A combined neutralization-reionization mass spectrometric and theoretical study of oxyallyl and other elusive [C3,H4,O] neutrals

FULL PAPER

A Combined Neutralization-Reionization Mass Spectrometric and Theoretical

Ƞ

Study of Oxyallyl and Other Elusive [C ,H ,O] Neutrals

3

4

a[°]

b

a[°°]

a

Christoph A. Schalley , Stephen Blanksby , Jeremy N. Harvey

, Detlef Schröder* , Waltraud Zummack ,

b

a

John H. Bowie , and Helmut Schwarz*

Institut für Organische Chemie der Technischen Universität^a,

Straße des 17. Juni 135, D-10623 Berlin, Germany

Department of Chemistry, University of Adelaide^b,

Adelaide, SA, 5006, Australia

Received January 30, 1998

Keywords: Neutralization-reionization mass spectrometry / Density functional theory / Oxyallyl / Ketocarbenes /

Methoxy vinylidene

Ϫ

,O]^•isomers, i.e. the radical

Five different anionic [C

3

,H

4

level of theory. For some species, notably the singlet state of

oxyallyl, the theoretical treatment is complemented by G2,

CASSCF, and MR-CI calculations. Theory and experiment

are in good agreement in that at the neutral stage (i) acrolein

does not react within the µsec timescale, (ii) acetyl and formyl

anions of acrolein, acetyl carbene, formyl methyl carbene,

methoxy vinylidene, and oxyallyl are generated in an ion

beam mass spectrometer and subjected to neutralization-

reionization (NR) mass spectrometric experiments including

neutral and ion decomposition difference (NIDD) mass spec- methyl carbenes isomerize to methyl ketene, (iii) methoxy

1

trometry; the latter allows for the examination of the neutralsЈ

vinylidene rearranges to methoxy acetylene, (iv) singlet A

1

unimolecular reactivity. Further, the anionic, the singlet and oxyallyl undergoes ring closure to cyclopropanone, and (v)

,O]^•

Ϫ/0/•+

potential-

3

triplet neutral, and the cationic [C

3

,H

4

2

triplet B oxyallyl may have a lifetime sufficient to survive a

energy surfaces are calculated at the B3LYP/6-311++G(d,p)

NR experiment.

Introduction

Chart 1

For isomeric [C ,H ,O] molecules, several unconventional

3

4

connectivities are conceivable due to the low degree of satu-

ration of the heavy-atom backbone, and the reader is in-

vited to complement the selection of structures depicted in

Chart 1. Besides more common isomers like acrolein 1,

methyl ketene 2, methoxy acetylene 3, or propargyl alcohol

4

, also hydroxy allene 5 or 1-hydroxy propyne 6 may be

[

1]

stable species . Experimental and theoretical evidence have

been provided for the existence of strained rings like allene

[

2]

[2a][2g][3]

[2g][4]

oxide 7 , cyclopropanone 8

, methyl oxirene 9

,

[

5]

[6]

and oxetene 10 . The ketocarbenes 11 and 12 were dis-

cussed with regard to the Wolff rearrangement of diazoke-

[7]

tones and the intermediacy of oxirenes in these reactions .

Further, unsaturated carbenes^[

idene

8]

like methoxy vinyl-

[

9][10]

13 may represent local minima on the [C ,H ,O]

3

4

potential-energy surface.

One of the most intriguing [C ,H ,O] species is, however,

3

4

the oxyallyl biradical 14. This biradical and its substituted

homologs have been proposed as intermediates or tran- Chart 2

sition structures in (i) the Favorskii rearrangement [Scheme

[

11]

1

, path (a)] , (ii) the disrotatory, degenerate isomerization

[12]

of cyclopropanone 8 [path (b)]

and the ring inversion in

bicyclo[n.1.0]alkanones^[, (iii) the isomerization of 7 to 8

13]

[

°]

Present address: The Skaggs Institute for Chemical Biology,

Scripps Research Institute, 10550 North Torrey Pines Road, [paths (b) and (c)]^[¹⁴^], (iv) the addition of diazomethane to

La Jolla, California, 92037, USA.

[15]

ketenes yielding cyclopropanones , and (v) in the photo-

[

°°]

Present address: Department of Physical Chemistry and the

Fritz Haber Research Center, The Hebrew University of Jeru-

salem, Jerusalem 91904, Israel.

rearrangements of 2,5-cyclohexadienones (Scheme 2)^[. A

16]

recent synthetic method even provides an experimental

Eur. J. Org. Chem. 1998, 987Ϫ1009



WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1998

1434Ϫ193X/98/0606Ϫ0987 $ 17.50ϩ.50/0

987

D. Schröder, H. Schwarz et al.

FULL PAPER

[4]

route to bicyclo[n,1,0]alkanones allowing for the study of formed upon dissociative ionization of diazoacetone .

the ring opening to the corresponding oxyallyl deriva- Similarly, several attempts to generate the oxyallyl radical-

[

17]

•؉

[2d][2e]

tives . The parent molecule 14, however, has not been iso- cation 14 failed

. However, NR experiments pro-

[

2d][2e][4]

lated yet. The analogy to isoelectronic trimethylenemethane vided evidence for the existence of neutral 7 and 9

5 (TMM)

Oxyallyl represents a non-Kekule molecule , which like isomers by NR mass spectrometry, we decided to use

TMM is predicted to have a triplet B ground state

.

[

18]

1

as well as 16 and 17 (Chart 2) is obvious.

To get access to neutral oxyallyl, acetyl carbene, and other

´

[19]

3

[20][21]

•Ϫ

[C ,H ,O] radical-anions rather than the corresponding

3 4

2

and is believed to be trapped chemically in cycloaddition cations as precursors.

[

22]

reactions [Scheme 1, path (d)] . The lowest singlet state

It is common knowledge that many positively charged

1

(

A ) of 14 was calculated to be a local minimum which is ions give rise to extensive isomerizations, while most anions

1

3

[20][21a]

only 1Ϫ2 kcal/mol above the B ground state

and do not. Further, most carbenes have positive electron affin-

2

[27]

was described as a genuine biradical (14) rather than a zwit- ities and do not rearrange as radical anions . For ex-

´

[

20][21a]

[28]

terion (14Ј)

. At the CASSCF(4,4)/6-31G(d) level of ample, the distonic

radical anions of the non-Kekule

theory, the transition structure for symmetry allowed dis- molecules TMM and tetramethyleneethane have been stud-

rotatory ring closure to cyclopropanone was found to be as ied^[. Similarly, the CH CO distonic radical-anion 16

29]

•

Ϫ

•؊

2

low as 0.3 kcal/mol^[relative to A oxyallyl. It has thus was examined by a variety of mass spectrometric tech-

23]

1

been concluded that isolation and identification of singlet niques^[, which also provided evidence that upon neu-

30]

oxyallyl by experimental means should be at least extremely tralization, 16 is formed in both the singlet and triplet

difficult, if not impossible.

states. Accordingly, negative ions appear as a promising ap-

proach to the isoelectronic oxyallyl system, rather than any

route using radical cations as precursors.

Scheme 1

In this paper, experimental data for the gas-phase exist-

ence of five neutral [C ,H ,O] isomers are provided, i.e. ac-

3

4

rolein 1, acetyl carbene 11, formyl methyl carbene 12, meth-

oxy vinylidene 13, and oxyallyl 14. The interpretation of the

experiments is aided by extensive B3LYP/6-311ϩϩG(d,p)

calculations on the anionic, the singlet and triplet neutral,

•

Ϫ/0/•ϩ

and the cationic [C ,H ,O]

potential-energy surfaces.

3

4

Due to some fundamental shortcomings of this theoretical

approach, additional calculations at higher levels of theory

are included. The results shed new light on the questions

concerning the existence, the singlet-triplet gap, and the

probability of spin inversion for singlet and triplet oxyallyl.

Scheme 2

Computational Details

•

Ϫ/0/•ϩ

In order to get an overview of the [C ,H ,O]

poten-

3

4

tial-energy surfaces, the economic hybrid DFT/HF B3LYP

One approach to experimentally generate and structur- method^[

31]

implemented in Gaussian 94

[32]

with 6-

ally characterize elusive and unconventional molecules uses 311ϩϩG(d,p) basis sets^[was applied first. Geometry op-

33]

the large potential of neutralization-reionization (NR) mass timizations were performed with gradient procedures, and

[

24]

spectrometry . In these experiments, a beam of ionic pre- vibrational frequencies were calculated in order to deter-

cursor molecules is first neutralized in a high-energy colli- mine the nature of stationary points as minima (no imagin-

sion and subsequently the fast moving neutrals are reion- ary frequency), transition structures (one imaginary fre-

ized in a second collision event. The experimental set-up is quency), or higher order saddle points. In certain cases, in-

typically such that the lifetimes of the neutrals between the trinsic reaction coordinates (IRCs)^[were calculated with

two collisions are in the range of microseconds. A recently B3LYP/6-31G(d) in order to relate minima and transition

developed method coined “neutral and ion decomposition structures; for the sake of brevity, we restrict this topic to

34]

difference” (NIDD) mass spectrometry^[allows to investi- the loss of CO from cyclopropanone as an example. Un-

25]

gate the unimolecular decay of the transient neutrals scaled^[

35]

frequencies were used to derive zero-point vi-

[

26]

formed . NIDD is based on the comparison of NR mass brational energy (ZPVE) corrections. Selected geometrical

spectra with charge reversal (CR) spectra, in which the data for calculated species are indicated in the Charts; a

charge of the projectile is inverted by two-electron transfer complete set of cartesian coordinates for each of the species

in a single collision.

is available upon request from the authors.

The NRMS approach to neutral molecules depends on

In general, the B3LYP approach is expected to provide

the availability of appropriate, well characterized precursor an accuracy of ca. ±5 kcal/mol for relative energies^[, and

36]

ions. According to earlier results, acetylcarbene radical-cat- even radical anions can be described quite well, provided

ion 11^•, for example, does not exist as a minimum on the the chosen basis set is sufficiently large . As will be shown

؉

[37]

[

4]

•؉

cation surface . Instead, ionized methyl oxirene 9

is by comparison of the theoretical results with known exper-

988

Eur. J. Org. Chem. 1998, 987Ϫ1009

3 4

Oxyallyl and Other [C ,H ,O] Neutrals

FULL PAPER

imental data for several key species (see below), the level of reference averaged coupled-pair functional (MR-ACPF)

theory and the corresponding basis set chosen are appropri- calculations, and multi-reference perturbational theory. For

ate for a qualitative, if not semiquantitative description of the multi-reference methods, the active space chosen was

these systems; note however, that the electron affinity of the (6,5) combination described above. The MR-ACPF cal-

[38]

CH is overestimated by ca. 12 kcal/mol . For several culations were repeated with the larger correlation-consitent

2

stationary points, the energetics were cross-checked by per- polarized valence triple zeta (cc-pVTZ) basis set, aug-

forming calculations with the highly reliable G2 method^[

39]

,

mented on oxygen, and omitting the d polarization func-

which is expected to give the energetics within an error of tions on hydrogen. (iv) Since the singlet and triplet states

2 kcal/mol. The G2 method involves geometry optimi- of oxyallyl are essentially degenerate at their minima, spin-

±

zation at the MP2/6-31G(d) level, and in all cases described inversion mediated surface-hopping could be expected to

further below these geometries were very similar to the occur. In order to estimate the probability of the intersys-

SO

B3LYP ones; accordingly, only the latter are reported.

tem crossing, the spin-orbit coupling matrix element (H )

The calculated relative energies were converted into heats between the two states was calculated. For this purpose, a

of formation by anchoring them to known experimental one-electron approximate spin-orbit Hamiltonian^[

45]

was

[

40]

data

while keeping the deviations as small as possible. In used, as implemented in the Gamess for PC program pack-

general, a reasonable agreement of experiment and theory age^[. The wavefunction of the singlet state was developed

46]

was found. Deviations larger than ±5 kcal/mol were, how- with the 6-31G(d) basis set as

a fully optimized

ever, obtained for the cumulenes such as methyl ketene or CASSCF(6,5) wavefunction. The triplet wavefunction was a

allene and their radical-cations (see below). These values CASSCF(6,5) wavefunction, where the orbitals of the active

were, therefore, not taken into account in the anchoring of space were fully optimized, but the inner electrons were de-

the heats of formation. The consistency of the present cal- scribed in a frozen-core approximation by the correspond-

culations with earlier computational results obtained using ing singlet-optimized orbitals. Tests using other types of

[

41][42]

MP2 and G2

formation of these cumulenes may need some refinement.

B3LYP was chosen, because it represents an economic charges were taken from the literature^[⁴⁵^a^]

and yet sufficiently accurate approach for most questions Surface hopping may also be important for the singlet

suggests that the experimental heats of wavefunctions to describe the two states lead to almost

SO

identical results with respect to H . The effective nuclear

.

connected to this project. However, it is essentially a one- and triplet states of the acetyl and methyl formyl carbenes.

determinant procedure and additionally contains an “ex- In this case, however, the minimum geometries are substan-

act” Hartree-Fock (HF) exchange contribution. In fact, tially different, which means that it is not immediately obvi-

B3LYP turns out to be inappropriate for a quantitatively ous that the system can reach a crossing region. It is there-

1

correct description of the biradicaloid, open-shell A state fore necessary to explicitly locate the minimum energy

1

of oxyallyl (see below)^[

20b][21a]

. Therefore, additional calcu- crossing point (MECP) between the two surfaces. This was

lations were performed using four approaches: (i) The BP86 done for the acetyl carbene system at the B3LYP level of

and BLYP density functionals implemented in Gaussian theory; the MECP in the formyl methyl carbene system is

were explored because these do not contain an HF contri- assumed to have similar properties. The method^[used to

bution. (ii) Complete active space self-consistent field locate the MECP involves a geometry optimization with

47]

(CASSCF) calculations were performed for the singlet state

BECKE3LYP/6-311ϩϩG(d,p) following a composite energy

using several combinations of basis sets and active spaces. gradient derived at each step from the energies and gradi-

In addition to calculations without active electrons (RHF ents of the singlet and triplet surfaces. This gradient is the

approach), the following combinations were considered: sum of one term which points towards the crossing hyper-

CASSCF(2,2) with the two unpaired electrons in the two line between the two surfaces, and one which points

valence orbitals of a₂and b₁symmetry, respectively, towards the minimum along the hyperline. The matrix ele-

CASSCF(4,4) with the four valence π electrons in the four ment H^S^Obetween the two states was calculated at the

valence π orbitals (in order, of b , b , a , and b symmetry), MECP using the same method as above. In this case, the

1

2

1

CASSCF(6,5) where the in-plane oxygen lone-pair (b sym- active space included four electrons in four orbitals (the two

2

*

metry) is also included in the active space, and orbitals at the carbene center and the π and π orbitals of

CASSCF(6,7) in which two virtual orbitals (a and b sym- the carbonyl group). Again, various other methods were

1

2

SO

metry) were added. This set of calculations was destined to used to determine H , all giving very similar results. Due

determine whether or not the singlet state is a minimum. to the C symmetry of this MECP, all three substates of the

1

(

iii) The relative energies of the singlet and triplet states triplet state can couple to the singlet state, and the reported

SO

at their respective CASSCF(6,5) geometries were examined

using several correlated methods. These calculations were

H

is the root-mean square of the coupling elements.

performed in Molpro94^[using the correlation-consistent

43]

Theoretical Results

polarized valence double zeta (cc-pVDZ) basis sets of Dun-

First, we shall consider the theoretical findings concern-

[

44]

ning et al.

For oxygen, the corresponding augmented ing the anionic, the singlet and triplet neutral, and the cat-

•

Ϫ/0/•ϩ

basis set including diffuse functions was used. The corre- ionic [C ,H ,O]

potential-energy surfaces. These re-

3

4

lation methods considered included coupled cluster, MR-CI sults will define the scope and serve as guidelines for the

with and without a correction for higher excitations, multi- experiments. At the outset, let us, however, make a general

Eur. J. Org. Chem. 1998, 987Ϫ1009

989

D. Schröder, H. Schwarz et al.

FULL PAPER

The C₂_vsymmetric oxyallyl radical-anion 14^•corre-

؊

comment on the anionic species. In addition to the re-

arrangements, anions may undergo vertical electron detach- sponds to a deep local minimum which is separated from

•

؊

•؊

ment and thus escape detection. This may happen, if the the isomeric structures 8 and 11 by large barriers. Frag-

•

؊

•Ϫ

relevant electron affinity (EA) is low enough and/or the in- mentation of 14 into H CCCH ϩ O or H CCO ϩ

2

•

Ϫ

ternal energy of the anion is sufficiently large. Further, even CH₂is quite energy demanding (>70 kcal/mol) and thus

anions which are stable against vertical electron detachment unlikely to occur. A 1,3-hydrogen migration via TS11

may correlate with neutrals of the same connectivity, but 14 leads to the acetyl carbene radical-anion 11 , but the

different geometry. Thus, some anions may exhibit negative barrier is more than 50 kcal/mol above 14 . In addition to

adiabatic EAs and may undergo autodetachment within the the fragmentation into propyne and O which is endo-

µs timescale of the mass spectrometric experiments. There- thermic by ca. 60 kcal/mol relative to 11 , loss of methyl

fore, related neutrals have to be taken into account in evalu- radical concomitant with formation of HCCO provides a

•

Ϫ

/

•

؊

•؊

•

؊

•

Ϫ

•

؊

Ϫ

ating the stability of anions^[

37c]

.

low-lying exit channel for 11 with an energy demand of

•؊

2

•Ϫ

Figure 1. Relevant parts of the anionic [C

3 4

,H ,O] potential-energy surface calculated at the B3LYP/6-311ϩϩG(d,p) level of theory.

Vertical arrows indicate adiabatic transitions to the corresponding neutrals of lowest heat of formation

2

,O]^•^؊Potential-Energy Surface

The Anionic [C

3

,H

4

only 7.7 kcal/mol above the minimum. The latter result is

consistent with the observation of methyl loss under ICR

2

•Ϫ

Several [C ,H ,O] isomers and their mutual transition

structures (Chart 3) were found at the B3LYP/6-

11ϩϩG(d,p) level of theory. The energetics (Table 1) and

the resulting potential-energy surface also include some

fragmentation channels as well as relevant neutral species.

The arrows connecting anions and neutrals in Figure 1 indi-

cate the adiabatic EA and, thus, provide information with

regard to the stability of the anions against electron detach-

ment.

3

4

conditions^[. Ring closure of 14 to the cyclopropanone

50]

•؊

•

؊

•Ϫ

•؊

radical-anion 8 via TS8 /14 has an energy demand of

3

•

؊

more than 40 kcal/mol relative to 14 and proceeds in a

^s^y^m^m^e^t^r^y^a^l^l^o^w^e^d^,^c^oⁿ^r^o^t^a^t^o^r^y^m^o^v^e^m^eⁿ^t^o^f^t^h^e^C^H2

groups. The disrotatory process is symmetry forbidden, and

all attempts to locate a transition structure for the disrota-

•

؊

•؊

tory ring closure of 14 failed. In C symmetric 8 the

s

oxygen is not located in the CCC plane due to the radical

2

•Ϫ

center at C(1). The negative adiabatic electron affinity of

Eight local [C ,H ,O] minima were found (Figure 1)

with three, however, being less stable than the correspond-

ing neutrals. While electron detachment from the radical

3

4

1

neutral cyclopropanone 8 implies, however, that ring clos-

•

؊

ure of 14 is most likely associated with electron detach-

•

؊

•

؊

•؊

ment instead of generating long-lived 8

.

anions of acrolein 1 , acetyl carbene 11 , methyl formyl

carbene 12^•, methoxy vinylidene 13 , and oxyallyl 14

؊

•؊

Among the isomers considered in this study, the radical

•

؊

is predicted to have considerable energy demands, electron anion of acrolein 1 corresponds to the deepest minimum.

•

؊

•؊

losses from the radical anions of methyl ketene 2 , meth-

1

is located ca. 10 kcal/mol below 14 and is connected

oxy acetylene 3^•, and cyclopropanone 8 are exothermic, to 8 via TS1 /8 which is probably a spurious tran-

؊

•؊

•Ϫ

•Ϫ •؊

though probably associated with kinetic barriers due to the sition structure in the sense discussed above, because neither

necessity of geometrical reorganization.

1 nor 8 have electron affinities large enough to allow the

990

Eur. J. Org. Chem. 1998, 987Ϫ1009

Oxyallyl and Other [C

Chart 3

3 4

,H ,O] Neutrals

FULL PAPER

Table 2. Calculated total energies (E

t

), zero-point vibrational ener-

gies (ZPVE), and experimental heats of formation (∆H

f

) of frag-

ments used in this study

[

a]

[b]

[c]

E

t

ZPVE

(hartree)

∆H

f

∆H

f

(

hartree)

(kcal/mol)

•

H

Ϫ0.502257

Ϫ

52.1

59.6

3

O

Ϫ75.089880

1

[d]

O

104.9

25.9

•

Ϫ

O

Ϫ75.149014

Ϫ39.130248

Ϫ39.148992

Ϫ39.175824

Ϫ39.825553

Ϫ76.616825

Ϫ76.731381

Ϫ77.329674

Ϫ76.916019

Ϫ77.264329

Ϫ76.850938

Ϫ77.892692

Ϫ78.564764

Ϫ78.185446

Ϫ78.448053

Ϫ113.344010 0.005041

Ϫ113.878377 0.012954

Ϫ113.567343 0.014473

Ϫ114.515350 0.026498

Ϫ114.117602 0.023838

Ϫ115.056175 0.036069

Ϫ

22.5

1

[e]

CH

2

0.016496

0.017125

0.015329

0.029635

0.014122

0.014512

0.026988

0.026085

0.023530

0.021604

0.036308

0.050773

0.048288

0.046475

104.8

103.5

3

2

93.0

88.4

•

Ϫ

CH

2

76.2

•

CH

3

34.8

•

HCC

134.2

65.7

Ϫ

62.3

HCCH

54.5

•

ϩ

314.1

104.0

363.4

317.4

H

2

CC:

•

ϩ

2

CC:

•

C

2

H

3

63.4

12.5

254.9

H

2

3

CCH

2

•

ϩ

2

250.5

85.7

CϪCH

CO

Ϫ26.4

10.7

207.9

Ϫ26.0

224.7

3.7

Ϫ32.5

42.4

Ϫ12.1

262.0

•

HCO

CH

HCCO

CH

ϩ

205.8

223.6

2

3

O

•

ϩ

O

Ϫ

Ϫ115.112312 0.034502 Ϫ31.5

Ϫ151.954398 0.018388

Ϫ152.040028 0.018373 Ϫ11.3

•

Ϫ

ϩ [f]

Ϫ151.587535 0.018300

Ϫ191.333376 0.046970

Ϫ152.620089 0.031510

Ϫ152.265089 0.031177

Ϫ116.637358 0.055491

Ϫ116.268204 0.050947

Ϫ116.640560 0.054844

Ϫ116.292902 0.052349

272.6

Ϫ

3

2

3

2

CCO

CO

Ϫ11.4

210.2

44.6

283.5

45.6

269.0

•

ϩ

211.4

276.2

261.7

CCH

•

ϩ

2

•

ϩ

Table 1. Calculated total energies (E

t

), zero-point vibrational ener-

CH CHCO Ϫ191.012770 0.050586

2

gies (ZPVE), and calculated as well as experimental heats of forma-

•

Ϫ

tion (∆H

f

) of anionic [C

3

,H

tion reactions

4

,O] isomers and relevant fragmenta-

[

a]

[a]

Total energies based on B3LYP/6-311ϩϩG(d,p) optimized struc-

[b]

tures including ZPVE corrections (unscaled). Ϫ

Reference data

[40]

[c]

as given in ref. . Ϫ Heats of formation of ions and excited state

species calculated from the difference in B3LYP/6-311ϩϩG(d,p)

total energies of these and the corresponding ground-state neutrals.

Ϫ ^[Since B3LYP calculations are problematic with the excited

singlet state of oxygen atoms, the excitation energy of 1.967 eV (see

E ^[

b]

ZPVE

∆H

[c]

∆H

f

[d]

t

f

(

hartree)

(hartree) (kcal/mol) (kcal/mol)

d]

CHCHO ₍₁•؊

•

؊

CH

2

3

)

Ϫ191.928179 0.056847 Ϫ21.0

Ϫ191.913221 0.057328 Ϫ11.6

[48]

_C_C_O•؊ ₍₂•؊

ref. ) relative to the triplet ground state has been added to the

[e]

)

_O_C_C_H• (3

؊

•؊

calculated energy of the triplet species. Ϫ This value is derived

)

Ϫ191.821696 0.056849

45.8

15.2

10.6

12.6

48.0

CO (8•؊₎Ϫ191.870588 0.057338

•

؊

from MR-CI calculations on the singlet-triplet gap of methylene

2

c-CH CH

2

[49a]

[f]

ϩ

•

؊

•؊

(ref.

). Ϫ HCCO is predicted to exhibit a triplet ground state

CH

3

^C^O^C^H• (11

)

Ϫ191.877830 0.056658

Ϫ191.874634 0.056760

Ϫ191.818321 0.057669

؊

with the singlet being ca. 30 kcal/mol higher in energy.

CCHO (12

؊

•؊

CH

OCHC:^•(13

•

Ϫ

•؊

_<₃₂[e]

CH_•

2

COCH

2

(14 ) Ϫ191.911810 0.055994 Ϫ10.7

rearrangement to compete with electron loss. Similar argu-

؊

•؊

TS1 /8

Ϫ191.831112 0.051170

Ϫ191.825790 0.051601

Ϫ191.847600 0.053444

Ϫ191.828888 0.050643

Ϫ191.865581 0.048008

Ϫ191.835633 0.046970

Ϫ191.789574 0.054844

Ϫ191.786372 0.055491

Ϫ191.795913 0.046839

Ϫ191.787556 0.050581

Ϫ191.729137 0.048624

39.9

43.2

29.6

41.3

18.3

37.1

66.0

68.0

69.8

67.3

103.9

ments apply to the 1,2-hydrogen migration via TS1^•/12

Ϫ

•؊

•

؊

•؊

TS1 /12

TS8^•/14

؊

in that electron detachment is predicted to precede the for-

TS11^•/14^•^؊

؊

•؊

mation of formyl methyl carbene anion 12 which is ca.

Ϫ

•

HCCO ϩ CH

3

22.7

•

؊

•؊

Ϫ

•

34 kcal/mol less stable than 1 . In analogy to 11 , the

CH

3

2

3

2

CCO ^ϩ^H•

Ϫ

•؊

Ϫ

CCH

2

ϩ O

71.5

70.5

77.0

69.8

101.7

carbene 12 can decompose to CH CCO anions and a

3

•

Ϫ

CCH ϩ O

CO ϩ CH

•؊

hydrogen atom. Although 13 represents the isomer of

•

Ϫ

2

Ϫ

•

HCC_•ϩ CH

3

O

highest energy studied here, the methoxy vinylidene radical-

Ϫ

•؊

HCC ϩ CH

3

O

anion 13 is predicted to be a stable species in the gas

phase^[, as the EA of the corresponding neutral exceeds

51]

[

a]

_T_h_e_t_o_t_a_l_e_n_e_r_g_i_e_s_o_f_t_h_e_f_r_a_g_m_e_n_t_s_a_r_e_g_i_v_e_n_i_n_T_a_b_l_e₂_._Ϫ[b]

Total energies based on B3LYP/6-311ϩϩG(d,p) optimized struc-

3

0 kcal/mol. Below the threshold for electron detachment

[c]

•؊

Ϫ

tures including ZPVE corrections (unscaled). Ϫ Calculated heats from 13 , C-O bond cleavage can proceed to yield HCC

of formation on the basis of literature data (see Computational

•

and CH O , also providing a potential route for the iso-

merization to 3 via an ion-dipole complex of the type

[HCC/CH O] . Passage from the methoxy vinylidene/

3

[

d]

[40]

[e]

Details). Ϫ Reference data as given in Table 2 and ref. . Ϫ

•

؊

_]_a_c_e_t_o_n_e_w_i_t_h_O_•Ϫ

3

Upper limit calculated from the reaction of [D

as described in ref.^[

50]

.

•Ϫ

3

Eur. J. Org. Chem. 1998, 987Ϫ1009

991

D. Schröder, H. Schwarz et al.

FULL PAPER

2

•Ϫ

methoxy acetylene part of the [C ,H ,O] potential-en- Chart 4

3

4

ergy surface to the other isomers requires extensive re-

arrangements, and we did not search for reaction pathways

connecting both parts, but assume that isomerizations other

than that of 13^•to 3 do not occur.

؊

•؊

From these calculations two major conclusions can be

drawn. (i) It should be possible to generate at least five dif-

2

•Ϫ

•؊

ferent [C ,H ,O] isomers, i.e. 1 , 11 , 12 , 13 , and

1

3

4

•

؊

4 , provided that appropriate precursors can be found.

This set of anions therefore defines the structural scope of

this study. (ii) Three of these isomers have fragmentation

channels below the threshold for electron detachment and

•

؊

are expected to undergo unimolecular dissociation. 11

[50]

•؊

can undergo facile loss of methyl radical , 12 can gener-

Ϫ

ate CH CCO anions by expulsion of a hydrogen atom,

3

and for 13^•the occurrence of CϪO bond cleavage is ex-

؊

Ϫ

pected to yield HCC anions concomitant with neutral

methoxy radicals. Instead, for the other two isomers, i.e.

•

؊

•؊

1

and 14 , no significant metastable ion decompositions

are expected to be observed experimentally.

1

The Neutral Singlet [C

3 4

,H ,O] Potential-Energy Surface

Owing to the fact that the singlet-triplet gaps of some

neutrals, e.g. oxyallyl^[

20][21a]

and the carbenes , are small,

[52]

singlet and triplet surfaces need to be considered. Before

discussing some particular questions concerning neutral ox-

yallyl in more detail, let us first provide an overview of the

1

3

[

C ,H ,O] and [C ,H ,O] potential-energy surfaces start-

3 4 3 4

ing with the singlet isomers.

1

8 stationary points on the [C ,H ,O] surface were opti-

3 4

mized (Chart 4). The minima of lowest energy (Figure 2)

1

are acrolein 1 (∆H ϭ Ϫ11.6) and methyl ketene 2 (∆H ϭ

f

Ϫ12.7), which are almost comparable in energy (Table

[

53]

3

)

. These two more conventional isomers are connected

1

by 1,3-hydrogen transfer via TS 1/ 2 having a substantial

energy demand (ca. 70 kcal/mol) but still lying below the

•

CO ϩ CH CH and CH ϩ HCCO exit channels. The

3

thermodynamically most favorable exit channel corre-

sponds to formation of carbon monoxide and ethene, but

1

the energy demand of the associated TS 2-CO is huge. Ac-

1

cordingly, the interconversion 3 q 2 is not facile, but

feasible provided that a sufficient amount of internal energy

is available.

1

Singlet oxetene 10 is ca. 30 kcal/mol higher in energy

1

than 1 and can be accessed via TS 1/ 10 with an energy

1

demand of 56 kcal/mol. It should be noted that acrolein

Methyl ketene 2 is also connected to acetyl carbene 11,

exists in s-cis or s-trans conformations. Of course, ring clo- which is higher in energy by 65 kcal/mol. The associated

1

sure to 10 involves the s-cis conformer as a starting point. TS 2/ 11 for the well-known Arndt-Eistert rearrange-

1

[6][7]

As the s-cis and s-trans structures of 1 do not differ much ment

in energy

be located below TS 1/ 10, we take the s-trans isomer as body of experimental studies

demands an energy of only 4.5 kcal/mol relative

[

54]

and the rotational barrier can be expected to to the carbene. This result is in agreement with the large

1

[6][7]

which describe the forma-

representative for both conformers and refrain from dis- tion of transient keto carbenes from diazo ketones to yield

1

cussing this aspect further. 10 appears to be a good candi- the corresponding ketenes as reaction products. Experi-

date for fragmentation into CH O ϩ C H , another exit ments with ¹³C labeled diazo ketones revealed that the

[6]

2

channel which is not too high in energy. However, this reac- label is distributed between the former carbene and car-

tion corresponds to a symmetry forbidden [2ϩ2] cyclorever- bonyl carbon atoms, and oxirenes have been postulated as

1

[2g]

1

sion proceeding via TS 10-CH O reflected by the sizable reaction intermediates . In fact, methyl oxirene 9 has

2

barrier of 71 kcal/mol.

been generated and identified by neutralization-reionization

992

Eur. J. Org. Chem. 1998, 987Ϫ1009

3 4

Oxyallyl and Other [C ,H ,O] Neutrals

FULL PAPER

Table 3. Calculated total energies (E

t

), zero-point vibrational energies (ZPVE), and calculated as well as experimental heats of formation

1

[a]

(

∆H ) of neutral singlet [C ,H ,O] isomers and relevant fragmentation reactions

f 3 4

(B3LYP)^[^b^]

hartree)

ZPVE

(hartree)

∆H

(B3LYP)^[^c^]

E

(G2)

∆H

(G2)

∆H

f

[e]

E

t

f

t

f

(

(kcal/mol)

(hartree)

(kcal/mol)

1

[d]

CH

2

3

CHCHO ( 1)

Ϫ191.913197

Ϫ191.914879

Ϫ191.844234

Ϫ191.861719

Ϫ191.880069

Ϫ191.805322

Ϫ191.861616

Ϫ191.810174

Ϫ191.768328

Ϫ191.802213

Ϫ191.824641

Ϫ191.803533

Ϫ191.789822

Ϫ191.814573

Ϫ191.822957

Ϫ191.798161

Ϫ191.823775

Ϫ191.747946

Ϫ191.908774

Ϫ191.845024

Ϫ191.792063

Ϫ191.779951

Ϫ191.779679

Ϫ191.771069

Ϫ191.750337

0.061062

0.060763

0.060446

0.061254

0.060537

0.057272

0.062463

0.057601

0.058537

0.055580

0.059951

0.056119

0.055123

0.057748

0.057756

0.056033

0.057725

0.054674

0.055814

0.053485

0.051516

0.048023

0.050027

0.049262

0.048012

Ϫ11.6

Ϫ12.7

31.7

17.6

9.1

Ϫ13.2

Ϫ18.0

Ϫ25.0

1

[d]

CHCO ( 2)

Ϫ12.6

1

OCCH ( 3)

1

c-OCH

2

C(

؍

CH

2

) ( 7)

Ϫ191.539543

Ϫ191.482789

Ϫ191.476214

17.4

7.2

1

[d]

c-CH

2

CH

2

CO ( 8)

4.0

1

c-OCHC(؊CH

3

) ( 9)

56.0

20.7

53.0

79.3

58.0

43.9

57.5

65.8

50.2

45.0

60.5

44.4

92.1

Ϫ8.8

31.2

64.4

72.0

72.2

77.6

90.5

148.3

1

c-OCHCHCH

2

( 10)

1

CH

3

COCH ( 11)

OCHC: ( 13)

53.0

1

3

1

TS 1/ 2

1

TS 1/ 10

1

TS 2/ 11

57.1

1

TS 2-CO

1

TS 7/ 8

•

Ϫ191.494922

Ϫ191.504859

Ϫ191.471029

Ϫ191.498198

45.4

39.1

60.4

43.3

1

2 2

TS 8/ 8 ( CH )

COCH ^•

1

TS 8/ 11

1

TS 8-CO

1

2

TS 10-CH O

CO ϩ H

CH O ϩ HCCH

CCH

2

CCH

2

Ϫ13.9

28.5

2

CO ϩ H

3

•

CH

3

ϩ HCCO

2

O ϩ H CC:

2

•

HCO ϩ C

2

H

3

1

CH

2

CO ϩ CH

2

1 [f]

2

CCH

2

ϩ O

150.5

[

a]

[b]

The total energies of the fragments are given in Table 2. Ϫ

Total energies based on B3LYP/6-311ϩϩG(d,p) optimized structures

[c]

including ZPVE corrections (unscaled). Ϫ Calculated heats of formation on the basis of literature data (see Computational Details).

Ϫ

B3LYP calculations are problematic with the excited singlet state of oxygen atoms, the excitation energy of 1.967 eV (see ref. ) relative

to the triplet ground state has been added to the calculated energy of the CH

[

d]

[39]

[e]

[40]

[f]

f

G2 data taken from ref. ; the data were adopted to the ∆H scale. Ϫ Reference data as given in Table 2 and ref. . Ϫ Since

[48]

1

3

2

ϭCϭCH

2

ϩ O exit channel.

1

Figure 2. Relevant parts of the singlet neutral [C

3

,H

4

,O] potential-energy surface calculated at the B3LYP/6-311ϩϩG(d,p) level of theory.

[39]

Values in brackets are heats of formation derived from G2 calculations (in part taken from ref.

)

Eur. J. Org. Chem. 1998, 987Ϫ1009

993

D. Schröder, H. Schwarz et al.

FULL PAPER

[

4]

mass spectrometry . At the B3LYP/6-311ϩϩG(d,p) level method; the very similar geometric and energetic results ob-

1

of theory 9 corresponds to a local minimum energetically tained with the G2 method are, however, comforting in

1

close to 11. Although we did not succeed in locating the this respect.

1

transition structure connecting 9 and 11, this part of the

potential-energy surface is assumed to be rather flat, be-

cause otherwise the exchange of the isotopic labels could

1

Figure 3. IRC calculation for the reaction 8 Ǟ TS 8-CO Ǟ C

ϩ CO

2 4

H

1

not compete with the isomerization to 2. Ring opening of

can either lead to acetyl carbene 11 or formyl methyl

carbene 12. While the Arndt-Eistert reaction of the former

1

9

1

involves a methyl migration with a 4.5 kcal/mol barrier, the

1

same reaction of 12 proceeds by a 1,2-hydrogen shift which

can be expected to be even more facile. All attempts to lo-

1

cate 12 at the B3LYP/6-311ϩϩG(d,p) level failed and led

to methyl ketene instead. This finding is in conflict with the

1

3

[6]

result that photolysis of 2-diazo-[2- C]propanal leads to

1

3

a distribution of the C-label over the C(1) and C(2) posi-

tions of the methyl ketene formed. Accordingly, these

experiments predict formyl methyl carbene as an intermedi-

ate with a lifetime long enough to undergo the isomeri-

zation 12 q 9 q 11, and it remains an open question,

whether or not 12 represents a genuine minimum.

1

With regard to singlet oxyallyl 14 (see below), we also

searched for a transition structure connecting it with acetyl

1

carbene 11 via a 1,3-hydrogen migration as found for the

radical anions. Indeed, a TS was found only 7.5 kcal/mol

higher in energy than the carbene, but an IRC calculation

1

revealed that this transition structure, TS 8/ 11, connects

1

1 with cyclopropanone 8 rather than oxyallyl. In fact,

this finding is already indicated by the short C(1)ϪC(3) dis-

˚

1

tance in TS 8/ 11 (1.861 A, Chart 4) as compared to the

˚

•

؊

•؊

much longer ones in TS11 /14 (2.207 A, Chart 3) and

˚

3

TS 11/ 14 (2.209 A, Chart 5, see below) of the radical anion

and the triplet neutral, respectively. As TS 2/ 11 and TS 8/

1 are quite close in energy, we performed G2 calcu-

1

[

55]

1

lations

(Figure 2, values in brackets) in order to verify

Cyclopropanone can also rearrange to allene oxide 7 via

1

the energetics. The G2 energies are in excellent agreement TS 7/ 8. Allene oxides have long been predicted to be inter-

with the B3LYP results. Thus, in line with experimental mediates in the formation of cyclopropanones by oxidation

[

6][7]

1

[2][14]

data

the Arndt-Eistert reaction 11 Ǟ 2 is more favor- of allenes with peracids

, and experimental evidence for

1

able than isomerization of 11 to cyclopropanone 8.

Cyclopropanone 8 may decompose unimolecularly via ionization mass spectrometry

TS 8-CO into carbon monoxide and ethene which is exo- lation techniques . The barrier for formation of cyclopro-

the existence of 7 has been provided by neutralization-re-

1

[2d][2e]

as well as matrix-iso-

1

[2g]

1

thermic by 18 kcal/mol and associated with a 35 kcal/mol panone TS 7/ 8 is higher than TS 8-CO by 6 kcal/mol at

barrier. According to the Woodward-Hoffmann rules^[

56]

,

the B3LYP/6-311Gϩϩ(d,p) level, but this difference dimin-

the cycloreversion corresponds to a concerted [2ϩ1] process ishes to ca. 1 kcal/mol using the G2 method. Therefore,

1

which is thermally allowed due to the particular orbital rearrangement to 7 starting with 8 might compete with

1

situation at the carbene-like CO subunit. Unlike 8 itself, the loss of CO, but 7 constitutes a dead end because the

TS 8-CO is, however, far from being C symmetric (Chart fragmentations into either H CO ϩ H CC, H CCO ϩ

1

2

v

2

[

57]

1

4

)

. In order to ensure that this transition structure in-

CH , or H CCCH ϩ O are quite high in energy (55,

2 2 2

1

deed connects 8 with the CO ϩ C H exit channel, an IRC 73, and 131 kcal/mol relative to 7, respectively). Methoxy

calculation was performed (Figure 3). Accordingly, loss of acetylene 3 and methoxy vinylidene 13 were found as local

2

4

1

CO from cyclopropanone occurs as a concerted reaction in minima. In accordance with the thermochemistry of other

[9]

1

that no intermediate 1,3-biradical is involved, nevertheless vinylidenes , 3 is more stable than 13 by ca. 48 kcal/mol.

it proceeds in an extremely asynchronous fashion. The first While all attempts failed to locate a TS connecting these

˚

CϪC bond is almost completely broken (r_C_Cϭ 2.116 A) in minima, it seems reasonable to assume that it is close in

1

the TS, while the second CϪC bond is exactly as long as in energy to 13 due to the presence of a hydrogen atom in the

˚

1

8

(r_C_Cϭ 1.469 A). At this sort of geometry, some biradical vinylidene unit.

1

character cannot be excluded, and some doubt could be

The parts of the [C ,H ,O] potential-energy surface de-

3 4

cast upon the accuracy of the single-reference B3LYP scribed so far are in reasonable agreement with previous

994

Eur. J. Org. Chem. 1998, 987Ϫ1009

3 4

Oxyallyl and Other [C ,H ,O] Neutrals

FULL PAPER

calculations and known experimental data. However, there Table 4. Calculated total energies (E

t

), zero-point vibrational ener-

gies (ZPVE), and calculated as well as experimental heats of forma-

exists one major problem which is directly connected to a

3

tion (∆H ) of neutral triplet [C ,H ,O] isomers and relevant frag-

f 3 4

1

central topic of this study. Thus, A oxyallyl 14 is pre-

[a]

1

mentation reactions

dicted to be a transition structure for the symmetry al-

lowed, disrotatory 8 Ǟ TS 8/ 8 Ǟ 8 stereomutation of

1

[b]

[c]

[d]

E

(

t

ZPVE

(hartree)

∆H

f

∆H

f

hartree)

(kcal/ (kcal/

mol) mol)

cyclopropanone rather than being a genuine minimum. As

1

a consequence, acetyl carbene 11 and allene oxide 7 are

also not connected to oxyallyl, but directly to cyclopro-

3

CH

2

3

CHCHO ( 1)

Ϫ191.825675

Ϫ191.815498

Ϫ191.819235

Ϫ191.732148

Ϫ191.839073

0.056509

0.057920

0.058379

0.058825

0.056839

43.3

49.6

1

3

panone. The prediction of A oxyallyl as a saddle point is

COCH ( 11)

1

3

_i_n_m_a_r_k_e_d_c_o_n_t_r_a_s_t_t_o_e_a_r_l_i_e_r_s_t_u_d_i_e_s[

20][21]

CCHO ( 12)

47.3

, shedding some

3

OCHC: ( 13)

102.0

34.9

doubt on the performance of B3LYP in accurately describ-

3 •

•

CH COCH

2 2

1

3

ing A oxyallyl (see below).

( B

2

) ( 14)

1

3

•

3

CH CH

2

CO

•

( 18) Ϫ191.826711

Ϫ191.750100

Ϫ191.779254

Ϫ191.757155

Ϫ191.805137

0.056385

0.052037

0.053445

0.051756

0.058083

42.6

90.6

72.4

86.2

56.0

Chart 5

3

TS 1/ 12

3

TS 1/ 18

3

TS 11/ 14

3

TS 18/ 18

3

(

c-CH

3

2

CH

2

CO)

•

CH

ϩ HCCO

Ϫ191.779951

Ϫ191.769081

Ϫ191.730440

0.048023

0.048635

0.054844

72.0

78.8 81.6

103.0 105.2

3

2

CO ϩ CH

2

3

2

CCH ϩ O

[

a]

[b]

The total energies of the fragments are given in Table 2. Ϫ

Total energies based on B3LYP/6-311ϩϩG(d,p) optimized structu-

[c]

res including ZPVE corrections (unscaled). Ϫ Calculated heats

of formation on the basis of literature data (see Computational

Details). Ϫ ^[Reference data as given in Table 2 and ref.

d]

[40]

.

3

Figure 4. Relevant parts of the triplet neutral [C

3

,H

4

,O] potential-

energy surface calculated at the B3LYP/6-311ϩϩG(d,p) level of

theory

ketene and triplet methylene provides an exit channel which

is directly accessible from triplet oxyallyl requiring 7 kcal/

mol less energy than TS 11/ 14.

3

The Neutral Triplet [C

3

,H

4

,O] Potential-Energy Surface

3

For the neutral triplet surface (Figure 4), six local min-

ima (Chart 5) have been located which correspond to acrol-

3

Rearrangement of formyl methyl carbene 12 into the

3

ein 1, acetyl carbene 11, formyl methyl carbene 12, meth- lowest triplet state of acrolein 1 is exothermic by ca. 4 kcal/

3

oxy vinylidene 13, oxyallyl 14, and the C-CO ring opened mol and proceeds via TS 1/ 12 having an energy demand

biradical ³18. Thus, exactly those five structures which of ca. 45 kcal/mol. Similarly, a 1,2-hydrogen shift via TS 1/

3

correspond to stable anions (see above) are local minima

on the triplet surface of the neutral species

18 connects 1 and 18 which are close in energy with a

mutual barrier of about 30 kcal/mol. Triplet cyclopro-

[58]

.

3

Surprisingly, B oxyallyl 14 represents the most stable panone does not correspond to a local minimum and can

2

of these, being separated by a barrier of 51 kcal/mol with rather be regarded as the transition structure for the de-

3

respect to 1,3-hydrogen shift via TS 11/ 14 to yield triplet generate exchange of the two methylene groups in 18 which

3

acetyl carbene 11 (Table 4). The lowest exit channel pro- is expected to be facile because TS 18/ 18 is located only

ceeds from 11 to CH ϩ HCCO with an energy demand 13 kcal/mol above 18.

3

•

3

of 22 kcal/mol relative to the carbene. The isomerization

Due to the significant singlet-triplet gaps of 53 and 23

3

1

4 q 11 is, however, unlikely because the formation of kcal/mol, respectively, the triplet isomers 1 and 13 are not

Eur. J. Org. Chem. 1998, 987Ϫ1009

995

D. Schröder, H. Schwarz et al.

FULL PAPER

1

likely to be formed by electron detachment of the corre- Table 7. Total energies (E

t

) of A

1

oxyallyl ( 14) and energies (Erel

)

3

1

of

B

2

oxyallyl ( 14) and cyclopropanone ( 8) relative to 14 at

various levels of theory

sponding radical anions in NR experiments. Instead, the

3

triplet state of acetyl carbene 11 is calculated to be ca. 4

kcal/mol more stable than the singlet. This result is in agree-

ment with previous calculations as well as experimental

work on π-acceptor substituted carbenes such as dicyano

methylene and others^[. Due to the energetic proximity of

singlet and triplet 11, spin inversion shall be considered too.

1_A

3_B

1

Oxyallyl

2

Cyclo-

Oxyallyl

propanone

[

a]

[b]

rel

[b]

E

t

E

rel

52]

ZPVE

CASSCF(6,5)/6-31G(d))

0.059988

0.060220 0.065149

(

Both states have geometries differing particularly with re- HF/cc-pVDZ

spect to the HCCO dihedral and the CCC and CCO angles.

Ϫ190.660187

Ϫ191.270594

Ϫ191.299715

Ϫ191.338351

Ϫ191.504859

d]

Ϫ40.0

Ϫ17.7

Ϫ13.5

Ϫ1.6

Ϫ0.8

Ϫ21.6

Ϫ10.1

Ϫ2.2

Ϫ3.3

Ϫ4.8

Ϫ3.8

Ϫ1.6

Ϫ1.2

Ϫ1.7

Ϫ1.1

Ϫ0.9

؊0.3

Ϫ51.2

Ϫ40.1

Ϫ39.7

Ϫ29.6

Ϫ31.9

Ϫ43.4

Ϫ35.9

Ϫ31.8

Ϫ36.7

Ϫ23.1

Ϫ28.8

Ϫ23.3

Ϫ28.9

Ϫ28.4

Ϫ28.2

Ϫ27.9

؊30.1

MP2/cc-pVDZ

CCSD/cc-pVDZ

Therefore, a reasonable description of the spin inversion re-

quires the determination of the minimum-energy crossing G2

CCSD(T)/cc-pVDZ

[c]

[

1

3

BHLYP/6-311ϩϩG(d,p) Ϫ191.751742

[d]

point (MECP) between these two states. MECP 11/ 11 was

B3LYP/6-311ϩϩG(d,p)

Ϫ191.839073

Ϫ191.826946

Ϫ191.882780

Ϫ190.734961

Ϫ190.758435

Ϫ191.291950

Ϫ191.317058

Ϫ191.252650

3

[d]

located ca. 9 kcal/mol above 11. The geometry (Chart 5) BLYP/6-311ϩϩG(d,p)

[d]

BP86/6-311ϩϩG(d,p)

corresponds almost to an averaged structure of both min-

CASSCF(6,5)/6-31G(d)

CASSCF(6,5)/cc-pVDZ

CASPT2/cc-pVDZ

ima. At the MECP, the matrix element for spin inversion

SO

Ϫ1

SO

was computed as H ϭ 5.5 cm . The magnitude of H

_t_e_n_d_s_t_o_i_n_c_r_e_a_s_e_w_i_t_h_t_h_e_a_t_o_m_i_c_n_u_m_b_e_r_,[

45][59]

CASPT3/cc-pVDZ

MR-CISD/cc-pVDZ

and the size

of the computed value can be regarded as being typical for

MR-CISD(ϩQ)/cc-pVDZ Ϫ191.325658

SO

[60]

carbon; for example, H of methylene was determined

MR-ACPF/cc-pVDZ

MR-ACPF/cc-pVTZ

Ϫ191.330166

؊191.492953

Ϫ1

as 8 cm . Although small, these matrix elements may still

account for an efficient spin crossing under the experimen-

tal conditions (see below).

[a]

1

Total energies of the A

1

state of oxyallyl from single-point calcu-

[b]

lations at CASSCF(6,5)/6-31G* geometry. Ϫ Energies relative to

1

the

1

A state of oxyallyl from single-point calculations at

CASSCF(6,5)/6-31G* geometries, and including the (unscaled)

[

c]

[d]

Table 5. Frequencies for the dis- and conrotatory movements of the

CASSCF zero-point energy. Ϫ Standard G2 procedure. Ϫ Ge-

1

CH

2

groups at the C2v optimized geometry of the A

glet oxyallyl 14

1

state of sin-

ometries fully optimized at the levels of theory given here.

ν

1

(disrot)

ν

2

(conrot)

Neutral Oxyallyl: A Challenge for Computational Chemistry

In view of the note by Osamura et al.^[²⁰^b^]that “the elec-

tronic state for this system is not necessarily best described

RHF/3-21G

410i

779 (ν

766 (ν

779 (ν

724 (ν

954 (ν

638 (ν

640 (ν

471 (ν

411

5

9

5

3

)

RHF/6-31G

329i

RHF/6-31G(d)

419i

by a one-configuration wave function”, the failure of

BHLYP/6-311ϩϩG(d,p)

B3LYP/6-311ϩϩG(d,p)

BLYP/6-311ϩϩG(d,p)

BP86/6-311ϩϩG(d,p)

CASSCF(2,2)/3-21G

402i

1

B3LYP in describing neutral A oxyallyl is not surprising,

405i

1

[

c]

1098i, 302i

1141i, 340i

140

as the method is essentially a one-determinant procedure

and contains an “exact” Hartree-Fock exchange contri-

bution. This conjecture is further supported by the fact that

RHF calculations with different basis sets (Table 5) lead to

similar results with respect to the imaginary frequency for

the disrotatory ring closure and predict singlet oxyallyl as a

saddle point.

[

a]

CASSCF(4,4)/6-31G(d)

132

[

b]

CASSCF(6,5)/3-21G

CASSCF(6,5)/6-31G

170

230

116i

374

128i

376

CASSCF(6,5)/6-31G(d)

CASSCF(6,5)/D95*

CASSCF(6,7)/6-31G(d)

87i

403

95i

404

76

449 (ν

3

Three questions shall be addressed in this section: (i) Is

A oxyallyl a transition structure or a local minimum? (ii)

1

[

a]

_T_a_k_e_n_f_r_o_m_r_e_f_.[23]. Ϫ From numerical differentiation of analy-

[b]

[

20b]

[c]

tical gradients. Taken from ref.

of C2v symmetric A

. Ϫ The vibrational analysis How large is the singlet-triplet splitting? (iii) What is the

oxyallyl with pure density functionals leads

to two imaginary frequencies. The larger of these corresponds to a

symmetric out of plane movement of the four hydrogen atoms.

1

3

1

probability of spin crossing from B to A oxyallyl? In

2

1

contrast to the hybrid method B3LYP, pure density func-

C

2

tional theory might be expected to correctly describe the

1

electronic structure of A oxyallyl, and therefore the BLYP

1

Table 6. Characteristic geometrical parameters of the C2v symme-

and BP86 functionals have been applied. As a reference, we

have chosen the frequencies (Table 5), geometries (Table 6),

3

1

2 1

tric B and A States of oxyallyl 14 at various levels of theory

1

3

1

and relative energies of A and B oxyallyl as well as

1 2

B

cc

2

A

1

r

co

α

cco

r

cc

r

co

α

cco

cyclopropanone (Table 7) obtained with CASSCF or MR-

CI methods (bold entries). In general, the geometrical pa-

3

BP86/6-311ϩϩG(d,p) 1.446 1.275 119.2 1.452 1.245 126.1

BLYP/6-311ϩϩG(d,p) 1.447 1.280 119.2 1.455 1.248 125.8

B3LYP/6-311ϩϩG(d,p) 1.436 1.267 119.1 1.438 1.238 126.1

BHLYP/6-311ϩϩG(d,p) 1.424 1.259 119.0 1.421 1.229 126.4

rameters of the B state are similar for all methods, while

2

1

for the A state all density functionals used show devi-

1

˚

ations of up to ca. 0.05 A and 3°, respectively, as compared

to the CASSCF results. As far as the singlet-triplet splitting

is concerned, the pure density functionals BLYP and BP86

are in reasonable agreement with CASSCF calculations

HF/6-31G(d)

1.425 1.275 118.6 1.413 1.231 126.8

1.443 1.243 119.5 1.451 1.236 126.5

MP2/6-31G(d)

CASSCF(6,5)/6-31G(d) 1.440 1.250 119.2 1.467 1.211 122.9

996

Eur. J. Org. Chem. 1998, 987Ϫ1009

3 4

Oxyallyl and Other [C ,H ,O] Neutrals

FULL PAPER

1

3

(

Table 7). The more Hartree-Fock exchange is taken into the A and B states diminishes to Ϫ1.6 and Ϫ1.2 kcal/

1 2

consideration, the larger the singlet-triplet splitting, i.e. mol upon inclusion of perturbation theory in CASPT2 and

Ϫ10.1 kcal/mol for B3LYP, Ϫ21.6 kcal/mol for BHLYP, CASPT3 calculations, respectively, to only Ϫ0.3 kcal/mol at

and Ϫ40.0 kcal/mol for a HF calculation. Similar argu- the MR-ACPF level of theory using the sizable cc-pVTZ

1

ments hold true for the energy difference between A oxy- basis set. In fact, both states are almost isoenergetic, al-

1

allyl and cyclopropanone, for which we consider a value of though the triplet remains the ground state at all levels ap-

ca. Ϫ30 kcal/mol as derived from MR-CI as a good esti- plied here.

1

3

mate. In accord with the change in the singlet-triplet split-

Thus, the A and the B states of oxyallyl are very simi-

1 2

1

ting, A₁oxyallyl becomes destabilized the more HF lar not only with respect to their geometries, but also as far

character is included. Interestingly, the BLYP and BP86 as their relative energy is concerned. Surface hopping from

density functionals even lead to two imaginary frequencies the triplet to the singlet surface may therefore play a role.

1

3

in the vibrational analysis of C₂_vsymmetric A oxyallyl, Accordingly, B oxyallyl could undergo spin inversion to

1

2

one corresponding to the disrotatory ring closure (302i the singlet surface, followed by ring closure to cyclopro-

Ϫ1

3

cm and 340i cm , respectively), the second, even larger panone. Provided that this process is efficient, also B oxy-

2

Ϫ1

one (1098i cm , 1141i cm ) describing a C symmetric allyl would represent a short-lived species. The highest

2

out-of-plane movement of the four hydrogen atoms. At probability for spin crossing can be expected at the mini-

1

[47][63]

these levels the A state is neither a local minimum, nor a mal-energy crossing point (MECP)

on the crossing

1

transition structure, but a higher-order saddle point. Hence, seam of both potential-energy surfaces. Unlike in the case

we conclude that these density functionals cannot describe of the acetyl carbene system, the triplet and singlet minima

singlet oxyallyl.

in the oxyallyl system are very close, and since their energies

From earlier vibrational analyses at the CASSCF(4,4)/6- are nearly degenerate, the MECP can be assumed to lie very

[

23]

3

1G(d)

and CASSCF(6,5)/3-21G levels with numerical close to both minima, and does not need to be located. The

[20b]

1

SO

differentiation of analytical gradients

, A oxyallyl has spin-orbit coupling element H calculated at the singlet

1

Ϫ1

been predicted to be a local minimum. The lowest, real fre- minimum is extremely small (ca. 0.05 cm ) whatever the

quency has been assigned to disrotatory movement of the wavefunction used to calculate it^[⁶⁴^]. The calculated H^S^Oof

two methylene groups. However, the transition structure as- 14 is two orders of magnitude smaller than that for acetyl

sociated with ring closure to cyclopropanone was only 0.33 carbene (see above), suggesting that interconversion of the

1

3

kcal/mol above the minimum. We could reproduce these re-

A and B states of oxyallyl may be inefficient in the µs

1 2

sults qualitatively, but obtain one imaginary frequency with timescale. In summary, the triplet state of oxyallyl is pre-

CASSCF(6,5)/3-21G (additional inclusion of the in-plane dicted to be experimentally accessible, because the thresh-

oxygen lone-pair (b ) in the active space), if the frequencies olds for isomerization and/or fragmentation are sufficiently

2

are calculated analytically. The situation is not very sensi- large and the triplet lifetime may suffice for its detection.

tive to the size of the basis set for a given active space,

i.e. all analytical frequency calculations at the CASSCF(6,5)

2

,O]^•^؉Potential-Energy Surface

The Cationic [C

3

,H

4

With respect to the NR experiments reported further be-

level (Table 5) predict imaginary frequencies of about 100i

Ϫ1

low, some knowledge about the cationic surface is required

cm . Upon further increase of the active space in a

(

Figure 5)^[

2d][2e][4][41]

. 16 stationary points (Chart 6) were

CASSCF(6,7)/6-31G(d) calculation (addition of two virtual

located and their energetics as well as those of a variety of

exit channels were evaluated (Table 8).

Ionized methyl ketene 2^•represents the most stable iso-

(

a , b ) orbitals in the active space) the frequency becomes

1 2

Ϫ1

real again (76 cm ), thus, predicting singlet oxyallyl as a

genuine intermediate. In conclusion, there is no obvious

؉

trend which correlates the size of either the active space or mer, and its geometry resembles much the corresponding

1

the basis set with the nature of A oxyallyl. In fact, these singlet neutral. Direct cleavage of a terminal CϪH bond in

1

•؉

ϩ

•

calculations cannot decide whether A oxyallyl is a local

2 , leading to H CCHCO ϩ H , is quite favorable in en-

1

2

•

ϩ

minimum or a transition structure, but only lead to the con- ergy and lies only ca. 9 kcal/mol above CO ϩ C H . Two

clusion that the [C ,H ,O] potential-energy surface is 1,2-hydrogen shifts lead from 2 via TS2 /18 to 18

rather flat around A oxyallyl. Moreover, even if A oxyal- a prototype distonic ion

lyl represents a real minimum on the potential-energy sur- 18 to ionized acrolein 1 . The s-cis form of 1 can un-

2

•؉

4

1

•؉

•ϩ

•؉

,

3

4

1

[28][65]

, and from 18 via TS1^•^ϩ/

•؉

1

•

؉

•؉

•

ϩ

•؉

face, rearrangement to cyclopropanone is expected to be dergo ring closure via TS1 /10 to the oxetene radical-

1

•؉

•ϩ

•؉

very facile. In conclusion, A oxyallyl is very likely to es- cation 10 . Due to the similar energetics of TS1 /10

1

cape all attempts of its generation and identification by neu- and TS1 /18 the isomerizations 1 Ǟ 10 and 1^•^؉

•

ϩ

•؉

Ǟ

•

؉

tralization-reionization mass spectrometry, although very 18 might compete to some extent. With respect to a mass

flat potential-energy surfaces with wells of less than a few spectrometric characterization, the formation of 10^•is,

kcal/mol have been successfully probed using this however, a dead end, as no low-lying, direct exit channels

are accessible for this radical cation. For example, the most

The singlet-triplet splitting was studied by a series of en- probable fragmentation into acetylene and ionized formal-

ergy calculations at higher levels using the optimized dehyde via TS10 ϪC H requires 50 kcal/mol more than

CASSCF(6,5)/6-31G(d) geometry (Table 7). Starting from TS1 /10 . Instead, energized 1 and 10 are likely to

؉

method^[⁶¹^]^[⁶²^].

•

ϩ

2

•

ϩ

•؉

ϩ

•

Ϫ4.8 kcal/mol with CASSCF(6,5)/6-31G(d), the splitting of reach the H CCHCO ϩ H exit channel.

2

Eur. J. Org. Chem. 1998, 987Ϫ1009

997

D. Schröder, H. Schwarz et al.

FULL PAPER

lower in energy than 14^•. This result is obviously wrong,

but still within the error margins of ±5 kcal/mol. Notwith-

standing, the qualitative conclusion remains unchanged in

that ionized oxyallyl is predicted to rapidly rearrange to al-

lene oxide radical-cations. The earlier experiments^[also

؉

Chart 6

2e]

revealed an intense loss of CH concomitant with the for-

2

mation of ionized ketene from allene oxide radical-cations

•

ϩ

7

. Owing to the fact that this exit channel is ca. 70 kcal/

•

ϩ

mol higher in energy than CO ϩ C H₄, appreciable bar-

2

riers and/or dynamic bottlenecks must prevent isomeri-

•

؉

•؉

zation of 7 and/or 14 to any of the other isomers which

•

ϩ

can decompose into CO and C H . Indeed, a concerted

,2-hydrogen shift via TS1 /7 was located which is more

than 60 kcal/mol above 7 and only 6 kcal/mol below the

threshold of the CH ϩ H CCO exit channel. If one

2

•؉

4

•

ϩ

1

•

؉

3

•ϩ

2

takes into account that rearrangements are entropically less

favorable as compared to direct bond cleavages, it is obvi-

ous that both pathways compete in high-energy collisions.

•

؉

•؉

Hence, it appears as if 7 and 14 are somewhat sepa-

rated from the other parts of the potential-energy surface.

Nevertheless, there might exist other feasible reaction path-

ways with lower energy demands. A plausible candidate

would be the formation of ionized cyclopropanone from

•

؉

•ϩ

7

, which then could decompose to CO ϩ C H₄. How-

2

ever, in analogy to the triplet neutral, ionized cyclopro-

panone does not correspond to a local minimum, but to

•

ϩ

•؉

TS18 /18 which describes the degenerate interchange of

the two methylene groups in the acyclic distonic ion 18^•

ϩ

.

Further, all attempts to locate a transition structure for

•

؉

CϪC ring closure from 14 or a concerted CϪO ring

•

؉

•ϩ

•؉

cleavage/CϪC closure from 7 failed. TS18 /18 is 10

•

ϩ

kcal/mol lower than the CO ϩ C H

exit channel which

2

2e]

4

is consistent with the observation^[of partial exchange of

the two methylene groups in unsymmetrically labeled 18^•

؉

.

With respect to the radical cations of acetyl and formyl

methyl carbenes, several attempts to optimize their geo-

•

؉

metries always lead only to ionized methyl oxirene 9 . This

[4]

result is again in accord with experimental studies , which

demonstrated that upon ionization of diazo acetone 9^•in-

stead of ionized acetyl carbene is formed. Finally, though

؉

•

؉

quite energetic the radical cations of methoxy acetylene 3

•

؉

and methoxy vinylidene 13 were located as local minima;

again, the transition structure connecting these isomers

could not be located.

Predictions Emerging from the Calculations

•

؉

A different situation is found for allene oxide 7 and

The presentation of the theoretical results so far de-

•

؉

oxyallyl radical-cations 14 which are located 39 and 60 scribed the potential-energy surfaces one by one in a kind

kcal/mol above 2 , respectively. In line with earlier re- of “horizontal” view. Neutralization-reionization (NR) and

sults , the oxyallyl radical-cation 14 represents a high- charge-reversal (CR) experiments involve, however, tran-

energy minimum on a rather flat part of the [C ,H ,O]

•

؉

[

2e]

•؉

2

•ϩ

sitions from one surface to another by electron transfer. Ac-

3

4

potential-energy surface. It is connected to the more favor- cordingly, we shall attempt to draw some conclusions from

•

؉

•ϩ

•؉

able allene oxide cation 7

by TS7 /14 , and calcu- the theoretical results in order to make predictions for the

lations at the MP4/6-31G(d) level have predicted a barrier experiments.

[2e]

of no more than 2 kcal/mol . At the B3LYP/6-

The electron-transfer processes occurring in NR and CR

11ϩϩG(d,p) level, TS7 /14 is unambiguously charac- experiments can be considered to occur as vertical processes

terized by a single imaginary frequency corresponding to which proceed without any significant geometrical

•

ϩ

•؉

3

•

؉

•؉

[66]

the ring closure 14 Ǟ 7 , but it is found to be slightly changes . Hence, Franck-Condon factors may play a cru-

998

Eur. J. Org. Chem. 1998, 987Ϫ1009

3 4

Oxyallyl and Other [C ,H ,O] Neutrals

FULL PAPER

Table 8. Calculated total energies (E

t

), zero-point vibrational energies (ZPVE), and calculated as well as experimental heats of formation

f 3 4

∆H ) of cationic [C ,H

,O]^•isomers and relevant fragmentation reactions

ϩ

[a]

(

[

b]

[c]

[d]

[e]

E

t

ZPVE

(hartree)

∆H

f

∆H

f

∆H

f

(

hartree)

(kcal/mol)

•

؉

•؉

CH

2

3

CHCHO (1

•؉

)

Ϫ191.549037

Ϫ191.589240

Ϫ191.504748

Ϫ191.526705

Ϫ191.515201

0.059192

0.059695

0.059186

0.060290

0.058519

0.062190

0.057865

0.059510

0.058848

0.052529

0.059647

0.054396

0.055826

0.057386

0.054536

0.058868

0.053329

0.050586

0.050781

0.050826

0.052583

0.047368

0.048102

0.048302

0.047673

0.052349

217.9

191.7

244.8

231.0

238.1

224.4

297.4

251.6

205.6

292.5

240.0

237.1

217.8

248.1

290.4

218.8

229.3

238.4

272.9

280.9

290.9

321.9

331.7

298.6

313.4

321.3

215.0

181.4

208.6(B), 219.2(M)

193.4(B), 190.7(M)

•

؉

CHCO (2

•؉

)

•

؉

OCCH (3

•

؉

•؉

c-OCH

2

C(

؍

CH

2

)

(7

)

228.5(T)

•

؉

•؉

c-OCHC(؊CH

3

(9

)

232.1(T)

•

ϩ

(10^•^؉) Ϫ191.537224

•؉

c-OCHCHCH

2

222.5(B); 223.0(M)

•

؉

CH

3

OCHC: (13

)

Ϫ191.420971

Ϫ191.493849

Ϫ191.567123

Ϫ191.428682

Ϫ191.512368

Ϫ191.517064

Ϫ191.532597

Ϫ191.499332

Ϫ191.432045

Ϫ191.546097

Ϫ191.529456

Ϫ191.515027

Ϫ191.460035

Ϫ191.447276

Ϫ191.431369

Ϫ191.381931

Ϫ191.366288

Ϫ191.414081

Ϫ191.395337

Ϫ191.382782

•

ϩ

•؉

CH

2

COCH

2

(14

)

251.7(T)

200.5(B,T), 201.8(M)

•

CH

CO (18^•^؉)

•؉

؉

2

•

؉

TS1 /7

•

؉

•؉

TS1 /10

•

؉

TS1 /18

•

؉

TS2 /18

227.3(M)

253.7(T)

•

؉

TS7 /14

•

؉

2 2

TS10 -C H

•؉

•

؉

[f]

TS18 /18

CO ϩ C

213.8

228.5

226.4(B), 221.8(T) 215.6(M)

224.2(T), 228.9(M)

•

ϩ

2

H

4

ϩ

•

CH CHCO ϩ H

2

ϩ

•

HCO ϩ C

2

H

3

271.3

279.2

291.0

•

ϩ

CH

2

O

ϩ HCCH

•ϩ

O ϩ HCCH

•

ϩ

O

ϩ H

2

CC:

•

ϩ

O ϩ H

2

CC:

•

ϩ

3

CO ϩ CH

2

303.2

328.6

•

ϩ

1

CO ϩ CH

2

•

ϩ

3

CCH

2

ϩ O

[

a]

[b]

The total energies of the fragments are given in Table 2. Ϫ

Total energies based on B3LYP/6-311ϩϩG(d,p) optimized structures

[c]

including ZPVE corrections (unscaled). Ϫ Calculated heats of formation on the basis of literature data (see Computational Details).

Ϫ

levels taken from ref. , ref. , and ref. , respectively. The data were adopted to the ∆H

corresponds to TS18 /18 , the vertical ionization energy of the neutral has been measured. See ref.

[

d]

[40]

[e]

Reference data as given in Table 2 and ref. . Ϫ Computated data at the UHF/6-31G(d) (B), MP4/6-31G(d) (T), and G2 (M)

[

4]

[39]

[41]

[f]

f

scale. Ϫ Although ionized cyclopropanone

•

ϩ

•؉

[40]

.

Figure 5. Relevant parts of the cationic ²[C

,H ,O] potential- ternal energy imparted in the species formed. As qualitative

•ϩ

3 4

energy surface calculated at the B3LYP/6-311ϩϩG(d,p) level of

rules, the efficiency of electron transfer decreases the larger

the geometry differences of the different charge states are,

and the amount of internal energy stored in the species

formed by vertical electron transfer increases with the ge-

ometry differences. Thus, significant geometry differences

will not only be associated with low efficiencies, but also

with the formation of excited products and thus enhanced

fragmentation. Having made these general remarks, let us

consider the behavior of the five stable radical-anions in

theory. Values in brackets are heats of formation derived from G2

_c_a_l_c_u_l_a_t_i_o_n_s_a_s_t_a_k_e_n_f_r_o_m_r_e_f_.[39]. For clarity, some high-energy

exit channels have been omitted (see Table 8)

Ϫ

ϩ

Ϫ

ϩ

NR and CR experiments; for the time being it seems

Ϫ

ϩ

to be too ambitious to predict NIDD spectra.

i) Due to the large singlet-triplet gap, singlet acrolein is

most probably formed upon neutralization of 1 . As the

(

•

؊

radical anion and the singlet neutral have similar geo-

1

metries, 1 will also be formed with little internal energy.

Further, the barriers associated with any isomerizations are

1

large. Hence, 1 is expected to survive as an intact entity

•

؉

in the µs regime and will be then reionized to 1 . These

considerations predict an intense survivor signal in the NR

1

mass spectrum due to reionized 1, accompanied by frag-

ϩ

mentation via α-cleavages to yield H CCHCO , HCO ,

2

ϩ

and C H₃fragments, respectively.

2

(

ii) Due to the small singlet-triplet gaps, acetyl carbene

1

1 and formyl methyl carbene 12 can be formed from their

cial role^[⁶⁷^]not only with respect to the efficiency of the radical anions in their triplet ground states or in the corre-

electron-transfer processes, but also for the amount of in- sponding singlet states. Both singlet carbenes are predicted

Eur. J. Org. Chem. 1998, 987Ϫ1009

999

D. Schröder, H. Schwarz et al.

FULL PAPER

1

to easily rearrange to methyl ketene 2. Due to the signifi- tive peaks in their CR mass spectra. Thus, one prerequisite

cant spin-orbit coupling constants, the triplet states may in- for a successful application of NR experiments is fulfilled

terconvert to the singlets also followed by isomerization to in that the anions under study are experimentally accessible

1

2

within the µs regime. All these four routes lead to 2 and do not give rise to extensive isomerizations. In addition

containing a significant amount of excess energy (more to the systems described further below, we examined a

than ca. 70 kcal/mol), such that within the lifetime of the series of other reagent combinations also including positive

1

neutral partial isomerization to acrolein 1 as well as frag- ions. These results are, however, not included here as the

mentation may occur despite the large energy demands. ions formed were either isomeric mixtures or dealt with

•

Ϫ/0/•ϩ

Due to these substantial differences between the anions and parts of the [C ,H ,O]

potential-energy surfaces

3

4

neutrals, weak survivor signals are expected for the radical which are not relevant for the oxyallyl system^[

anions of both carbenes. Instead, the structural features of

2d][2e][4][41]

.

Ϫ

ϩ

Figure 6. (a) NR mass spectrum (O

2

, 80% T, O , 80% T) and

1

energized

2

imply formation of the fragments

Ϫ

ϩ

(

b) CR mass spectrum (O , 80% T) of acrolein radical-anions

2

ϩ

•؊

H CCHCO , H CO (n ϭ 0, 1), and C H (n ϭ 0Ϫ4)

1

2

generated by NICI of a mixture of acrolein and N O

2

n

2

n

upon reionization to cations. In contrast to case (i), how-

ever, a significant amount of methyl radical loss concomi-

ϩ

tant with the HCCO fragment is also expected. Based on

•

؊

•؊

these arguments, 11 and 12 are predicted to have simi-

Ϫ

ϩ

lar, though certainly not identical NR spectra.

(

iii) The situation is less complicated for the radical-anion

•

؊

of methoxy vinylidene 13 . The size of the singlet-triplet

gap implies that neutralization leads predominantly to the

singlet neutral, which subsequently suffers a facile 1,2-hy-

1

drogen shift to yield methoxy acetylene 3. Due to the sig-

nificant exothermicity of this rearrangement, extensive frag-

1

Ϫ

ϩ

mentation of 3 appears likely. Accordingly, the NR and

Ϫ

ϩ

CR spectra are expected to be dominated by the

HCCO fragment, along with signals characteristic for a

methoxy moiety, i.e. H CO (n ϭ 0Ϫ3) , as well as a

ϩ

[63]

n

weak recovery signal.

(

iv) For neutral oxyallyl 14, the near degeneracy of the

singlet and triplet states suggests that both neutrals are

formed upon electron detachment from the radical anion.

1

For the A state the calculations predict isomerization to

1

cyclopropanone, which can subsequently decompose to

carbon monoxide and ethene due to the excess energy de-

1

liberated in the rearrangement 14 Ǟ 8. As far as the

formation of singlet oxyallyl as transient neutral is con-

Ϫ

ϩ

cerned, we thus expect a NR spectrum which hardly

•

ϩ

shows a recovery signal, but is dominated by CO and

Acrolein Radical-Anion

ϩ

C H (n ϭ 0Ϫ4) fragments formed either by reionization

2

n

Ϫ

The radical anion 1^•can easily be generated by electron

of the neutral fragments or by fragmentation of the dis-

tonic ion 18^•. However, spin inversion of triplet oxyallyl capture in negative ion chemical ionization (NICI) of acro-

ϩ

3

to the singlet surface may be slow such that the B ground lein with N O as reagent gas. As expected from the calcu-

2

3

state, 14, may survive at the neutral stage of a NR experi- lations, significant metastable ion decompositions are not

•

؊

Ϫ

ϩ

Ϫ

ϩ

ment. Reionization of this species to the radical-cation observed for 1 . The NR and CR mass spectra of

•

؊

surface should then inter alia give rise to loss of methylene

1

(Figure 6) are very similar. The most important features

•

ϩ

together with ionized ketene, H CCO , as a characteristic are: (i) a large recovery signal, (ii) the base peak corre-

2

fragmentation arising from the oxyallyl structure. The sponding to the loss of hydrogen, and (iii) less intense sig-

Ϫ

ϩ

Ϫ

ϩ

•ϩ

overall CR and NR spectra are then predicted to be nals for HCO and C H₄or isobaric CO , respectively.

2

Ϫ

ϩ

superpositions of the processes described for each of the In fact, the minor differences between the NR and

two states.

Ϫ

ϩ

CR mass spectra are insignificant in terms of the NIDD

scheme^[, suggesting that the neutral does not rearrange

25]

Experimental Results

extensively within the µs timescale as predicted above. Thus,

Ϫ

ϩ

Ϫ

ϩ

Ϫ

ϩ

•؊

1

•؉

NR , CR , and NIDD methods have been applied the equilibrium structures of 1 , 1, and 1 are similar to

•

؊

to the radical-anions of acrolein 1 , acetyl and formyl each other, such that the vertical electron-transfer processes

•

؊

•؊

methyl carbenes 11 and 12 , methoxy vinylidene 13

,

are expected to proceed without imparting large amounts

and oxyallyl 14^•. It turns out that these radical anions are of internal energy into the neutral or the cation for the ma-

؊

3

distinguishable from each other by some structurally indica- jority of these species. Formation of the triplet neutral 1

1000

Eur. J. Org. Chem. 1998, 987Ϫ1009

3 4

Oxyallyl and Other [C ,H ,O] Neutrals

FULL PAPER

Ϫ

ϩ

can rigorously be excluded due to the large singlet-triplet Figure 8. CR (O

2

, 80% T) mass spectrum of [1-D

1

]acetyl carbene

radical-anions 11a generated by NICI of a mixture of [1,1,1-

]acetone and N

•

؊

splitting. As a consequence, fragmentations as well as re-

arrangements are minor for acrolein, accounting for the in-

tense recovery signals observed in both spectra. Among the

D

3

2

O

fragment ions, those arising from the energetically and en-

ϩ

tropically favored α-cleavages prevail, i.e. H CCHCO ,

2

ϩ

HCO , and C H . Though low in energy, the entropically

disfavored rearrangement to 18 , which subsequently de-

composes to C H₄ϩ CO, can only compete to a smaller

2

3

•

؉

•

ϩ

2

extent. Overall, acrolein represents a system which is stable

in the anionic, neutral and cationic forms and we consider

1

as a reference for the other [C ,H ,O] isomers.

3 4

Ϫ

ϩ

2 2

Figure 7. (a) NR mass spectrum (O , 80% T, O , 80% T), (b)

Ϫ ϩ

Ϫ

ϩ

CR mass spectrum (O

acetyl carbene radical-anions 11 generated by NICI of a mixture

of diazoacetone and N

2

, 80% T), and (c) NIDD spectrum of

•؊

2

O

for 11^•which exhibits characteristic methyl losses in its

؊

[50]

MI and CA spectra . As far as connectivity is concerned,

Ϫ

ϩ

the CR mass spectrum (Figure 7b) shows losses of CH

ϩ

and CH fragments, leading to CH CO and HCCO cat-

3

ions, respectively. In particular, the acetyl ion is character-

•

؊

istic for 11 as it involves loss of the high-energy methin

unit and is only observed as a major fragment ion for this

Ϫ

ϩ

ion. However, the CR spectrum contains some frag-

ϩ

ments, e.g. the reasonably large HCO signal, which imply

the occurrence of some rearrangements (see below).

In order to further cross-check the structural assignment

and also to study possible intramolecular rearrangements,

we took advantage of the following procedure. It is well

known^[

50][68]

that the reaction of acetone with O yields a

•Ϫ

•

؊

ca. 1:1 mixture of acetyl carbene 11 and oxyallyl radical-

•

؊

anions 14

via competing 1,1- and 1,3-abstractions of

H₂” from acetone. Thus, acetyl carbene radical-anion

can be made by NICI of [1,1,1-D ]acetone with N O as

•

ϩ

“

3

2

•

Ϫ

reagent gas. Reaction of O should accordingly lead to the

,1-abstractions of “D₂” and “H₂” leading to the D₁-

and D -labeled ions 11a and 11b , respectively. Due to

the labeling, the 1,3-abstraction of “HD ” to yield [1,1-

D ]oxyallyl, 14a , does not interfere any more with the

,1-abstraction products. While loss of “H₂” yields 11b

•

ϩ

•ϩ

1

•

؊

•؊

3

•

ϩ

•

؊

2

•

ϩ

•؊

1

which is isobaric with the D -labeled acetone enolate, “D

3

•

ϩ

•؊

”

loss should lead to pure 11a . Except for the label,

the corresponding CR spectrum (Figure 8) is indeed in

2

Ϫ

ϩ

complete agreement with that of the unlabeled ion gener-

ated from diazoacetone/N O. Thus, the expulsions of CD

2

ϩ

and CH moieties lead to intense signals due to CH CO

3

ϩ

and DCCO , respectively, while loss of CH is practically

absent. In summary, we conclude that both precursors lead

to ion beams consisting of the genuine radical anions of

acetyl carbene.

In agreement with the predictions derived from the theo-

Ϫ

ϩ

retical results (see above), the NR spectrum (Figure 7a)

•

؊

Ϫ

ϩ

of 11 differs completely from the corresponding CR

spectrum. While the methyl loss corresponds to the base

Acetyl Carbene Radical-Anion

The generation of a pure ion beam of 11^•is a straight- peak in the CR spectrum, the most abundant signal in

؊

Ϫ

ϩ

Ϫ

ϩ

•ϩ

forward matter if NICI of diazoacetone with N O as a re- the NR spectrum is due to CO /C H₄. Further, the

2

ϩ

Ϫ

ϩ

agent gas is applied. Electron capture of the diazo com- HCO signal becomes less intense in the NR spectrum

Ϫ

ϩ

pound followed by loss of N gives rise to an intense peak as compared to the CR experiment. Most significantly,

2

Eur. J. Org. Chem. 1998, 987Ϫ1009

1001

D. Schröder, H. Schwarz et al.

FULL PAPER

1

the abundant, and structurally indicative CH loss in the When formed from 11 via TS 2/ 11, neutral 2 contains an

CR spectrum vanishes almost completely, if the NR

Ϫ

ϩ

Ϫ

ϩ

excess energy of about 70 kcal/mol which is stored in vi-

procedure is applied. These findings indicate that the neu- brations and rotations. This amount of energy not only ac-

Ϫ

ϩ

tral carbene initially formed upon collisional electron de- counts for the weak recovery signal in the NR spectrum,

1

tachment of its radical anion is no longer present in the but it may also suffice for a partial isomerization of 2 to

Ϫ

1

neutral beam. This conclusion is confirmed by the NR

1, which must, nevertheless, occur to a minor extent due

•

؊

Ϫ

ϩ

spectrum of 11 , in which HCCO represents the base to the much more intense recovery signals in the CR and

peak while the recovery signal for 11 hardly exceeds the

signal-to-noise ratio. Due to the positive electron affinities

of carbenes, one would instead expect a fair recovery signal non-fragmenting methyl ketene or to the triplet species 11

for acetyl carbene if reionized to anions . The qualitative (see below). As mentioned above, the HCO fragment indi-

differences of the NR and CR spectra are expressed cates some type of structural rearrangement which is, how-

•

؊

Ϫ

ϩ

•؊

NR spectra of 1 . Indeed, the recovery signal in the

Ϫ

ϩ

•؊

NR spectrum of 11 could also be completely due to

3

[4]

ϩ

Ϫ

ϩ

Ϫ

ϩ

Ϫ

ϩ

1

quantitatively in the NIDD spectrum (Figure 7c). Thus, ever, not accounted for by the formation of 2. The obser-

•

ϩ

Ϫ

ϩ

the structurally indicative peaks, i.e. the losses of CH and vation of HCO as a negative signal in the NIDD spec-

•

ϩ

•ϩ

•؊

CH , are negative, while the H CCHCO and C H₄

/

trum of 11 points towards an origin from ionic species.

3

2

CO^•appear as intense positive peaks. In conjunction with Indeed, formation of HCO can be rationalized by iso-

the weak recovery signal, the NIDD pattern very much sug- merization to methyl oxirene radical-cations 9^•^؉upon

gests the occurrence of an isomerization at the neutral stage. charge inversion, because acetyl carbene is not a minimum

ϩ

Scheme 3

[

4]

•؉

As outlined above, the theoretical treatment suggests re- on the cation surface . Subsequently, 9 undergoes frag-

1

ϩ

arrangement of singlet acetyl carbene 11 to methyl ketene mentation to afford HCO . This interpretation is not only

(Scheme 3). Single-point calculations reveal an internal in line with the negative NIDD signal for HCO , but

energy content of at least 19 kcal/mol for 11 if formed also accounts for the shift of this fragment to DCO for

upon vertical electron detachment of the radical anion. Ac- the D -labeled ion 11a (see Figure 8).

The electronic ground state of acetyl carbene is, however,

1 can easily be surmounted, and formation of 2 at the predicted to be a triplet, and there is no reason why 11

1

Ϫ

ϩ

2

1

ϩ

•

؊

1

cordingly, the barrier associated with methyl shift via TS 2/

1

3

1

neutral stage can account for the experimental findings. should not be formed upon electron detachment. The intact

Upon reionization methyl ketene gives rise to losses of carbene structure would, however, (i) yield recovery signals

Ϫ

ϩ

atomic hydrogen atom as well as carbon monoxide. The in NR and NR experiments, and (ii) lead to some loss

Ϫ

ϩ

corresponding fragments give rise to positive signals in the of CH in the NR experiment. These criteria are not met

Ϫ

ϩ

1

NIDD spectrum and identify the 11 Ǟ 2 rearrange- experimentally, and in accord with the theoretical predic-

3

ment as occurring at the neutral stage. The fact that the tion we conclude that 11 is unstable on the microsecond

fragmentations occur after reionization is not at all an ob- time-scale of the experiment, presumably due to crossing

jection against the NIDD method, because these dis- from the triplet to the singlet state. In fact, a single point

1

3

sociations can only proceed, if the 11 Ǟ 2 rearrangement calculation reveals that 11 is formed with an excess energy

occurs at the neutral stage of the NR experiment^[

25a][25c]

.

of ca. 9 kcal/mol upon vertical electron detachment from

1002

Eur. J. Org. Chem. 1998, 987Ϫ1009

3 4

Oxyallyl and Other [C ,H ,O] Neutrals

FULL PAPER

1

1^•such that the neutral may overcome the MECP to the the absence of some fragments which are observed for the

؊

3

1

singlet surface. Overall, surface hopping from 11 to 11 other radical anions under study, i.e. the losses of CH and

1

seems to occur easily followed by the isomerization 11 Ǟ CH in particular, confirms this structural assignment.

2

1

Ϫ

ϩ

Ϫ

ϩ

•؊

2

, such that neither state of 11 exhibits a lifetime sufficient

The CR and NR spectra of 12 (Figure 9) show

ϩ

for its unequivocal detection by NRMS.

some pronounced differences. In particular, the HCO sig-

nal becomes much smaller in the NR experiment, while the

Ϫ

ϩ

Figure 9. (a) NR mass spectrum (O

2

, 80% T, O

2

, 80% T), (b)

•

Ϫ

ϩ

Ϫ

ϩ

losses of H , CH , and C H /CO, respectively, gain in

3 2 4

CR mass spectrum (O

formyl methyl carbene radical-anions 12 generated by by NICI intensity. The NIDD spectrum (Figure 9c) provides some

of a mixture of [2,2-D ]-propanal and N

2

, 80% T), and (c) NIDD spectrum of

•؊

Ϫ

ϩ

2

O

quantitative information in this respect. Thus, the dominant

ϩ•

ϩ

positive signal is due to C H₄/CO , while HCO pre-

2

vails on the negative scale. In analogy to acetyl carbene (see

above), we conclude that formyl methyl carbene formed

upon vertical electron detachment undergoes the rearrange-

1

ment 12 Ǟ 2 at the neutral stage and further assume that

3

1

the spin inversion 12 Ǟ 12 is efficient. Thus, the carbenes

1 and 12, while being separated on the anion surface, are

1

predicted to by and large collapse to the same neutral upon

electron detachment (Scheme 3). This conjecture is fully in

Ϫ

ϩ

Ϫ

ϩ

line with the comparison of the CR and NR spectra

Ϫ

ϩ

of the carbenes (Figures 7 and 9). Thus, the CR spectra

of 11 and 12 are completely different from each other,

while the corresponding NR spectra are rather similar,

though not identical . The slight difference of the two

NR spectra may be attributed to the amount of internal

energy stored in 2 depending on its generation from 11

or 12 together with some minor contributions from other

•

؊

•؊

Ϫ

ϩ

[69]

Ϫ

ϩ

1

isomers, e.g. 1 or residual amounts of ЈcoldЈ triplet species

below the respective MECPs. In conclusion, the two keto

carbenes 11 and 12 represent well characterized, stable spe-

cies as radical anions, but neither spin state of the neutral

carbenes is long-lived enough to be detected by NRMS due

to the facility of spin-inversion as well as rearrangement at

the neutral stage.

Methoxy Vinylidene Radical-Anions

For the generation of 13^•, methyl vinyl ether was sub-

؊

jected to NICI with N O. This route follows earlier stud-

2

ies^[

“

51]

which have demonstrated that the abstraction of

•

ϩ

•Ϫ

H₂” by O

proceeds as a 1,1-elimination from the

methylene group; the methyl group was found not to be

involved. While the presence of a methyl group is apparent

Ϫ

ϩ

from the CR mass spectrum (Figure 10b), we cannot rig-

orously exclude the presence of isomeric methoxy acetylene,

ϩ

because the structurally indicative C and CH fragments

are obscured by those arising from the methyl group. Not-

withstanding, we exclude formation of the radical anion 3^•

؊

Formyl Methyl Carbene Radical-Anions

[9b][70]

due to the negative electron affinities of acetylenes

.

•

؊

•؊

A pure ion beam of 12 can be generated by reacting Formation of 13 is also suggested by the CA mass spec-

2,2-D ]propanal with O , leading to loss of D O con- trum in which the three most intense peaks correspond to

•

Ϫ

[

2

•

؊

Ϫ

ϩ

Ϫ

•Ϫ

Ϫ

comitant with formation of unlabeled 12 . The CR

HCC (100%), H CC (60%), and CH O (10%). These

2 3

mass spectrum (Figure 9b) exhibits some signals which are signals can be attributed to the initial formation of an

•

Ϫ

indicative for the anticipated anion structure. The intense [HCC/CH O] ion/dipole complex by CϪO bond cleavage

3

ϩ

•؊

•

HCO and C H₃signals point to the occurrence of an α- of 13 . The different electron affinities of HCC (2.97 eV)

2

•

cleavage of the CϪC bond adjacent to the carbonyl group and CH O (1.57 eV) are reflected in the intensities of the

3

upon charge inversion, and also hydrogen-atom loss can be corresponding fragment ions. In addition, the [HCC/

•

Ϫ

attributed α-CϪH bond cleavage. Further, the presence of CH O] ion/dipole complex can undergo hydrogen trans-

3

weak signals due to methyl loss and methyl cation reveal fer followed by the loss of formaldehyde in which the

the presence of methyl as a structural subunit. Moreover, anionic product must be the radical anion of vinylidene^[

9b]

Eur. J. Org. Chem. 1998, 987Ϫ1009

1003

D. Schröder, H. Schwarz et al.

FULL PAPER

Ϫ

ϩ

Figure 10. (a) NR mass spectrum (O

2

, 80% T, O

2

, 80% T), (b) appear on the negative scale. Consequently, the reactivity

Ϫ

ϩ

Ϫ

ϩ

CR mass spectrum (O

methoxy vinylidene radical-anions 13 generated by by NICI of a

mixture of methyl vinyl ether and N

2

, 80% T), and (c) NIDD spectrum of

1

of neutral 13 is characterized by an intense methyl loss.

•

؊

According to the NIDD scheme^[, such a situation im-

plies that a substantial fraction of the neutrals formed upon

electron detachment of the radical anion undergoes dis-

sociation into a specific channel, i.e. methyl loss, rather than

structural rearrangements etc.

25b]

2

O

These result are in pleasing agreement with the theoreti-

3

cal predictions. Formation of 13 can be excluded due to

1

the size of the singlet-triplet gap, and 13 is expected to

1

rearrange easily into methoxy acetylene 3 which is more

stable by ca. 47 kcal/mol. Assuming that the barrier con-

1

[9] 1

necting 13 with 3 is in the range of a few kcal/mol , 3

is generated with an internal energy of about 50 kcal/mol.

This amount of energy is more than sufficient to promote

fragmentation which has an energy demand of ca. 40 kcal/

1

•

mol relative to 3 to afford CH₃ϩ HCCO radicals. Sub-

sequently, these neutral fragments are reionized giving rise

ϩ

Ϫ

ϩ

to the intense HCCO signal in the NR spectrum. It

ϩ

should be noted that the CH₃signal is expected to appear

on the positive scale of the NIDD spectrum like the

HCCO signal, but yet the CH₃peak is negative. As dis-

Ϫ

ϩ

[

72]

cussed previously for a different example , this apparent

failure of NIDD to detect both neutrals formed as positive

signals can be traced back to differences in collection ef-

ficiencies of the complementary fragments. Thus, two-elec-

Ϫ

ϩ

tron transfer in the CR experiment is likely to afford

ϩ

comparable amounts of CH₃and HCCO fragments, due

to the similar ionization energies of CH₃(9.8 eV) and

HCCO (9.9 eV). Instead, reionization of the neutral frag-

ments is much more efficient for HCCO as compared to

•

the methyl radical which has only about third the mass and

kinetic energy. Thus, methyl is much more discriminated in

Ϫ

ϩ

Ϫ

ϩ

NR than in CR , such that the latter overcompensates

ϩ

the NR component and CH appears as a negative signal

in the CR spectrum.

The question remains whether the existence of 13 can be

3

Ϫ

ϩ

1

Ϫ

ϩ

deduced from the experimental data alone. In the NR

spectrum, a recovery signal is observed which is, unfortu-

nately, too weak for further characterization aimed at pro-

rather than that of acetylene. The presence of a methoxy bing the presence of the vinylidene structure in the reion-

group in the ion formed upon NICI of methyl vinyl ether ized species^[. The fragmentation pattern in the NR

10b]

Ϫ

ϩ

with N O is further confirmed by the characteristic pattern spectrum is also in accord with structure 13 and separates

2

Ϫ

ϩ

observed in the CR mass spectrum. Thus, a minor signal this from the other isomers except for a clear-cut distinction

ϩ

for [C,H ,O] is accompanied by an intense HCO and two from isomeric 3 which definitely cannot be made using the

3

•

ϩ

•ϩ

smaller CH O and CO signals. This pattern is typical experimental data. Single-point calculations indicate that a

2

for the methoxy cation more or less irrespective of the way few kcal/mol excess energy are deposited in the molecule

of its formation in CR, CR/CA, or CIDI experi- in the consecutive one-electron oxidations from the radical

[

25a][25c][63][71]

ments

. In conclusion, we assume that NICI of anion to the neutral and further to the radical cation, the

methyl vinyl ether with N O leads to a ion beam of gen- barrier associated with rearrangement to the acetylene can

2

uine 13^•

؊

.

be assumed to be rather low for the neutral and cationic

While the NR spectrum of 13^•^؊differs very much species. However, methyl loss from 3 should proceed rather

from the CR spectrum, the fragmentation pattern is in- fast due to the internal energy imparted in 3, if formed

deed quite similar except that in Figure 10a the HCCO

Ϫ

ϩ

1

Ϫ

ϩ

1

ϩ

1

from 13 (see above). Nevertheless, we cannot provide any

signal is drastically increased at the expense of all other definitive evidence to decide whether the weak recovery sig-

Ϫ

ϩ

1

signals. This is also reflected in the NIDD spectrum nal is due to genuine 13 or to small amounts of “hot”

(

ϩ

1

Figure 10c) in which the HCCO is the only significant

3 which survive the passage from the first to the second

signal with a positive intensity while no pronounced peaks collision cell.

1004

Eur. J. Org. Chem. 1998, 987Ϫ1009

Oxyallyl and Other [C

3 4

,H ,O] Neutrals

FULL PAPER

Ϫ

ϩ

Oxyallyl Radical-Anions

Figure 12. (a) NR mass spectrum (O , 80% T, O , 80% T), (b)

2

Ϫ

ϩ

Ϫ

ϩ

CR mass spectrum (O

1,1-D ]oxyallyl radical-anions 14a generated by by NICI of a

mixture of [1,1,1-D ]acetone and N

2

, 80% T), and (c) NIDD spectrum of

•

Ϫ

The reaction of acetone with O is known to yield a

mixture of acetyl carbene 11 and oxyallyl 14 radical-

anions via formal 1,1- and 1,3-abstractions of “H₂

•؊

[

2

•

؊

•؊

3

2

O

•

ϩ [50]

”

.

Therefore, other ways of generating a pure beam of oxyallyl

radical-anions 14^•had to be explored. One possible way

is the complete desilylation of 1,3-bis(trimethylsilyl)acetone

by NICI in a NF /F plasma. In analogy to the generation

؊

3

2

of TMM radical-anion^[

29b][29c]

, this process can be assumed

•

Ϫ

to proceed very specifically to yield pure 14 . In fact, the

Ϫ

ϩ

CR mass spectrum (Figure 11) reveals loss of CH con-

2

comitant with the generation of ketene radical-cations as

•

؊

the most abundant process. Contribution of 11 to the ion

beam can be ruled out, due to the practical absence of the

characteristic loss of CH in Figure 11. Unfortunately, inten-

sity of oxyallyl radical-anions achieved upon desilylation of

bis(trimethylsilyl)acetone was not sufficient for reliable NR

Ϫ

ϩ

experiments and a NIDD analysis. Nevertheless, the CR

•

؊

spectrum of unlabeled 14 shall serve as a reference in the

following discussion.

Ϫ

ϩ

Figure 11. CR (O

2

, 80% T) mass spectrum of oxyallyl radical-

anions 14 generated by NICI of a mixture of 1,3-bis(trimethylsi-

lyl)acetone, NF , and F . Due to low intensities, the spectrum was

•

؊

3

2

recorded using B(1) only for mass selection followed by charge in-

version in the subsequent field-free region. As the products were

monitored with E(1), the peaks are broader as compared to the

other spectra, which were obtained under double-focussing condi-

tions.

The further arguments are derived from experiments car-

•

؊

ried out with [1,1-D ]oxyallyl anions 14a generated by are still present as distinct signals. Vice versa, the peaks due

2

•

ϩ

•ϩ

NICI of [1,1,1-D ]acetone with N O as reagent gas (see to CO and C H D

above). Except for the mass shifts due to labeling, the two fragments appear as positive signals in the NIDD

CR spectrum of 14a (Figure 12b) is in good agreement spectrum (Figure 12c), indicating the formation of CO and

with that of the unlabeled oxyallyl. Thus, CH and CD₂C H D at the neutral stage of the experiment . Accord-

increase. Consequently, the latter

3

2

Ϫ

ϩ

Ϫ

ϩ

•؊

[73]

2

ϩ•

losses are observed as abundant processes, while no signal ingly, H CCO and D CCO give rise to large negative

2

due to the expulsion of CH is observed. The oxyallyl struc- peaks suggesting that loss of methylene occurs after reioni-

ture is further confirmed to be intact by the fact that the zation to the cation radical. Nevertheless, it is notable that

Ϫ

ϩ

signal for CHD loss is marginal and, thus, no abundant H/ loss of methylene is observed at all in the NR spectrum

D exchange is observed between the two methylene groups. (see below). Moreover a decent recovery signal is observed

In conclusion, 14^•and 14a are well characterized with in the NR spectrum of 14a , and also the NR spec-

؊

•؊

Ϫ

ϩ

•؊

Ϫ

•

؊

respect to their ion structures and can be examined by NR trum of 14a exhibits a signal due to the reionized parent

experiments. In line with the theoretical predictions both ion.

anions do not show significant metastable ion decompo-

sitions.

The theoretical analysis presented above suggests that

•

؊

1

electron detachment from 14 leads to both the singlet A₁

Ϫ

ϩ

•؊

3

As predicted by theory, the NR spectrum of 14a

and triplet B states of oxyallyl. Under the experimental

2

Ϫ

ϩ

(

Figure 12a) differs very much from the CR spectrum. conditions, ¹14 is predicted to immediately collapse to

1

Thus, the losses of CH and CD decrease in intensity, but cyclopropanone 8. Due to the excess energy of about 30

2

Eur. J. Org. Chem. 1998, 987Ϫ1009

1005

D. Schröder, H. Schwarz et al.

FULL PAPER

Scheme 4

1

kcal/mol deliberated in the rearrangement 14 Ǟ 8, a sub- detachment from the radical anion could provide an

1

stantial fraction of the neutral 8 is expected to reach the alternative for the methylene losses. Nevertheless, there ex-

3

CO ϩ C H exit channel. These fragments are subsequently ists no reason why 14 should not be formed in the neu-

2

4

reionized and thus show up as positive peaks in the tralization step and this would Ϫ if spin inversion occurs at

Ϫ

ϩ

1

NIDD spectrum. Notwithstanding, 8 cannot be the all Ϫ rearrange to 8 rather than 7. In summary, the CR

Ϫ

ϩ

•؊

only neutral species formed in the NR experiment, be- and NR spectra of 14 can be rationalized by a super-

cause neither the present calculations nor earlier experimen- position of fragments due to the formation of neutral sin-

[

2e]

1

3

tal data

can explain the distinct loss of methylene upon glet A and triplet B oxyallyl, but other explanations are

1 2

1

reionization by involving 8 alone. Triplet oxyallyl, however, possible and cannot rigorously be excluded.

can easily account for the fragmentation into CH

H CCO . Theory predicts (i) formation of 14 as the

2

ϩ

2

•

ϩ

3

Conclusions

ground state neutral upon electron detachment from the

This study underlines the conceptual advantage of using

anion, (ii) inefficient intersystem crossing to the singlet sur- anionic precursors in probing the structure and reactivity of

SO

face due to the small magnitude of H , and (iii) reioni- elusive neutrals by mass spectrometric means. In particular,

3

zation of 14 to the cation radical of allene oxide for which rearrangements are much less frequent for anions than for

[2e]

loss of methylene is characteristic . In summary, the cations while neutralization of anions, viz. electron detach-

Ϫ

ϩ

Ϫ

ϩ

•؊

NR and NIDD spectra of 14a are in agreement ment, is more facile than neutralization of cations. This ap-

with the prediction that these represent superpositions of proach has enabled the identification of the radical anions

components arising from singlet and triplet oxyallyl of acrolein, acetyl carbene, formyl methyl carbene, methoxy

(Scheme 4).

vinylidene, and oxyallyl. Further, the recently developed

It is therefore tempting to attribute the recovery signals NIDD approach allows for the examination of the unimole-

3

in the NR spectra to the formation of B oxyallyl which cular reactivity of the neutral species generated in NR

2

survives the NR procedure intact. Unfortunately, however, experiments. Thus, evidence is presented for the rearrange-

detailed consideration of the respective potential-energy ment of acetyl and formyl methyl carbenes to methyl ke-

surfaces reveals that none of the recovery signals provides tene. The rearrangement of methoxy vinylidene to methoxy

3

unambiguous evidence for intermediate 14. In particular, acetylene and the loss of CO from singlet oxyallyl via cyclo-

the depth of the potential-energy well of neutral cyclopro- propanone as an intermediate constitute Ϫ although pre-

panone implies that asides fragmentation via the CO ϩ dicted by theory Ϫ hitherto unknown reactions.

1

C H exit channel, some 8 may remain intact. Therefore,

The present results also pose some questions to the NR

2

4

1

neutral 8 could account for the recovery signals observed methodology. The mere observation of a recovery signal has

in both types of NR experiments, because it is expected to often been considered as significant evidence for the exist-

undergo reionization to the cation and anion surfaces, i.e. ence of the neutral under study. The present study reveals,

1

•؉ 1 •؉

•؉

1

•؊

8

Ǟ TS18 / 18 Ǟ 18 and 8 Ǟ 8 , respectively; the however, that the assignment of the neutral structure can

latter process would involve formation of a metastable be rather difficult, if a further characterization of the sur-

anion, but due to its non-planar geometry (see above), it vivor ions cannot be made. For example, there is no doubt

1

may be accessible from excited 8 and electron detachment that acrolein exists as a neutral, but for the acetyl and for-

may be prevented by a kinetic barrier. The losses of CH

myl-methyl carbenes it is quite certain that the survivor ions

do not correspond to the reionized carbenes, but rather to

2

•؊

Ϫ

ϩ

and CD observed in the NR mass spectrum of 14a

2

1

cannot arise from neutral 8a, and these signals provide ionized methyl ketene formed via isomerization at the neu-

strong, but indirect evidence for the formation of long-lived, tral stage. Similarly, the survivor ions observed for methoxy

3

1

neutral 14. However, formation of neutral 7 upon electron vinylidene anion-radicals cannot provide direct evidence for

1006

Eur. J. Org. Chem. 1998, 987Ϫ1009

3 4

Oxyallyl and Other [C ,H ,O] Neutrals

FULL PAPER

the existence of the corresponding neutral, and also the between the two collision chambers. Subsequent reionization to

3

cations occurred in the second cell by collision with oxygen (80%

T). The resulting mass spectra were recorded by scanning B(2).

possible existence of B oxyallyl can only be deduced from

2

a combination of theory and experiment. Thus, just as the

absence of a recovery signal does not necessarily mean that

the neutral under study is non-existent, the observation of a

recovery signal does not prove the existence of the neutral,

because it may be due to rearranged species.

Finally, it should be pointed out that the analysis of NR

and CR spectra benefits much from theoretical models of

the potential-energy surfaces involved. In fact, some exper-

Charge-reversal mass spectra^[of the anions to cations ( CR )

were obtained by colliding the ion beam with oxygen (80% T) in

the field-free region preceding B(2). Under these conditions, the

79]

Ϫ

ϩ

Ϫ

ϩ

CR process can be treated as a vertical, two-electron oxidation

occurring in a single step, although some species may undergo mul-

Ϫ

ϩ

tiple collisions. For direct comparison with the NR spectra, the

Ϫ

ϩ

Ϫ

ϩ

CR experiments were also recorded under NR conditions (i.e.

Ϫ

ϩ

2 2

O /O , each 80% T), but no severe differences to the CR spectra

imental data can only be interpreted by explicit reference to obtained under single-collision conditions were found for the sys-

tems under study. The spectra were accumulated and on-line pro-

cessed with the AMD-Intectra data system; usually, 10 to 50 scans

were averaged to improve the signal-to-noise-ratio.

the theoretical predictions, e.g. spin inversions. For the bulk

of our calculations, the B3LYP/6-311ϩϩG(d,p) level of the-

ory was chosen on pragmatic grounds due to its moderate

expense. This method is, however, not appropriate for all

Ϫ

NIDD spectra^[²⁵^]^[²⁶^]were obtained by subtracting the nor-

ϩ

Ϫ

ϩ

malized peak heights of the CR spectra from those of the corre-

parts of the potential-energy surface. In particular, the sin-

Ϫ

ϩ

1

sponding NR spectra according to equation (1).

(NIDD) ϭ [I (NR) / Σ (NR)] Ϫ [I (CR) / Σ

In this normalization scheme positive intensities in the difference

glet biradicaloid A state of oxyallyl is not accurately de-

1

scribed with B3LYP as demonstrated by the rather different

results obtained using other methods. In general, com-

plementary calculations are therefore recommended in any

study where the performance of DFT is either unknown or

expected to be poor.

I

i

I

i

I

i

(CR)]

(1)

spectra indicate products whose formation involves either de-

compositions or rearrangements of the intermediate neutral spe-

cies^[, while decompositions of the projectile and/or recovery ions

which do not involve processes at the neutral stage appear as nega-

25c]

Financial support by the Deutsche Forschungsgemeinschaft, the

Fonds der Chemischen Industrie, and the Volkswagen Stiftung is

acknowledged. C. A. S. thanks Dr. M. Aschi for an introduction

into the methods for determining minimal-energy crossing points.

J. N. H. is grateful to the Alexander von Humboldt-Stiftung for a

research scholarship. We thank M. Schmidt and A. Granovsky for

providing us with a copy of Gamess for PC and are indebted to T.

Steinke for the generous allocation of computer time on the Cray

J932 at the Konrad-Zuse-Institut, Berlin, as well as excellent ser-

vice. Finally, we appreciate helpful discussion and comments by W.

T. Borden, W. Sander, and R. R. Squires.

Ϫ

ϩ

tive peaks in the NIDD spectra.

Diazoacetone was synthesized from acetyl chloride and diazo-

[80]

methane as described in the literature . 1,3-Bis(trimethylsilyl)ace-

tone was obtained by oxidation of the corresponding alcohol with

[81]

Ag

CD

2

CO

3

/celite . [1,1,1-D

3

]acetone was prepared by reaction of

3

MgI with acetaldehyde followed by Jones oxidation of the al-

cohol formed. Immediate distillation and usage of the product is

recommended in order to obtain highest isotopic purities; other-

wise the deuterium label may undergo complete intermolecular

equilibration within a few hours or days depending on the purity

Experimental Section

The experiments were performed with a modified VG ZAB/HF/

AMD four-sector mass spectrometer of BEBE configuration (B

of the sample. [2,2-D

2

]propanal was made by repeated isotopic ex-

O in refluxing pyridine.

O-labeled acetone was generated by hydrolysis of the correspond-

changes of the unlabled aldehyde with D

2

1

8

stands for magnetic and E for electric sectors) which has been de- ing cyclohexylimine^[with H

82]

18

O in the inlet system of the mass

2

scribed in detail previously^[. In brief, the anions of interest were

generated by negative ion chemical ionization (NICI, repeller volt-

age ca. 0 V, electron energy 50Ϫ100 eV) of mixtures of the corre-

74]

spectrometer. All other substances were commercially available and

used without further purification.

sponding precursors with N

of oxyallyl radical-anions, 1,3-bis(trimethylsilyl) acetone was sub-

jected to NICI with mixture of NF

fluorine was generated inside the ion source by thermolysis of sil-

ver(II) fluoride.

2

O. For the generation of a pure beam

Ƞ

Dedicated to Professor M. H. Zenk on the occasion of his

6

5th birthday.

[29b][29c][75]

[1]

[1a]

and F

2

; elemental

For 1,3-hydrogen shifts in neutral and cationic enols, see:

G. Frenking, N. Heinrich, J. Schmidt, H. Schwarz, Z. Natur-

3

[

1b]

forsch. 1982, 37b, 1597. Ϫ

Y. Apeloig, B. Ciommer, G.

Depke, G. Frenking, M. Karni, S. Meyn, J. Schmidt, H.

[

1c]

Schwarz, Int. J. Mass Spectrom. Ion Proc. 1983, 59, 21. Ϫ

The ions were accelerated to 8 keV translational energy and

mass-selected by means of B(1)/E(1) at mass resolutions of

m/∆m ϭ 2000Ϫ5000. Unimolecular decompositions of metastable

ions (MI) occurring in the field-free section preceding B(2) were

recorded with this sector. For collisional activation (CA) experi-

[1d]

H. Schwarz, Adv. Mass Spectrom. 1985, 13. Ϫ

G. Frenking,

N. Heinrich, W. Koch, H. Schwarz, J. Am. Chem. Soc. 1986,

[1e]

1

08, 593. Ϫ

N. Heinrich, H. Schwarz in Ion and Cluster-Ion

Spectroscopy and Structure (Ed.: J. P. Maier), Elsevier, Amster-

dam, 1989.

´

[

2] [2a]

[

76]

G. LЈabbe, Angew. Chem. 1980, 92, 277; Angew. Chem., Int.

ments , the ions were collided in the field-free region between

[2b]

Ed. Engl. 1980, 19, 276. Ϫ

E. N. Marvell, Thermal Electro-

E(1)/B(2) with helium at 80% transmission (T) of the incident

[2c]

cyclic Reactions, Academic Press, New York, 1980. Ϫ

H. H.

beam; these conditions approximate single-collision conditions^[

77]

.

Wassermann, D. R. Berdahl, T.-J. Lu in The Chemistry of the

Cyclopropyl Group (Ed.: Z. Rappoport), Wiley, Chichester,

Ϫ

ϩ [78]

In neutralization-reionization ( NR )

experiments, the rad-

[2d]

1

987. Ϫ

F. Turecek, D. E. Drinkwater, F. W. McLafferty, J.

[2e]

ical anions were neutralized by high-energy collisions with molecu-

lar oxygen (80% T) in the first of two differentially pumped colli-

sion cells located in the field-free region between E(1) and B(2).

Unreacted ions were deflected away from the beam of neutral spe-

cies by applying a voltage of 1 kV on a deflector electrode located

Am. Chem. Soc. 1990, 112, 5892. Ϫ

F. Turecek, D. E. Drink-

water, F. W. McLafferty, J. Am. Chem. Soc. 1991, 113, 5950. Ϫ

[2f]

M. Nakata, H. Frei, J. Am. Chem. Soc. 1992, 114, 1363. Ϫ

W. Sander, A. Kirschfeld in Strain in Organic Chemistry, Vol.

[2g]

4

, JAI Press, 1995, p. 1.

[

3] [3a]

[3b]

N. J. Turro, Acc. Chem. Res. 1969, 2, 25. Ϫ

H. J. Rodri-

1007

Eur. J. Org. Chem. 1998, 987Ϫ1009

D. Schröder, H. Schwarz et al.

FULL PAPER

[

22] [22a]

[22b]

guez, J.-C. Chang, T. F. Thomas, J. Am. Chem. Soc. 1976, 98,

N. J. Turro, Pure Appl. Chem. 1971, 27, 679. Ϫ

H. M.

[

3c]

2

027. Ϫ

A. Sevin, E. Fazilleau, P. Chaquin, Tetrahedron

R. Hoffmann, Angew. Chem. 1973, 85, 877; Angew. Chem., Int.

[22c]

1

981, 22, 3831.

Ed. Engl. 1973, 12, 819. Ϫ

R. Noyori, Acc. Chem. Res. 1979,

[

4]

5]

[22d]

F. Turecek, D. E. Drinkwater, F. W. McLafferty, J. Am. Chem.

Soc. 1991, 113, 5958.

12, 61. Ϫ

C. J. Samuel, J. Chem. Soc., Perkin Trans. 2 1981,

[

22e]

736. Ϫ

A. P. Masters, M. Parvez, T. S. Sorensen, F. Sun, J.

[22f]

H. T. Yu, W. T. Chan, J. D. Goddard, J. Am. Chem. Soc. 1990,

Am. Chem. Soc. 1994, 116, 2804. Ϫ

Also see: R. Noyori,

1

12, 7529.

K. Yokoyama, Y. Hayakawa, J. Am. Chem. Soc. 1973, 95, 2723.

D. Lim, D. A. Hrovat, W. T. Borden, W. L. Jorgensen, J. Am.

6] [6a]

[23]

[24]

J. Fenwick, G. Fater, K. Ogi, O. P. Strausz, J. Am. Chem.

[6b]

Soc. 1973, 95, 124. Ϫ

A. Krantz, J. Chem. Soc., Chem. Com-

Chem. Soc. 1994, 116, 3494.

[6c]

[24a]

mun. 1973, 630. Ϫ

K.-P. Zeller, Liebigs Ann. Chem. 1979,

For reviews dealing with NR techniques, see:

C. Wesdemi-

[24b]

[

6d]

[6e]

2

036. Ϫ

K.-P. Zeller, Chem. Ber. 1979, 112, 678. Ϫ

G.

otis, F. W. McLafferty, Chem. Rev. 1987, 87, 485. Ϫ

J. K.

Maier, H. P. Reisenauer, T. Sayrac, Chem. Ber. 1982, 115, 2192.

Terlouw, H. Schwarz, Angew. Chem. 1987, 99, 829; Angew.

[

6f]

[24c]

Ϫ

G. Maas in Houben-Weyl, Methoden der organischen

Chem., Int. Ed. Engl. 1987, 26, 805. Ϫ

J. L. Holmes, Mass

[24d]

Chemie (Ed.: M. Regitz), Vol. E19b, Thieme, Stuttgart, 1989.

Spectrom. Rev. 1989, 8, 513. Ϫ

A. W. McMahon, S. K.

[

7] [7a]

[7b]

M. Berthelot, Bull. Soc. Chim. Fr. 1870, 14, 113. Ϫ

L.

M. Torres, E.

M. Lown, H. E. Gunning, O. P. Strausz, Pure Appl. Chem.

Chowdhury, A. G. Harrison, Org. Mass Spectrom. 1989, 24,

[7c]

[24e] [24f]

Wolff, Liebigs Ann. Chem. 1902, 325, 129. Ϫ

620. Ϫ

Turecek, Org. Mass Spectrom. 1992, 27, 1087. Ϫ

berg, H. Schwarz, Acc. Chem. Res. 1994, 27, 347. Ϫ

F. W. McLafferty, Science 1990, 247, 925. Ϫ

F.

[

24g]

N. Gold-

[

7d]

[24h]

1

980, 52, 1623. Ϫ

E. G. Lewars, Chem. Rev. 1983, 83, 519.

D. V.

8] [8a]

[8b]

P. J. Stang, Chem. Rev. 1978, 78, 383. Ϫ

H. F. Schaefer

Zagorevskii, J. L. Holmes, Mass Spectrom. Rev. 1994, 13, 133.

[8c]

[25] [25a]

III, Acc. Chem. Res. 1979, 12, 288. Ϫ

P. J. Stang, Acc. Chem.

P. J. Stang in Houben-Weyl, Methoden

C. A. Schalley, Gas-Phase Ion Chemistry of Peroxides, Dis-

sertation, Technische Universität Berlin D83, Shaker, Herzog-

[

8d]

Res. 1982, 15, 348. Ϫ

[25b]

der organischen Chemie (Ed.: M. Regitz), Vol. E19a, Thieme,

Stuttgart, 1989.

enrath/Germany, 1997. Ϫ

C. A. Schalley, G. Hornung, D.

Schröder, H. Schwarz, Int. J. Mass Spectrom. Ion Processes,

[

9]

[25c]

For calculations on the singlet and triplet vinylidene parent sys-

1998, 1721173, 181. Ϫ

G. Hornung, C. A. Schalley, M.

[

9a]

tems, see:

Y. Osamura, H. F. Schaefer III, S. K. Gray, W. H.

Dieterle, D. Schröder, H. Schwarz, Chem. Eur. J. 1997, 3, 1866.

For a review on methods for the investigation of the neutrals’

reactivity by NRMS, see: C. A. Schalley, G. Hornung, D.

Schröder, H. Schwarz, Chem. Soc. Rev., 1998, 27, 91.

[27] [27a]

[9b]

[26]

Miller, J. Am. Chem. Soc., 1981, 103, 1904. Ϫ

T. P. Hamilton, H. F. Schaefer III, J. Am. Chem. Soc. 1990, 112,

M. M. Gallo,

[

9c]

8

714. Ϫ

Jr., J. Am. Chem. Soc., 1992, 114, 6133. Ϫ

Thomas, B. J. DeLeeuw, Y. Yamaguchi, H. F. Schaefer III, J.

G. A. Peterson, T. G. Tensfeldt, J. A. Montgomery,

[9d]

G. Vacek, J. R.

C. Wesdemiotis, F. W. McLafferty, J. Am. Chem. Soc. 1987,

[27b]

109, 4760. Ϫ

R. Feng, C. Wesdemiotis, F. W. McLafferty,

[27c]

[9e]

Chem. Phys. 1993, 98, 4766. Ϫ

N. Y. Chang, M. Y. Shen,

J. Am. Chem. Soc. 1987, 109, 6521. Ϫ

C. Wesdemiotis, B.

[9f]

C.-h. Yu, J. Chem. Phys. 1997, 106, 3237. Ϫ

C.-h. Yu, Chem. Phys. Lett. 1997, 277, 245.

W.-C. Chen,

Leyh, A. Fura, F. W. McLafferty, J. Am. Chem. Soc. 1990, 112,

8655. Ϫ Ketocarbenes are even predicted to give rise to stable

[

10]

[10a]

[27d]

Only a few vinylidenes have been identified experimentally:

S. M. Burnett, A. E. Stevens, F. S. Feigerle, W. C. Lineberger,

anions in solution, see:

Res. 1988, 21, 400.

D. Bethell, V. D. Parker, Acc. Chem.

[

10b]

[28]

[28a]

Chem. Phys. Lett. 1983, 100, 124. Ϫ

Chem. Phys. Lett. 1989, 156, 397. Ϫ

D. Sülzle, H. Schwarz,

K. M. Ervin, J. Ho,

For reviews on distonic ions, see:

S. Hammerum, Mass

[10c]

[28b]

Spectrom. Rev. 1988, 7, 123. Ϫ

K. M. Stirk, L. K. M. Ki-

[10d]

[28c]

W. C. Lineberger, J. Chem. Phys. 1989, 91, 5974. Ϫ

M. K.

minkinen, H. I. Kenttämaa, Chem. Rev. 1992, 92, 1649. Ϫ

[28d]

Gilles, W. C. Lineberger, K. M. Ervin, J. Am. Chem. Soc. 1993,

H. I. Kenttämaa, Org. Mass. Spectrom. 1994, 29, 1. Ϫ

R.

[

10e]

1

12, 8714. Ϫ

J. Breidung, H. Bürger, C. Kötting, R. Ko-

L. Smith, P. K. Chou, H. I. Kenttämaa in The Structure, Ener-

getics and Dynamics of Organic Ions (Eds.: T. Baer, C. Y. Ng, I.

Powis), Wiley, New York, 1996.

[29] [29a]

pitzky, W. Sander, M. Senzlober, W. Thiel, H. Willner, Angew.

Chem. 1997, 109, 2072; Angew. Chem. Int. Ed. Engl. 1997, 36,

[

10f]

1983. Ϫ

C. Kötting, W. Sander, J. Breidung, W. Thiel, M.

J. Lee, P. K. Chou, P. Dowd, J. J. Grabowski, J. Am. Chem.

[29b]

Senzlober, H. Bürger, J. Am. Chem. Soc. 1998, 120, 219.

Soc. 1993, 115, 7902. Ϫ

P. G. Wenthold, J. Hu, R. R.

[29c]

[

11] [11a]

[11b]

P. J. Chenier, J. Chem. Educ. 1978, 55, 286. Ϫ

L. J.

Schaad, B. A. Hess, Jr., R. Zahradnık, J. Org. Chem. 1981,

6, 1909.

Squires, J. Am. Chem. Soc. 1994, 116, 6961. Ϫ

P. G. Went-

´

hold, R. R. Squires, J. Am. Chem. Soc. 1994, 116, 11890. Ϫ For

[29d]

4

a theoretical study of the dianions, see:

K. B. Wiberg, J.

12] [12a]

A. Liberles, A. Greenberg, A. Lesk, J. Am. Chem. Soc.

Am. Chem. Soc. 1990, 112, 4177.

[12b]

[30]

1

972, 94, 8685. Ϫ

A. Liberles, S. Kang, A. Greenberg, J.

D. Schröder, N. Goldberg, W. Zummack, H. Schwarz, J. C.

Poutsma, R. R. Squires, Int. J. Mass Spectrom. Ion Processes,

1997, 165/166, 71.

31] [31a]

Org. Chem. 1973, 38, 1922.

[

13]

D. A. Hrovat, A. Rauk, T. S. Sorensen, H. K. Powell, W. T.

Borden, J. Am. Chem. Soc. 1996, 118, 4159.

[

Density Functional Methods in Chemistry, (Eds.: J. Lab-

14] [14a]

[31b]

J. K. Crandall, W. H. Machleder, M. J. Thomas, J. Am.

anowski, J. Andzelm), Springer, Heidelberg, 1991. Ϫ

A. D.

[14b]

[31c]

Chem. Soc. 1968, 90, 7346. Ϫ

J. K. Crandall, W. H. Mach-

Becke, J. Chem. Phys. 1993, 98, 5648. Ϫ

P. J. Stevens, F. J.

Devlin, C. F. Chablowski, M. J. Frisch, J. Phys. Chem. 1994,

98, 11623.

[

14c]

leder, J. Am. Chem. Soc. 1968, 90, 7347. Ϫ

Org. Chem. 1983, 48, 4744.

J. V: Ortiz, J.

[

15] [15a]

[32]

M. H. J. Cordes, J. A. Berson, J. Am. Chem. Soc. 1996, 118,

GAUSSIAN94, Revision B 3: M. J. Frisch, G. W. Trucks, H. B.

Schlegel, P. M. W. Gill, B. G. Johnson, M. A. Robb, J. R. Chee-

seman, T. Keith, G. A. Peterson, J. A. Montgomery, K. Raghav-

achari, M. A. Al-Laham, V. G. Zakrzewski, J. V. Ortiz, J.B.

Foresman, C. Y. Peng, P.Y. Ayala, W. Chen, M. W. Wong, J. L.

Andres, E. S. Replogle, R. Gomperts, R. L. Martin, D. J. Fox,

J. S. Binkley, D. J. Defrees, J. Baker, J.P. Stewart, M. Head-

Gordon, C. Gonzales, J. A. Pople, GAUSSIAN Inc., Pittsburgh

PA, 1995.

33]

[

15b]

6

241. Ϫ

T. S. Sorensen, F. Sun, J. Am. Chem. Soc. 1997,

1

19, 11327.

16]

K. Schaffner, M. M. Demuth in Rearrangements in Ground and

Excited States (Ed.: P. deMayo), Vol. 3, Academic Press, New

York, 1980.

17] [17a]

C. Black, P. Lario, A. P. Masters, T. S. Sorensen, F. Sun, F.

[17b]

Can. J. Chem. 1993, 71, 1910. Ϫ

T. S. Sorensen, F. Sun,

Also, see: T. S. Sorensen, F.

[17c]

Can. J. Chem., 1996, 74, 79. Ϫ

[

Sun, J. Am. Chem. Soc. 1995, 117, 5592.

J. A. Berson in Diradicals (Ed.: W. T. Borden), Wiley, New

York, 1982.

R. Krishnan, J. S. Binkley, R. Seeger, J. A. Pople, J. Chem. Phys.

1980, 72, 650.

[

18]

34] [34a]

[34b]

K. Fukui, Acc. Chem. Res. 1981, 14, 363. Ϫ

C. Gonza-

19] [19a]

[34c]

H. C. Longuet-Higgins, J. Chem. Phys. 1950, 18, 265. Ϫ

W. T. Borden, I. Iwamura, J. A. Berson, Acc. Chem. Res.

lez, H. B. Schlegel, J. Chem. Phys. 1989, 90, 2154. Ϫ

Gonzalez, H. B. Schlegel, J. Phys. Chem. 1990, 94, 5523.

C.

[

19b]

[

35] [35a]

[35b]

1

994, 27, 109.

G. Rauhut, R. Pulay, J. Phys. Chem. 1995, 99, 3093. Ϫ

A. P. Scott, L. Radom, J. Phys. Chem. 1996, 100, 16502.

[

20] [20a]

[18]

[20b]

W. T. Borden in ref. . Ϫ

Y. Osamura, W. T. Borden,

36]

K. Morokuma, J. Am. Chem. Soc. 1984, 106, 5112.

P. M. W. Gill, B. G. Johnson, J. A. Pople, Chem. Phys. Lett.

1992, 197, 499.

21]

Substituted oxyallyls have been predicted by theory to have sin-

[21a]

[37] [37a]

glet ground states. For example, see:

M. B. Coolidge, K.

J. M. Galbraith, H. F. Schaefer III, J. Chem. Phys. 1996,

[37b]

Yamashita, K. Morokuma, W. T. Borden, J. Am. Chem. Soc.

105, 862. Ϫ

R. A. King, J. M. Galbraith, H. F. Schaefer

[37c]

[21b]

1

990, 112, 1751. Ϫ

A. S. Ichimura, P. M. Lahti, A. R.

III, J. Phys. Chem. 1996, 100, 6061. Ϫ

C. A. Schalley, D.

[21c]

Matlin, J. Am. Chem. Soc. 1990, 112, 2868. Ϫ

W. T. Borden, J. Org. Chem. 1995, 60, 2654.

H. K. Powell,

Schröder, H. Schwarz, K. Möbus, G. Boche, Chem. Ber./Recueil

1997, 130, 1085.

1008

Eur. J. Org. Chem. 1998, 987Ϫ1009

3 4

Oxyallyl and Other [C ,H ,O] Neutrals

FULL PAPER

[

38]

[58]

•؊

D. G. Leopold, K. K. Murray, A. E. S. Miller, W. C. Lineberger,

A distonic ion 18 could not be located on the anion surface,

and it is likely to undergo autodetachment in conjunction with

cyclization to neutral 8.

J. Chem. Phys. 1985, 83, 4849.

39] [39a]

L. A. Curtiss, K. Raghavachari, G. W. Trucks, J. A. Pople,

[39b]

[59]

[60]

J. Chem. Phys. 1991, 94, 7221. Ϫ

J. A. Pople, J. Chem. Phys. 1991, 95, 4040. Ϫ

K. Raghavachari, J. A. Pople, J. Chem. Phys. 1993, 98, 1293. Ϫ

L. A. Curtiss, L. D. Kock,

S. R. Langhoff, J. Chem. Phys. 1980, 73, 2379.

[39c]

L. A. Curtiss,

J. N. Harvey, D. Schröder, H. Schwarz, Bull. Soc. Chim. Belg.

1997, 106, 447.

ˇ

[

39d]

[61]

For a modification of the G2 approach using DFT, see: C.

D. Schröder, C. A. Schalley, J. Hrusak, N. Goldberg, H.

W. Bauschlicher, Jr., H. Partridge, J. Chem. Phys. 1995, 103,

788.

Schwarz, Chem. Eur. J. 1996, 2, 1235.

[62]

1

D. Schröder, J. N. Harvey, M. Aschi, H. Schwarz, J. Chem.

Phys., in press.

[63]

[

40]

If not stated otherwise, all thermochemical data have been

[

40a]

taken from:

974, 39, 123. Ϫ

J. F. Liebman, A. Greenberg, J. Org. Chem.

M. Aschi, J. N. Harvey, C. A. Schalley, D. Schröder, H.

Schwarz, J. Chem. Soc., Chem. Commun., 1998, 531.

[

40b]

1

S. G. Lias, J. E. Bartmess, J. F. Liebman,

[64]

3

1

J. L. Holmes, R. D. Levin, W. G. Mallard, Gas-Phase Ion and

Neutral Thermochemistry, J. Phys. Chem. Ref. Data, Suppl. 1,

Group symmetry rules predict the coupling of a B

2

and A

1

SO

state to be non-zero. In this case, the low value of the H is

not due to symmetry constraints, but due to the fact that the

[

40c]

1

988. Ϫ

S. G. Lias, J. F. Liebman, R. D. Levin, S. A. Kafafi,

1

3

NIST Standard Reference Database, Positive Ion Energetics, Ver-

1 2

orbital occupations of the A and B state are almost iden-

[40d]

sion 2.01, Gaithersburg, MD, 1994. Ϫ

J. E. Bartmess, NIST

tical. Therefore, the change of orbital angular momentum

Standard Reference Database, Negative Ion Energetics, Version

necessary to assist spin inversion is negligible. Vibrational exci-

[40e]

3

.01, Gaithersburg, MD, 1993. Ϫ

S. G. Lias, J. F. Liebman,

tation of 14 is associated with symmetry reduction to C

1

, but

SO

R. D. Levin, J. Phys. Chem. Ref. Data 1984, 13, 695.

G. Bouchoux, J. Mol. Struct. (Theochem) 1987, 151, 107.

M. L. McKee, L. Radom, Org. Mass Spectrom. 1993, 28, 1238.

Molpro 96.4 is a package of ab initio programs written by H.-

J. Werner and P. J. Knowles, with contributions from: J. Almlöf,

R. D. Amos, M. J. O. Deegan, S. T. Elbert, C. Hampel, W.

Meyer, K. Peterson, R. Pitzer, A. J. Stone, P. R. Taylor, R.

Lindh, 1996.

H

is not expected to increase very much, because the nature

of the electronic structure remains unchanged.

65]

[

41]

42]

43]

[

This distonic ion has earlier been characterized experimentally.

[65a]

For example, see:

J. C. Traeger, C. E. Hudson, D. J. McA-

[65b]

doo, Org. Mass Spectrom. 1989, 24, 230. Ϫ

C. E. Hudson,

[

65c]

D. J. McAdoo, Org. Mass Spectrom. 1992, 27, 1384. Ϫ

K.

M. Stirk, J. C. Orlowski, D. T. Leeck, H. I. Kenttämaa, J. Am.

Chem. Soc. 1992, 114, 8604.

66] [66a]

[

44] [44a]

_T_._H_._D_u_n_n_i_n_g_,_J_r_._,_J_._C_h_e_m_._P_h_y_s_.₁₉₈₉_,₉₀_,₁₀₀₇_._Ϫ[44b]

P. Fournier, J. Appell, F. C. Fehsenfeld, J. Durup, J. Phys.

[66b]

1

972, B5, L58. Ϫ

F. C. Fehsenfeld, J. Appell, P. Fournier,

[66c]

R. A. Kendall, T. H. Dunning, Jr., R. J. Harrison, J. Chem.

Phys. 1992, 96, 6796.

J. Durup, J. Phys. 1973, B6, L268. Ϫ

J. C. Lorquet, B. Leyh-

45] [45a]

Nihaut, F. W. McLafferty, Int. J. Mass Spectrom. Ion Processes,

1990, 100, 465.

S. Koseki, M. W. Schmidt, M. S. Gordon, J. Phys. Chem.

[45b]

1992, 96, 10768. Ϫ

S. Koseki, M. S. Gordon, M. W.

[

67]

68]

69]

V. Q. Nguyen, F. Turecek, J. Mass Spectrom. 1996, 31, 843, and

literature cited therein.

Schmidt, N. Matsunaga, J. Phys. Chem. 1995, 99, 12764.

M. W. Schmidt, K. K. Baldridge, J. A. Boatz, S. T. Elbert, M.

S. Gordon, J. H. Jensen, S. Koseki, N. Matsunaga, K. A.

Nguyen, S. Su, T. L. Windus, M. Dupuis, J. A. Montgomery, J.

Comput. Chem. 1993, 14, 1347. The version for PC was com-

piled by A.A. Granovsky, Moscow State University.

J. N. Harvey, M. Aschi, H. Schwarz, W. Koch, Theoret. Chem.

Acc., in press.

46]

For a review, see: J. Lee, J. J. Grabowski, Chem. Rev. 1992, 92,

1

611.

•؊

Due to the differences of the CR and NR spectra of 11 and

•

؊

1

2 , the resulting NIDD spectra differ quantitatively. While

[

47]

48]

this is inherent to the negative part of the NIDD scale, the

essence of the positive part remains, however, the same. Thus,

both carbenes rearrange to methyl carbene for which positive

G. Herzberg, Molecular Spectra and Molecular Structure Ϫ I.

Spectra of Diatomic Molecules, Krieger, Malabar/Florida, 1989.

•

NIDD signals due to losses of H and CO/C

2

H

4

are observed.

[70]

[71]

K. D. Jordan, P. D. Burrow, Acc. Chem. Res. 1978, 11, 341.

49] [49a]

P. Saxe, H. F. Schaefer III, N. C. Handy, J. Phys. Chem.

[71a]

For example, see:

P. C. Burgers, J. L. Holmes, Org. Mass

[49b]

1

981, 85, 745. Ϫ For reviews, see:

P. P. Gaspar, G. S. Ham-

[71b]

Spectrom. 1984, 19, 452. Ϫ

H. W. Zappey, S. Ingemann, N.

mond in Carbenes (Eds.: M. Jones, R. A. Moss) Vol. 1, Wiley,

M. M. Nibbering, J. Am. Soc. Mass Spectrom. 1992, 3, 515. Ϫ

[49c]

New York, 1973. Ϫ

W. T. Borden, E. R. Davidson, Ann.

[71c]

C. A. Schalley, A. Fiedler, G. Hornung, R. Wesendrup, D.

Rev. Phys. Chem. 1979, 30, 125. Ϫ ^[

49d]

E. R. Davidson in ref.

[18]

Schröder, H. Schwarz, Chem. Eur. J. 1997, 3, 626.

[

50]

51]

J. H. J. Dawson, A.J. Noest, N. M. M. Nibbering, Int. J. Mass

Spectrom. Ion Phys. 1979, 30, 189.

[72]

[73]

•ϩ

A similar effect was earlier found for the CH

2

signal in the

•؊

•

Ϫ

NIDD spectrum of the CH

2

CO

2

distonic radical-anion 16 .

J. H. J. Dawson, K. R. Jennings, J. Chem. Soc., Faraday Trans.

[30]

See ref.

.

2

1976, 72, 700.

The formation of both CO and ethene is confirmed by the

52] [52a]

E. Wasserman, L. Barash, W. A. Yager, J. Am. Chem. Soc.

Ϫ

ϩ

18

NIDD spectrum of O labeled oxyallyl, which shows posi-

tive signals at m/z 28 and m/z 30.

[74] [74a]

[52b]

1

968, 90, 1485. Ϫ

Am. Chem. Soc. 1977, 99, 13. Ϫ

J. Am. Chem. Soc. 1978, 100, 1333. Ϫ

Okuno, Y. Izawa, J. Chem. Soc., Perkin Trans. 2 1980, 1636. Ϫ

R. R. Lucchese, H. F. Schaefer III, J.

[52c]

N. C. Baird, K. F. Taylor,

R. Srinivas, D. Sülzle, T. Weiske, H. Schwarz, Int. J. Mass

[52d]

H. Tomioka, H.

[74b]

Spectrom. Ion Processes 1991, 107, 368. Ϫ

R. Srinivas, D.

Sülzle, W. Koch, C. H. DePuy, H. Schwarz, J. Am. Chem. Soc.

[

52e]

P. H. Mueller, G. N. Rondan, K. N. Houk, J. F. Harrison,

[74c]

1

991, 113, 5970. Ϫ

Int. J. Mass Spectrom. Ion Processes 1996, 153, 173.

[75] [75a]

C. A. Schalley, D. Schröder, H. Schwarz,

D. Hooper, H. P. Willen, J. F. Liebman, J. Am. Chem. Soc. 1981,

[

52f]

1

03, 5049. Ϫ

A. C. Hopkinson, M. H. Lien, Can. J. Chem.

S. W. Froelicher, B. S. Freiser, R. R. Squires, J. Am. Chem.

[

52g]

1

985, 63, 3582. Ϫ

J. Farras, S. Olivella, A. Sole, J. Vilarrasa,

[75b]

Soc. 1986, 108, 2853. Ϫ

Chem. Soc. 1988, 110, 607.

S. T. Graul, R. R. Squires, J. Am.

J. Comput. Chem. 1986, 7, 428.

[

53]

54]

55]

The calculated heats of formation of acrolein and methyl ketene

[76]

K. L. Busch, G. L. Glish, S. A. McLuckey, Mass

Spectrometry/Mass Spectrometry: Techniques and Applications

of Tandem Mass Spectrometry, VCH Publishers, Weinheim,

f

deviate significantly from the literature values, i.e. ∆H (1) ϭ

(2) ϭ Ϫ25 kcal/mol (ref.^[), and rein-

40b]

Ϫ18 kcal/mol and ∆H

vestigation of these data seems indicated.

T. Nakano, K. Morihashi, O. Kikichi, J. Mol. Struct.

f

1

988.

[77]

[78]

J. L. Holmes, Org. Mass Spectrom. 1985, 20, 169.

(

Theochem) 1992, 253, 161. The energy difference between neu-

[24d]

As suggested in ref.

, the charges of projectile and recovery

tral s-cis and s-trans-acrolein calculated in this study amounts

to 2.2 kcal/mol.

ions are indicated by superscripts.

M. M. Bursey, Mass Spectrom. Rev. 1990, 9, 555.

[79]

[

[80] [80a]

[80b]

Some of the G2 energies given in Figures 2 and 4 have been

F. Arndt, J. Amende, Chem. Ber. 1928, 61, 1122. Ϫ

M.

[

39]

taken from ref.

A. McKervey, P. Ratananukul, Tetrahedron Lett. 1983, 24, 117.

I. Fleming, A. Pearce, J. Chem. Soc., Perkin Trans. 1 1981, 251.

56] [56a]

[81]

[82] [82a]

R. B. Woodward, R. Hoffmann, The Conservation of Or-

[56b]

bital Symmetry, Verlag Chemie, Weinheim, 1970. Ϫ

kui, Acc. Chem. Res. 1971, 4, 57.

57]

K. Fu-

H. Weingarten, J. P. Chupp, W. A. White, J. Org. Chem.

[82b]

1967, 32, 3246. Ϫ

C. A. Schalley, D. Schröder, H. Schwarz,

[

S. Yamabe, T. Minato, Y. Osamura, J. Am. Chem. Soc. 1979,

J. Am. Chem. Soc. 1994, 116, 11089.

101, 4525.

[98037]

Eur. J. Org. Chem. 1998, 987Ϫ1009

1009

Article Doi

A combined neutralization-reionization mass spectrometric and theoretical study of oxyallyl and other elusive [C3,H4,O] neutrals

DOI: 10.1002/(SICI)1099-0690(199806)1998:6<987::AID-EJOC987>3.0.CO;2-G

Source and publish data:

Authors:

Article abstract of DOI:10.1002/(SICI)1099-0690(199806)1998:6<987::AID-EJOC987>3.0.CO;2-G

Full text of DOI:10.1002/(SICI)1099-0690(199806)1998:6<987::AID-EJOC987>3.0.CO;2-G

Products guided by the article

R&D Labs maybe for 39493-21-5

Relevant to this article

Hot Product