Gas-phase spectroscopy of the unstable acetonitrile N-oxide molecule, CH3CNO

Article abstract of DOI:10.1021/jp002851z

The unstable acetonitrile N-oxide molecule, CH₃CNO, has been thermolytically generated in very high yield in the gas phase from its stable ring dimer, dimethylfuroxan, and studied by ultraviolet photoelectron spectroscopy, photoionization mass spectroscopy, and mid-infrared spectroscopy. The individual spectroscopies provide a detailed investigation into the vibrational and electronic character of the molecule, and are supported by both conventional ab initio calculations and density functional theory. The ground-state structure is also investigated by theory at the B3-LYP, MPn (n = 2-4), QCISD, and QCISD(T) levels with medium to large basis sets, and illustrates the need for a precise description of electron correlation. Given that both isomerization and dimerization are feasible loss processes for this unstable molecule, the relative stability of CH₃CNO with respect to the known cyanate (CH₃OCN), isocyanate (CH₃NCO), and fulminate (CH₃ONC) isomers and the mechanism of the dimerization processes were studied with density functional theory.

Full text of DOI:10.1021/jp002851z

1244
J. Phys. Chem. A 2001, 105, 1244-1253
Gas-Phase Spectroscopy of the Unstable Acetonitrile N-Oxide Molecule, CH3CNO
Tibor Pasinszki*,† and Nicholas P. C. Westwood*,‡
Department of Inorganic Chemistry, Budapest UniVersity of Technology and Economics, H-1521 Budapest,
Gelle´rt te´r 4, Hungary, and Guelph-Waterloo Centre for Graduate Work in Chemistry, Department of
Chemistry and Biochemistry, UniVersity of Guelph, Guelph, Ontario, Canada N1G 2W1
ReceiVed: August 7, 2000; In Final Form: December 8, 2000
The unstable acetonitrile N-oxide molecule, CH3CNO, has been thermolytically generated in very high yield
in the gas phase from its stable ring dimer, dimethylfuroxan, and studied by ultraviolet photoelectron
spectroscopy, photoionization mass spectroscopy, and mid-infrared spectroscopy. The individual spectroscopies
provide a detailed investigation into the vibrational and electronic character of the molecule, and are supported
by both conventional ab initio calculations and density functional theory. The ground-state structure is also
investigated by theory at the B3-LYP, MPn (n ) 2-4), QCISD, and QCISD(T) levels with medium to large
basis sets, and illustrates the need for a precise description of electron correlation. Given that both isomerization
and dimerization are feasible loss processes for this unstable molecule, the relative stability of CH3CNO with
respect to the known cyanate (CH3OCN), isocyanate (CH3NCO), and fulminate (CH3ONC) isomers and the
mechanism of the dimerization processes were studied with density functional theory.
Introduction
phase by ultraviolet photoelectron (PE), photoionization mass
(PIM), and mid-IR spectroscopies, and consider potential loss
processes involving isomerization and/or dimerization. Rela-
tively stable large aromatic nitrile oxides are known to isomerize
to isocyanates upon heating, whereas the primary loss process
of small nitrile oxides is dimerization.5 Thus, 1 is unstable at
room temperature, dimerizing within 1 min at 18 °C to the
dimethylfuroxan ring, 2 (3,4-dimethyl-1,2,5-oxadiazole 2-oxide),
thereby inhibiting investigation in the gas phase. It is stated,
however,18 that 1 can be stored at temperatures well below 0
°C. For solution chemistry 1 is normally generated in situ by
dehydrohalogenation of acetohydroxamyl chloride (R-chloro-
acetaldoxime)19 or by dehydration of nitroethane.5 Of relevance
to the present gas-phase work is the observation that 1 can be
generated from its stable ring dimer, 2, using the flash vacuum
thermolysis (FVP) method (600 °C, 10-3 Torr).20 Nonetheless,
the IR spectrum of 1 has only been investigated in solutions of
CCl4, CS2, and cyclohexane,18 and matrix IR has identified 1
following photolysis of acetonitrile and O(3P) atoms.21 The only
published gas-phase spectroscopy is the microwave (MW)22,23
and millimeter wave spectra24 of samples obtained by the
standard route of HCl elimination from R-chloroacetaldoxime.19
The analyzed rs structure indicates that 1 is a symmetric top
(C3V) with a large dipole moment (4.49 D).
Several open-chain isomers of simple pseudohalide molecules
containing a “CNO” group can be drawn, some of which are
known experimentally. The methyl derivatives represent the only
example of small pseudohalides where the three lowest energy
isomers (isocyanate (-NCO), cyanate (-OCN), and nitrile oxide
(-CNO) in order of decreasing thermodynamic stability) are
known and can be cleanly generated in the gas phase. CH3-
NCO is a liquid at room temperature (boiling point, 44-46 °C),1
the generation and gas-phase spectroscopy of the semistable
CH3OCN molecule was recently reported,2,3 and a gas-phase
investigation of pure CH3CNO is discussed in the present work.
A fourth open-chain isomer (excluding the unreasonable valence
structures -NOC and -CON) is the much higher energy
fulminate, CH3ONC, observed by low-temperature infrared (IR)
spectroscopy.4
Substituted nitrile oxides, XsCtNfO, the archetypal 1,3-
dipolarophiles, have proven to be useful reagents in synthetic
organic chemistry. Because of their high reactivity and propen-
sity to dimerize, they are not isolated, but rather are generated
in situ for various dipolar cycloaddition reactions including
formation of substituted heterocycles such as 2-isoxazolines,
isoxazoles, 1,3,4-dioxazoles, 1,2,4-oxadiazoles,5 and many
important biologically active compounds.6 More recently, the
smaller derivatives (X ) Br-,7,8 Cl-,9,10 NC-,11 and ONC-
12) have been generated as unstable molecules in the gas phase.
Of interest are their structures and the associated spectros-
copy,8,10,13-15 bringing into focus issues of quasi-linearity, the
efficacy of quantum mechanical calculations to predict structures
and barriers to linearity,7,9,11,12,16,17 and for those species
containing the relatively abundant elements C, N, O, and H
possible candidacy for astrophysical observation.
Early ab initio calculations limited to the Hartree-Fock (HF)
level with modest (STO-3G, 4-31G, 6-31G*) basis sets25,26
suggested that 1 had a linear CCNO frame, in agreement with
the MW results. This was fortuitous since it is now well-known
that large basis sets and the inclusion of electron correlation
are of crucial importance in the prediction of the structures of
nitrile oxides,7,9,11,16,17,27 In the present work ab initio calcula-
tions (including density functional theory (DFT) methods) are
used to assess the effect of basis set size and electron correlation
on the calculated versus MW structure of 1, and to assist in the
analysis of the PE and IR data. Since 1 and the better known
isomers CH3NCO and CH3OCN can potentially interconvert and
are of interest as model compounds for substituted R(CNO)
species, comparison is made to the spectroscopy and structure
In the present work we discuss the gas-phase generation and
first characterization of the CH3CNO molecule, 1, in the gas
* To whom correspondence should be addressed. E-mail: pasinszki.inc@
chem.bme.hu, westwood@chembio.uoguelph.ca.
† Budapest University of Technology and Economics.
‡ University of Guelph.
10.1021/jp002851z CCC: $20.00 © 2001 American Chemical Society
Published on Web 02/02/2001

Gas-Phase Spectroscopy of CH3CNO
J. Phys. Chem. A, Vol. 105, No. 8, 2001 1245
SCHEME 1
the pressure was kept constant between 450 and 500 mTorr.
The spectra were recorded at 0.5 cm-1 resolution.
Computational Methods
Ab initio calculations for the structure of 1 were carried out
at the MP2, MP3, MP4SDQ, MP4 ()MP4SDTQ), QCISD, and
QCISD(T) levels using standard basis sets, viz., 6-31G**,
6-311G(2d,2p), or cc-pVTZ (Dunning’s correlation-consistent
basis set; [4s,3p,2d,1f] on C, N, and O atoms, and [3s,2p,d] on
H atoms). All electrons were included in the correlation energy
calculations (i.e., “full”). DFT, in the form of B3-LYP (Becke’s
three-parameter exchange functional in combination with the
Lee, Yang, and Parr correlation functional), was also used to
calculate the equilibrium structure of 1. Isomerization of 1, the
interconnecting transition states (TSs), and the dimerization
mechanism were investigated at the cost-effective B3LYP/6-
31G** level. Equilibrium molecular geometries were fully
optimized and harmonic vibrational frequencies calculated at
the minimum-energy geometries to confirm them as real points
on the potential energy surface (zero imaginary frequencies).
TSs were characterized by one imaginary frequency. IPs for 1
were calculated using the HAM/332 and the outer valence
Green’s function (OVGF) methods.33 All calculations were
performed with the Gaussian-92 and Gaussian-94 quantum
chemistry packages34,35 implemented on Silicon Graphics Inc.
Challenge/XL and Origin 200 workstations.
of these isomers, including DFT calculations of their bending
potentials, and the unimolecular isomerization processes of 1
to the other isomers. The dominant loss process of 1, dimer-
ization to the ring furoxan 2, is also considered using DFT.
Experimental Section
The specifics of the generation of 1 from the stable dimer
precursor 2 are given below in the Results and Discussion.
Several methods of synthesizing 2 are known. Many are
multistep procedures often utilizing the oxidizing power of liquid
dinitrogen tetroxide, N2O4. One of the simplest (and oldest)
methods describes the oxidation of dimethylglyoxime with
N2O4.28 Since it has been noted that dichloroglyoxime can be
oxidized to dichlorofuroxan using white fuming HNO3,29 and
since the handling of liquid nitric acid, HNO3, is easier than
that of N2O4, we adopted and modified this latter method using
concentrated HNO3 to prepare 2 from dimethylglyoxime (Al-
drich), Scheme 1.
Results and Discussion
A 40 g sample of dimethylglyoxime was added slowly (3 h),
with stirring, to 300 mL of concentrated nitric acid (68% m/m)
at room temperature. The evolution of brown nitrogen oxides
was observed instantly, the solution turning brown. Upon
completion of the addition of dimethylglyoxime, the solution
was stirred for 1 h, poured onto ice, and extracted with 3 ×
100 mL of dichloromethane. The extract was washed with ice
water until neutral, dried over magnesium sulfate, and evapo-
rated. A 25 mL sample of a pale green raw product was
obtained, which was distilled four times through a 15 cm long
Vigreux column, giving 2 as an almost colorless liquid with a
boiling point of 50-52 °C at ca. 2.5 mmHg. The purity was
confirmed by IR spectroscopy. The repeated distillations were
of crucial importance in obtaining pure 2, since the boiling point
of the green side product (not identified in this work) was only
4 °C lower at 2.5 mmHg than that of 2. Due to the loss of 2
during the repeated purification process, the final yield was low,
ca. 30%, but this could certainly be improved using more
sophisticated purification methods.
HeI (21.2 eV) PE spectra were recorded on a fast-pumping
spectrometer specifically designed to study reactive and unstable
species.30,31 The resolution was ca. 45 meV during the experi-
ments, with spectra calibrated using the known ionization
potentials (IPs) of O2 and N2. Mass analysis of ions is achieved
with a quadrupole mass analyzer (Hiden Analytical, 320 amu)
mounted directly above the photoionization point. With the
conventional EI source removed, ionization is provided by
single-wavelength HeI or unfiltered HLR,â,γ (10.2-12.7 eV)
radiation. PE and PIMS spectra, although not done in coinci-
dence, can be recorded within seconds of each other under
identical conditions; thus, it is assumed that for a given PE
spectrum the subsequent PIMS is of the same compound.
Generation, Identification, and Stability of CH3CNO, 1.
For direct gas-phase generation of 1 the methods used for in
situ 1,3-dipolar cycloadditions in solution5,19 are inappropriate
due to unwanted elimination products (e.g., HCl from aceto-
hydroxamoyl chloride). An approach based on the thermolytic
cycloreversion of 220 offers a direct route capable of generating
1 as the sole product for spectroscopic investigation. Precedents
for this approach also include our recent demonstration of the
clean generation of NCCNO from its stable furoxan dimer11
and the use of this method by other groups.36 By attaching the
thermolysis tube directly to the spectrometers, we could monitor
the efficiency of the thermolytic cycloreversion, and directly
observe potential intramolecular rearrangements to isomers of
1 (e.g., the isocyanate), or possible dimerization.
Thermolysis of 2 was carried out in a quartz tube (8 mm
i.d.) heated along 15 cm, packed with quartz chips to improve
efficiency. The FVP of 2 led to smooth formation of 1 at 600
°C. From the various spectroscopies (below) which confirmed
the identity of the molecule, the direct gas-phase yield of 1 can
be almost quantitative, with only trace amounts of CO, CO2,
NO, CH3NCO, and CH3CN as side products. Purer samples
could be obtained by condensing 1 at -196 °C in a U-trap
connected to the IR cell or PE spectrometer and then gradually
warming, pumping off traces of side products below -20 °C.
1 had sufficient vapor pressure at -20 °C for PES and PIMS,
and at -10 °C for IR investigation. After the spectra were
recorded, the temperature was increased to 0 °C; only gaseous
1 was observed at this temperature, indicating that dimerization
of the trapped material is slow, unlike the case of BrCNO.7
This trapping/revaporizing result suggests that 1 could be trapped
and stored at liquid N2 temperatures, and then recovered upon
warming. There was no indication of isomerization of 1 under
these conditions.
Mid-IR spectra of 1 were collected on a Nicolet 20SXC
interferometer equipped with a 20 cm single-pass cell. The cell,
with KBr windows, gave a spectral range down to 350 cm-1
.
Computational Investigation of the Structure of 1. Since
the inclusion of electron correlation is de rigeur for calculations
on simple nitrile oxides,7,9,11,16,17,27 the geometry of 1 was
The effluent from the pyrolysis tube or sample container was
pumped continuously through the cell using a rotary pump while

1246 J. Phys. Chem. A, Vol. 105, No. 8, 2001
Pasinszki and Westwood
TABLE 1: Calculated Barriers to Linearity, Equilibrium Structures, Total Energies, Rotational Constants, and Dipole
Moments of CH3CNO, 1a
MP2
MP3
MP4SD Q
MP4
QCISD
QCISD(T )
B3-LYP
exptl MW22
Barrier to Linearityb
basis set
cc-pVTZ
6-311G(2d,2p)
6-31G**
0.0
(19.3)
(75.5)
0.0
(0.0)
(0.0)
0.0
(0.0)
0.0
(0.0)
126.6
(226.4)
0.0
(0.0)
0.0
(0.0)
Structure for the Largest Basis Set and Experimental Structure
C-C
1.453
1.174
1.198
1.085
1.466
1.148
1.210
1.084
1.468
1.159
1.216
1.086
1.477
1.191
1.213
1.087
1.089
1.469
1.156
1.221
1.086
1.471
1.166
1.221
1.088
1.456
1.156
1.210
1.091
1.442
C-N
N-O
C-H
C-H′′
1.169
1.217
no data for
H params
CCN
CNO
HCC
H′′CC
180.0
180.0
110.6
180.0
180.0
110.2
180.0
180.0
110.3
154.5
180.0
180.0
110.2
180.0
180.0
110.3
180.0
180.0
110.7
180.0
180.0
172.7
110.4
110.5
total energy
-207.588165 -207.520477 -207.536622 -207.582668 -207.536014 -207.570355 -207.981613
Ac
B
161.9197
3.9180
3.9180
5.276
161.4988
3.9312
3.9312
5.406
161.1145
3.8926
3.8926
5.460
121.6789
3.9500
3.9196
5.002
160.9994
3.8890
3.8890
5.482
160.5556
3.8608
3.8608
5.436
160.5408
3.9266
3.9266
4.271
3.914796
4.49
C
µd
a Bond angles in deg, bond lengths in Å, total energies in au. For the structures, the cc-pVTZ basis set was used at the B3-LYP and MP2 levels
and the 6-311G(2d,2p) basis set was used at the MP3, MP4, and QCI levels. All of the electrons were included in the correlation energy calculations
(“full”). b Barriers to linear CCNO frame in cm-1. Values in parentheses refer to barriers obtained with smaller basis sets. c Rotational constants in
GHz. d Dipole moments in D.
calculated at various levels of post-HF theory, with increasing
basis set size from 6-31G** through 6-311G(2d,2p) to cc-pVTZ
to study basis set effects.
(T) level,38 the linear geometry of the molecule is reproduced
and the calculated geometry is in better agreement with the
experimental structure. Clearly, the electron-rich nature of nitrile
oxides poses a serious computational challenge due to large and
sensitive electron correlation effects, and demands the use of
extra large basis sets and very high levels of theory when
accurate structural and spectroscopic properties are calculated.
This has been demonstrated recently in calculations for the
structure of HCNO27 and in calculating energy levels in the
region of the barrier for BrCNO16 and ClCNO.17
We note here that if very high level conventional ab initio
calculations are not feasible, DFT (e.g., using B3-LYP) can
provide a reliable cost-effective alternative, giving, for nitrile
oxides, reasonable results for structures (Table 1) and good
agreement with vibrational frequencies. For this reason we have
used the B3-LYP method in calculations of the related isomers
and their interconversions, and for evaluating the dimerization
process (below).
The known structural differences between 1 and the CH3-
NCO and CH3OCN isomers suggest different electronic struc-
tures and chemical reactivities. To highlight these before
discussing the spectroscopy, we have calculated the (B3-LYP/
cc-pVTZ) structures of these two other isomers (where MW
structures are also known39,40) and compared the potentials of
their lowest frequency deformations with that of 1. Figure 1
demonstrates that CH3OCN has a steep harmonic well, which
becomes progressively more anharmonic in CH3NCO and CH3-
CNO. The fitted potentials are as follows:
The calculated equilibrium geometry of 1 is shown in Table
1, and concurs (but see below) with the experimental rs geometry
of the heavy atoms obtained from the MW data,22 which suggest
a symmetric-top molecule with a linear CCNO frame. From
the l-type doubling constant a value of 150(50) cm-1 was
obtained for the lowest frequency deformation, insufficiently
low enough to indicate quasi-linear behavior,24 but suggesting
that, as with other simple nitrile oxides, difficulties with
calculated structures might be expected. Indeed, the calculated
structures fall into the familiar pattern,7,9,11 with respect to both
the size of the basis set and the level of electron correlation.
With the smallest basis set (6-31G**) the MP3, MP4SDQ,
QCISD, QCISD(T), and B3-LYP levels predict a linear CCNO
frame, whereas MP2 and MP4SDTQ predict a bent molecule.
Increases in basis set size decrease the barrier to linearity, with
the calculated structure becoming linear at MP2/cc-pVTZ.
Correspondingly, the CtN bond, which is strongly correlated
to the CCN angle, is shortest at linearity, e.g., with MP2, 1.193
Å (6-31G**) f 1.181 Å (6-311G(2d,2p) f 1.174 Å (cc-pVTZ).
Resource limitations precluded further increases in basis set size
at the MP4SDTQ level, where 1 is still bent, but we expect the
same trend toward linearity to hold.
The strong electron correlation effects on calculated properties
noted previously (BrCNO,7 ClCNO,9 and HCNO27) are dem-
onstrated by an oscillatory behavior in the MPn expansion series
(MP1)HF, MP2, MP3, MP4), for example, the CtN bond
length using the 6-311G(2d,2p) basis set (1.123, 1.181, 1.148,
and 1.191 Å, respectively), barrier height (0, 19.3, 0, and 126.6
cm-1, respectively), and total energy (Table 1). The other effect
is that triple excitations markedly lengthen the CtN bond
length, tending to bend the structure; compare MP4SDQ with
MP4SDTQ and QCISD with QCISD(T). MP4SDTQ clearly
overestimates the CsC and CtN bond lengths and barrier
height, in accord with the known fact that the method exag-
gerates triple contributions to the correlation energy.37 By
including a better description of triple excitations at the QCISD-
CH3OCN: V (cm-1) ) 7830.60 + 22.751F - 6.360F2 +
0.064F3
CH3NCO: V (cm-1) ) 595.95 + 4.646F - 1.191F2 +
0.014F3 + 0.00010F4
CH3CNO: V (cm-1) ) 0.581F2 + 0.00016F4
where F is the deviation from 180° at the central atom. What

Gas-Phase Spectroscopy of CH3CNO
J. Phys. Chem. A, Vol. 105, No. 8, 2001 1247
Figure 3. Gas-phase mid-infrared spectrum of CH3CNO at ca. 450
mTorr and 0.5 cm-1 resolution. Insets show details of the ν2, ν7 and
ν3, and ν4 bands.
Figure 1. Calculated (B3-LYP/cc-pVTZ) potentials (constructed by
optimizing every 5°) of the lowest frequency deformations in (a) CH3-
CNO, (b) CH3NCO, and (c) CH3OCN.
pseudohalide (X ) ONC-, NC-) group by electropositive
-CH3. Other fragments also observed for 1 are (M + 1)+,
HCNO+, CNO+, CH2C+, and CH3+. The (M + 1)+ fragment
also seen with the CH3NCO and CH3OCN isomers2 may result
from secondary ion reactions in the quadrupole. This may also
explain the observation of HCNO+ and CNO+ fragment pairs.
Fragmentation is usually reduced in the HLR,â,γ PIMS due to
the lower energy light source and is reflected in the strong
increase in the parent ion signal and negligible NO+ and CH3
+
fragments.
Infrared Spectroscopy. The gas-phase IR spectrum of 1 is
shown in Figure 3 with the vibrational frequencies of the
fundamentals provided in Table 2 together with the calculated
(B3-LYP/cc-pVTZ and QCISD/6-31G**), unscaled, vibrational
frequencies and intensities. For comparison, Table 2 also
includes the data on this molecule from the only other known
(solution18 and matrix21) IR work. The observed spectrum
confirms the identity of 1, the absence of the precursor/dimer
Figure 2. (a) HeI and (b) HLR,â,γ photoionization mass spectra of CH3-
CNO.
molecule 2, which has its strongest band at 1614 cm-1 41
and
,
that there is no evidence for the CH3OCN and CH3NCO
isomers. 1 is the sole species in the gas phase, with almost all
bands (weak and strong) being assignable to either fundamentals
or overtone/combination bands.
emerges is that CH3OCN is strongly bent and rigid (small
amplitude deformation), CH3CNO is linear and floppy (large
amplitude deformation at the central carbon atom), and CH3-
NCO is in between, its frame being less bent, but more flexible
than that of CH3OCN. These pronounced structural differences
are reflected in the spectroscopy.
Photoionization Mass Spectroscopy. The HeI and HLR,â,γ
PIMS spectra of 1 (Figure 2) are relatively simple, confirm its
identification, and clearly distinguish it from the cyanate and
isocyanate isomers (see Figure 2 in ref 2), the cracking patterns
being quite different. The base peaks in the HeI PIMS of 1,
CH3NCO, and CH3OCN are NO+, CO+, and CH3+, respectively.
NO+ is observed only with 1, quite typical for nitrile ox-
ides.7,11,12 CO+ is not observed in the PIMS of 1, and is
relatively small for CH3OCN. The CH3+ fragment, with a strong
relative intensity in the HLR,â,γ PIMS of CH3OCN, is not
observed in the corresponding PIMS of CH3NCO and is
negligible for 1.
The 15 normal modes of 1 transform in C3V symmetry as
5A1 + 5E, and all but the lowest frequency bending mode (ν10),
are observed in this work. For such a symmetric top, parallel-
type bands of a1 symmetry (A1 r A1 transition) should exhibit
P, Q, and R structure arising from overlapping K ) 0, 1, 2, ...
subbands. However, in a case such as 1 where IA/IB is small
(i.e., A . B), the Q branch is predicted to have a low intensity.42
Perpendicular transitions (E r A1) in a symmetric top can lead
to the classic, and often prominent, rQ and pQ subbands (due to
different K values) with a separation given by 2[A(1 - úi) -
B], where úi is the Coriolis coupling constant. Both types of
bands are observed, the subbands in the perpendicular transitions
(in those cases where úi ≈ 0) being clearly resolved even at 0.5
cm-1 resolution, and four of the five parallel transitions
exhibiting unresolved P and R envelopes with a separation of
band maxima of ca. 15-17 cm-1
.
The three strongest peaks in the HeI PIMS of 1, the molecular
ion (M+), and NO+ and XC+ (X ) -CH3) fragments are also
In the CH stretching region near 3000 cm-1, the most
prominent band exhibiting PR structure is centered at 2943 cm-1
and is assigned to ν1, the symmetric CH3 stretch of a1 symmetry.
The antisymmetric CH3 stretch, ν6(e), to higher wavenumber,
is much weaker, as predicted by the calculations. Such an E r
A1 perpendicular transition has its intensity distributed into the
rQ and pQ subbands, which show evidence of intensity alterna-
tion (..., strong, weak, weak, strong, ...) due to nuclear spin
statistics, despite the poor signal/noise ratio. The observed
seen in the spectra of other XCNO molecules (X ) Br-,7 NC-
11
,
ONC-12). Their relative intensities are different, however,
as the NO+ peak dominates the HeI spectra of BrCNO, NCCNO,
and ONCCNO, with only a small-intensity XC+ fragment. For
1, the relative intensity (compared to that of M+) of NO+ has
strongly diminished while that of XC+ has increased, showing
that the probability of neutral NO loss from the excited M+ ion
increases by replacing the electronegative halide (X ) Br-) or

1248 J. Phys. Chem. A, Vol. 105, No. 8, 2001
Pasinszki and Westwood
TABLE 2: Experimentala and Calculatedb Vibrational Frequencies (cm-1) of the Fundamental Vibrations of CH3CNO
experimental
solnc
B3-LYP/cc-pVTZ
QCISD/6- 31G**
assignment and
description
gas
matrixd
freq (sym)
intense
freq (sym)
intense
3022.5f vvw
2967
2924
3089 (E)
4.5
3213 (E)
5.2
ν6 CH as str
2950 R
w
3028 (A1)
17.7
3125 (A1)
17.5
ν1 CH sym str
}
2935 P
2311g vs
2315
1441
2309
1381
2455 (A1)
1478 (E)
429.6
8.7
2493 (A1)
1540 (E)
324.5
7.3
ν2 CNO as str
ν7 CH3 as def
1453.2h vw
1401 R
vw
1381
1428 (A1)
21.2
1478 (A1)
0.1
ν3 CH3 sym def
}
1386 P
1356 R
1348.9 Q
1346.3 Q
1340 P
1034 vvw (see the
text re the
band at 1162)
791 R
is
1319
1177
1332
1389 (A1)
1051 (E)
195.2
1.0
1379 (A1)
1083 (E)
140.2
0.5
ν4 CNO sym str
ν8 CH3 rock
}
788.1 Q
781.9 Q
777 P
iw
778
780
795 (A1)
20.4
799 (A1)
26.1
ν5TC-C str
}
477.3 vw j
478
297
510 (E)
137 (E)
4.0
1.4
493 (E)
112 (E)
6.8
0.9
ν9 CNO bend
ν10 CCN bend
(150 (50))k
a Band positions (gas) taken from the maxima or sharp features of the IR bands. b Unscaled harmonic frequencies. c From ref 18. d From ref 21.
e In km/mol. f Sharp rQ/pQ subbands (cm-1) at 3022.5 (most intense), 3012.8, 3003.9, 2993.8(s), 2984.3, 2973.9, 2965.1(s), and 2936 (s) (overlapping
ν1) (see the text). g Strongest peak of a complex band (see the text). h Sharp Q/pQ subbands (cm-1) at 1453.2 (most intense), 1442.1, 1429.3,
r
1416.1(s), 1403 and 1388 (overlapping ν3), and 1373.8 (overlapping ν4) (see the text). i See the text regarding possible difference bands. j Central
spike. k From ref 24 obtained from the microwave spectrum, ref 22.
subband spacings are about 9 cm-1, the line positions being
given as a footnote to Table 2. We cannot unambiguously
identify the band origin; the center of the four strongest observed
subbands is at about 2980 cm-1, although the most intense
subband at 3022.5 cm-1 is listed in the table as the nominal
band center.
The two most prominent bands in the spectrum, characteristic
of all nitrile oxides, are centered at 2311 and 1348 cm-1
corresponding to the antisymmetric (ν2) and symmetric (ν4)
CNO stretches, both of a1 symmetry. Their relative intensities
are well predicted by the ab initio calculations. The symmetric
stretch shows PR structure with what looks like two Q branches
(labeled as such in Table 2), although one of these may arise
from a difference band of the type ν4 + ν10 - ν10. The
antisymmetric stretch shows a complicated rotational profile
consistent with the concurrent excitation of low-frequency
bending modes. The value of 2311 cm-1 is assigned to the most
intense feature, although Q branches at 2305, 2318, and 2333
cm-1 are observed. Such hot band structure was observed
previously on the antisymmetric CNO stretches of BrCNO,7
ClCNO,9 NCCNO,11 and ONCCNO.12,13
To the high wavenumber side of the symmetric stretch
centered at 1348 cm-1 are two additional weak bands with quite
different rotational profiles (see the insets of Figure 3), predicted
by the ab initio calculations to correspond to the antisymmetric
ν7(e) and symmetric ν3(a1) CH3 deformations. These are
identified, respectively, at 1453.2 cm-1 (most intense subband
of the well-spaced rQ and pQ subbands of an E r A1
perpendicular transition) and 1394 cm-1 (center of the PR
structure), mimicking in profile the close e and a1 pair (ν6 and
ν1) in the CH3 stretching region. The degenerate mode exhibits
the expected intensity alternation due to nuclear spin statistics;
the positions of these subbands, with an average spacing of about
13 cm-1, are given as a footnote to Table 2. This spacing is
somewhat higher than that observed for ν6 (ca. 9 cm-1), but
understandable in terms of the possible allowed values (between
+1 and -1) of úi. In the absence of the Coriolis force (úi ) 0),
the calculated rotational constants (Table 1) predict a value of
ca. 10.5 cm-1 for the subband separation. A curious feature
(possibly arising from an A-E Coriolis interaction and a
resulting intensity perturbation) is the sudden decrease of the
subband intensity to the high-wavenumber side, although weaker
structure extending up to ca. 1570 cm-1 can be observed. As
with ν6 this makes location of the ν7 band center problematic,
and although we list the strongest subband peak at 1453.2 cm-1
in Table 2, the actual band center could be higher. A likely
combination band, ν4 + ν10 of E symmetry, would also appear
in this region, potentially leading to a Fermi resonance.
Below 1200 cm-1 are four weak but distinct bands, two of
which, on the basis of their band structure, can be assigned to
the fundamentals ν5(a1) (785 cm-1) and ν9(e) (477 cm-1),
respectively, the CC stretch and the CNO bend. The positions
of these normal modes are in good agreement with the calculated
frequencies, as are the intensities, with the more intense ν5
showing PR structure and two central Q branches again possibly
resulting from difference bands. The perpendicular band ν9
exhibits K structure, albeit not as clear as in ν6 and ν7, with
extra “spiked” intensity in the center of the band, suggesting
that ú ≈ 1, and/or A′ * A′′ and B′ * B′′. The remaining
fundamental in our range, the CH3 rock ν8(e), is more
problematic. On the basis of the solution work,18 which places
ν8 at 1177 cm-1, we should assign the observed band at 1162
cm-1 to this mode. However, we note (Table 2) that the
calculated intensity for this band is exceedingly low, 20-50
times less than for the adjacent ν5, and the position calculated
by both computational methods is 110-80 cm-1 lower, the only
time that the (unscaled) calculations give a lower value than
experiment. Perhaps more convincing is the fact that the
comparable CH3 rock in the similar molecule CH3CN43 is
located at 1041 cm-1. There is a very, very weak band at 1034
cm-1 in the spectrum, which we believe is a more likely
contender for ν8. The lowest bending mode ν10 is not observed
in this work, having been estimated24 from the l-type doubling
constant q obtained in the microwave spectrum22 to be 150(50)
cm-1, although the solution work places this band at 297 cm-1
(see also below).
A comparison of the present gas-phase data with the solution18
and matrix work21 shows good correspondence (apart from ν8),
the main discrepancies arising from the precise location of the
degenerate modes ν6 and ν7, where, as noted above, the position

Gas-Phase Spectroscopy of CH3CNO
J. Phys. Chem. A, Vol. 105, No. 8, 2001 1249
TABLE 3: Experimental Vibrational Frequencies (cm-1) of
the Very Weak Bands in the IR Spectrum of CH3CNO
Assigned to Overtone, Combination, and Other Bands
tentative
assignment
(exptl (gas)
qualifier
a
b
3634
3096
2880 R
2865 P
2743 R
2727 P
2694 R
2685 Q
2682 Q
2679 P
2533
2422 R
2415 Q
2179
ν2 + ν4 (A1)
ν2 + ν5 (A1)
? depends on band center of ν2
? depends on band center of ν7
ν3 + ν7 (E)
}
? but not predicted to have PR structure
ν3 + ν4 (A1)
}
2ν4 (A1)
}
Figure 4. HeI photoelectron spectrum of CH3CNO and schematics of
the corresponding molecular orbitals.
c
2ν8 + ν9 (A1, E) ? depends on band center of ν8
c
? depends on band center of ν8
ν3 + ν8 (E)
}
? but not predicted to have RQ structure
TABLE 4: Experimental and Calculated Ionization
Potentials (eV) of CH3CNO
ν3 + ν5 (A1) or
ν2 - ν10 (E)
predicts ν10 at ca. 132 cm-1
2057 R
2042 P
1788
1162
c
calcd
orbital
character
2ν8 (A1, E)
? depends on band center of ν8
}
exptl IPa
HAM/3b
OVGFb
KTc
?
c
ν8 + ν10 (A1, E) ? depends on band center of ν8
9.92
14.75
16.3
17.0
17.68
9.71 (e)
14.75 (e)
16.26 (e)
16.77 (a1)
17.98 (a1)
23.55 (a1)
9.82
15.32
16.57
16.84
18.02
25.41
10.37
16.62
18.28
18.59
19.82
28.44
πnb(CNO)
πCH3
predicts ν10 at ca. 128 cm-1
910
615
ν5 + ν10 (E)
ν9 + ν10 (A1, E) predicts ν10 at ca. 135 cm-1
predicts ν10 at ca. 125 cm-1
πb(CNO)
σ(C-C)
σ(nO)
a Hot bands on ν2 (see the text) preclude a precise determination of
the band center. b A perpendicular-type band structure precludes a
precise determination of the band center. c Depends on precise location
of the band center for the very, very, weak ν8 band.
σ(CH3)
a Vertical ionization potentials. b The B3-LYP/cc-pVTZ calculated
geometry was used. c Koopmans’ theorem IEs (HF//B3-LYP/cc-pVTZ).
forward. In contrast to the PE spectra of the structurally different
CH3NCO46,47 and CH3OCN2 molecules, that of 1 is consistent
with C3V symmetry. We note that both HAM/3 and OVGF give
good agreement for the IPs, experiment and calculation agreeing
within 0.6 eV. HAM/3 (hydrogenic atoms in molecules), which
is known to give accurate representations of IPs of molecules
containing first-row atoms,48 is in particularly good agreement
(within 0.3 eV). The OVGF calculations give a decided
improvement over the Koopmans’ values (Table 4). The
sequence of molecular orbitals (MOs) deduced for 1 is
of the band centers is uncertain. In Table 2 we have picked the
most intense rQ or pQ subbands as the band identifiers, although
in both modes weaker subbands march to lower and higher
energy. The unscaled harmonic calculations do a reasonable job
for the band positions and intensities, and directed us to
considering an alternative location for ν8. Overall the B3-LYP
method performs better than QCISD, although appropriate
scaling factors could diminish this difference. The CN and NO
stretches, which we have labeled as νas(CNO) and νs(CNO),
respectively, do show large calculated deviations, evidence for
strong coupling in these modes.
...(8a1)2(9a1)2(1e)4(2e)4(3e)4
Table 3 provides a summary of the remaining weak features
in the spectrum, which are assigned to either overtones or
combination bands or are unknown. The point should be made
that these bands seem to track in intensity with the fundamentals,
and so we have some confidence in assigning them to bands of
1; some exhibit a ∆PR of ca. 15-17 cm-1 consistent with the
fundamentals of 1. Several of the combination bands are
assigned in Table 3 with no commentary. Others, which are
perhaps more insecure, have associated comments. These arise,
in the main, because of the uncertainty in the band centers of
ν2 and ν7 and the predilection for placing ν8 at 1034 cm-1 rather
than the more obvious 1162 cm-1 position (vide supra), as well
as the uncertainty in locating its band center. 2ν4 exhibits the
similar multiple Q branches of the fundamental. There is a
certain internal consistency which arises, and indeed the lowest
three combination bands predict that the unobserved lowest ν10
fundamental occurs between 125 and 135 cm-1, a prediction
that is not outrageous given the close agreement with the
calculations and the estimate24 from microwave spectroscopy
but, nevertheless, disagreeing with the solution IR result of 297
Molecular orbital plots are shown as insets in Figure 4. The
observed PE spectrum up to 20 eV corresponds to five ionic
states with the sequence X(2E), A(2E), B(2E), C(2A1), and
D(2A1). The first three bands at 9.92, 14.75, and 16.3 eV can,
in the Koopmans’ approximation, be assigned to the degenerate
nonbonding πnb(CNO), pseudo-π(CH3), and bonding πb(CNO)
molecular orbitals, respectively. The first band at 9.92 eV has
a narrow Franck-Condon profile in agreement with its assign-
ment as a nonbonding orbital in a high-symmetry molecule, and
is lower than those of the isomers CH3NCO46,47 (10.61 (a′′) eV)
and CH3OCN2 (11.30 (a′′) eV). Notwithstanding the breaking
of the degeneracy due to bending in CH3NCO and CH3OCN,
the lower IP of 1 can be rationalized by the following simple
orbital model; the π system of the -CNO, -NCO, or -OCN
pseudohalide group is delocalized over the three atoms. The
bonding orbitals (πb) are delocalized over all three, whereas
the nonbonding orbitals (πnb) have a nodal plane at the central
atom, perpendicular to the frame of the group. Thus, the πnb
orbital is localized on the two outer atoms, which for -NCO
and -OCN are the two most electronegative (O and N), leading
to a higher IP than in the case of -CNO. For comparison, the
first IPs of the hydrogen derivatives HNCO47 and HCNO44 are
11.60 and 10.83 eV, respectively.
cm-1 18
.
Photoelectron Spectroscopy. The PE spectrum of 1 is shown
in Figure 4 with experimental and calculated IPs given in Table
4. From a comparison with the known PE spectra of HCNO,44
BrCNO,7 NCCNO,11 and ONCCNO12 (see also ref 45), as well
as from the calculated IPs, the assignment is relatively straight-
The second band in the PE spectrum at 14.75 eV with a broad
Franck-Condon profile is assigned to the degenerate CH3

1250 J. Phys. Chem. A, Vol. 105, No. 8, 2001
Pasinszki and Westwood
TABLE 5: Calculated Geometries, Total and Relative Energies of CH3(CNO) Isomers, and Connecting Transition Statesa
rel energy
total
energy (au),
rel energy
(kcal/mol)e
CCSD(T)(full)//
moleculeb
(1234)
R24
R14
R342,
d
(kcal/mol),e
c
c
d
no.
typeb R12
R23
R34
R123
Energy Minima
139.6 172.9
115.4 177.8
180.0 180.0
R234 R341
δ1234 B3-LYP/6-31G ** B3-LYP+ZPE B3-LYP+ZPE
3
4
1
5
6
7
CNCO
A
A
A
B
A
B
1.440 1.205 1.182
1.456 1.295 1.166
1.460 1.165 1.218
180.0
180.0
-207.991966
-207.944522
-207.903433
-207.877618
-207.854887
-207.824792
0
0
COCN
CCNO
CC(NO)
CONC
CN(OC)
29.7
54.9
70.4
85.1
102.8
25.1
57.4
64.4
82.0
97.8
1.477 1.262 1.769 1.314 141.8 47.9 45.4 180.0
1.450 1.330 1.183 110.7 174.7 180.0
1.468 1.563 1.291 1.439 109.7 59.7 69.6 105.2
Transition States
8
9
CCNO (5-3)
COCN (4-3)
B
C
B
B
A
B
C
C
1.541 1.284 2.249 1.248 106.5 26.9 27.8 180.0
2.126 1.262 1.201 2.491 63.5 176.3 49.0 180.0
1.448 1.565 1.249 1.704 111.5 73.6 61.8 101.9
1.386 2.138 1.162 2.008 105.2 67.6 80.0 101.5
-207.849783
-207.839871
-207.811224
-207.803761
-207.802820
-207.783494
-207.741824
-207.734376
86.2
92.9
86.9
91.6
10 CONC (6-4)
11 CN(OC) (7-)f
12 CCNO (5-1)
13 CNCO (7-1)
14 CONC (6-1)
15 CCNO (6-5)
111.2
113.6
115.1
128.4
153.5
158.2
106.3
110.6
117.2
128.3
150.9
153.7
1.484 1.276 1.282
120.0 103.4
85.7
1.513 1.272 2.341 1.283 108.4 23.7 23.5 180.0
2.451 1.244 1.208 2.774 75.6 166.9 62.3
1.926 1.253 1.435 1.981 117.2 83.1 105.3
0.0
0.0
a Geometries obtained at the B3LYP/6-31G** level. Bond lengths (R) are in Å, and bond angles (R) and dihedral angles (δ) are in deg. b See
Figure 6 for the generalized shape of the structures and the text for explanations. The numbers in Figure 6 refer to the numbers in the table. c In
f
the case of type B. d In the case of type C. e Energies relative to that of CH3NCO. The 6-31G** basis set was used in the calculations. Decomposition
of 7 into CO and H2CdNH.
pseudo-π orbital localized on the methyl group, which is mixed
to some extent with πb(CNO) (insets of Figure 4). Such CH3 π
orbitals typically occur between 14 and 16 eV in the analogous
CH3NCO and CH3OCN molecules. The third band at 16.3 eV
corresponds to the πb(CNO) MOs and has a substantially broader
Franck-Condon profile than that of πnb(CNO), which reflects
its assignment to the bonding CNO π orbitals. The third and
fourth bands in the spectrum overlap, with the fourth observed
as a shoulder at 17.0 eV. It originates from the highest lying σ
orbital, which is delocalized over the whole CCNO frame,
having a dominant contribution from the C-C bond (see the
insets). The last observed band at 17.68 eV is assigned to the
oxygen terminal lone pair orbital nO(8a1), such an orbital giving
rise to a sharp Franck-Condon profile in the spectra of HCNO44
Figure 5. Calculated (B3-LYP/6-31G**) minima and transition states
(17.79 eV), NCCNO11 (18.75 eV), and BrCNO7 (17.14 eV).
for isomerization processes for the four main open-chain isomers, CH3-
No other bands are observed, in agreement with the calculations
(Table 4) predicting the next IP in the inner valence region
beyond the HeI (21.2 eV) range.
NCO, 3, CH3OCN, 4, CH3CNO, 1, and CH3ONC, 6.
On the basis of the positive inductive effect (+I) of the CH3
group, bands corresponding to the CNO group (πnb(CNO), πb-
(CNO), and nO) are expected to destabilize and shift toward
lower IP compared to those of HCNO. Hence, the πnb(CNO)
orbital, which does not mix with the methyl group pseudo-π
orbital, shifts from 10.83 eV in HCNO to 9.92 eV in 1. The
πb(CNO) MOs, however, shift in the opposite direction (15.92
eV f 16.3 eV), demonstrating a substantial interaction between
the methyl pseudo-π and CNO bonding π orbitals. This
interaction overcompensates for the methyl group +I effect, thus
stabilizing the πb(CNO) orbitals.
Isomerization of 1. Figure 5 shows the calculated (B3-LYP/
6-31G**) minima and linking TSs for isomerization processes
in CH3CNO (1), CH3NCO (3), CH3OCN (4), and CH3ONC (6).
The calculated total and relative energies (including the zero-
point energy (ZPE)) of the minima and TSs are given in Table
5, with the structural labeling scheme presented in Figure 6.
The present DFT calculations concur with the sequence of
stabilities predicted previously using HF/STO-3G25 and find two
other minima, 5 and 7, each being a methyl group attached to
a three-membered ring, through either C (5) or N (7). Structure
5 is somewhat similar to both methyloxazirine and methylcar-
bonylnitrene, which the HF calculations identified as energeti-
Figure 6. Structural configurations and labeling scheme for the
calculated (B3-LYP/6-31G**) structures in Table 5.
cally close isomers with a small barrier between them, 11.0 and
23.8 kcal/mol from the nitrene and oxazirine sides, respectively.
At the current higher level, these two minima have collapsed
into 5, which, from a comparison with typical C-N, C-O, and
N-O single and double bonds,49 has a C-N bond length akin
to that of a double bond, a C-O between a single and double
bond (49% double bond character), and an N-O much longer
(ca. 0.32 Å) than a single bond. Such a structure for ring CH3C-
(NO), 5, is in keeping with analogous structures calculated for
HC(NO)50,51 and NCC(NO).52 The minimum-energy structure
has a methyl CH eclipsing the CN bond, while the other possible
conformer with methyl CH eclipsing the CO bond, noted

Gas-Phase Spectroscopy of CH3CNO
J. Phys. Chem. A, Vol. 105, No. 8, 2001 1251
previously,25 is a TS corresponding to rotation of the CH3 group
around the C-C single bond. The other cyclic minimum, 7, is
energetically high lying, and was previously suggested25 as an
intermediate in the H3CONC f H3CNCO rearrangement. The
present calculations show that 7 easily decomposes to CO and
H2CdNH (via TS 11 and methylnitrene) with a small barrier
of 12.8 kcal/mol (CCSD(T)//B3LYP + ZPE), Table 5). It can
be regarded as a potential intermediate in the decomposition of
1 via the sequence 1 f 13 f 7 f 11 (Figure 5).
The room temperature stability of the methyl pseudohalides
can be discussed in the context of the lowest energy path leading
to isomerization. We apply here a well-known rule of thumb
which states that a chemical reaction can proceed at room
temperature if the barrier to the process is less than 20 kcal/
mol. Molecules 1, 3, and 4 lie in deep wells on the potential
energy surface (Figure 5), with the lowest isomerization barrier
being 59.8, 86.9, and 66.5 kcal/mol, respectively. As far as
intramolecular isomerization is concerned, the calculations
confirm that these molecules are stable, consistent with the
experimental observations of longevity in the dilute gas phase
or in inert solid matrixes. 3 is a liquid at room temperature,1
stable even in the condensed phase, and 4 can be cleanly
generated in solid-gas reactions at room temperature, being
semistable in the dilute gas phase.2,3 1 is observed in this work
to be stable to isomerization in the gas phase and in the
condensed phase at low temperatures.
Figure 7. Calculated (B3-LYP/6-31G**) least energy surface for the
dimerization of CH3CNO to the ring furoxan, and comparison with
the corresponding dimerization process for ClCNO.
barriers for 6 f 1 and 1 f 6 of 68.9 and 93.5 kcal/mol and for
4 f 3 and 3 f 4 of 66.5 and 91.6 kcal/mol. This type of
isomerization can proceed without, e.g., 6 f 4, or with the
intermediacy of a non-pseudohalide intermediate (1 f (5) f
3). The latter is a standard loss process for large nitrile oxides
to isocyanates at elevated temperatures,5 and was proposed as
a potential loss route for photochemically induced RCNO f
RNCO rearrangements in low-temperature matrixes.53 For 1 at
room temperature, however, this is not favored, the alternative
loss route for such small nitrile oxides being intermolecular
dimerization, a much lower energy process.
Molecules 5 and 6 are in much shallower wells, with 5
isomerizing to 3 via TS 8 with a barrier of 22.5 kcal/mol, and
6 to 4 via TS 10 with a barrier of 24.3 kcal/mol, much less
than the barrier of 43 kcal/mol suggested by the HF calculations,
which also predicted an intermediate in a shallow well not found
at the present level of theory. The barriers, however, are high
enough to suggest that spectroscopic investigation of 5 and 6
may be feasible at room temperature or lower, although they
can isomerize at higher temperatures. As far as we know 5 has
not been identified, but 6 has, as a minor component among
the cold (-196 °C) trapped reaction products of the FVP of
3-methyl-4-(O-methyloximino)isoxazol-5(4H)-one,4 with the
suggested intramolecular decomposition process being that
predicted here. Molecule 7, as discussed above, is not stable at
room temperature.
Dimerization of Acetonitrile N-Oxide, 1, to Dimethyl-
furoxan, 2. The short lifetime of 1, and indeed other small nitrile
oxides, has been long known to be due to dimerization to the
furoxan, with presumed low activation barriers. Such dimer-
izations are a special type of [3+2] dipolar cycloaddition, where
both the 1,3- and the 1,2-dipoles are nitrile oxides. We have
calculated this surface for 1 at the B3-LYP/6-31G** level
following a procedure similar to that used for the dimerization
of ClCNO.9 Mechanisms for such 1,3-dipolar cycloaddition
reactions have engendered considerable discussion because a
definitive choice between the two main mechanisms could not
be made (see, for example, ref 54), the issues relating to a
simultaneous one-step process,55,56 or a sequential two-step
mechanism57 in which only one σ bond forms in the first step.
Figure 7 shows the lowest energy reaction path on the
potential surface for this process, comparing it to that for the
ClCNO case. A Firestone-type process is indicated, a two-step
mechanism, where the first step proceeds via TS1 with an energy
barrier of 12.8 kcal/mol, where the two dipoles are in an
extended conformation (NCCN dihedral angle, 180°; C-C bond
length, 1.875 Å) to an intermediate (IM) with a close to planar
(NCCN dihedral angle, 150°) structure and a shorter CC bond
(1.550 Å). This may be identified as the Firestone intermediate.
In the second step the extended intermediate IM closes with
rotation around the established C-C single bond to the final
five-membered furoxan ring 2 after passing over TS2 with an
energy barrier of 7.4 kcal/mol. Since TS2 is higher in energy
than TS1, the second step is rate determining. The overall barrier
of 16.6 kcal/mol at TS2 compared to the energy of two
monomers is relatively low, predicting that dimerization of 1
must be rapid at room temperature, but high enough to prevent
dimerization at lower temperatures. This concurs with our
experimental observation (PE and IR spectra) that 1 can be
The geometries of the TSs (Figure 5, Table 5) clearly
represent the structural changes that the molecules make during
the isomerization processes. The pseudohalide “normal” T “iso”
(viz., cyanate 4 T isocyanate 3, and fulminate 6 T isofulminate
(nitrile oxide) 1) rearrangements proceed via four-membered
planar rings (structures 9 and 14), where the methyl group is
simply transferred from one end of the pseudohalide frame to
the other. The barrier for these processes is high, 66.5 and 68.9
kcal/mol for 4 f 3 and 6 f 1, respectively, the latter having
been previously calculated at the HF/STO-3G level25 (66 kcal/
mol), in good agreement with the present results. The well-
known condensed-phase cyanate f isocyanate rearrangement
at room temperature cannot be explained by intramolecular
isomerization processes, and so bi- or polymolecular reactions
may be involved.
The type of isomerization where the pseudohalide group
bends strongly to form a three-membered ring which consecu-
tively opens up to form another isomer has been discussed
above. We note here that this is a typical process in the normal
T normal (fulminate 6 T cyanate 4) and iso T iso (isofulminate
1 T isocyanate 3) isomerization processes with barriers (6 f
4, 24.3 kcal/mol; 1 f (5) f 3, 59.8 kcal/mol) smaller than
those of the corresponding normal f iso rearrangements with

1252 J. Phys. Chem. A, Vol. 105, No. 8, 2001
Pasinszki and Westwood
(6) Diana, G. D.; Cutcliffe, D.; Volkots, D. L.; Mallamo, J. P.; Bailey,
T. R.; Vescio, N.; Oglesby, R. C.; Nitz, T. J.; Wetzel, J.; Giranda, V.; Pevear,
D. C.; Dutko, F. J. J. Med. Chem. 1993, 36, 3240. Amici, M. De; Micheli,
C. De; Misani, V. Tetrahedron 1990, 46, 1975.
(7) Pasinszki, T.; Westwood, N. P. C. J. Phys. Chem. 1995, 99, 6401.
(8) Gillies, C. W.; Gillies, J. Z.; Lichau, H.; Winnewisser, B. P.;
Winnewisser, M. Chem. Phys. Lett. 1998, 285, 391.
(9) Pasinszki, T.; Westwood, N. P. C. J. Phys. Chem. A 1998, 102,
4939.
(10) Lichau, H.; Winnewisser, B. P.; Winnewisser, M.; Gillies, C. W.;
Gillies, J. Z. To be published.
isolated at low temperatures and revaporized below 0 °C without
significant loss to the dimer. As Figure 7 illustrates, the
dimerization mechanisms of 1 and ClCNO9 are very similar,
following a two-step process, but with significantly different
barriers. That for ClCNO is a mere 4.3 kcal/mol, indicative of
a process that can go to completion very easily, thereby
explaining the spontaneous and rapid dimerization of ClCNO
observed experimentally at room temperature.
(11) Pasinszki, T.; Westwood, N. P. C. J. Phys. Chem. 1996, 100, 16856.
(12) Pasinszki, T.; Westwood, N. P. C. J. Am. Chem. Soc. 1995, 117,
8425.
(13) Guo, B.; Pasinszki, T.; Westwood, N. P. C.; Bernath, P. F. J. Chem.
Phys. 1995, 103, 3335.
(14) Guo, B.; Pasinszki, T.; Westwood, N. P. C.; Zhang, K.-Q.; Bernath,
P. F. J. Chem. Phys. 1996, 105, 4457.
(15) Brupbacher, Th.; Bohn, R. K.; Ja¨ger, W.; Gerry, M. C. L.; Pasinszki,
T.; Westwood, N. P. C. J. Mol. Spectrosc. 1997, 181, 316.
(16) Koput, J. J. Phys. Chem. A 1999, 103, 2170.
Conclusion
In this work we have sought to generate CH3CNO, 1, cleanly
in the gas phase for spectroscopic observation, to compare
structural and spectroscopic data with conventional ab initio
methods and DFT methods, and to describe, computationally,
potential loss processes via either isomerization or dimerization
pathways. The method chosen, thermolytic cycloreversion of
the ring dimer dimethylfuroxan, 2, turns out to be an excellent
approach permitting the acquisition of clean gas-phase photo-
ionization mass, mid-infrared, and HeI photoelectron spectra.
The gas-phase IR spectrum, assigned on the basis of band shapes
and a comparison with computational results, provides a clear
picture of a symmetric-top molecule and should provide a useful
starting point for high-resolution investigation.58 The HeI
photoelectron spectrum was assigned on the basis of C3V
symmetry and is in excellent agreement with both HAM/3 and
OVGF calculations for the ionization energies. The structure
investigated with medium- to large-scale ab initio calculations
demonstrates the continued difficulty for conventional ab initio
calculations due to strong correlation effects. Nonetheless, DFT
can be a cost-effective approach for such molecules, and this
approach was used to investigate differences in the potential
surfaces of 1, CH3NCO, and CH3OCN reflective of their
structural differences, and isomerization processes to chain and
ring species. The conclusion is that 1 has a linear or quasi-
linear heavy atom chain and is stable to unimolecular decom-
position, and as demonstrated by DFT calculations, dimerization
to the stable furoxan is the dominant loss process involving a
two-step mechanism. Nonetheless, loss through dimerization is
shown to be not as significant as in the case of the halosubsti-
tuted nitrile oxide ClCNO and, by extension, BrCNO.
(17) Koput, J. J. Phys. Chem. A 1999, 103, 6017.
(18) Isner, W. G.; Humphrey, G. L. J. Am. Chem. Soc. 1967, 89, 6442.
(19) Zinner, G.; Gu¨nther, H. Angew. Chem., Int. Ed. Engl. 1964, 3, 383.
(20) Mitchell, W. R.; Paton, R. M. Tetrahedron Lett. 1979, 26, 2443.
(21) Mielke, Z.; Hawkins, M.; Andrews, L. J. Phys. Chem. 1989, 93,
558.
(22) Bodenseh, H. K.; Morgenstern, K. Z. Naturforsch. 1970, 25a, 150.
(23) Blackburn, P. B.; Brown, R. D.; Burden, F. R.; Crofts, J. G.; Gillard,
I. R. Chem. Phys. Lett. 1970, 7, 102.
(24) Winnewisser, M.; Pearson, E. F.; Galica, J.; Winnewisser, B. P. J.
Mol. Spectrosc. 1982, 91, 255. Galica, J.; Winnewisser, M.; Winnewisser,
B. P. J. Mol. Struct. 1984, 114, 243.
(25) Poppinger, D.; Radom L. J. Am. Chem. Soc. 1978, 100, 3674.
(26) Wiberg, K. B.; Breneman, C. M. J. Am. Chem. Soc. 1990, 112,
8765.
(27) Koput, J.; Winnewisser, B. P.; Winnewisser, M. Chem. Phys. Lett.
1996, 255, 357.
(28) Scholl, Chem. Ber. 1890, 23, 3499.
(29) Ungnade, H. E.; Kissinger, L. W. Tetrahedron 1963, 19 (Suppl.
1), 143.
(30) Frost, D. C.; S. T. Lee, S. T.; McDowell, C. A.; Westwood, N. P.
C. J. Electron Spectrosc. Relat. Phenom. 1977, 12, 95.
(31) Pasinszki, T.; Westwood, N. P. C. J. Mol. Struct. 1997, 408/409,
161.
(32) Åsbrink, L.; Fridh, C.; Lindholm, E. Chem. Phys. Lett. 1977, 52,
69. The HAM/3 program (PC version) is available from the QCPE, Indiana
University, Bloomington, IN (Chong, D. P. QCMP005, 1985).
(33) Niessen, W. von; Schirmer, J.; Cederbaum, L. S. Comput. Phys.
Rep. 1984, 1, 57.
(34) Frisch, M. J.; Trucks, G. W.; Head-Gordon, M.; Gill, P. M. W.;
Wong, M. W.; Foresman, J. B.; Johnson, B. G.; Schlegel, H. B.; Robb, M.
A.; Replogle, E. S.; Gomperts, R.; Andres, J. L.; Raghavachari, K.; Binkley,
J. S.; Gonzalez, C.; Martin, R. L.; Fox, D. J.; Defrees, D. J.; Baker, J.;
Stewart, J. J. P.; Pople, J. A. Gaussian 92, Revision E.1, Gaussian, Inc.,
Pittsburgh, PA, 1992.
(35) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.;
Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.; Keith, T. A.; Petersson,
G. A.; Montgomery, J. A.; Raghavachari, K.; Al-Laham, M. A.; Zakrzewski,
V. G.; Ortiz, J. V.; Foresman, J. B.; Cioslowski, J.; Stefanov, B. B.;
Nanayakkara, A.; Challacombe, M.; 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. and Pople, J. A.: Gaussian 94, Revision D.4,
Gaussian, Inc., Pittsburgh, PA, 1995.
(36) Flammang, R.; Barbieux-Flammang, M.; Gerbaux, P.; Wentrup,
C.; Wong, M. W. Bull. Soc. Chim. Belg. 1997, 106, 545.
(37) Pople, J. A.; Head-Gordon, M.; Raghavachari, K. Int. J. Quantum
Chem., Quantum Chem. Symp. 1988, 22, 377. Kraka, E.; Gauss, J.; Cremer,
D. J. Mol. Struct.: THEOCHEM 1991, 80, 95.
(38) He, Z.; Cremer, D. Int. J. Quantum Chem., Quantum Chem. Symp.
1991, 25, 43. He, Z.; Cremer, D. Theor. Chim. Acta 1993, 85, 305.
(39) Calculated equilibrium geometry for CH3NCO using B3-LYP/cc-
pVTZ and experimental MW (r0) results (Koput, J. J. Mol. Spectrosc. 1986,
115, 131) in parentheses: CsN ) 1.439 (1.4507) Å, NdC ) 1.197 (1.214)
Å, CdO ) 1.173 (1.166) Å, CsNsC ) 139.9° (135.6°), NsCsO )
173.7° (170.3°).
(40) Calculated equilibrium geometry for CH3OCN using B3-LYP/cc-
pVTZ and experimental MW (rs) results (ref 3) in parentheses: CsO )
1.456 (1.455) Å, OsC ) 1.287 (1.302) Å, CtN ) 1.156 (1.146) Å, Cs
OsC ) 115.6° (113.8°), OsCsN ) 177.9° (178.4°).
(41) Pasinszki, T.; Westwood, N. P. C. To be published.
(42) Teller, E. Hand-und Jahrb. d. Chem. Phys. 1934, 9 (II), 43.
Acknowledgment. We thank the Hungarian Scientific
Research Fund (OTKA Grant F022031) and the Natural
Sciences and Engineering Research Council of Canada (NSERC)
for research and equipment grants in support of this work. T.P.
thanks the Hungarian Academy of Sciences for the award of a
Janos Bolyai research scholarship.
References and Notes
(1) Kricheldorf, R.; Leppert, E. Synthesis 1976, 329.
(2) Pasinszki, T.; Westwood, N. P. C. J. Phys. Chem. 1995, 99, 1649.
(3) Sakaizumi, T.; Mure, H.; Ohashi, O.; Yamaguchi, I. J. Mol.
Spectrosc. 1990, 140, 62. Sakaizumi, T.; Sekishita, K.; Furuya, K.; Tetsuda,
Y.; Kaneko, K.; Ohashi, O.; Yamaguchi, I. J. Mol. Spectrosc. 1993, 161,
114. Sakaizumi, T.; Namikawa, M.; Ohashi, O. J. Mol. Struct. 1995, 345,
189.
(4) Wentrup, C.; Gerecht, B.; Laqua, D.; Briehl, H.; Winter, H.-W.;
Reisenauer, H. P.; Winnewisser, M. J. Org. Chem. 1981, 46, 1046. The
choice of nomenclature for compounds of the type RONC is variable. As
noted in this reference, and as used in the present paper, we use the IUPAC
name “fulminate” for RONC (NOC rule C-833.1; GNOC recommendation
R-5.9.7.2).
(5) Grundmann, Ch.; Grunanger, P. The Nitrile Oxides, Versatile Tools
of Theoretical and PreparatiVe Chemistry; Springer-Verlag: New York,
Heidelberg, Berlin, 1971. Torssell, K. B. G. In Nitrile Oxides, Nitrones,
and Nitronates in Organic Synthesis; Feuer, H., Ed.; Organic Nitro
Chemistry Series; VCH Publishers Inc.: New York, 1988.

Gas-Phase Spectroscopy of CH3CNO
J. Phys. Chem. A, Vol. 105, No. 8, 2001 1253
(43) Parker, F. W.; Nielsen, A. H.; Fletcher, J. Mol. Spectrosc. 1957, 1,
107. Nakagawa, I.; Shimanouchi, T. Spectrochim. Acta, A 1962, 18, 513.
(44) Bastide, J.; Maier, J. P. Chem. Phys. 1976, 12, 177.
(45) Pasinszki, T.; Westwood, N. P. C. J. Electron Spectrosc. Relat.
Phenom. 1998, 97, 15.
(46) Cradock, S.; Ebsworth, E. A. V.; Murdoch, J. D. J. Chem. Soc.,
Faraday Trans. 2 1972, 68, 86. Pasinszki, T.; Yamakado,; H.; Ohno, K. J.
Phys. Chem. 1993, 97, 12718.
(47) Eland, J. D. H. Philos. Trans. R. Soc. London, A 1970, 268, 87.
(48) Chong, D. P. Theor. Chim. Acta 1979, 51, 55.
(49) Calculated CN, CO, and NO bond lengths in H2CdNH, H3Cs
NH2, H2CdO, H3CsOH, and H2NsOH at the B3LYP/6-31G** level are
1.270, 1.464, 1.207, 1.418, and 1.448 Å, respectively.
(50) Mebel, A. M.; Luna, A.; Lin, M. C.; Morokuma, K. J. Chem. Phys.
1996, 105, 6439.
(52) Fehe´r, M.; Pasinszki, T.; Veszpre´mi, T. Inorg. Chem. 1995, 34,
945.
(53) Maier, G.; Teles, J. H. Angew. Chem., Int. Ed. Engl. 1987, 26,
155.
(54) Houk, K. N.; Yamaguchi, K. In 1,3-Dipolar Cycloaddition Chem-
istry; Padwa, A., Ed.; Wiley-Interscience: New York, 1984; Vol. 2.
(55) Huisgen, R. Angew. Chem., Int. Ed. Engl. 1963, 2, 565; J. Org.
Chem. 1968, 33, 2291; 1976, 41, 403.
(56) Huisgen, R. In 1,3-Dipolar Cycloaddition Chemistry; Padwa, A.,
Ed.; Wiley-Interscience: New York, 1984; Vol. 1.
(57) Firestone, R. A. J. Org. Chem. 1968, 33, 2285; 1972, 37, 2181; J.
Chem. Soc. A 1970, 1570; Tetrahedron 1977, 33, 3009.
(58) B. P. Winnewisser and M. Winnewisser have indicated (private
communication to N.P.C.W.) that they have some unpublished high-
resolution IR data on 1.
(51) Jursic, B. S.. J. Phys. Chem. A 1999, 103, 1880.