Experimental and theoretical studies on the thermal decomposition of heterocyclic nitrosimines

Article abstract of DOI:10.1021/ja010659k

A series of substituted 2-nitrosiminobenzothiazolines (2) were synthesized by the nitrosation of the corresponding 2-iminobenzothiazolines (6). Thermal decomposition of 2a-f and of the seleno analogue 7 in methanol and of 3-methyl-2-nitrosobenzothiazoline (2a) in acetonitrile, 1,4-dioxane, and cyclohexane followed first-order kinetics. The activation parameters for thermal deazetization of 2a were measured in cyclohexane (ΔH? = 25.3 ± 0.5 kcal/mol, ΔS? = 1.3 ± 1.5 eu) and in methanol (ΔH? = 22.5 ± 0.7 kcal/mol, ΔS? = -12.9 ± 2.1 eu). These results indicate a unimolecular decomposition and are consistent with a proposed stepwise mechanism involving cyclization of the nitrosimine followed by loss of N₂. The ground-state conformations of the parent nitrosiminothiazoline (9a) and transition states for rotation around the exocyclic C=N bond, electrocyclic ring closure, and loss of N₂ were calculated using ab initio molecular orbital theory at the MP2/6-31G* level. The calculated gas-phase barrier height for the loss of N₂ from 9a (25.2 kcal/mol, MP4(SDQ, FC)/6-31G*//MP2/6-31G* + ZPE) compares favorably with the experimental barrier for 2a of 25.3 kcal/mol in cyclohexane. The potential energy surface is unusual; the rotational transition state 9a-rot-ts connects directly to the orthogonal transition state for ring-closure 9aTS. The decoupling of rotational and pseudopericyclic bond-forming transition states is contrasted with the single pericyclic transition state (15TS) for the electrocyclic ring-opening of oxetene (15) to acrolein (16). For comparison, the calculated homolytic strength of the N-NO bond is 40.0 kcal/mol (MP4(SDQ, FC)/6-31G* + ZPE).

Full text of DOI:10.1021/ja010659k

VOLUME 123, NUMBER 31
AUGUST 8, 2001
© Copyright 2001 by the
American Chemical Society
Experimental and Theoretical Studies on the Thermal Decomposition
of Heterocyclic Nitrosimines1
Richard A. Bartsch,* Yeh Moon Chae, Sihyun Ham, and David M. Birney*
Contribution from the Department of Chemistry and Biochemistry, Texas Tech UniVersity,
Lubbock, Texas 79409-1061
ReceiVed March 12, 2001
Abstract: A series of substituted 2-nitrosiminobenzothiazolines (2) were synthesized by the nitrosation of the
corresponding 2-iminobenzothiazolines (6). Thermal decomposition of 2a-f and of the seleno analogue 7 in
methanol and of 3-methyl-2-nitrosobenzothiazoline (2a) in acetonitrile, 1,4-dioxane, and cyclohexane followed
first-order kinetics. The activation parameters for thermal deazetization of 2a were measured in cyclohexane
(∆Hq ) 25.3 ( 0.5 kcal/mol, ∆Sq ) 1.3 ( 1.5 eu) and in methanol (∆Hq ) 22.5 ( 0.7 kcal/mol, ∆Sq )
-12.9 ( 2.1 eu). These results indicate a unimolecular decomposition and are consistent with a proposed
stepwise mechanism involving cyclization of the nitrosimine followed by loss of N2. The ground-state
conformations of the parent nitrosiminothiazoline (9a) and transition states for rotation around the exocyclic
CdN bond, electrocyclic ring closure, and loss of N2 were calculated using ab initio molecular orbital theory
at the MP2/6-31G* level. The calculated gas-phase barrier height for the loss of N2 from 9a (25.2 kcal/mol,
MP4(SDQ, FC)/6-31G*//MP2/6-31G* + ZPE) compares favorably with the experimental barrier for 2a of
25.3 kcal/mol in cyclohexane. The potential energy surface is unusual; the rotational transition state 9a-rot-ts
connects directly to the orthogonal transition state for ring-closure 9aTS. The decoupling of rotational and
pseudopericyclic bond-forming transition states is contrasted with the single pericyclic transition state (15TS)
for the electrocyclic ring-opening of oxetene (15) to acrolein (16). For comparison, the calculated homolytic
strength of the N-NO bond is 40.0 kcal/mol (MP4(SDQ, FC)/6-31G*//MP2/6-31G* + ZPE).
Introduction
explored as anticancer and antithrombotic drugs.7 The most
widely studied reactions of nitroso compounds have been the
homolysis to form NO.2b However, other reactions are known;
for example, N-nitrosoaziridenes lose N2O to form alkenes in a
The chemistry of nitroso compounds has been a subject of
great interest to a broad audience. The most well-known
reactions are ones that generate and trap nitric oxide (NO).2
These reactions have recently been shown to be important in a
variety of biochemical signaling pathways,3 and some of these
pathways have captured the attention of nonchemists as well.4
NO is also believed to be involved in the nitrogen cycle.5
Historically, nitrosimines were studied because they were
potential carcinogens.6 More recently N-nitrosoureas have been
(2) (a) Reszka, K. J.; Bilski, P.; Chignell, C. F. J. Am. Chem. Soc. 1996,
118, 8719-8720. (b) Cheng, J.-P.; Xian, M.; Wang, K.; Zhu, X.; Yin, Z.;
Wang, P. G. J. Am. Chem. Soc. 1998, 120, 10266-10267. (c) Richter-
Addo, G. B. Acc. Chem. Res. 1999, 32, 529-536. (d) Williams, D. L. H.
Acc. Chem. Res. 1999, 32, 869-876. (e) Turjanski, A. G.; Leonik, F.; Estrin,
D. A.; Rosenstein, R. E.; Doctorovich, F. J. Am. Chem. Soc. 2000, 122,
10468-10469. (f) Xian, M.; Xhu, X.-Q.; Lu, J.; Wen, Z.; Cheng, J.-P. Org.
Lett. 2000, 2, 265-268. (g) Lu¨, J.-M.; Wittbrodt, J. M.; Wang, K.; Wen,
Z.; Schlegel, H. B.; Wang, P. G.; Cheng, J.-P. J. Am. Chem. Soc. 2001,
123, 2903-2904. (h) Zhu, X.-Q.; He, J.-Q.; Li, Q.; Xian, M.; Lu, J.; Cheng,
J.-P. J. Org. Chem. 2000, 65.
(1) Taken in part from the MS thesis of Chae, Y. M. Mechanism of the
Thermal Decomposition of Heterocyclic Nitrosimines; Washington State
University, 1972; p 48.
10.1021/ja010659k CCC: $20.00 © 2001 American Chemical Society
Published on Web 07/11/2001

7480 J. Am. Chem. Soc., Vol. 123, No. 31, 2001
Bartsch et al.
cheletropic fragmentation.8 In contrast, N-nitrosimines (1)
spontaneously lose nitrogen rather than NO.6b,9 This gives the
corresponding ketone, as illustrated in eq 1. Loss of nitrogen
has been shown to be quantitative.9a In one earlier study,
nitrogen evolution was shown to proceed via first-order kinetics
(t1/2 ) 4.75 h) for 1 with R1 ) 4-MePh and R2 ) 2-MePh.9b
Steric bulk and conjugation stabilize nitrosimines. For example,
1 with R1 ) t-Bu and R2 ) 2-MePh is stable for weeks, but no
dialkylnitrosimines are known.9b The 3-substituted 2-nitrosimi-
nobenzothiazonines (e.g., 2a) are among the more stable
derivatives known and require heating, for example in refluxing
methanol, for deazetization.6b
to the radical pathway. The differences in the products observed
in eq 2 as compared to eq 3 require that the two mechanisms
be different and are consistent with a concerted reaction for the
thermal deazetization of unhindered nitrosimines.
The work reported herein was undertaken to address several
questions arising from the previous results and to explore the
mechanism of the reaction. First, a detailed kinetic study was
performed to obtain activation energies and to confirm that the
formation of ketones is indeed first-order, in agreement with
the observed first-order formation of nitrogen. Second, solvent
effects were studied to give insight into the changes in charge
distribution in going to the transition state. Last, ab initio
calculations were used to explore the viability and the details
of the proposed mechanism (eq 2). These predict that the
reaction follows an unusual two-stage mechanism with two
sequential transition states (CdN rotation followed by C-O
bond formation) for the formation of the spiro intermediate (e.g.,
3).
For the thermal reaction, a unimolecular, two-step mechanism
has been proposed as shown in eq 2.9b In this mechanism, a
concerted electrocyclization is envisioned to form the strained
four-membered ring in 3, followed by a presumably forbidden,10
but highly exothermic, deazetization to give 4. It was this
unusual mechanism that first drew our attention to this reaction.
The electrocyclic ring closure is, at first glance, a 4-electron
process, analogous to the cyclization of butadiene10 or ac-
rolein.10,11 This would be expected to involve rotation around
the CdN bond coupled with C-O bond formation.11 Interest-
ingly, the calculated pathway reported herein finds those two
motions to be uncoupled.
Experimental Results and Discussion
A series of 3-substituted 2-nitrosiminobenzothiazolines
(2a-e) as well as the disubstituted analogue 2f were prepared
following literature precedents, as shown in eq 4. The reaction
of an arylamine hydrochloride with potassium thiocyanate gave
the corresponding unsymmetrical thiourea 5.14 Oxidative cy-
clization with Br2 gave the iminobenzothiazolines (6).9a,14a,15
Finally, treatment with sodium nitrite in acetic acid gave the
nitrosiminobenzothiazoline (2).9a,b,15b The nitrosiminobenzo-
selenazoline (7) was similarly prepared. However, an attempt
to synthesize the oxygen analogue. 8, gave only the correspond-
ing ketone.
Photolysis of the strongly colored nitrosimines, such 2a, gives
rise to a variety of products that appear to have come from loss
of NO and subsequent radical reactions (eq 3).12 Similar
products, indicative of radical reactions, are also observed in
the thermolysis of sterically hindered nitrosimines (e.g., 1, with
R1 ) t-Bu and R2 ) 2-MePh).13 Steric constraints were proposed
disfavor the cyclic pathway (eq 2), thus diverting the reaction
(3) (a) Palmer, R. M. J.; Ferrige, A. G.; Moncada, S. Nature 1987, 327,
524. (b) Butler, A. R.; Williams, D. L. H. Chem. Soc. ReV. 1993, 22, 233-
241.
(4) For the biochemical pathway involved in the biological activity of
Viagra, see: (a) Terrett, N. K.; Bell, A. S.; Ellis, P. Bioorg. Med. Chem.
Lett. 1996, 6, 1819. (b) Wilson, E. K. Chem. Eng. News 1998, 76, June 29,
29-33.
(5) Averill, B. A. Chem. ReV. 1996, 96, 2951-2964.
(6) (a) Anselme, J. P. N-Nitrosamines; American Chemical Society:
Washington, DC, 1979. (b) Challis, B. C.; Challis, J. A. In N-Nitrosimines
and N-Nitrosoimines; Patai, S., Ed.; John Wiley and Sons: New York, 1982;
Vol. 2, pp 1151-1223. (c) Olszewska, T.; Milewska, M. J.; Gdaniec, M.;
Malusznska, H.; Polonski, T. J. Org. Chem. 2001, 66, 510-506.
(7) (a) Gnewuch, C. T.; Sosnovsky, G. Chem. ReV. 1997, 97, 829-1013.
(b) Rehse, K.; Ciborski, T. Arch. Pharm. (Weinheim, Ger.) 1995, 328, 71.
(8) (a) Rundel, W.; Mu¨ller, E. Chem. Ber. 1963, 96, 2528-2531. (b)
Clark, R. D.; Helmkamp, G. K. J. Org. Chem. 1964, 29, 1316-1320.
(9) (a) Besthorn, E. Chem. Ber. 1910, 43, 1519-1526. (b) Thoman, C.
J.; Hunsberger, S. J.; Hunsberger, I. M. J. Org. Chem. 1968, 33, 2852-
2857. (c) Akiba, K.-y.; Ishidawa, K.; Inamoto, K. Bull. Chem. Soc. Jpn.
1978, 51, 535-539.
Thermal decomposition of 3-methyl-2-nitrosiminobenzothia-
zoline (2a) has previously been shown to quantitatively evolve
nitrogen.9a To confirm the identity and yield of the organic
product as 4a, 2a was refluxed in methanol at 62.3° for 5.1
half-lives. A quantitative yield of 3-methyl-2-benzothiazolinone
(4a) was estimated from the UV spectrum in methanol, based
on independently synthesized 4a. (See Supporting Information
for details.)
(10) Woodward, R. B.; Hoffmann, R. The ConserVation of Orbital
Symmetry; Verlag Chemie, GmbH: Weinheim, 1970.
(11) Yu, H.; Chen, W.-T.; Goddard, J. D. J. Am. Chem. Soc. 1990, 112,
7529-7537.
(14) (a) Passing, H. J. Prakt. Chem. 1939, 153, 1-25. (b) Fairful, A. E.
S.; Peak, D. A. J. Chem. Soc. 1955, 796-802.
(15) (a) Fuchs, K.; Grauaug, E. Chem. Ber. 1928, 61, 57-65. (b) Tsuda,
K.; Fukushima, S. J. Pharm. Soc., Jpn. 1942, 62, 64.
(12) Akiba, K.; Fukawa, I.; Mashita, K.; Inamoto, N. Tetrahedron Lett.
1968, 2859-2862.
(13) Jappy, J.; Preston, P. N. Tetrahedron Lett. 1970, 1157-1160.

Thermal Decomposition of Heterocyclic Nitrosimines
J. Am. Chem. Soc., Vol. 123, No. 31, 2001 7481
Table 2. Rates of Decomposition of
Table 1. Rate of Decomposition of 3-Substituted
2-Nitrosiminobenzothiazolines 2a-f and of 7 in Methanol at 62.3°
3-Methyl-2-nitrosiminobenzothiazoline (2a) in Various Solvents at
62.3°
k1 × 105
relative
rate
3-substituent 6-substituent
sec-1
σ*
solvent
k1 × 105, sec-1
relative rate
Z, kcal/molb
2a
2b
2c
2d
2e
2f
7 (Se)
CH3
H
H
H
H
H
OCH3
H
2.06a
1.00
0.54
0.49
2.84
0.99
0.45
0.57
0.00
-0.10
-0.12
+0.60
+0.22
acetonitrile
methanol
1,4-dioxane
cyclohexane
0.94 ( 0.01
0.46
1.00
1.61
9.32
71.3
83.6
66c
C2H5
n-C3H7
Ph
CH2Ph
CH3
1.10 ( 0.03
1.00 ( 0.03
5.86 ( 0.39
2.04 ( 0.03
0.93 ( 0.04
1.17 ( 0.01
2.06a
3.35 ( 0.05
19.2 ( 0.6
54
a Calculated from the Arrhenius plot (see Supporting Information).
b Z value, as a measure of solvent polarity, ref 17. c Estimated from
the data for acetone.
CH3
a Calculated from the Arrhenius plot (see Supporting Information).
Table 3. Activation Energies (Ea), Enthalpies (∆Hq) and Entropies
(∆Sq) for the Decomposition of
3-Methyl-2-nitrosiminobenzothiazoline (2a) in Cyclohexane and
Methanol
solvent
Ea, kcal/mol
∆Hq, kcal/mol
∆Sq, eua
methanol
cyclohexane
23.1 ( 0.71
26.0 ( 0.50
22.5 ( 0.7
25.3 ( 0.5
-12.9 ( 2.1
1.3 ( 1.5
a Calculated at 60°.
substitutent in 2d and the substitution of Se for S in 7. However,
the observed substituent effects are very small. Furthermore,
the products are not consistent with those expected from a
radical pathway, such as appears to be operative in the
photochemical deazetization of 2.12 Finally, the activation
entropies, ∆Sq of 1.3 eu in cyclohexane and of -12.9 eu in
methanol, are inconsistent with the large and positive activation
entropy anticipated for a fragmentation reaction. Thus, all of
these results are consistent with the mechanism presented in eq
2.
Changing from a nonpolar to a polar solvent makes only small
changes in the rate. Qualitatively, the reaction is faster in
cyclohexane, the least polar solvent. This indicates that the
transition state is less polar than the reactants 2a. This is
reasonable, given that polar, delocalized resonance structures
can be drawn for nitrosimines. However, the reaction is only
21 times slower in acetonitrile at 62.3°, indicating that there
are not significant changes in polarity between 2a and the
transition state. Quantitatively, the rate for decomposition of
2a does not correlate well with the Z value as a measure of the
solvent polarity (Table 2).17 Methanol is the most polar of the
four solvents studied, but the rate is slowest in acetonitrile and
fastest in cyclohexane. It could be that there is some hydrogen
bonding between the nitrosimine and methanol that is enhanced
in the transition state, and this increases the rate. This would
be consistent with the more negative value of ∆Sq (-12.9 eu)
measured for the deazetization in methanol as compared to that
in cyclohexane (1.3 eu)
Figure 1. Rate of decomposition of nitrosimines 2 in methanol at 62.5°
versus Taft σ* steric parameter. R2 ) 0.92.
The primary focus of this experimental work was the kinetics
of the decomposition of nitrosiminobenzothiazolines 2a-f. The
kinetics were measured by monitoring the disappearance of the
nitrosimine absorptions at 340-360 nm over time. The raw
kinetic data are presented and shown graphically in the
Supporting Information. Control experiments demonstrated that
3-methyl-2-nitrosiminobenzothiazoline (2a) in methanol follows
Beer’s law over the range of concentrations (2.97-9.35) × 10-5
M and this method gave reproducible experimental rates. (See
Figures S1 and S2, Supporting Information.) The decomposition
of the substituted nitrosiminobenzenothiazolines 2a-f and the
selenium analogue 7 followed good first-order kinetics. The
measured rate constants are reported in Table 1.
The rate of deazetization is only slightly affected by changes
in the 3- and 6-substituents. The largest change is for phenyl
substitution (2d) which increases the rate by a factor of 2.84
compared with 2a, while addition of a 6-methoxy substituent
(2f) reduces the rate by a factor of 0.45. This admittedly limited
range of substituents showed a good correlation with the Taft
σ* (steric) parameter,16 as shown in Figure 1. Replacement of
sulfur in 2a by selenium in 7 likewise makes only a modest
effect, reducing the rate by a factor of 0.57.
Changing the solvent has only a modest effect on the rate of
decomposition of 2a (Table 2). The rate is fastest in cyclohexane
and slowest in acetonitrile. The temperature dependence of the
rate of decomposition of 2a was determined in both methanol
and cyclohexane. The derived activation energies (Ea), enthalpies
(∆Hq), and entropies (∆Sq) are presented in Table 3.
The decompositions of 2a-f and 7 follow first-order kinetics.
This would be consistent either with the proposed mechanism
in eq 2 or with a radical fragmentation mechanism (eq 3). In a
radical mechanism, the substituents would be expected to have
a larger influence on the rate, particularly comparing the phenyl
As originally proposed, deazetization of 2 might proceed via
the intermediate 3 (eq 2). This is a unimolecular reaction, in
agreement with the observed first-order kinetics. The intermedi-
ates 3 would be expected to be less polar than the nitrosimines
2, consistent with the slower rates measured in more polar
solvents. The change in hybridization from sp2 to sp3 would
increase steric crowding between R1 and R2 in acyclic nitros-
imines 1 and would disrupt the heteroaromaticity of 2. Thus,
this proposed mechanism is also consistent with the solvent
effects and the stability trends.
Theoretical Studies
An ab initio study of the two-step pathway was undertaken
with the expectation that it would provide additional insight
(16) Lowry, T. H.; Richardson, K. S. Mechanism and Theory in Organic
Chemistry, 3rd ed.; Harper & Row: New York, 1987.
(17) Kosower, E. M. J. Am. Chem. Soc. 1958, 80, 3235.

7482 J. Am. Chem. Soc., Vol. 123, No. 31, 2001
Bartsch et al.
Table 4. Relative Energies (kcal/mol) for the Reactions of 9a, 9b, and 13, Optimized at the MP2/6-31G* level
structure
RHF/6-31G*a
MP2b
MP2b∆G
MP3b
MP4(SDQ)b
MP4b+ZPEc
R ) H
0.00
9a
9a(out)
9a′
0.00
-2.56
0.94
0.00
2.62
3.17
0.00
-0.57
2.25
0.00
0.35
2.66
0.00
0.23
2.67
1.67
2.62
9a′(out)
∆E(rot)d
∆E(rot)-oute
9aTS
10a
10aTS
11a + N2
-0.98
4.52
3.62
1.30
2.43
2.34
17.36
17.64
17.39
23.95
19.78
28.57
-89.95
17.01
16.32
23.13
18.71
26.29
-102.96
18.06
18.33
24.00
14.35
28.78
-90.89
17.29
17.68
22.28
15.97
27.29
-92.76
16.61
16.74
21.51
15.00
25.22
-95.53
18.59
27.99
18.74
36.65
-107.76
R ) CH3
9b
9b′
9bTS
10b
10bTS
11b + N2
0.00
10.80
27.42
20.51
36.37
0.00
11.55
22.59
20.94
28.28
0.00
0.00
11.08
22.70
15.96
27.93
0.00
10.96
21.00
17.25
26.60
0.00f
9.91f
20.32f
17.18f
24.76f
-94.72f
-105.73
-87.43
-92.61
-92.33
Homolytic Fragmentationg
12 + NO
13 + NO
55.7
47.8
45.8
35.5
38.5
36.6
38.7
33.3
40.0
33.7
Oxetene Ring Opening
15
15TS
16
33.50
68.64
0.00
27.31
55.61
0.00
28.88
55.77
0.00
26.71
60.47
0.00
28.60
58.85
0.00
29.46
58.39
0.00
a Optimized at the RHF/6-31G* level. b Calculated using the 6-31G* basis set and the frozen core approximation. c Zero-point vibrational energy
correction from MP2/6-31G* frequencies and scaled by 0.9646. d 9a-rot-ts, barrier height between 9a and 9a′. e 9a-rot-ts-out, barrier height between
9a(out) and 9a(out)′. f Zero-point vibrational energy correction from RHF/6-31G* frequencies and scaled by 0.9135. g Energies relative to 9a.
Table 5. CHELPG Charges for the Decomposition of Nitrosimines
into the nature of the reaction. Nitrosimines 9a and 9b (eq 5)
9a and 9b Using RHF/6-31G* Wave Functions at the MP2/6-31G*
Optimized Geometries
were chosen as simplified models of the experimental system.
These retain the essential electronic features, namely the
9a
0.095 -0.060
0.847 0.652
9a′
9a-rot-ts 9aTS
10a
10aTS
11a
nitrosimine functionality and the thiazole ring, without the
computational complexity of the fused benzene ring and
additional nitrogen substituents. The experimental activation
barrier for the deazetization of 3-methyl-2-nitrosiminobenzothia-
zoline (2a) is 25.3 kcal/mol in cyclohexane and 22.5 in
methanol. Since there is only a small solvent effect on the rate
of the decomposition, the calculated gas-phase activation barriers
for the reactions of 9 might be expected to be comparable to
the experimental values, particularly in cyclohexane. The steric
and electronic effects of an alkyl substituent on the ring nitrogen
(N3) were probed by the methyl derivative 9b. The loss of NO
and the ring opening of the resulting radical were modeled by
12 and 13, respectively.
S1
C2
0.101
0.732
0.045 -0.088 -0.061 -0.057
0.953
1.029
0.907
0.590
0.036
N3 -0.554 -0.136 -0.492 -0.624 -0.703 -0.521
C4 -0.033 -0.315
0.005 -0.035
0.017 -0.067 -0.102
C5 -0.398 -0.133 -0.298 -0.339 -0.321 -0.290 -0.250
N6 -0.633 -0.543 -0.681 -0.649 -0.466 -0.328
N7
0.158
0.135
0.166
0.202
0.143
0.145
O8 -0.357 -0.391 -0.391 -0.516 -0.457 -0.610 -0.564
9b
9b′
9bTS
10b
10bTS
11b
S1
0.053
0.732
0.018
0.608
0.044 -0.049 -0.027 -0.064
C2
0.760
0.813
0.726
0.476
0.041
N3 -0.200 -0.064 -0.228 -0.299 -0.184
C4 -0.119 -0.156 -0.068 -0.031 -0.092 -0.226
C5 -0.338 -0.266 -0.343 -0.364 -0.336 -0.243
N6 -0.509 -0.562 -0.553 -0.441 -0.302
N7
O8 -0.328 -0.338 -0.479 -0.401 -0.544 -0.532
C9 0.040 -0.011 0.081 0.026 -0.004 -0.089
0.086
0.108
0.165
0.130
0.125
calculated at the MP4(FC,SDQ)/6-31G* level with the
MP2(FC)/6-31G* optimized geometries. The zero-point vibra-
tional energy (ZPE) corrections were obtained by scaling the
RHF/6-31G* ZPE by 0.9135, or the MP2/6-31G*(FC) ZPE by
0.9646 as recommended by Pople et al.19 Unless otherwise
indicated, all energies discussed in the text are MP4(SDQ,FC)/
Computational Methods
The geometries and energies of the ground-state conforma-
tions as well as transition states along the reaction pathway were
calculated with ab initio molecular orbital theory using Gaussian
94.18 Geometries were optimized first at the RHF/6-31G* level
and then reoptimized at the MP2(FC)/6-31G* level. Frequency
calculations verified the identity of each stationary point as a
minimum or transition state; these were done at both levels for
the hydrogen substituent and at the RHF/6-31G* level for the
methyl substituent. Single-point energies of each structure were
(18) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.;
Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.; Keith, T.; Petersson, G.
A.; Montgomery, J. A.; Raghavachari, K.; Al-Laham, M. A.; Zakrzewski,
V. G.; Ortiz, J. V.; Foresman, J. B.; Peng, C. Y.; Ayala, P. Y.; Chen, W.;
Wong, M. W.; Andres, J. L.; Replogle, E. S.; Gomperts, R.; Martin, R. L.;
Fox, D. J.; Binkley, J. S.; Defrees, D. J.; Baker, J.; Stewart, J. P.; Head-
Gordon, M.; Gonzalez, C.; Pople, J. A. Gaussian 94, revision B.3; Gaussian,
Inc.: Pittsburgh, PA, 1995.
(19) Pople, J. A.; Scott, A. P.; Wong, M. W.; Radom, L. Isr. J. Chem.
1993, 33, 345-350.

Thermal Decomposition of Heterocyclic Nitrosimines
J. Am. Chem. Soc., Vol. 123, No. 31, 2001 7483
Figure 2. MP2/6-31G* optimized geometries of conformations of 9a, rotational transition states 9a-rot-ts and 9a-rot-ts-out, transition states
9aTS and 10aTS, and structures 10a and 11a. Distances are in Å, angles in degrees. The atom numbering and shadings are as shown for 9a.
6-31G*//MP2(FC)/6-31G* level with the ZPE correction.
Absolute energies are reported in the Supporting Information,
and relative energies are reported in Table 4. Geometries
optimized at the MP2(FC)/6-31G* level are shown in Figures
2, 6, and S9. CHELPG charges are given in Table 5.20 Atom
numbering follows 9a in Figure 2 or 15 in Figure 5.
calculated to be relatively close in energy. The anti conforma-
tions (9a′ and 9a′(out)) are consistently higher in energy, which
is also consistent with steric crowding in these conformations.
The four possible conformations of 9a were calculated as
shown in Figure 2. The C2dN6-N7dO8 moiety is, in most
cases, planar and thus can be either s-cis or s-trans.21 The NO
can be either syn or anti to the sulfur; the anti conformation (to
the sulfur) is designated 9′, while s-trans is designated as out.
The geometries of the conformations of 9a were calculated and
are shown in Figure 2. It is clear from inspection of 9a′ that
there is some steric crowding and that there is not room to
accommodate replacement of the N3H with an N3-CH3 group.
The 9b′ conformation is distorted from planarity because of
steric crowding between the nitrosimine and the N-methyl
groups (Figure S9). The bond angles of the in conformations
are more open than those in the out conformations, again
reflecting crowding. The out conformations show slightly more
pronounced alternation in the C2dN6, N6-N7, and N7dO8 bond
lengths; there is no obvious reason for this subtle trend. Although
there is some variation, 9a is calculated to be the most stable
conformation at most levels of theory, while 9a(out) is
The homolytic bond strength of the N-NO bond was
calculated to be 40.0 kcal/mol, (Table 4) on the basis of the
formation of NO and 12 from 9a (eq 5). For comparison, the
experimental bond strengths for N-nitrosoureas are in the range
of 23.1-33.1 kcal/mol.2b Although accurate calculation of
radicals is difficult,22 these results suggest that homolysis is not
a facile process. The ring-opened radical 13 is calculated to be
6.3 kcal/mol more stable than 12, consistent with the observation
of ring-opened products in the photochemistry of 2 (eq 3).12
We located transition states for the isomerization of 9a to
9a′ and of 9a(out) to 9a(out)′. These both correspond to
rotations around the C2dN6 bond. The barriers for these
rotations are comparable, and both are fairly low, 16.61 and
16.69 kcal/mol, respectively. The transition states are very close
to symmetrical; the N3-C2dN6-N7 and S1-C2dN6-N7 dihe-
dral angles in 9a-rot-ts are 89.0° and -93.8°, respectively, and
90.0° and -96.0° in 9a-rot-ts-out, respectively. Rotation does
(20) Breneman, C. M.; Wiberg, K. B. J. Comput. Chem. 1990, 11, 361-
373.
(21) Nitrosamines also are planar, with hindered rotation.6c In contrast,
nitrosiminoformaldehyde (H2CdN-NdO) is nonplanar with a C-N-N-O
dihedral angle of 71.5° at the MP2/6-31G* level.
(22) Hrovat, D., A.; Morokuma, K.; Borden, T., Weston. J. Am. Chem.
Soc. 1994, 116, 1072-1076.

7484 J. Am. Chem. Soc., Vol. 123, No. 31, 2001
Bartsch et al.
Figure 3. Selected basis orbitals of 9a-rot-ts.
not dramatically change the C2dN6 bond length, although it is
shortened from 1.321 or 1.328 Å in 9a(out) or 9a′(out) to 1.307
Å in 9a-rot-ts-out. Similarly, the bond length of 1.335 Å in
9a-rot-ts compares with bond lengths of 1.334 and 1.340 Å in
9a and 9a′, respectively. This reflects, at least in part, a change
toward sp hybridization; as the bond rotates, the C2dN6-N7
angles open from 116.3° and 107.0° in 9a and 9a(out),
respectively, to 116.7° and 118.5° in 9a-rot-ts and 9a-rot-ts-
out, respectively. Clearly the π-bond is not completely broken
at the transition state. This is not surprising; partial π-bonding
is possible between the N6-lone pair and the C2-p orbital as
shown in Figure 3. There is also a relatively small charge
redistribution at the transition states; only -0.07 esu are
transferred from the N6-N7dO8 moiety on going from 9a to
9a-rot-ts (Table 5). There does not appear to be substantial
bonding between the C2 and O8 in 9a-rot-ts; the C2-O8 bond
distance is 2.512 Å, and the N6-N7dO8 bond angle of 115.8°
is actually wider than in 9a-rot-ts-out, where it is 113.8°.
The spirobicyclic intermediate 10a was located. There is
nothing particularly notable about the geometry of this structure;
the bond lengths and angles are as would be expected for this
strained intermediate. It is less polar than 9a; N6-N7-O8 moiety
bears a -0.780 esu charge in 10a as compared to 0.832 esu in
9a (Table 5). This decrease in polarity is consistent with the
observed solvent effects for 2a, for which deazetization is faster
in less polar solvents (Table 2).
Figure 4. Potential energy surface for the deazetization of 9a.
Geometries are optimized at the MP2/6-31G* level. Relative energies
are at the MP4(SDQ, FC)/6-31G* + ZPE level. The in-plane reaction
coordinate corresponds to in-plane motions leading to cyclization and
deazetization, while the out-of-plane coordinate corresponds to rotation
around the C2-N6 bond. See text for a discussion of the relationship
between 9a-rot-ts and 9aTS.
existence of two transition states does not require that the actual
reaction trajectory of a given molecule pass through 9a-rot-ts,
but it will pass close to 9aTS.
A transition structure (9aTS) for the electrocyclization of 9a
to generate 10a (eq 5) was also located. The geometry of 9aTS
is remarkably similar to the rotational transition state 9a-rot-ts
described above. The C2-N6 bond remains close to the plane
of the ring. As would be expected, the C2-N6 and N7-O8 bonds
lengthen (by 0.106 and 0.052 Å, respectively), while the N6-
N7 distance shortens by 0.74 Å as compared to 9a-rot-ts. The
major difference between the two structures is the C2-O2
distance, which is reduced from 2.512 Å in 9a-rot-ts to 1.988
Å in 9aTS. One might question whether two distinct transition
states with such similar geometries can exist for such different
reactions, rotation and cyclization. The frequency calculations
confirm that both are true transition states. The potential energy
surface is represented in Figure 4. The transition structure 9aTS
connects two stationary points, 10a and 9a-rot-ts; it just happens
that the latter of these stationary points is itself a transition state
in an orthogonal direction, namely rotation about the C2dN6
bond.23 This unusual potential energy surface has an analogy
in the potential energy surface of cyclooctatetraene (14), for
which the delocalized D8h structure (14TS′) is a transition state
connecting the localized D4h structures (14TS), which are
themselves transition states for the ring-inversion process (eq
6).24 The symmetry constraints on this system in particular and
of degenerate reactions in general have recently been discussed
by Schaad and Hu.25 Transition structure 9aTS, although not
degenerate, is consistent with their analysis. We note that the
The overall barrier to formation of 10a from 9a is 21.5 kcal/
mol. However, most of this barrier arises from the barrier to
rotation from 9a to 9a-rot-ts (16.6 kcal/mol). The barrier for
the ring closure from 9a-rot-ts to 10a is 4.9 kcal/mol and for
the ring-opening from 10a to 9a is 6.5 kcal/mol. When viewed
from this perspective, the calculated barriers for this electrocyclic
reaction are remarkably low! This is because the transition state
is pseudopericyclic.26 The C2-O8 bond distance changes, but
the C2dN6-N7dO8 system remains essentially planar and there
are no changes in orbital overlap in the N6-N7dO8 π-system.
Without a loop of interacting orbitals, electron-electron repul-
sion can be minimized. This is indeed the case, since O8 and
C2 are attracted to each other with calculated charges of -0.516
and 0.953 esu, respectively, in 9aTS.
The ring-opening reaction of oxetene (15, eq 4) might be
viewed as an analogous model system for the ring-opening of
10a via 9aTS. Although the reaction of oxetene to give acrolein
(16) has previously been calculated at the CASSCF/6-31G*//
RHF/6-31G* level11 and the geometry of acrolein has been
reported at the MP2/6-31G* level,27 the geometries of 15 and
15TS were reoptimized at the MP2/6-31G* level for the
comparison with the reaction of 10a to 9a and are shown in
Figure 5. The activation barrier calculated here for the ring-
(23) Rotation at the geometry of 9aTS raises the energy of the system
because it breaks the partially formed C2-O8 bond.
(24) Wenthold, P. G.; Hrovat, D. A.; Borden, W. T.; Lineberger, W. C.
Science 1996, 272, 1456-1459.
(25) Schaad, L. J.; Hu, J. J. Am. Chem. Soc. 1998, 120, 1571-1580.
(26) (a) Birney, D. M.; Wagenseller, P. E. J. Am. Chem. Soc. 1994, 116,
6262-6270. (b) Birney, D. M.; Xu, X.; Ham, S. Angew. Chem., Int. Ed.
1999, 38, 189-193.

Thermal Decomposition of Heterocyclic Nitrosimines
J. Am. Chem. Soc., Vol. 123, No. 31, 2001 7485
to be 10.2 kcal/mol. The Hammond postulate31 would predict
that this should have a low barrier because it is very exothermic.
Indeed, the exothermicity of the retro [2 + 2] reaction to form
11a is calculated to be 95.5 kcal/mol, and the exoergonicity is
calculated to be 103.0 kcal/mol (unscaled ∆G at the MP2/6-
31G* level). This reflects the relief of ring strain in the transition
state, as well as the greater strengths of the CdO π-bond and
of the second NdN π-bond. There is also a significant, favorable
entropic contribution to the exoergonicity of the reaction simply
due to the formation of two products. It is noteworthy that the
barrier of 10.2 kcal/mol for the formation of 11a and N2 from
10a is higher than the 6.5 kcal/mol for the ring-opening reaction
of 10a to give 9a via 9aTS. Indeed, if 9aTS is considered to
be the transition state connecting 10a to 9a-rot-ts, then this is
indeed endothermic by 1.6 kcal/mol, yet has a lower barrier.
This is not surprising, because the [2 + 2]-cycloreversion is
orbital-symmetry-forbidden, while the electrocyclization is
allowed and is pseudopericyclic.29a Thus, the overall gas-phase
barrier for this deazetization reaction (9a to 10aTS) is calculated
to be 25.2 kcal/mol. This compares very favorably with the
experimental barrier of 25.3 kcal/mol for the deazetization of
3-methyl-2-nitrosiminobenzothiazoline (2a) in cyclohexane.
Furthermore, 10aTS is calculated to be less polar than 9a; the
N6-N7-O8 moiety in 10aTS is calculated to have a -0.793
esu charge as compared to -0.832 esu in 9a (from Table 5).
This is consistent with the observed solvent effects for the
deazetization of 2a (Table 2).
The overall entropy of the deazetization reaction (9a to
10aTS) was calculated to be + 0.9 eu This ignores statistical
contributions from the multiple conformations of 9, as well as
the existence of enantiomeric transition states 10aTS. This small,
positive entropy of activation is consistent with the experimental
value of +1.3 eu measured for 2a in cyclohexane.
Similar results were obtained with methyl substitution on the
ring nitrogen, 9b, 9b′, 9bTS, 10b, 10bTS, and 11b. These
structures are shown in Figure S9 in the Supporting Information.
With the exception of 9b′, there are not significant differences
in geometries between the hydrogen and methyl series. As
discussed above, steric hindrance in 9b′ between the N6-N7d
O8 and the methyl group force N6-N7dO8 to twist out of plane
(the N3-C2dN6-N7 dihedral angle is 48.3 °). The methyl group
does not make a significant difference to the calculated barrier
of 24.8 kcal/mol for 10bTS as compared to 25.2 kcal/mol for
10aTS. This is consistent with the experimental results which
do not show significant changes in rate as the ring nitrogen
(N3) substituent is varied.
Figure 5. MP2/6-31G* optimized geometries for 15, 15TS, and 16.
See Figure 2 for key; atom numbering is as shown for 15.
opening reaction of oxetene via 15TS is 28.9 kcal/mol (MP4-
(SDQ)/6-31G*//MP2/6-31G* + ZPE). This compares more
favorably with the experimental activation enthalpy of 24.1 (
1.5 kcal/mol,28 than does the CASSCF/6-31G*//RHF/6-31G*
barrier of 32.2 kcal/mol. The barrier calculated here (15TS) is
22.4 kcal/mol higher than for the comparable reaction of 10a.
Significantly, the exothermicity for the ring-opening of oxetene
is calculated to be 29.5 kcal/mol, whereas for the reaction of
10a to 9a-rot-ts it is 1.7 kcal/mol. The remarkably lower barrier
for the ring-opening of 10a, especially in light of the modest
calculated exothermicity, is consistent with the low barriers
found for other pseudopericyclic reactions when the geometry
and electrostatic interactions are favorable.29 Comparing the
geometries of both transition structures optimized at the MP2/
6-31G* level, 9aTS is essentially planar with a C2N6N7O8
dihedral angle of 2.3°. The corresponding dihedral angle in the
transition state of oxetene ring-opening (15TS) is 14.4°. This
reflects the need for rotation of the CH2 moiety to accommodate
developing π-overlap in the transition state, as is typical in
pericyclic reactions.10,30 In principle, oxetene ring-opening could
proceed via an analogous two-stage process, with a planar,
pseudopericyclic transition state connecting to a transition state
for rotation around the C4-C3 bond. The difference between
the two systems is that the barriers for bond breaking in 15 and
for rotation around the C4dC3 bond in acrolein (16) are much
higher than for 10a and 9a; thus, there is an energetic benefit
to twisting in the ring-opening of oxetene, which allows for
partial p-overlap in the transition state 15TS. From this
perspective then, the ring closure of 9a to 10a may be considered
to have decoupled the two fundamental processes, rotation and
bond formation, which are required for the overall pericyclic
process.
Conclusions
Ab initio calculations suggest that the deazetization of
nitrosiminothiazolines can be understood as a three-stage
reaction. Beginning from one of the inward conformations, for
example, 9a or 9a′, the first stage involves rotation of N6-
N7dO8 out of the plane to a rotational transition state, 9a-rot-
ts. This transition state positions these atoms so that they can
cyclize in the second stage, without passing through a separate
intermediate, via a low-barrier, pseudopericyclic transition state
(9aTS) to the spirobicyclic intermediate 10a. This two-stage
process is qualitatively different from the single-stage process
for the pericyclic ring-opening of oxetene (15). However, the
pericyclic process can be understood as a compromise between
the two fundamental steps, rotation and pseudopericyclic ring
closure. Next, in the rate-determining step, this undergoes an
orbital-symmetry-forbidden, but extremely exothermic, deazeti-
Finally, transition structure 10aTS for the [2 + 2]-cyclo-
reversion of 10a to form 11a and N2 was located (eq 5). The
activation barrier for this elementary reaction step is calculated
(27) Wiberg, K. B.; Rosenberg, R. E.; Rablen, P. R. J. Am. Chem. Soc.
1991, 113, 2890-2898.
(28) Martino, P. C.; Shevlin, P. B. J. Am. Chem. Soc. 1980, 102, 5429-
5430.
(29) (a) Birney, D. M.; Ham, S.; Unruh, G. R. J. Am. Chem. Soc. 1997,
119, 4509-4517. (b) Birney, D. M. J. Am. Chem. Soc. 2000, 122, 10917-
10925.
(30) Spellmeyer, D. C.; Houk, K. N. J. Am. Chem. Soc. 1988, 110, 3412-
3416.
(31) Hammond, G. S. J. J. Am. Chem. Soc. 1955, 77, 334-338.

7486 J. Am. Chem. Soc., Vol. 123, No. 31, 2001
Bartsch et al.
kinetic data and the absorption maxima for various nitrosimines in
methanol and for 3-methyl-2-nitrosiminobenzothiazoline (2a) in various
solvents are given in Tables S1, S2, and S3 in the Supporting
Information. Linear regression least-squares analysis of ln absorption
versus time gave the first-order rate constants. The standard deviations
of the slopes gave the estimated errors in the first-order rate constants.
zation via 10aTS to form the products 11a and N2. The overall
calculated gas-phase barrier of 25.2 kcal/mol compares favorably
with the experimental barrier of 25.3 kcal/mol for the deazeti-
zation of 3-methyl-2-nitrosiminobenzothiazoline (2a) in cyclo-
hexane. This proposed mechanism is also consistent with the
observed modest solvent and substitutent effects on the rate.
Homolytic fragmentation of 9a to 12 and NO is calculated to
be 40.0 kcal/mol endothermic; the larger barrier explains why
homolysis is not generally observed under thermal conditions.
Calculation of Activation Parameters. Since first-order rate
constants for the thermal decomposition of 2a were determined at
several different reaction temperatures in methanol and cyclohexane,
activation parameters could be evaluated. Linear regression least-squares
analysis of ln k versus 1/T gave the Arrhenius activation energy. The
error in the activation energy was taken to be the standard deviation of
the slope.
Experimental Procedures
The substituted iminobenzothiazolines 6a (mp 124-125 °C), 6f (mp
94-94.5 °C), and 3-methyl-2-iminobenzoselenazoline (mp 101-105
°C) were obtained from the laboratories of Professor S. Hu¨nig of the
Product Study of the Thermal Decomposition of 3-Methyl-2-
nitrosiminobenzothiazoline (2a) in Methanol. By using the procedure
outlined above for the kinetic studies, solutions of 3-methyl-2-
nitrosiminobenzothiazoline (2a) in methanol were prepared. The
resulting solutions were sealed in ampules and placed in a 62.3°
constant-temperature bath for 58 h. This period corresponded to 5.1
half-lives of the kinetic reaction.
1
University of Wu¨rtsburg. H NMR spectra were recorded at 60 MHz
with a Varian A-60 spectrometer, in CDCl3 as solvent. UV spectra
were recorded using a Cary 15 UV/vis spectrometer.
Preparation of 3-Methyl-2-nitrosiminobenzothiazoline (2a). Into
a pear-shaped flask, fitted with a pressure-equilibrating addition funnel
and magnetic stirrer, were placed 1.00 g (6.1 mmol) of 3-methyl-2-
iminobenzothiazoline (6a, obtained from the laboratories of S. Hu¨nig,
mp 124-125 °C) and 7.4 mL of glacial acetic acid. The resulting
solution was cooled with an ice bath. A sodium nitrite solution, 3.0
mL (8.7 mmol of sodium nitrite), was added dropwise. The color of
the solution changed from colorless to yellow-gold to orange, and an
orange precipitate of the nitrosimine 2a formed. After filtering the
solution, the nitrosimine was washed several times with water and dried
in vacuo over silica gel. The chemically pure nitrosimine (0.98 g,
75.6%) mp 152-153 °C (reported 152 °C).9a IR 1540, 1390; 1H NMR
3.90 (s, 3H), 7.27-7.88 (m, 4H).
The ultraviolet spectra of the solutions were examined for absorption
at λmax ) 288 nm which is due to 3-methyl-2-benzothiazolinone. By
comparison of the absorbance with that of known concentrations of
3-methyl-benzothiazolinone (4a) in methanol (Figure S1 and Table S4,
Supporting Information) at λmax ) 288 nm, the yield of 3-methyl-2-
benzothiazolinone (4a) was calculated (Table S5, Supporting Informa-
tion). Thus, a 101.8 ( 0.5% yield of 3-methyl-2-benzothiazolinone was
obtained. The slightly higher than 100% yield of 3-methyl-2-benzothia-
zolinone (4a) is probably due to errors introduced by the low absorbance
of the product when compared to that of the reactant.
Preparation of nitrosimines 2b-f and 7 and of intermediates 5 and
6 are described in the Supporting Information.
Acknowledgment. D.M.B. thanks the Robert A. Welch
Foundation for support of this research and Paul Wenthold for
helpful discussions.
Kinetic Studies of the Thermal Decomposition of 3- and 3,6-
Substituted 2-Nitrosiminobenzothiazolines. A solution of the desired
heterocyclic nitrosimine-solvent combination was prepared in a
volumetric flask by dissolving the nitrosimine in the requisite solvent.
Dissolution in cyclohexane required stirring with a magnetic stirring
bar for several hours at room temperature. During this period, the
volumetric flasks were wrapped in aluminum foil.
Portions of the solutions were sealed in nitrogen-flushed glass
ampules, and the ampules were wrapped in aluminum foil. The ampules
were immersed in a constant-temperature bath. After the desired reaction
time, ampules were removed and cooled in an ice-water bath to quench
the reaction.
Supporting Information Available: Experimental details
for the preparation and characterization of nitrosimines 2b-f
and 7 and of intermediates 5 and 6, kinetic data and molar
absorptivities and Beer’s law, log absorbance vs time and
Arrhenius plots, MP2/6-31G* optimized Cartesian coordinates,
vibrational frequencies, and a table of absolute energies for all
calculated stationary points are provided (PDF). This material
is available free of charge via the Internet at http://pubs.acs.org.
The ultraviolet spectrum of the solution was then examined for
absorption of the nitrosimine in the region of 345 nm. Summaries of
JA010659K