4,5-Dihydro-1,2,3-oxadiazole: A Very Elusive Key Intermediate in Various Important Chemical Transformations

Article abstract of DOI:10.1002/chem.201502326

4,5-Dihydro-1,2,3-oxadiazoles are postulated to be key intermediates in the industrial synthesis of ketones from alkenes, in the alkylation of DNA in vivo, and in the decomposition of N-nitrosoureas; they are also a subject of great interest for theoretical chemists. In the presented report, the formation of 4,5-dihydro-1,2,3-oxadiazole and the subsequent decay into secondary products have been studied by NMR monitoring analysis. The elusive properties evading characterization have now been confirmed by ¹H, ¹³C, and ¹⁵N NMR spectroscopy, and relevant 2D experiments at very low temperatures. Our experiments with suitably substituted N-nitrosoureas using thallium(I) alkoxides as bases under apolar conditions answer important questions on the existence and the secondary products of 4,5-dihydro-1,2,3-oxadiazole. Short-lived! 4,5-Dihydro-1,2,3-oxadiazole is generated by treatment of N-(2-chloroethyl)-N-nitrosourea with thallium alkoxides at low temperatures, but it is not formed by nucleophilic dealkylation of the corresponding N-methyloxadiazolinium salt. Even at -90 °C, 4,5-dihydro-1,2,3-oxadiazole cannot be detected directly. However, it was identified by analysis of its decay products, acetaldehyde, ethylene oxide, and diazomethane (see scheme).

Full text of DOI:10.1002/chem.201502326

DOI: 10.1002/chem.201502326
Communication
&
Reactive Intermediates
4
,5-Dihydro-1,2,3-oxadiazole: A Very Elusive Key Intermediate in
Various Important Chemical Transformations
[a]
[a]
[b]
[b]
[c]
Klaus Banert,* Neeraj Singh, Benjamin Fiedler, Joachim Friedrich,* Marcus Korb, and
Heinrich Lang
[c]
Dedicated to Prof. Harald Günther on the occasion of his 80th birthday
Chem. Eur. J. 2015, 21, 15092 – 15099
15092
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communication
utilized to synthesize cyclopentanone and ketones with
[
2c]
Abstract: 4,5-Dihydro-1,2,3-oxadiazoles are postulated to
be key intermediates in the industrial synthesis of ketones
from alkenes, in the alkylation of DNA in vivo, and in the
decomposition of N-nitrosoureas; they are also a subject
of great interest for theoretical chemists. In the presented
report, the formation of 4,5-dihydro-1,2,3-oxadiazole and
the subsequent decay into secondary products have been
studied by NMR monitoring analysis. The elusive proper-
ties evading characterization have now been confirmed
twelve-membered rings (30000 tons per year) by means of
1
,3-dipolar cycloaddition of N O to the alkene. Recently, Panov
2
[
2a,b]
et al. described
an alternative liquid-phase oxidation of cy-
clohexene and cyclopentene by nitrous oxide with nearly
00% selectivity. In these processes, 4,5-dihydro-1,2,3-oxadi-
1
azoles have been hypothesized to be intermediates that imme-
diately, under the reaction conditions, extrude dinitrogen to
give ketones in high yields. The use of N O in organic chemis-
2
try has been rather unexplored because of the relative chemi-
cal inertness and the high-energy input required. This low re-
activity can be overcome by utilizing reaction partners with
1
13
15
by H, C, and N NMR spectroscopy, and relevant 2D ex-
periments at very low temperatures. Our experiments
with suitably substituted N-nitrosoureas using thallium(I)
alkoxides as bases under apolar conditions answer impor-
tant questions on the existence and the secondary prod-
ucts of 4,5-dihydro-1,2,3-oxadiazole.
[
6]
[7]
ring strain, N-heterocyclic carbenes, or metalorganic re-
[
8]
[9,10]
agents. Quantum chemical calculations performed on 1
[
3,11]
and similar heterocycles
predict various secondary products
by different modes (Scheme 2), including the formation of di-
azomethane and formaldehyde as one of the most favorable
routes on the basis of inherent weakness of the NÀO bond.
On the other hand, the decomposition of suitably substitut-
ed N-nitrosoureas at physiological pH has also been postulat-
[1]
4
,5-Dihydro-1,2,3-oxadiazoles of type 1 are postulated to be
intermediates in the synthesis of ketones from alkenes under
[
2,3]
[4]
drastic conditions,
the decomposition of N-nitrosoureas at
ed to proceed via 1. In these cases, quantum chemical calcu-
lations likewise predict the intermediacy and the secondary
[
4]
physiological pH, and the alkylation of DNA and other rele-
[5]
[9]
vant molecules in vivo (Schemes 1 and 2). We became inter-
products of 1 (Scheme 2). The role of 3-methyl-4,5-dihydro-
ested in 1 from a process used by BASF Ludwigshafen wherein
1,2,3-oxadiazolium salts of type 2 in DNA alkylation and alkyl-
ation in general also suggests the transient formation of 1.
[
5]
N O (a waste product and highly potent greenhouse gas) is
2
There are reports on the methylation of nucleophilic com-
[
5a]
pounds by 2 in very good to moderate yields. In spite of
many discussions involving 2 as the methylating agent, no at-
tempts have been made to characterize 1 or its secondary
[
5]
products directly. However, contrary results with main or ex-
[2c]
Scheme 1. Industrial synthesis of cyclic ketones employed by BASF.
clusive nucleophilic attack at C-5 of 2 have also been publish-
[
5a,c,e,f,i,12]
ed.
Armed with such a great deal of information about the im-
portant role played by 1 in various branches of chemistry, we
present our results regarding to the existence and the forma-
tion of secondary products of 1.
Previous studies have been done on the decomposition of
[
4]
N-nitrosoureas in aqueous buffered solutions, but these con-
ditions are not conclusive to comment on the stability of
1
and the formation of its primary decay products. We chose
[
13]
thallium(I) alkoxides as bases, because of their high solubility
in various apolar solvents even at very low temperatures,
which is quintessential for monitoring the reaction by NMR
spectroscopy under such conditions. First, we utilized 2-chloro-
Scheme 2. Possible modes of formation and subsequent products of 1.
[
4d]
ethyl-N-nitrosourea (3) as a substrate to investigate the for-
mation of 1 in non-aqueous solutions (Scheme 3). Tosylate 4
was also synthesized for NMR studies, but it was not suitable
owing to its limited solubility in dichloromethane at low tem-
peratures. The generation of 1 was initially attempted by treat-
ing 3 with TlOEt (1.0 equiv) in CD Cl within the range À40 to
[
a] Prof. Dr. K. Banert, N. Singh
Organic Chemistry, Technische Universität Chemnitz
Strasse der Nationen 62, 09111 Chemnitz (Germany)
E-mail: klaus.banert@chemie.tu-chemnitz.de
Homepage: https://www.tu-chemnitz.de/chemie/org/index.html.en
[
b] B. Fiedler, Prof. Dr. J. Friedrich
2
2
[
14]
Theoretical Chemistry, Technische Universität Chemnitz
Strasse der Nationen 62, 09111 Chemnitz (Germany)
E-mail: joachim.friedrich@chemie.tu-chemnitz.de
À208C under constant monitoring by NMR spectroscopy. Im-
1
mediately, we could observe the H NMR signal of ethylene
oxide (15% yield) and traces of acetaldehyde along with
strong signals of 1-ethoxyethyl carbamate (5) (67% yield) and
a trace signal of diazomethane (Table 1). The identity of ethyl-
ene oxide and 5 was further confirmed by their chemical shift
[
c] M. Korb, Prof. Dr. H. Lang
Inorganic Chemistry, Technische Universität Chemnitz
0
9107 Chemnitz (Germany)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201502326.
13
values in the C NMR spectrum. The observation of ethylene
Chem. Eur. J. 2015, 21, 15092 – 15099
www.chemeurj.org
15093 ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communication
To ensure that acetaldehyde was not directly resulting from
TlOEt by transfer and loss of hydride (as in the Meerwein–
Ponndorf–Verley reduction), we used thallium propoxide,
[
13a]
TlOPr
(1.1 equiv), and the reaction with 3 was monitored by
NMR spectroscopy from À608C to room temperature. Here,
we could clearly observe the signals of ethylene oxide, acet-
aldehyde, and diazomethane, whereas those of the carbamic
acid derivative 6 were very weak. The higher yield of acetalde-
hyde and diazomethane in this reaction can be attributed to
[
13a]
the different association nature
of various thallium(I) alkox-
ides. We also utilized conventional bases such as potassium hy-
droxide, diisopropylamine, and 2,2,6,6-tetramethylpiperidine to
[
14]
study the formation and decomposition of 1 (Table 1).
To confirm our results and especially the detection of diazo-
15
methane, we synthesized the N-labeled substrate 3a and
monitored the reaction with TlOEt (1.1 equiv) in CD Cl from
2
2
15
À80 to À508C (Scheme 4) by N NMR spectroscopy (Figure 1).
15
Scheme 3. Mechanisms to explain the formation of 1 and the experimentally
observed secondary products.
Even at À808C, we could easily detect a signal of NꢀN at d=
[15] 15
À69.7 ppm in the N NMR spectrum. From À80 to À508C,
we observed an increase in the
15
intensity of NꢀN and a conse-
[a]
15
Table 1. Yield of the products formed from 3 or 14 and suitable bases in CD
2
Cl .
2
quent decrease in the
NO
signal of 3a. Subsequently, we
studied the reaction of 3a with
TlOPr in the temperature range
Substrate
Base
T [8C]
CH
%]
3
CHO
Ethene
oxide [%]
CH
[%]
2
N
2
Carbamates
5 or 6 [%]
trans-2-
butenal [%]
[
3
3
3
3
3
3
TlOEt
TlOPr
KOH
Aq. KOH
DIPA
À20
À20
RT
trace
32
6
15
3
15
28
9
17
4
trace
67
trace
–
–
–
4
–
–
–
[
14]
of À80 to À108C. In this case,
we observed a clearly visible N
3
–
–
–
15
–
–
–
–
–
15
À20
NMR signal of CH N N at d=
2
[
c]
RT
1
4.2 ppm (Figure 1) and the cor-
[d]
TEMP
TlOPr
RT
À20
4
7
3
19
trace
4
[
b]
responding doublet with d=
1
4
3
15
1
3
.34 ppm and J( N, H)=1.1 Hz
1
[
a] Yields were measured by H NMR spectroscopy using an internal standard. [b] Incomplete reaction because
1
[16,17]
in the H NMR spectrum,
of the low solubility of 14; yields are based on consumed 14. [c] DIPA=diisopropylamine. [d] TEMP=2,2,6,6-
which indicated a yield of diazo-
methane of about 3%. The iden-
tetramethylpiperidine.
oxide and 5 can be understood by abstraction of an amide
proton by ethoxide and subsequent release of isocyanic acid
followed by an intramolecular S 2 reaction, which leads to
N
transient compound 1. This species immediately loses dinitro-
gen to give ethylene oxide and acetaldehyde. Generation of di-
azomethane is rationalized by 1,3-dipolar cycloreversion of 1.
The formation of 5 can be explained by a nucleophilic attack
of the thallium hemiacetal, resulting from the reaction of acet-
1
5
Scheme 4. Reaction of N-labeled 3a with TlOR.
aldehyde and TlOEt, on the urethane, which originates from
the reaction of isocyanic acid and TlOEt. Alternatively, 5 may
be directly produced from isocyanic acid and the thallium-
hemiacetal. Since most of the acetaldehyde is transformed into
an acetal-type secondary product, it is not surprising that the
even more electrophilic formaldehyde, which should be
formed in small amounts by the cycloreversion of 1, could not
be detected in any experiment.
1
5
1
tity of diazomethane was further confirmed by N, H long-
range correlation (gHMBCAD) 2D NMR spectroscopy and by
15
15
the synthesis of CH N N for comparison from NO-labeled N-
2
methyl-N-nitrosourea and KOH in H O/CDCl . The very different
2
3
proportions of diazomethane in the experiments with 3/3a
Doubling the amount of TlOEt did not make any appreciable
change. Encouraged by the clean reaction, we decided to
monitor the reaction from À908C to room temperature by
NMR spectroscopy. However, even at À908C, we could easily
detect the same secondary products in lesser yields along with
and TlOEt or TlOPr might be attributed to the various associa-
[
13a]
tion natures
of thallium(I) alkoxides, or can be explained by
the instability of diazomethane in the presence of TlOEt. In
1
a H NMR control experiment with diazomethane/CDCl pre-
3
pared from N-methyl-N-nitrosourea, addition of TlOEt led to
92% decay of the diazomethane after 15 min at À108C,
[
14]
unreacted 3.
Chem. Eur. J. 2015, 21, 15092 – 15099
www.chemeurj.org
15094
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communication
Figure 2. ORTEP representation (50% probability level) of the molecular
structure of 2 with atom numbering scheme; hydrogen atoms and further
+
À
[14,19]
3
co-crystallized species (H O , TsO ) are omitted for clarity.
[
18]
other oxadiazole compounds, for which XRD data were im-
perative for unambiguous structure elucidation. Therefore, to
be sure of the structure and chemistry of 2, we synthesized
[
5c,i]
2
(
and determined its structure by single-crystal XRD
[
14,19]
Figure 2).
When we repeated the treatment of 2 with 3,4-dichlorothio-
phenol, we did not detect even a trace of the methyl thioether,
and the reaction gave a mixture of many different products,
1
5
15
15
Figure 1. N NMR spectra with detection of NꢀN and CH
2
N N after treat-
ment of 3a with TlOEt or TlOPr at À808C. MeNO
2
was used as external refer-
but none of the secondary products of 1 were observed
ence.
[20]
(
Scheme 5). We then utilized QN3
(Q=n-hexadecyltributyl-
whereas an analogous sample without TlOEt showed negligi-
ble decomposition. In an additional control experiment, we
confirmed that the stability of diazomethane in the presence
of thallium propoxide is significantly greater than that with
thallium ethoxide.
In our experiments with 3 or 3a and TlOEt or TlOPr, we did
15
15
[15]
not detect a N NMR signal of N NO or any NMR signal of
ethene. Thus, the 1,3-dipolar cycloreversion of 1 to give these
products is excluded. Although the transformation of 3/3a in
the presence of thallium(I) alkoxides already started at À908C,
we were not able to detect any NMR signal of 1 in the range
of À90 to À808C or at higher temperatures. Instead, only sec-
ondary products, such as dinitrogen, diazomethane, ethylene
oxide, and acetaldehyde or its acetal-type succeeding prod-
ucts, were identified. This result seems to supply clear evi-
dence of the inherent short-lived properties of the heterocyclic
compound 1. However, we cannot exclude that the fragmenta-
tion of 1 is efficiently catalyzed by thallium cations. On the
other hand, monitoring the reaction of 3 with potassium hy-
droxide at À208C also did not lead to the observance of NMR
signals of elusive intermediate 1.
3
Scheme 5. Products of nucleophilic attack on 2, Q=n-C16H33Bu P.
phosphonium), which is an azide salt with unique properties
and especially useful in monitoring nucleophilic substitution
reactions. Substrate 2 was treated with QN3 (2.0 equiv) at
À458C, and the reaction was monitored by NMR spectrosco-
[
14]
py. Immediately, we could identify the open-chain N-nitros-
amine rotamers 7 along with CH N (17% yield). The formation
3
3
of 7 can be explained by S 2 reaction at C-5 of 2. The detec-
N
tion of methyl azide suggested that some of the needle-like
azide nucleophile had probably attacked the N-methyl group
of 2. However, as we did not observe any of the secondary
products of 1, the generation of methyl azide was probably oc-
curring through another mechanism. After complete consump-
tion of 2 and additional 24 h at À208C, we measured the
Although reports in medicinal chemistry on the possible role
[
5]
of salt 2 in the methylation of nucleophilic molecules suggest
the parallel formation of 1, no attempts have been made to
detect intermediate 1, and no sequential products of 1 were
identified (Scheme 2). For example, accounts of obtaining the
corresponding methyl thioether from 2 and 3,4-dichlorothio-
1
H NMR spectrum again and found that the yield of methyl
azide had increased to 23%. When the solvent and all volatile
products, such as methyl azide, were removed in vacuum after
total transformation of 2, another treatment of the residue of
7 with QN /CDCl at À308C slowly led to renewed formation
phenol in >90% yield were explained by S 2 attack of the
N
[5a]
sulfur nucleophile at the N-methyl group, which must lead
to the demethylated heterocycle 1. However, 1 or its decay
products were not even mentioned in this case. There are also
3
3
1
13
of methyl azide, which was identified by H and C NMR spec-
troscopy. Owing to these results, we assume that the open
chain rotamers 7 are not stable and slowly yield methyl azide
[
5c,d,12]
a few reports
doubting the role of 2 as a methylating
agent. The structure of 2 is an intriguing one and cannot be
in the presence of QN . Similar results were obtained with
3
fully assigned by NMR spectroscopy alone. This is similar to
TMGA (tetramethylguanidinium azide), which indicate that the
Chem. Eur. J. 2015, 21, 15092 – 15099
www.chemeurj.org
15095
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communication
size or the nature of the cation had little role to play. We also
carbon atoms and diazo compounds through cleavage of the
carbon–carbon bond is also known in the case of special het-
treated 2 with QI hoping that the polarizable nature and the
[
24]
large size of iodide would promote an S 2 reaction at the N-
erocycles of type 11. However, there are also differences in
the succeeding reactions of 1 and the known chemistry of 11,
that included 1,3-dipolar cycloreversion to produce azides RN3
N
methyl group of 2. But also in this case, we observed a very
clean reaction which gave only open chain rotamers 8 with no
[
21]
[25]
trace of CH3I.
We then synthesized compound 10 as a yellow oil by mesyl-
and alkenes. We were not able to find any evidence of the
analogous transformation of 1, which should lead to the gen-
[
22]
ation of known
2-methyl-1-[methyl(nitroso)amino]-2-propa-
eration of N O and ethene. Whereas more than one thousand
2
nol followed by thermal ring closure of mesylate
9
examples of stable 4,5-dihydro-1,2,3-triazoles are reported in
literature, and even heterocyclic compounds 11 without sub-
stituents at C-4 and C-5 have been isolated at room tempera-
[
14]
(
Scheme 6).
The idea was that the two geminal methyl
[
26]
ture, we have obtained evidence for the elusive properties
of 1 that prevent direct detection of 1 even at low tempera-
tures. We assume that the very weak NÀO bond is responsible
for the low stability of 1.
Since we favor heterolytic cleavage of the NÀO bond in
1
via zwitterionic intermediate 12, we also investigated the
Scheme 6. Synthesis of 10 (Ms=methanesulfonyl).
base-induced decay of substrate 14 (Scheme 7). This reaction
was previously studied in aqueous buffers at physiological pH
[
27]
groups at C-5 would discourage the S 2 reaction at that posi-
and 378C. Under these conditions, however, at least diazo-
N
tion and the direct displacement reaction at the N-methyl
group of 10 might be feasible. However, 10 did not give any
methane, if formed, would not persist. Owing to its limited sol-
ubility, 14 cannot be reacted with thallium propoxide in CD Cl2
2
reaction with QN , QI, or other nucleophiles. Thus, our results
at very low temperatures. Even at À208C, we had to use a sus-
pension of 14, and thus the consumption of this substrate was
slower. Nevertheless, we were able to detect the products
acetaldehyde, ethene oxide, and diazomethane (Table 1). We
assume that these compounds are formed via betain 12. This
species is also a plausible intermediate to explain all three
types of products after in situ generation of 1 from 3, al-
though the presence of diazomethane might be alternatively
rationalized by synchronous 1,3-dipolar cycloreversion of 1.
Homolytic cleavage of the NÀO bond in 1 to generate dirad-
ical 15 is another first step to initiate the decay of the five-
membered heterocycle. But the hydrogen shift in 15 seems to
be less likely as a possible route for the formation of acetalde-
hyde.
3
do not confirm S 2 attack at the methyl group of N-3, not only
N
in the case of substrate 10, but also for the reactions of salt 2.
Therefore, the methylation of DNA in vivo and methylation of
other molecules by treatment with 2 certainly does not involve
1
as an intermediate and occurs by another mechanism.
The decay of short-lived heterocycle 1 can be compared
with the well-known chemistry of 4,5-dihydro-1H-1,2,3-triazoles
[
23]
1
1 (Scheme 7). In the case of 1, the decomposition is domi-
We investigated the plausibility of a homolytic NÀO bond
cleavage by quantum chemical calculations. For an adequate
quantum chemical description of the open-shell singlet state
of 15, a multi-reference method such as CASPT2 is required. In
order to capture the dynamic correlation at a higher level of
theory, we used CCSD(T)(F12) to compute the energies of
1
and the triplet state of 15. CCSD(T) is adequate for these
Scheme 7. Cleavage reactions of the intermediates 12 and 13 generated
from the precursors 1, 11, and 14.
computations, since the CASSCF calculations indicate a single-
reference character.
The homolytic cleavage of the NÀO bond can be computed
using the CCSD(T)(F12) energies and the CASPT2 singlet-triplet
splitting of 15 [Eq. (1)].
nated by cleavage of the very weak NÀO bond followed by
loss of dinitrogen and ring closure to form a three-membered
CCSDðTÞðF12Þ
CCSDðTÞðF12Þ
CCSDðTÞðF12Þ
CASPT2
15Àsinglet
CASPT2
15Àtriplet
ring (S i). Alternatively, 1,2-hydride shift leads to a carbonyl
DE
¼ E15Àtriplet
À E1Àsinglet
þ E
À E
N
compound, or carbon–carbon bond cleavage generates anoth-
er carbonyl compound with a lower number of carbon atoms
and a diazo compound. All three types of reactions are also rel-
evant in the thermolysis of nitrogen heterocycles 11. After fis-
sion of the NÀN single bond, simple 4,5-dihydro-1,2,3-triazoles
mainly yield aziridines along with imines, which result from
ð1Þ
The results of the CASPT2(6,6) calculations and those follow-
ing Equation (1) are collected in Table 2 for the two different
conformers 15a and 15b (Figure 3). Since the dissociation en-
ergies of Equation (1) and CASPT2 agree very well, we con-
clude that the calculations provide the required accuracy to
[
23]
a 1,2-hydride shift. However, formation of imines with less
Chem. Eur. J. 2015, 21, 15092 – 15099
www.chemeurj.org
15096
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communication
Table 2. ZPVE- and COSMO-corrected reaction energies for the formation
of 15a and 15b from 1 in the singlet state via homolytic cleavage of the
N-O-bond, calculated with CASPT2(6,6) (def2-TZVPP basis set) and
CCSD(T)(F12) (cc-pVDZ-F12 basis set).
CASPT2
solv
solv
Conformer
DE
kJmol
+D
]
E.q. (1)+D
À1
À1
[
[kJmol
]
1
1
5a
5b
+145.2
+151.8
+155.7
+162.5
1
[35b,e]
[35d]
Scheme 8. H NMR
and MS data
of alleged 4,5-dihydro-1,2,3-oxadi-
azoles, which do not support the structure assignment of these heterocy-
cles.
1
unit in the five-membered ring of 16b should give H NMR sig-
[
35b]
nals at clearly lower field than d=2.45–2.96 ppm.
When ap-
propriate 4,5-dihydro-1,2,3-triazoles 11 are utilized as model
[
26]
1
compounds, the H NMR signals of the CH ÀN=N group in
2
Figure 3. Geometries of the DFT-optimized conformers 15a (left) and 15b
right).
1
6b should resonate in the range of d=4.0–4.3 ppm. The MS
(
data of 16c include m/z=457 as the highest mass signal,
+
[35d]
which was attributed to [MÀN ] .
However, it can also be
2
+
draw our conclusion: The dissociation energy is at least
interpreted as the [M] signal of a compound with an oxirane
ring instead of a 4,5-dihydro-1,2,3-oxadiazole unit. All reports
on stable 4,5-dihydro-1,2,3-oxadiazoles have in common the
fact that irrefutable structure proof, for example, by single-crys-
tal X-ray diffraction, is missing. Furthermore, thermolysis fol-
lowed by analysis of the succeeding products were not per-
formed in any case, and references to other papers with alleg-
edly isolable 4,5-dihydro-1,2,3-oxadiazoles or to previous at-
tempts to generate elusive 1 were not cited. Thus, we have
slight doubts that any heterocycle of type 1 has been directly
detected or even isolated, although different substitution pat-
terns may stabilize (or destabilize) such compounds. Structure
proof by X-ray diffraction or at least analysis of the thermal
decay products would dispel our doubts in the case of isolable
4,5-dihydro-1,2,3-oxadiazoles.
À1
1
45.2 kJmol , which is significantly higher than the activation
barriers for a concerted formation of acetaldehyde (being in
À1
the range of 75.7–97.5 kJmol ) or diazomethane (being in the
À1 [3]
range of 81.8–105.5 kJmol ). Therefore we can exclude a ho-
molytic cleavage of the NÀO bond from quantum chemical cal-
culations.
The energy-minimized structures for all compounds were
obtained by RI-DFT geometry optimizations with the TURBO-
[
28]
[29]
MOLE program package using the PW6B95 functional, the
[
30]
[31]
TZVPP basis set, Grimme’s D3 dispersion correction, and
[32]
the COSMO model (e=8.9). The geometry optimization of
was carried out as closed-shell singlet and those of com-
pound 15a,b as triplet. The CASPT2 calculations were per-
1
[
33]
formed with MOLPRO
and the CCSD(T)(F12) calculations
[
28,34]
with TURBOMOLE.
In summary, we have found strong evidence for the inter-
vention of 1 in our experiments with N-(2-chloroethyl)-N-nitro-
The resulting single-point energies of 1, 15a and 15b were
[
5]
corrected by the zero-point vibrational energies (ZPVE) and
sourea (3). In contrast to previous reports, our attempts to
demethylate 2 by attack of nucleophiles failed to produce 1. In
our conditions, 1 is too elusive to be detected directly, but pro-
ceeds to give oxirane, acetaldehyde, and diazomethane. Most
likely, these products arise from the zwitterionic intermediate
12 (Scheme 7) by way of intramolecular displacement, 1,2-hy-
dride shift, and fragmentation, respectively. A direct approach
to 12, starting from N-(2-hydroxyethyl)-N-nitrosourea (14) gave
product distributions similar to those obtained with 3. We
cannot exclude that synchronous cycloreversion of 1 contrib-
utes to the formation of diazomethane. The alternative cyclore-
solv
a solvation contribution (D ), both obtained from the DFT cal-
culations. The solvation correction was computed from DFT
single-point energies with and without using the COSMO
model.
Several authors have claimed that they successfully synthe-
sized single examples of 4,5-dihydro-1,2,3-oxadiazoles that can
[35]
be isolated at ambient temperature. Some of these reports
[
35a]
are very old,
but the structures were published also in
recent papers without NMR data or other convincing spectro-
[
35h]
1
scopic proof.
In some cases, H NMR data are evidently not
compatible with the suggested structures as shown in
version of 1, leading to N O and ethylene, was not observed in
2
Scheme 8. For example, the carbon-bonded methyl group of
our experiments.
1
1
6a should lead to a H NMR signal with a significantly lower
In earlier reports on the decay of N-(2-chloroethyl)-N-nitro-
soureas in aqueous solutions, acetaldehyde and ethylene
[35e,36]
chemical shift value than d=3.10 ppm,
and the CH ÀN=N
2
Chem. Eur. J. 2015, 21, 15092 – 15099
www.chemeurj.org
15097
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communication
glycol were detected and interpreted as succeeding products
Acknowledgement
[4]
of the postulated intermediate 1. Since we have been able to
reveal three different pathways for the decomposition of
We thank Dr. Henrique Teles (BASF) for stimulating discussions.
We are grateful to Dr. M. Hagedorn for his assistance with low-
temperature NMR experiments. N.S. thanks DAAD (Deutscher
Akademischer Austauschdienst) for a Ph.D. fellowship. M.K. is
grateful to the Fonds der Chemischen Industrie for a Ph.D. fel-
lowship.
1
(Scheme 7), including the first proof of the 1,3-dipolar cyclo-
reversion with formation of diazomethane, evidence of the ex-
istence of short-lived 1 is now unequivocal.
In previous quantum chemical calculations, which analyzed
the decay of 4,5-dihydro-1,2,3-oxadiazoles, it has been shown
that the competition between 1,2-hydride shift to give ketones
and cycloreversion to produce diazo and carbonyl compounds
is highly dependent on the substitution pattern of the cyclic or
Keywords: alkylation · cycloreversion · NMR spectroscopy ·
nucleophilic substitution · reactive intermediates
[
3]
bicyclic substrates. Generation of intermediates, similar to 12
or 15, and formation of epoxides are not investigated in this
[
3]
[1] Reviews: a) L. B. Clapp in Comprehensive Heterocyclic Chemistry, Vol. 6,
Pergamon (Eds.: A. R. Katritzky, C. W. Rees), Oxford, 1984, pp. 365–391;
b) T. L. Gilchrist, P. M. O’Neill, in Comprehensive Heterocyclic Chemistry II,
Vol. 4 (Eds.: A. R. Katritzky, C. W. Rees, E. F. V. Scriven), Elsevier, Oxford,
1996, pp. 165–178.
[2] a) G. I. Panov, K. A. Dubkov, E. V. Starokon, V. N. Parmon, React. Kinet.
Catal. Lett. 2002, 76, 401–406; b) K. A. Dubkov, G. I. Panov, E. V. Staro-
kon, V. N. Parmon, React. Kinet. Catal. Lett. 2002, 77, 197–205; c) J.
Teles, B. Rçßler, R. Pinkos, T. Genger, T. Preiss, PCT Int. Appl.
WO2005030689A2, 2005.
article, although the latter compounds were already suggest-
ed as secondary products of 4,5-dihydro-1,2,3-oxadiazoles,
when the reaction of alkenes with N O to give carbonyl com-
2
pounds as main products was reported for the first time in
[
37]
1
951. In another quantum chemical study, no reaction chan-
nel connecting the 4,5-dihydro-1,2,3-oxadiazole, derived from
the reaction of cyclohexene with N O, and cyclohexene oxide
2
[
11]
could be found. In sharp contrast, the favored transforma-
tion into ethylene oxide and dinitrogen, besides the slower cy-
cloreversion to afford formaldehyde and diazomethane, were
discussed as exclusive pathways in the decomposition of
[3] I. Hermans, B. Moens, J. Peeters, P. Jacobs, B. Sels, Phys. Chem. Chem.
Phys. 2007, 9, 4269–4274.
[
4] a) J. W. Lown, S. M. S. Chauhan, J. Org. Chem. 1982, 47, 851–856;
b) J. W. Lown, S. M. S. Chauhan, J. Org. Chem. 1981, 46, 2479–2489;
c) R. B. Brundrett, J. Med. Chem. 1980, 23, 1245–1247; d) J. W. Lown,
S. M. S. Chauhan, J. Org. Chem. 1981, 46, 5309–5321; e) J. W. Lown,
S. M. S. Chauhan, R. R. Koganty, A.-M. Sapse, J. Am. Chem. Soc. 1984,
[
38]
1
and described in a previous quantum chemical report.
However, we observed formation of acetaldehyde and ethyl-
ene oxide as major succeeding products of 1, whereas diazo-
methane is generated as minor component. These results are
not far from the product distribution in the industrial process-
1
06, 6401–6408; f) J. W. Lown, L. W. McLaughlin, J. A. Plambeck, Bio-
chem. Pharmacol. 1979, 28, 2115–2121; g) D. C. Chatterji, R. F. Greene,
J. F. Gallelli, J. Pharm. Sci. 1978, 67, 1527–1532.
[
5] a) C. J. Michejda, S. R. Koepke, M. B. Kroeger-Koepke, L. Hernandez, ACS.
Sym. Ser. 1994, 553, 195–210; b) C. J. Michejda, A. W. Andrews, S. R.
Koepke, Mutat. Res. 1979, 67, 301–308; c) R. N. Loeppky, L. Yu, F. Gu, Q.
Ye, J. Am. Chem. Soc. 1996, 118, 10995–11005; d) M. B. Kroeger-Koepke,
C. J. Michejda, P. P. Roller, L. K. Keefer, Cancer Res. 1985, 45, 2973–2975;
e) C. J. Michejda, S. R. Koepke, J. Am. Chem. Soc. 1978, 100, 1959–1960;
f) R. N. Loeppky, Q. Ye, P. Goelzer, Y. Chen, Chem. Res. Toxicol. 2002, 15,
es of treating cyclic olefins with N O, which lead mainly to ke-
2
tones, while decay products of diazo compounds are detected
[2c]
in small amounts only (Scheme 1). We assume that epoxides,
as decomposition products of 4,5-dihydro-1,2,3-oxadiazoles,
might be produced also in these industrial processes, but they
are possibly isomerized under the drastic reaction conditions
4
8
70–482; g) R. N. Loeppky, A. Srinivasan, Chem. Res. Toxicol. 1995, 8,
17–820; h) R. N. Loeppky, E. Erb, A. Srinivasan, L. Yu, ACS Symp. Ser.
[39]
(
250–2808C) to furnish ketones (Scheme 7). However, in the
case of the betain species of type 12 derived from cyclopen-
tene via a bicyclic 4,5-dihydro-1,2,3-oxadiazole intermediate
1994, 553, 334–336; i) S. R. Koepke, R. Kupper, C. J. Michejda, J. Org.
Chem. 1979, 44, 2718–2722.
6] K. Banert, O. Plefka, Angew. Chem. Int. Ed. 2011, 50, 6171–6174; Angew.
Chem. 2011, 123, 6295–6298.
[7] A. G. Tskhovrebov, B. Vuichoud, E. Solari, R. Scopelliti, K. Severin, J. Am.
Chem. Soc. 2013, 135, 9486–9492.
[
(
see Scheme 1), hydride shift to give cyclopentanone is per-
haps favored, and formation of cyclopentene oxide by S i reac-
N
tion is repressed because of the fixed cis configuration and ste-
reoelectronic reasons.
[
8] G. Kiefer, L. Jeanbourquin, K. Severin, Angew. Chem. Int. Ed. 2013, 52,
302–6305; Angew. Chem. 2013, 125, 6422–6425.
6
Considering the central role of 1 in such important chemical
transformations, the present results not only advance the un-
derstanding of the mechanism involved, but are also expected
to provide a basis for the development of novel synthetic pro-
cedures for generating 1.
[
9] a) M. B. Kroeger Koepke, A. M. Schmiedekamp, C. J. Michejda, J. Org.
Chem. 1994, 59, 3301–3306; b) A.-M. Sapse, E. B. Allen, J. W. Lown, J.
Am. Chem. Soc. 1988, 110, 5671–5675.
[
[
[
[
10] M.-D. Su, H.-Y. Liao, W.-S. Chung, S.-Y. Chu, J. Org. Chem. 1999, 64,
710–6716.
11] V. I. Avdeev, S. P. Ruzankin, G. M. Zhidomirov, Chem. Commun. 2003, 42–
3.
12] L. Zhao, R. Zhong, X. Yuan, Y. Cui, Q. Dai, Chin. Sci. Bull. 2004, 49, 1450–
452.
13] a) P. J. Burke, R. W. Matthews, D. G. Gillies, J. Chem. Soc. Dalton Trans.
980, 1439–1442; b) C. Janiak, Coord. Chem. 1997, 163, 107–216;
6
4
1
Experimental Section
1
c) L. K. Keefer, S.-m. Wang, T. Anjo, J. C. Fanning, C. S. Day, J. Am. Chem.
Soc. 1988, 110, 2800–2806.
14] See Supporting Information for details.
For experimental details and supporting spectra, please see the
Supporting Information.
[
[
15] S. Berger, S. Braun, H.-O. Kalinowski, NMR-Spektroskopie von Nichtmetall-
15
en, Band 2, N NMR-Spektroskopie, Thieme, Stuttgart, 1992.
1
5
[
16] For d( N) of diazomethane, quite different values, depending on the
solvent, were reported, see references [4e,15].
Chem. Eur. J. 2015, 21, 15092 – 15099
www.chemeurj.org
15098
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communication
1
3
15
1
[
[
[
17] For identical values of d( H) and J( N, H) of diazomethane, see: J. P. Ja-
[32] A. Klamt, G. Schüürmann, J. Chem. Soc. Perkin Trans. 2 1993, 799–805.
[33] a) H.-J. Werner, P. J. Knowles, G. Knizia, F. R. Manby, M. Schütz, WIREs
Comput. Mol. Sci. 2012, 2, 242–253; b) MOLPRO, version 2012.1, a pack-
age of ab initio programs, H.-J. Werner, P. J. Knowles, G. Knizia, F. R.
Manby, M. Schütz, and others, see http://www.molpro.net.
cobsen, K. Schaumburg, J. T. Nielsen, J. Magn. Reson. 1974, 13, 372–
3
78.
18] a) S. F. Nelsen, M. F. Teasley, M. Kaftory, J. Org. Chem. 1988, 53, 5930–
933; b) N. Boukamcha, R. Gharbi, M.-T. Martin, A. Chiaroni, Z. Mighri, A.
5
Khemiss, Tetrahedron 1999, 55, 449–460.
[34] a) S. Ten-no, Chem. Phys. Lett. 2004, 398, 56–61; b) D. P. Tew, W. Klopper,
J. Chem. Phys. 2005, 123, 074101; c) K. Peterson, T. Adler, H.-J. Werner, J.
Chem. Phys. 2008, 128, 084102; d) J. G. Hill, S. Mazumder, K. A. Peterson,
J. Chem. Phys. 2010, 132, 054108; e) K. E. Yousaf, K. A. Peterson, J. Chem.
Phys. 2008, 129, 184108; f) H. Fliegl, W. M. Klopper, C. Hättig, J. Chem.
Phys. 2005, 122, 084107; g) W. Klopper, C. C. M. Samson, J. Chem. Phys.
2002, 116, 6397–6410; h) S. Ten-no, J. Chem. Phys. 2004, 121, 117–129.
[35] a) L. I. Smith, W. B. Pings, J. Org. Chem. 1937, 2, 95–111; b) A. M.
Zvonok, N. M. Kuz’menok, L. S. Stanishevskii, Khim. Geterotsikl. Soedin.
1982, 5, 679–683; c) A. M. Zvonok, N. M. Kuz’menok, E. B. Okaev, I. G.
Tishchenko, Vestsi Akad. Navuk BSSR Ser. Khim. Navuk 1987, 4, 55–58;
d) H. Saito, A. Sato, T. Ashizawa, M. Morimoto, T. Hirata, Chem. Pharm.
Bull. 1990, 38, 3202–3210; e) M. Botta, R. Saladino, D. Lamba, R. Nicolet-
ti, Tetrahedron 1993, 49, 6053–6070; f) R. Saladino, L. Stasi, C. Crestini,
R. Nicoletti, M. Botta, Tetrahedron 1997, 53, 7045–7056; g) O. Illa, X.
Bagan, A. M. Cazorla, C. Lyon, A. Baceiredo, V. Branchadell, R. M. OrtuÇo,
J. Org. Chem. 2006, 71, 5320–5327; h) E. A. El-Sawi, T. B. Mostafa, H. A.
Radwan, Chem. Heterocycl. Compd. 2009, 45, 981–989.
19] CCDC 1061632 contains the supplementary crystallographic data for
this paper. These data can be obtained free of charge from The Cam-
bridge Crystallographic Data Centre
[
[
20] K. Banert, Synthesis 2007, 3431–3446.
21] We tried quite different reaction conditions and also utilized other nu-
cleophiles, such as bromide and thiocyanate. In all cases, however, we
found attack at C-5 with ring opening or no consumption of 2.
22] D. Seebach, D. Enders, Chem. Ber. 1975, 108, 1293–1320.
23] a) H. Wamhoff, in Comprehensive Heterocyclic Chemistry, Vol. 5, Part 4A,
Pergamon (Eds.: A. R. Katritzky, C. W. Rees), Oxford, 1984, pp. 669–732;
b) W.-Q. Fan, A. R. Katritzky, in Comprehensive Heterocyclic Chemistry II,
Vol. 4, Elsevier (Eds.: A. R. Katritzky, C. W. Rees, E. F. V. Scriven), Oxford,
[
[
1996, pp. 1–126.
[
[
24] a) M. Regitz, G. Himbert, Justus Liebigs Ann. Chem. 1970, 734, 70–85;
b) G. Bianchetti, D. Pocar, P. Dalla Croce, A. Vigevani, Chem. Ber. 1965,
9
8, 2715–2724; c) P. Dalla Croce, R. Stradi, Tetrahedron 1977, 33, 865–
8
67.
25] a) K. Hirakawa, T. Ito, Y. Okubo, S. Nakazawa, J. Org. Chem. 1980, 45,
668–1672; b) M. S. Ouali, M. Vaultier, R. CarriØ, Bull. Soc. Chim. Fr.
979, 633–643; c) D. Pocar, M. C. Ripamonti, R. Stradi, P. Trimarco, J.
[36] The value d=3.10 ppm might be a typing error and should read d=
1.30 ppm. However, when we repeated the simple reaction of 1,3-di-
1
1
methyl-5 formyluracil with diazomethane in Et O/MeOH, we did not
2
Heterocycl. Chem. 1977, 14, 173–174; d) Y. M. Saunier, R. Danion-
Bougot, D. Danion, R. CarriØ, Tetrahedron 1976, 32, 1995–1999; e) F.
Texier, J. Bourgois, J. Heterocycl. Chem. 1975, 12, 505–508.
get 16a, which should be formed in 80% yield, or 1,3-dimethyl-5-(2,4-
dihydro-1,3,4-oxadiazol-2-yl)uracil that should result with 53% yield. In-
stead, we always obtained a complicated mixture with a high number
of products after the reaction was performed several times and by dif-
ferent co-workers.
[
26] For several examples, see: K. Banert, J. Lehmann, H. Quast, G. Meichs-
ner, D. Regnat, B. Seiferling, J. Chem. Soc. Perkin Trans. 2 2002, 126–
1
34.
[37] F. S. Bridson-Jones, G. D. Buckley, L. H. Cross, A. P. Driver, J. Chem. Soc.
1951, 2999–3008.
[
[
27] S. S. Singer, J. Med. Chem. 1985, 28, 1088–1093.
28] TURBOMOLE V6.5 2013, a development of University of Karlsruhe and
Forschungszentrum Karlsruhe GmbH, 1989–2007, TURBOMOLE GmbH,
since 2007; available from http://www.turbomole.com.
29] Y. Zhao, D. G. Truhlar, J. Phys. Chem. A 2005, 109, 5656–5667.
30] a) K. Eichkorn, F. Weigend, O. Treutler, R. Ahlrichs, Theor. Chem. Acc.
[38] V. Branchadell, E. Muray, A. Oliva, R. M. OrtuÇo, C. Rodríguez-García., J.
Phys. Chem. A 1998, 102, 10106–10112.
[39] With other examples, the uncatalyzed isomerization of epoxides to give
carbonyl compounds was observed, for example, during distillation: H.
Hopf, R. Wandeler, Chimia 1962, 16, 86–87.
[
[
1
997, 97, 119–124; b) F. Weigend, M. Häser, H. Patzelt, R. Ahlrichs,
Chem. Phys. Lett. 1998, 294, 143–152; c) F. Weigend, Phys. Chem. Chem.
Phys. 2006, 8, 1057–1065.
[
31] a) S. Grimme, J. Antony, S. Ehrlich, H. Krieg, J. Chem. Phys. 2010, 132,
1
2
54104/1–19; b) S. Grimme, S. Ehrlich, L. Goerigk, J. Comput. Chem.
011, 32, 1456–1465.
Received: June 16, 2015
Published online on August 10, 2015
Chem. Eur. J. 2015, 21, 15092 – 15099
www.chemeurj.org
15099
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim