Role of the P-F bond in fluoride-promoted aqueous VX hydrolysis: An experimental and theoretical study

Article abstract of DOI:10.1021/jo301549z

Following our ongoing studies on the reactivity of the fluoride ion toward organophosphorus compounds, we established that the extremely toxic and environmentally persistent chemical warfare agent VX (O-ethyl S-2-(diisopropylamino)ethyl methylphosphonothioate) is exclusively and rapidly degraded to the nontoxic product EMPA (ethyl methylphosphonic acid) even in dilute aqueous solutions of fluoride. The unique role of the P-F bond formation in the reaction mechanism was explored using both experimental and computational mechanistic studies. In most cases, the "G-analogue" (O-ethyl methylphosphonofluoridate, Et-G) was observed as an intermediate. Noteworthy and of practical importance is the fact that the toxic side product desethyl-VX, which is formed in substantial quantities during the slow degradation of VX in unbuffered water, is completely avoided in the presence of fluoride. A computational study on a VX-model, O,S-diethyl methylphosphonothioate (1), clarifies the distinctive tendency of aqueous fluoride ions to react with such organophosphorus compounds. The facility of the degradation process even in dilute fluoride solutions is due to the increased reactivity of fluoride, which is caused by the significant low activation barrier for the P-F bond formation. In addition, the unique nucleophilicity of fluoride versus hydroxide toward VX, in contrast to their relative basicity, is discussed. Although the reaction outcomes were similar, much slower reaction rates were observed experimentally for the VX-model (1) in comparison to VX.

Full text of DOI:10.1021/jo301549z

Article
pubs.acs.org/joc
Role of the P−F Bond in Fluoride-Promoted Aqueous VX Hydrolysis:
An Experimental and Theoretical Study
Daniele Marciano,§,† Ishay Columbus,§,† Shlomi Elias,† Michael Goldvaser,† Ofir Shoshanim,‡
Nissan Ashkenazi,*,† and Yossi Zafrani*,†
‡
†Department of Organic Chemistry and Department of Environmental Physics, Israel Institute for Biological Research,
Ness-Ziona, 74100, Israel
S
* Supporting Information
ABSTRACT: Following our ongoing studies on the reactivity of the
fluoride ion toward organophosphorus compounds, we established that the
extremely toxic and environmentally persistent chemical warfare agent
VX (O-ethyl S-2-(diisopropylamino)ethyl methylphosphonothioate) is
exclusively and rapidly degraded to the nontoxic product EMPA (ethyl
methylphosphonic acid) even in dilute aqueous solutions of fluoride. The
unique role of the P−F bond formation in the reaction mechanism was
explored using both experimental and computational mechanistic studies. In
most cases, the “G-analogue” (O-ethyl methylphosphonofluoridate, Et-G)
was observed as an intermediate. Noteworthy and of practical importance is
the fact that the toxic side product desethyl-VX, which is formed in
substantial quantities during the slow degradation of VX in unbuffered water,
is completely avoided in the presence of fluoride. A computational study on
a VX-model, O,S-diethyl methylphosphonothioate (1), clarifies the distinctive tendency of aqueous fluoride ions to react with such
organophosphorus compounds. The facility of the degradation process even in dilute fluoride solutions is due to the increased
reactivity of fluoride, which is caused by the significant low activation barrier for the P−F bond formation. In addition, the unique
nucleophilicity of fluoride versus hydroxide toward VX, in contrast to their relative basicity, is discussed. Although the reaction
outcomes were similar, much slower reaction rates were observed experimentally for the VX-model (1) in comparison to VX.
affected by pH, reaching as much as 50% at pH 7−12 and ∼13%
above pH 12.4b Desethyl-VX has similar toxicity to that of VX
and is “environmentally stable” and far more persistent toward
further hydrolysis.4,5 Therefore, intense efforts are directed to
the development of easy, applicable, and gentle methods for VX
degradation to nontoxic products, preferably by noncorrosive
aqueous media and in a catalytic manner.
INTRODUCTION
■
The reactivity of the fluoride ion toward phosphyl moieties has
been investigated in various disciplines of organophosphorus
chemistry.1 Kirby, Nome, and co-workers found that the
nucleophilicity of fluoride toward phosphorus is greater than
its basicity. In addition, the fluoride ion is nearly 3 orders of
magnitude more reactive than an oxyanion of the same
basicity.1d,f The unique strength of the P−F bond, as indicated
by the enhanced thermodynamic affinity of fluoride toward the
phosphoryl group, is reflected by the stability of the transi-
tion states containing partial P−F bonds.2 Interestingly, this
feature was utilized not only for synthetic purposes but also for
fluoride-induced reactivation of the phosphorylated acetylcho-
line esterase enzyme inhibited by the nerve agent sarin
(O-isopropyl methylphosphonofluoridate, GB).3
Among the chemical warfare agents (CWAs), the nerve agent
VX (O-ethyl S-2-(diisopropylamino)ethyl methylphosphono-
thioate) is a target of prior importance in the development of
decontamination methods, due to its high toxicity, environmental
persistence, and low volatility.4 In neutral to basic pH, the
hydrolysis of VX leads to a mixture of products, namely ethyl
methylphosphonic acid (EMPA) and S-2-(diisopropylamino)-
ethyl methylphosphonothioic acid (desethyl-VX) resulting from
the cleavage of the P−S or P−O bonds, respectively. In aqueous
solutions, the relative amount of desethyl-VX was found to be
Recently, we published our results on the detoxification of
both nerve agents VX and GB under the action of water-swelled
polystyrene-supported ammonium fluorides (Amberlite IRA
900 F−).6 It was found that during the process, VX is degraded
catalytically by the heterogeneous aqueous ammonium fluoride
system to the nontoxic product EMPA, via a “G-analogue” inter-
mediate (Et-G), which is more prone to hydrolysis (Scheme 1).
In the same study, we have shown that desethyl-VX is not
formed and that EMPA and 2-(diisopropylamino)ethanethiol
Received: July 22, 2012
Published: October 18, 2012
© 2012 American Chemical Society
10042
dx.doi.org/10.1021/jo301549z | J. Org. Chem. 2012, 77, 10042−10049

The Journal of Organic Chemistry
Article
media of the water-swelled resin.6 Namely, in the first step, a
mixture of Et-G and EMPA was obtained, and at the end of the
process the sole phosphorus-containing product was EMPA.
Curve-fitting of the results obtained in most of the experi-
ments led to the following minimal kinetic model (eq 1) with
calculated pseudo-first-order half-life times summarized in
Table 1. To compare different reaction conditions, half-life
Scheme 1. Catalytic VX Detoxification Process with
Amberlite IRA 900 F−
Table 1. Degradation Rate of VX (66 mM) in Different
Media
(DAT) are the sole products. In the past, we have investigated
the outcome of chemical warfare agents including VX on wet
KF/Al2O3, enriched by fluoride ions. We found that on this
active support, EMPA was formed exclusively.7 The present
results, together with those obtained with Amberlite IRA
900 F−,6 suggest that Et-G may also be produced but immediately
hydrolyzed to EMPA on wet KF/Al2O3.
products (%)
t1/2 (VX) t1/2 (Et-G)
a
run
salt
solvent
(min)
(min)
EMPA desethyl-VX
1
2
3
4
5
6
7
8
9
KF
water
water
12
46
102
580
158
84
100
100
100
100
100
100
100
68
−
−
−
−
−
−
−
32
a
KF
KF
b
methanol
water
69
NaF
CsF
TBAF
11
In the present study, two questions are addressed regarding
the facile degradation of VX promoted by water-solvated fluoride
ions. The first is related to the mechanism of the processes,
namely, to the effectiveness of the fluoride in the detoxification
of VX and to the driving force for these reactions. The second
question is related to the potential utility of these processes,
namely, their ability to take place in homogeneous aqueous
dilute solutions containing soluble fluoride ions with nearly
neutral pH. Regarding the first issue, the capability to accurately
elucidate reaction mechanisms involving nucleophilic transforma-
tions of P(V) compounds using ab initio and DFT methods,
both in the gas phase and in solution, had been demonstrated
many times during the past decade.8,9 Therefore, we presumed
that by using these methodologies and conducting a comple-
mentary mechanistic study, we shall gain a better understanding
of these unique transformations (i.e., formation and subsequently
hydrolysis of Et-G). Regarding the second question, from a
CWAs detoxification point of view, the homogeneous nature of
the decontaminant solution is critical for optimal decontamina-
tion of polluted areas. In our opinion there will be no limitation
for the use of fluoride solutions in the decontamination of CWAs.
Moreover, fluoride has low corrosiveness and low toxicity, and
in many countries it is used as an additive in drinking water
(fluoridation of water) for the prevention of tooth decay.
We disclose here our findings on the fluoride-catalyzed
degradation of VX in homogeneous neutral dilute aqueous
solutions. The mechanistic study relies on both experimental
and computational results.
water
11
108
149
nd
water
18
a
TBAF
KCl
water
53
water
3465
3438
−
−
none
water
<50
50
a
In all runs, 5 equiv of fluoride were used except for runs 2 and 7
b
where only 1 equiv was used. EMMP (94%) is the main product, and
EMPA (6%) is also formed.
times for both VX and Et-G degradation were calculated [t1/2
(VX) or t1/2 (Et-G)].
k1
k2
(1)
A → B → C
Figure 1 describes graphically a process in which VX (0.066M)
was dissolved in an aqueous solution containing potassium
Figure 1. Curve fitting for the degradation profiles of VX (66 mM),
Et-G formation and degradation, and EMPA formation in an aqueous
solution containing 5 equiv of KF (run 1).
RESULTS AND DISCUSSION
■
Kinetic Study of VX Degradation. The degradation of VX
in a series of dilute (0.066−0.33 M) aqueous fluoride salt (NaF,
KF, CsF, or TBAF) solutions was examined, and the effect of
the fluoride ion was determined in basic buffers, unbuffered
water, and methanol. A control experiment was also conducted
with KCl in unbuffered water. A fixed volume of a salt solution
was placed in an NMR tube, and an appropriate amount of VX
was dissolved in it. The degradation of VX was monitored
under various conditions and its outcome followed mainly by
31P NMR. The identity of the products was confirmed by
fluoride (5 equiv, Table 1, run 1). VX rapidly disappeared (t1/2
=
12 min), but only after full hydrolysis of Et-G (t1/2 = 102 min)
does the solution become free of toxic compounds. This
represents a typical reaction profile of most of the processes
studied in this paper (for data concerning other experiments, see
Figures S2−S14, Supporting Information). In most cases, before
ca. 20% conversion was reached, the reaction was relatively slow
as compared with the accelerated degradation observed later.
However, even in this time period the amount of EMPA was
always superior to that of Et-G (Figure 1). We hypothesized
that these effects may be due to gradual changes in pH that
may occur during the degradation process. This issue will be
discussed later.
comparison with authentic samples using 31P NMR, H NMR,
1
and 19F-NMR.6 We were satisfied to find that even in those
dilute fluoride solutions, VX was fully degraded to the final
nontoxic EMPA and DAT products. Moreover, the composi-
tion of the product mixtures and the reaction profiles were
similar to those found in the concentrated heterogeneous
10043
dx.doi.org/10.1021/jo301549z | J. Org. Chem. 2012, 77, 10042−10049

The Journal of Organic Chemistry
Article
Inspection of the data presented in Table 1 emphasizes the
dramatic effect of the fluoride ion on the degradation process.
To elucidate whether the reaction profile depends on solvation
properties of the species involved in the process, a similar
experiment was performed in methanol as an organic protic polar
solvent. A similar degradation profile was observed in methanol,
but the degradation rate of VX was 5-fold slower (t1/2 ∼ 69 min,
run 3). In addition, Et-G, which did not accumulate over 6%, was
converted to ethyl methyl methylphosphonate (EMMP) at a rate
similar to that found in water for the generation of EMPA. These
results imply that Et-G undergoes very fast methanolysis and/or
that direct methanolysis of VX is preferred over the formation
of Et-G. Our calculations unambiguously show that both
hypotheses are valid (vide infra).
pH Effect. Another parameter which may influence the
reaction profile is the pH. The initial pH of the fluoride solution
was ∼8.0. With the addition of the amino basic compound,
VX, the pH instantaneously increased to 10.7. A continuous
measurement of the pH during the degradation of VX (5 equiv
of KF in water, run 1) showed that the pH decreased from 10.7
to 7.0 toward the end of the reaction (see Figure 3). This
The degradation of VX in unbuffered water is quite slow (t1/2
∼
3438 min, run 9), and a significant amount of the toxic product
desethyl-VX is obtained (50%).4b,10 When only 1 equiv of KF
was used, VX was degraded in less than 6 h (t1/2 ∼ 46 min) and
the full hydrolysis of Et-G to EMPA took almost 2 days (t1/2
∼
580 min) (Table 1, run 2). Once the KF concentration was
increased by 5 fold, the degradation rates of VX and of Et-G
were 4 and 5 times faster, respectively. This result emphasizes
the importance of the fluoride salt concentration. It also implies
that VX disappears mainly through the reaction with the
fluoride ion, and that the basic medium induced by KF and VX
itself favors the hydrolysis of Et-G.
To learn about the role of the counterion, KF was replaced
by NaF, CsF, or tetrabutylammonium fluoride (TBAF) (runs
4−6). The rates obtained were of the same order of magnitude,
and no correlation could be found between the cation pro-
perties and either VX degradation or hydrolysis of Et-G (see
Figure 2). These results are well correlated with the prediction
Figure 3. pH change during the degradation of VX in the presence of
KF (run 1).
Figure 2. VX (66 mM) degradation profile in various media. The solvent
is water, and 5 equiv of fluoride salt are used unless otherwise specified.
decrease was fast at the beginning of the process and became
slower after full disappearance of VX (pH 8.11). Because Et-G
began to accumulate as the pH decreased, we hypothesized that
if the pH is maintained at its original value (∼10.5), any Et-G
formed will be immediately hydrolyzed to EMPA. Accordingly,
an experiment in which KF was dissolved in a carbonate−
bicarbonate buffer solution (0.1 M, pH 10.5) was performed.
As expected, no Et-G could be detected during the rapid
degradation of VX (t1/2 = 32 min, run 10), desethyl-VX was not
formed, and EMPA was the only product. We concluded that
at basic pH, the degradation of Et-G is favored in such a way
that its hydrolysis is faster than its formation. Interestingly,
in a buffered aqueous solution (pH 10.5) lacking fluoride, the
degradation rate of VX was very slow (t1/2 ∼ 2760 min, run 14)
and desethyl-VX was formed (40%) together with EMPA (50%)
and traces of methyl phosphonothioate ethyl ester (6%). Other
buffered solutions with pHs of 7.45, 8.84, and 9.65 were also
examined. As expected, while the medium becomes less basic,
the VX degradation rate increases, Et-G hydrolysis becomes
slower, and so more Et-G accumulates (see Table 2). At pH
7.45, we observed for the first time a situation in which the
amount of Et-G was superior to that of EMPA until VX almost
disappeared (see Figure 4). This profile entirely fits the above-
mentioned minimal kinetic model in eq 1, even at the beginning
of the reaction.
made by Churchill and Lee for the chelation of Me-VX (methyl
ester VX) with Na+ and K+ in the gas phase.11 They found a
minor P−S bond length contraction of 0.01 Å for K+ which
renders the P−S bond scission slightly harder than that for
Na+. This would account for the small difference seen in the
hydrolysis rate (runs 1 and 4). In the case of TBAF, two
concentrations were examined (1 and 5 equiv, runs 6 and 7),
and for the higher concentration, the degradation rate for VX
was increased by a factor of 3.
To evaluate the role of the fluoride ion, a control experiment
with KCl (5 equiv, run 8) was conducted. The reaction was
very slow (t1/2 ∼ 2.5 d), and desethyl-VX (32%) was formed
as in unbuffered water (run 9). This result emphasizes the
unique role of fluoride in contrast to other nucleophiles,
and particularly halides. As was previously shown with the
fluoride solid-supports (KF-Al2O3 and Amberlite IRA 900 F−),
the main advantage of using fluorides, in contrast to basic
media, is to prevent the desethyl-VX formation and to signifi-
cantly reduce the corrosivity.6,7 The hydrolysis of VX by
hydroxide solutions (KOH or NaOH) was intensively explored
and well documented.4 These reactions were found to be
strongly dependent on both the pH and the temperature and
always yielded a mixture of EMPA and desethyl-VX. The
present study shows that in basic buffers the presence of
fluoride not only improves the degradation rate of VX but also
prevents the formation of desethyl-VX.
In light of these results, it is clear that the reactions that occur
during the degradation process of VX begin with a very fast
10044
dx.doi.org/10.1021/jo301549z | J. Org. Chem. 2012, 77, 10042−10049

The Journal of Organic Chemistry
Article
Table 2. Degradation Rate of VX (66 mM) at Different pH
with KF (5 equiv) or without Fluoride
Scheme 2. Calculated Pathways for the Reaction of
O,S-Diethyl Methylphosphonothioate (1)
t1/2 (VX) t1/2 (Et-G) EMPA
desethyl-
VX (%)
run salt
media
(min)
(min)
(%)
a
10
11
12
13
14
KF buffer 10.50
32
17
−
17
20
44
100
100
−
−
−
−
a
KF buffer 9.65
b
c
KF buffer 8.84
20
100
100
54
b
c
KF buffer 7.45
5
a
−
buffer 10.50
2760
−
40
a
b
c
Carbonate−bicarbonate buffer (0.1 M). Tris buffer (0.1 M). Part of
EMPA is esterified by the triol present in Tris buffer.
were obtained from frequency calculations using B3LYP/
6-311G+(d,p) level of theory (Table S21, Supporting Informa-
tion). To check the validity of this DFT method, single-point
energies were also calculated at the MP2/6-31G+(d) level
(Table 3). Some relatively large discrepancies were found
a
Table 3. Relative Free Energies (kcal mol−1) for the
Reactions Shown in Scheme 2 and Transition States
relative free energies MP2/6-31G+(d)//
B3LYP/6-311G+(d,p)
reactions and transition states
gas phase
water
methanol
Figure 4. VX (66 mM) degradation profile in a pH 7.45 buffer
solution containing 5 equiv of KF (run 13).
TS1
−10.7
−12.0
−29.9
−37.5
−29.8
−32.9
−6.8
−10.7
−12.5
−16.2
−22.1
−40.7
−40.6
16.8
17.9
9.9
17.3
17.2
TS2
TS3(H)
nucleophilic substitution of the aminoethylthiolate group by
fluoride ion (k1) followed by hydrolysis (k2). The rates of both
steps depend on the pH. In basic media (pH > 10), in the
presence of fluoride, the hydrolysis is so fast that Et-G could
not be detected (k1 ≪ k2). As the pH is decreased to almost
neutral, the hydrolysis rate is not sufficient to immediately
decompose Et-G once it is formed (k1 > k2). Between basic and
neutral pH, the hydrolysis is fast enough so that the amount of
EMPA is superior to that of Et-G, but it is low enough to allow
the accumulation of Et-G. In nonbuffered solutions, the pH
decreases as EMPA is formed. The beginning of the process has
a profile similar to that found for buffer 9.65 (run 11), but later
the reaction is slowed down by the decrease in pH.
TS3(Me)
1.7
3.6
TS4(H)
10.6
19.1
16.9
TS4(Me)
TS5(H)
TS5(Me)
15.3
TS6(H)
TS6(Me)
13.3
PR1 (Et-G + EtS−)
PR2 (EMPA + EtS−)
PR3 (EMMP + EtS−)
−2.1
−24.0
−7.9
−33.1
a
Energies are relative to starting materials which are defined as zero.
Quantum Mechanical Study. Quantum mechanical
methods were previously shown to yield accurate results in
theoretical studies of analogous nucleophilic attacks on various
phosphates8 and phosphonates9 including CWAs.12 As a model
for VX we used O,S-diethyl methylphosphonothioate (1) and
studied the reactions’ pathways which may evolve from nucleo-
philic attacks by fluoride, hydroxide, or methoxide anions.
Similar analogy was previously used successfully by Patterson
and co-workers13 who produced results which were in excellent
agreement with the available experimental data.14 It should
also be noted that, theoretically, attack at the ester/thio ester
carbon is also possible, as shown to occur for several phos-
phate triesters in the gas phase.8c,g,h,15 Nevertheless, we9h,16
and others9e,13,17 showed that in solution only attack at the
phosphorus takes place as expected from hard nucleophiles.18
Therefore, we neglected processes which comprise an attack of
the nucleophiles on atoms other than P.
between the two methods. The most pronounced difference was
found for the energy values of the species that evolved from
fluoride attack in water. Moreover, recalculating these species at
the B3LYP level of theory using SMD as the solvation model,19
we obtained results much closer to those obtained at MP2 level
with IEF-PCM solvation model. Therefore, we use in our
discussion the more reliable MP2 results.
In principle, two modes of attack should be considered: one
is opposite to the thioester group or to the fluorine in Et-G (2),
and the second is opposite to the ester group (see odd and even
numbers for TSs, respectively, Figure 5). The preference
between the two modes should be dictated by the relative
stability of the trigonal bipyramidal transition states formed,
namely, according to the apicophilicity of the ligands around
the phosphorus center.20 Two possible pathways (paths 1
and 2, Scheme 2) for the formation of the final products
were studied. Path 1 is a two-step reaction in which there is a
nucleophilic attack by F− anion via either TS1 or TS2 on 1 to
initially form Et-G (2). The final product 3 (EMPA or EMMP)
is obtained via TS3 or TS4 by hydrolysis/methanolysis using
the corresponding nucleophile. Path 2, on the other hand,
would not involve the formation of Et-G but rather the direct
hydrolysis/methanolysis of 1, via TS5 or TS6. Calculating the
To carefully evaluate all plausible reaction pathways which
may evolve from an initial attack of either fluoride, hydroxide,
or methoxide nucleophile at the P center of the model
compound (Scheme 2), all starting materials, transition states,
and products were fully optimized in the gas phase and in
solution (water and methanol), and their thermodynamic data
10045
dx.doi.org/10.1021/jo301549z | J. Org. Chem. 2012, 77, 10042−10049

The Journal of Organic Chemistry
Article
Figure 5. Three-dimensional structures of the calculated transition states. (Orange = phosphorus, red = oxygen, gray = carbon, yellow = sulfur,
cyan = fluorine).
kinetics of path 2 in comparison to those of path 1 would shed
more light on our experimental results. In studying the
structure of all the possible transition states (Figure 5) it is
clear that TS1, TS3, and TS5, in which the corresponding P−S
and P−F bonds are apical, have the correct geometry needed
for the formation of products 2 or 3 following the cleavage of
the appropriate bonds. However, even if the favored transition
states are TS2, TS4, and TS6, the products 2 and 3 should be
obtained after a pseudorotation process of an initial trigonal
bipyramid intermediate, which is formed during the nucleo-
philic attack. Such pseudorotations are well-known processes
and are usually low in energy demand,9h,21 although in the case
of VX and its model, Patterson found them to be substantial
Figure 6. Reaction profiles for the degradation of VX-analogue 1 in
water.
and energetically costly.13 Nevertheless, by keeping in mind
that experimentally we did not detect by NMR any trace of
desethyl-VX (P−O cleavage), it is clear that only two modes
slightly exothermic by 2.1 kcal/mol. Once 2 is formed, it would
of attack involving P−S cleavage are relevant. The first should
comprise an attack opposite to the thiolate ligand (TS1 or
TS5) followed by cleavage of the P−S bond. The second would
involve pseudorotation of the intermediate which evolves from
attack opposite to the OEt ligand (TS2 or TS6), to form a new
intermediate in which the SEt and the CH3 fragments are both
apical. The latter intermediate would next undergo P−S bond
scission to give 2 or 3. In any case, Patterson established13c that
the initial attack is the rate-determining step. Therefore, on
the basis of his results, we feel confident to neglect any other
intermediates formed.
undergo an attack by OH−, via TS3(H) or TS4(H). We
calculated the activation free energies of these transition states
in water to be 12.0 and 12.7 kcal/mol, respectively, again in
good agreement with previously reported values for sarin.13b
Notably, although the energy difference between these two
transition states was calculated to be only 0.7 kcal/mol, both
should eventually lead to the formation of 3. The analysis leads
to the conclusion that in path 1, the attack of fluoride is the
rate-determining step. All the above explain very well the experi-
mental fact that Et-G is initially accumulating in substantial
amounts (for example, see run 1 and Figure 1) unless the pH is
basic (see run 10 and Figure S9, Supporting Information). We
cannot exclude the possibility of some equilibrium between 1
and 2 at room temperature and low pH, although a substantial
difference in activation barriers is expected, as PR1 is 2.1 kcal/
mol lower in energy than the starting materials (Table 3).
Calculated Reaction in Methanol. Comparing the
calculated values in water with those in methanol, a somewhat
different picture emerges (Figure 7). The calculated activation
energies for the formation of Et-G (path 1, TS1 or TS2) and
the direct methanolysis of 1 [path 2, TS5(Me) or TS6(Me)]
are 17.3, 17.2, 15.3, and 13.3 kcal/mol, respectively. Namely,
direct methanolysis by methoxide, either via attack opposite to
Calculated Reaction in Water. From examining the four
possible transition states (TS1, TS2, TS5, and TS6) which
evolve from a nucleophilic attack on 1 (Table 3 and Figure 6),
it is clear that the nucleophilicity of F− and OH− is similar with
a slight preference for the fluoride. This is unexpected if one
considers their relative basicity, as was also previously observed
by others.1 The calculated free energy for the formation of
these transition states are 16.8, 17.9, 19.1, and 16.9 kcal/mol,
respectively. It should also be noted that the barriers for the
OH− attack are in good agreement with previously reported
calculations by Patterson and co-workers.12b,c Thermodynami-
cally, we found that the formation of 2 from 1 in water is
10046
dx.doi.org/10.1021/jo301549z | J. Org. Chem. 2012, 77, 10042−10049

The Journal of Organic Chemistry
Article
Table 4. Degradation Rate of 1 (66 mM) with KF (5 equiv)
or without Fluoride in Aqueous Solutions
t1/2 (Et-G) EMPA desethyl-
run salt
media
t1/2 (1) (min)
(min)
(%)
1 (%)
15 KF water
861
571
−
−
100
100
−
−
9
a
16 KF buffer 10.50 876
a
b
b
17
−
buffer 10.50
>2.4 × 104
35
a
b
Carbonate−bicarbonate buffer. After 333 h reaction.
Figure 7. Reaction profiles for the degradation of VX-analogue 1 in
methanol.
1 is much slower than that of VX itself (compare run 15 versus
run 1, or run 16 versus run 10). We believe that the steric effect
mentioned above cannot explain by itself this remarkable
difference. This effect has to be thoroughly studied in a more
general perspective.
the thiolate or via attack opposite to the OEt followed by a fast
pseudorotation, is preferred by at least 2 kcal/mol over an
attack by F− (providing of course that the pH is basic).
Moreover, inspection of path 1 reveals that the formation of
Et-G is the rate-determining step, as the calculated free energy
of activation for the formation of TS3(Me) and TS4(Me) are
9.6 and 11.5 kcal/mol, respectively. The barrier for the second
step is significantly much lower in methanol than in water.
According to our calculations, both mechanisms (direct metha-
nolysis versus formation of Et-G followed by its methanolysis)
may coexist in methanol, albeit to a lesser extent than in the
parallel process in alkaline water. Thermodynamically, we found
that the formation of 2 from 1 in methanol is more exothermic
by 7.9 kcal/mol (in contrast with 2.1 kcal/mol found in water),
and the entire process is exothermic by 33.1 kcal/mol (compared
to 24.0 kcal/mol found in water). These computations support
the observation made in methanol (run 3) in which the level of
Et-G was very low all along the process because methanolysis
of VX was slightly preferred to fluoride attack, and in parallel,
methanolysis of Et-G was a fast process compared to its forma-
tion (Figure S3, Supporting Information). Moreover, it explains
the kinetic results in water compared with methanol (runs 1 and 3
and Figure 2).
Kinetics of VX-Model 1 Degradation. The calculated
energies in this work fit well with the above-mentioned
experimental results both in terms of kinetics and products
identity. In their comparative theoretical mechanistic study on
the hydrolysis of VX-model 113b versus VX,13c Patterson and
co-workers concluded that “the same mechanism is elucidated
by either the model or VX and it is apparent that nucleophilic
chemistry at the phosphorus atom in VX can be accurately
predicted using small model systems”. However, they also
reported some differences between these two phosphono-
thioate analogues. Among them, the most important is that the
bulky aminothiolate moiety in VX may pose a significant steric
effect on the stability of the trigonal bipyramide intermediate.
To clarify this point in our fluoride/methylphosphonothioate
system from both kinetics and products identity points of view,
we explored the reactions of VX-model 1 with fluoride under
the same conditions mentioned above. We found that this less
hindered amine-free methylphosphonothioate compound re-
acts with fluoride ions in a similar manner, in which Et-G is
formed as an intermediate that eventually hydrolyzed to EMPA
(Table 4). The increased reactivity of fluoride versus hydroxide
seems yet again, unambiguous, when inspecting the reactions
data of 1 with and without fluoride in basic buffer (runs 16 and 17).
However, we were quite surprised to find that the reaction of
CONCLUSION
■
We have shown by both theoretical and experimental studies
that the fluoride ion is very reactive toward the phosphorus
atom in compounds such as VX, especially in water.
Degradation experiments in dilute aqueous fluoride solutions
show that VX, an extremely toxic and environmentally stable
compound, rapidly converts to the nontoxic product EMPA via
the intermediate Et-G (5 equiv of fluoride; t1/2[VX] 12 min;
t
1/2[Et-G] 102 min). The toxic side product desethyl-VX, which
forms in substantial quantities during the degradation of VX in
unbuffered water, is completely avoided in this aqueous fluoride
homogeneous environment. In methanol, the reaction rates are
slower for the degradation of VX and Et-G by a factor of 6 and
1.5, respectively. Typically the hydrolysis of Et-G is the rate-
determining step, and it strongly depends on the pH evolving
during the reaction. When the pH is maintained at its initial
level (10.5) by a buffer, the hydrolysis of Et-G is so fast that no
Et-G can be detected. Our quantum mechanical calculations on
a VX-model clearly demonstrate the unique role of the fluoride
ion in this facile degradation process. The similar reactivity of
fluoride versus hydroxide is reflected by the significant low transi-
tion state energy in the case of fluoride which was unexpected
considering their relative basicity. We are currently extending our
study to other general organophosphorus compounds such as
phosphites, phosphonates, and phosphates, and more partic-
ularly, the reasons for the remarkable difference in reaction rates
of VX versus VX-model 1 with fluoride are under investigation.
EXPERIMENTAL SECTION
■
Materials. The organophosphorus compounds were of high purity
(>99%) and did not contain any of the hydrolysis products. VX was
obtained locally at IIBR using standard methods.22 The different
fluoride salts were purchased from Sigma-Aldrich.
NMR. 31P and 19F MAS NMR spectra were obtained at 202 and
471 MHz, respectively, on a 11.7 T (500 MHz) spectrometer.
Chemical shifts for 31P and 19F were referenced to external trimethyl
phosphate (TMP) and CFCl3, respectively, as 0 ppm. For 31P spectra,
the pulse delay was 2 s. The frequency offset was set between the
signals of VX and EMPA. For comparison purposes, spectra were
recorded under identical conditions.
Curve Fitting. The kinetic models for the degradation reactions
were solved by orthonormalization of the species population’s vectors
according to the Gram−Schmidt process. The time-dependent popula-
tion vectors were extracted from the NMR experiments by
10047
dx.doi.org/10.1021/jo301549z | J. Org. Chem. 2012, 77, 10042−10049

The Journal of Organic Chemistry
Article
mathematical integration over a best fitting to Gaussian functions. The
fit was preceded by the removal of the experimental noise using
singular value decompostion (SVD). The Gaussian function was
determined because of better convergence to ∼0.505 Hz spectral
resolution,23 this in comparison to Lorentz and Voigt functions which
give overestimation. The extracted populations were normalized to
one at each time sequence and then were fed into the mathematical
model.24,25 In all cases, the finite instrument response determined
by the interferometry time is included in the fitting procedure by
convolving the model matrix by Gaussian width of fwhm ∼2 min.
Description of the Computational Methods. Optimized
geometries and harmonic frequencies for all the reported species
(starting materials, transition states, intermediates, and products) were
obtained by density functional calculations using the B3LYP hybrid
functional and 6-311+G(d,p) basis set (corrected for unscaled ZPVE).26a
Gas-phase optimized structures were used as input for the calculations
in solution which were reoptimized at the same level of theory with
the integral equation formalism polarized continuum solvation model
(IEF-PCM).27 Energies from these calculations were converted to
thermodynamic data based on the frequency calculations. More accurate
electronic energies were obtained via single-point calculations at the
MP2/6-31G+(d) level of theory. These calculations were performed on
the converged geometries in the gas phase and in solution, respectively.26
Thermal correction to Gibbs free energy was obtained from frequency
calculations in the gas phase and in solution, respectively. All energies in
solution (at MP2 level) include both electrostatic and nonelectrostatic
corrections.26b All transition states were verified by calculating the
intrinsic reaction coordinates (IRC) and/or by examining the imaginary
frequency’s normal mode.28
Sample Preparation. Caution: These experiments should only be
performed by trained personnel using applicable safety procedures. Salt
solutions were prepared by dissolving the appropriate amounts of salt
(KF, NaF, CsF, TBAF, KCl) or KOH in water or buffered aqueous
solution or methanol. Two kinds of solutions were prepared containing
1 or 5 equiv of salts per OP equivalent. An amount of 1 or 5.3 μL
(13.2 mM or 66 mM) of the OP compound (VX or VX-model 1) was
applied via a syringe to a Teflon NMR tube containing 300 μL of the
salt solution. The tube was sealed with a Teflon cap and inserted in
a glass NMR tube. 31P and 19F MAS NMR spectra were measured
periodically to determine the amount of remaining starting material and
identify degradation products. The buffered solutions (0.1 M each)
were prepared by standard procedures,29 carbonate−bicarbonate buffer
for pH 10.5 and 9.65 and Tris buffer for pH 8.64 and 7.45. Tris buffer
contains 3-amino-3-(2-hydroxyethyl)pentane-1,5-diol, which reacts
with EMPA to give the corresponding ester detected by 31P NMR at
δ 32 ppm. It was assumed that it does not dramatically affect the
outcome of the processes.
REFERENCES
■
(1) Selected articles: (a) Kirby, A. J.; Lima, M. F.; da Silva, D.;
Roussev, C. D.; Nome, F. J. Am. Chem. Soc. 2006, 128, 16944−16952.
(b) Kirby, A. J.; Lima, M. F.; da Silva, D.; Nome, F. J. Am. Chem. Soc.
2004, 126, 1350−1351. (c) Admiraal, S. J.; Herschlag, D. J. Am. Chem.
Soc. 1999, 121, 5837−5845. (d) Herschlag, D.; Jencks, P. J. Am. Chem.
Soc. 1990, 112, 1951−1956. (e) Kirby, A. J.; Younas, M. J. Chem. Soc.
(B) 1970, 1165−1172. (f) Khan, S. A.; Kirby, A. J. J. Chem. Soc. B
1970, 1172−1182.
(2) (a) Fey, N.; Garland, M.; Hopewell, J. P.; McMullin, C. L.;
Mastroianni, S.; Orpen, A. G.; Pringle, P. G. Angew. Chem., Int. Ed.
2012, 51, 118−122. (b) Grant, D. J.; Matus, M. H.; Switzer, J. R.;
Dixon, D. A.; Francisco, J. S.; Christe, K. O. J. Phys. Chem. A 2008,
112, 3145−3156. (c) Kirby, A. J.; Warren, S. G. In The Organic
Chemistry of Phosphorus; Elsevier Publishing Co.: London, 1966; p 5.
(3) Van der Schans, M. J.; Polhuijs, M.; Van Dijk, C.; Degenhardt, C.
E. A. M.; Pleijsier, K.; Langenberg, J. P.; Benschop, H. P. Arch. Toxicol.
2004, 78, 508−524.
(4) Selected articles: (a) Kim, K.; Tsay, O. G.; Atwood, D. A.;
Churchill, D. G. Chem. Rev. 2011, 111, 5345−5403. (b) Groenewold,
G. S. Main Group Chem. 2010, 9, 221−244. (c) Brevett, C. A. S.;
Sumpter, K. B. Main Group Chem. 2010, 9, 205−219. (d) Vayron, P.;
Renard, P.-Y.; Taran, F.; Creminon, C.; Frobert, Y.; Grassi, J.;
Mioskowski, C. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 7058−7063.
(e) Yang, Y.-C. Acc. Chem. Res. 1999, 32, 109−115.
(5) Cragan, J. A.; Ward, M. C.; Mueller, C. B. J. Hazard. Mater. 2009,
170, 72−78.
(6) Marciano, D.; Goldvaser, M.; Columbus, I.; Zafrani, Y. J. Org.
Chem. 2011, 76, 8549−8553.
(7) (a) Zafrani, Y.; Yehezkel, L.; Goldvaser, M.; Marciano, D.;
Waysbort, D.; Gershonov, E.; Columbus, I. Org. Biomol. Chem. 2011,
9, 8445−8451. (b) Gershonov, E.; Columbus, I.; Zafrani, Y. J. Org.
Chem. 2009, 74, 329−338.
(8) Examples for phosphates: (a) Dejaegere, A.; Karplus, M. J. Am.
Chem. Soc. 1993, 115, 5316−5317. (b) Dejaegere, A.; Liang, X.;
Karplus, M. J. Chem. Soc., Faraday Trans. 1994, 90, 1763−1770.
(c) Chang, N.-Y.; Lim, C. J. Phys. Chem. A 1997, 101, 8706−8715.
(d) Chang, N.-Y.; Lim, C. J. Am. Chem. Soc. 1998, 120, 2156−2167.
(e) Florian, J.; Warshel, A. J. Phys. Chem. B. 1998, 102, 719−734.
(f) Hu, C.-H.; Brinck, T. J. Phys. Chem. A 1999, 103, 5379−5386.
(g) Lopez, X.; Dejaegere, A.; Karplus, M. J. Am. Chem. Soc. 2001, 123,
11755−11763. (h) Menegon, G.; Loos, M.; Chaimovich, H. J. Phys.
Chem. A 2002, 106, 9078−9084. (i) Arantes, G. M.; Chaimovich, H. J.
Phys. Chem. A 2005, 109, 5625−5635. (j) Lop
Lera, A. R.; York, D. M. Chem.Eur. J. 2005, 11, 2081−2093.
(k) Iche-Tarrat, N.; Barthelat, J.-C.; Rinaldi, D.; Vigroux, A. J. Phys.
́
ez, C. S.; Faza, O. N.; de
́
Chem. B 2005, 109, 22570−22580. (l) Lopez, X.; Dejaegere, A.;
Leclerc, F.; York, D. M.; Karplus, M. J. Phys. Chem. B 2006, 110,
11525−11539.
ASSOCIATED CONTENT
* Supporting Information
■
S
(9) Examples for phosphonates: (a) Thatcher, G. R. J.; Campbell, A.
S. J. Org. Chem. 1993, 58, 2272−2281. (b) Zheng, F.; Zhan, C.-G.;
Ornstein, R. L. J. Chem. Soc,. Perkin Trans. 2 2001, 2355−2363.
(c) Bell, A. J.; Citra, A.; Dyke, J. M.; Ferrante, F.; Gagliardi, L.; Watts,
P. Phys. Chem. Chem. Phys. 2004, 6, 1213−1218. (d) Barbosa, A. C. P.;
Borges, I., Jr.; Lin, W. O. J. Mol. Struct. (THEOCHEM) 2004, 712,
187−193. (e) Doskocz, M.; Roszak, S.; Majumdar, D.; Doskocz, J.;
Gancarz, R.; Leszczynski, J. J. Phys. Chem. A 2008, 112, 2077−2081.
(f) McAnoy, A. M.; Paine, M. R. L.; Blanksby, S. J. Org. Biomol. Chem.
2008, 6, 2316−2326. (g) Mandal, D.; Mondal, B.; Das, A. K. J. Phys.
Chem. A 2010, 114, 10717−10725. (h) Ashkenazi, N.; Segall, Y.;
Chen, R.; Sod-Moriah, G.; Fattal, E. J. Org. Chem. 2010, 75, 1917−
1926.
Kinetic data of the reactions, NMR spectra, absolute energies,
and number of imaginary frequencies and optimized geometries
for all calculated species performed in different solutions. This
material is available free of charge via the Internet at http://
pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
*E-mail: yossiz@iibr.gov.il; nissan.ashkenazi@iibr.gov.il.
■
Author Contributions
§These authors contributed equally.
(10) Szafraniec, L. J.; Szafraniec, L. L.; Beaudry, W. T.; Ward, J. R.
ADA-250773; U.S. Army Chemical Research, Development and
Engineering Center, Aberdeen Proving Ground: Aberdeen, MD, 1990.
(11) Bandyopadhyay, I.; Kim, M. J.; Lee, Y. S.; Churchill, D. G. J.
Phys. Chem. A 2006, 110, 3655−3661.
(12) (a) Zheng, F.; Zhan, C.-G.; Ornstein, R. L. J. Chem. Soc., Perkin
Trans. 2 2001, 2355−2363. (b) Menke, J. L.; Patterson, E. V. J. Mol.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
This work was internally funded by the Israeli Prime Minister’s
office.
■
10048
dx.doi.org/10.1021/jo301549z | J. Org. Chem. 2012, 77, 10042−10049

The Journal of Organic Chemistry
Article
Struct. (THEOCHEM) 2007, 811, 281. (c) McAnoy, A. M.; Williams,
J.; Paine, M. R.; Rogers, M. L.; Blanksby, S. J. J. Org. Chem. 2009, 74,
9319−9327. (d) Bera, N. C.; Maeda, S.; Morokuma, K.; Viggiano, A.
A. J. Phys. Chem. A 2010, 114, 13189−13197. (e) Midey, A. J.; Miller,
T. M.; Viggiano, A. A. Int. J. Mass Spectrom. 2012, 315, 1−7.
(13) (a) Patterson, E. V.; Cramer, C. J. J. Phys. Org. Chem. 1998, 11,
̌
232−240. (b) Seckute, J.; Menke, J. L.; Emnett, R. J.; Patterson, E. V.;
̌
̇
Cramer, C. J. J. Org. Chem. 2005, 70, 8649−8660. (c) Daniel, K. A.;
Kopff, L. A.; Patterson, E. V. J. Phys. Org. Chem. 2008, 21, 321−328.
(14) Yang, Y.-C. Acc. Chem. Res. 1999, 32, 109−115.
(15) (a) Hodges, R. V.; Sullivan, S. A.; Beauchamp, J. L. J. Am. Chem.
Soc. 1980, 102, 935−938. (b) Lum, R. C.; Grabowski, J. J. J. Am. Chem.
Soc. 1992, 114, 8619−8627.
(16) Ashkenazi, N.; Zade, S. S.; Segall, Y.; Karton, Y.; Bendikov, M.
Chem. Commun. 2005, 5879−5881.
(17) (a) Haake, P. C.; Westheimer, F. H. J. Am. Chem. Soc. 1961, 83,
1102−1109. (b) Kluger, R.; Covitz, F.; Dennis, E.; Williams, L. W.;
Westheimer, F. H. J. Am. Chem. Soc. 1969, 91, 6066−6072. (c) Kluger,
R.; Thatcher, G. R. J. Am. Chem. Soc. 1985, 107, 6006−6011.
(d) Kluger, R.; Thatcher, G. R. J. Org. Chem. 1986, 51, 207−212.
(18) Bunton, C. A. Acc. Chem. Res. 1970, 3, 257−265.
(19) Cramer, C. J.; Truhlar, D. G. Acc. Chem. Res. 2008, 41, 760−768.
(20) (a) Trippett, S. Pure Appl. Chem. 1974, 40, 595−604.
(b) Trippett, S. Phosphorus, Sulfur Relat. Elem. 1976, 1, 89−98.
(21) (a) Westheimer, F. H. Acc. Chem. Res. 1968, 1, 70−78.
(b) Gillespie, P.; Hoffman, P.; Klusacek, H.; Marquarding, D.; Pfohl,
S.; Ramirez, F.; Tsolis, E. A.; Ugi, I. Angew. Chem., Int. Ed. 1971, 10,
687−715. (c) Gillespie, P.; Ramirez, F.; Ugi, I.; Marquarding, D.
Angew. Chem., Int. Ed. 1973, 12, 91−119. (d) Hoffman, R.; Howell, J.
M.; Muetterties, E. L. J. Am. Chem. Soc. 1972, 94, 3047−3058.
(22) Black, R. M.; Harrison, J. M. The chemistry of organo-
phosphorus chemical warfare agents. In The chemistry of organo-
phosphorus compounds; Hartley, F. R., Ed.; John Wiley and Sons Ltd.:
Chichester, England, 1996; Vol 4, pp 783−798.
(23) The procedure effect is depicted in Figure S1 in the Supporting
Information.
(24) Ernsting, N. P.; Kovalenko, S. A.; Senyushkina, T.; Saam, J.;
Farztdinov, V. J. Phys. Chem. A 2001, 105, 3443−3453.
(25) Shoshana, O.; Pe’rez Lustres, J. L.; Ernsting, N. P.; Ruhman, S.
Phys. Chem. Chem. Phys. 2006, 8, 2599−2609.
(26) (a) Gaussian 03, Revision D.01; Gaussian, Inc., Wallingford, CT,
2004. (b) Gaussian 09, Revision B.01; Gaussian, Inc., Wallingford, CT,
2010. For full citation, see Supporting Information.
(27) (a) Cossi, M.; Barone, V.; Cammi, R.; Tomasi, J. J. Chem. Phys.
Lett. 1996, 255, 327−335 and references cited therein. (b) Cances, E.;
Mennucci, B.; Tomasi, J. J. Chem. Phys. 1997, 107, 3032−3041.
(28) Gonzalez, C.; Schlegel, H. B. J. Chem. Phys. 1989, 90, 2154−
2161.
(29) Mohan, C. Buffers - A guide for the preparation and use of buffers
in biological systems; CALBIOCHEM, EMD Biosciences, Inc., an
Affiliate of Merck KGaA: Darmstadt, Germany, 2003.
10049
dx.doi.org/10.1021/jo301549z | J. Org. Chem. 2012, 77, 10042−10049