R. Spectrochim.
Acta 1988, 44A, 631.
The individual intermediates in the hydrolysis of phosphoryl
trichloride (POCl3) were characterized: phosphorodichloridic
acid (2, HPO2Cl2) by a combination of 31P NMR and X-ray
crystallography, and phosphorochloridic acid (3, H2PO3Cl) by
31P NMR. Two complementary analytical methods were
developed to monitor the completeness of the hydrolysis of
phosphoryl trichloride: 31P NMR and in situ Raman spectros-
(19) Rudolph, W. W.; Irmer, G. Appl. Spectrosc. 2007, 61, 1312.
(20) Thiel, O. R.; Achmatowicz, M.; Bernard, C.; Wheeler, P.; Savarin,
C.; Correll, T. L.; Kasparian, A.; Allgeier, A.; Bartberger, M. D.; Tan,
H.; Larsen, R. D. Org. Process Res. DeV. 2009, 13, 230.
(21) The progress of the P-Cl bond quenching was quantitatively monitored
by 31P NMR.
1498
•
Vol. 14, No. 6, 2010 / Organic Process Research & Development

Figure 12. ARC of reaction solution immediately after the quench.
Figure 13. ARC of the final filtrate.
copy. 31P NMR provides a reliable tool to quantify the different
phosphorous species. Raman spectroscopy can be used as an
in situ tool, but the method needs to be individually developed
in order to ensure no interference with the solvent or substrate.
The successful employment of this technique may also be
complicated by the presence of solids in the reaction mixture,
leading to an unpredictable increase of the Raman scattering
due to solids in the interrogation volume of the probe.
conditions can serve as a starting point for other reactions, a
full hazard evaluation of the reaction, quench, and filtrates using
calorimetry is strongly recommended for all reactions involving
POCl3. Overall our studies emphasize the importance of
confirming the completeness of a quench prior to a scale-up to
ensure safe operations.
Experimental Section
The mechanistic studies shown have allowed the design of
a safe quench for the specific reaction (Scheme 7). In general,
an inverse quench at 20 °C in presence of base was shown to
be preferred. When applying the recommended quench condi-
tions to specific reactions, additional parameters such as solvent
and product formation need to be taken into account. In the
example highlighted within this manuscript, 1.5 equiv of NaOH
was co-dosed with the reaction solution during the quench in
order to avoid accumulation of intermediates containing P-Cl
bonds, which was confirmed via 31P NMR spectroscopy and
ARC. It needs to be emphasized that, while these quench
Synthesis of 2-Chloroquinoline-3-carbaldehyde II Us-
ing an Improved Quench Protocol. A dry 2.0-L reactor was
equipped with an addition funnel and overhead stirring and was
flushed with N2. Acetanilide I (330 mmol) was charged to the
reactor, and the reactor temperature was set to 0 ( 5 °C.
Phosphoryl trichloride (183 mL, 1.96 mol) was charged to the
reactor, stirring was initiated, and the mixture was stirred until
homogeneous. The addition funnel was charged with N,N-
dimethylformamide (63.5 mL, 820 mmol), which was slowly
added to the reactor, keeping an internal temperature of 0 ( 5
°C. The solution was heated to 70 ( 5 °C and held at that
Vol. 14, No. 6, 2010 / Organic Process Research & Development
•
1499

Scheme 7. Comparison of the initial workup and isolation conditions (red arrows) with the improved protocol (blue arrows)
temperature for >16 h. The reaction solution was then cooled
to 20 ( 5 °C and assayed for progress. Acetonitrile (5 vol)
was added slowly to the 2.0-L reactor and stirred until
homogeneous. A 5.0-L reactor was equipped with an addition
funnel, pH probe and overhead stirring and was placed under
N2. The 5.0-L reactor was charged with acetonitrile (10 vol)
and water (10 vol) and set to an internal temperature of 20 (
5 °C. The addition funnel for the 5.0-L reactor was charged
with 5 N NaOH (2.94 mol). The contents of the 2.0-L reactor
were charged into the 5.0-L reactor simultaneously with the 5
N NaOH, keeping the internal temperature at 20 ( 5 °C and
the pH < 7. The mixture was held at 20 ( 5 °C for 1 h with
stirring after the addition was complete. The filtrates were
assayed for product loss, the contents of the 5.0-L reactor were
filtered, and the solids were collected on an M-grade frit. The
filtrates were charged back into the 5.0-L reactor and analyzed
by 31P NMR prior to disposal for complete POCl3 hydrolysis.
The solids were washed with a 50:50 v/v acetonitrile/water
mixture (5 vol) and water (5 vol). The isolated compound II
was dried to a constant mass under vacuum/N2 at 20 °C.
Acknowledgment
We thank Dr. Tiffany Correll for invaluable assistance related
to the Raman spectrometry experiments. Sadia Terranova and
Justin Tvetan are acknowledged for their part in the hazard
evaluation. Dr. Margaret Faul, Dr. Janet Cheetham, Dr. David
Semin, and Dr. Jerry Murry are gratefully acknowledged for
support of this project.
Supporting Information Available
Description of experimental methods including Raman,
NMR, calorimetry (ARC and DSC), and X-ray crystal-
lography. Crystallographic data for 3,5,6-triphenyl-2,3,5,6-
tetraza-bicyclo[2.1.1]-hex-1-ene phosphorodichloridic acid
complex (2 · C20H16N4). This material is available free of
charge via the Internet at http://pubs.acs.org.
Received for review May 27, 2010.
OP1001484
1500
•
Vol. 14, No. 6, 2010 / Organic Process Research & Development