Retro-Diels-Alder reaction in mass spectrometry

Retro-Diels-Alder reaction in mass spectrometry FrantiSek Turetek and Vladimir HanuS Jaroslau Heyrous@ lnstitute of Physical Chemistry and Electrochemistry, Mlichova 7, 121 38 Praha 2, Czechoslovakia

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Thermochemistry and kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Orbital symmetry considerations ................................. B. Stereochemical approach . . . . . . . . . . ....... ..... IV. Scope and limitations . . . . . . . . . . . . . . . . ......................... A. Substituted cyclohexenes ....................................... B. Norbornene derivatives ......................................... C. Other bridged systems . . . . . . . . . . . . ............. D. Ortho-condensed systems . . . . . . . . . . ........................ E. Intramolecular RDA . . . . . . . . . . . . . . . ............. F. Heterocyclic compounds ........................................ V. Natural products ..................... ...................... A. Steroids ..................... ...................... B. Pentacyclic triterpenes . . . . . . . . ................. C. Carotenoids ................................................... D. Alkaloids ...................................................... E. Miscellaneous . . . . . . . . . . . . . . . . . . .................... VI. Acknowledgments . . . . . . . . . . . . . . . . .................... VII. References ....................... ..............

85 86 92 93 97 101 101 109 114 117 119 121 128 128 134 135 142 145 145

I. INTRODUCTION

The formation of conjugated-diolefinic (diene) and mono-olefinic (ene) fragments upon electron ionization (EI) of molecules containing a cyclohexene unit was recognized by Biemann (1)as formally analogous to the familiar retro-Diels-Alder (RDA) reaction (2). The term was recommended for use in the mass spectrometric nomenclature (3) and has been widely used. From the analytical point of view, RDA fragmentation appeared very promising for structure elucidation purposes, mainly for two reasons:

(1) Specific formulation of a diene and an ene fragment ion from a substituted cyclohexene would enable reliable location of the double bond (Scheme A). (2) Provided orbital symmetry control operates, the presence or absence of an RDA reaction would distinguish between annulated isomers in Mass Spectrometry Reviews 1984, 3, 85-152 0 1984 John Wiley & Sons, Inc.

CCC 0277-7037/84/010085-68$04.00

TURECEK AND HANUS

86 -

t.

1+*

R3

Scheme A

bicyclic or polycyclic systems, for only cis isomers would undergo a symmetry-allowed, supra-supra fragmentation (Scheme B). Soon it turned out, however, that neither the occurrence nor stereospecificity of the RDA reaction follows clear-cut rules and perhaps this gave an impetus to extensive investigations of the reaction mechanism. The early results were summarized in a comprehensive review (4). The fragmentation of Tetralin and its heteroanalogs, a special type of RDA, has been reviewed recently by Kuhne and Hesse (5). The present review covers the period 1965-1982. Although the RDA reaction has received much attention from mass spectrometrists, a large body of data has come from synthetic work in which mass spectrometry merely served as an analytical method. Reports of this kind have also been abstracted even though the results were often not confirmed by accurate mass measurements, detection of metastable ions, or isotope labeling. 11. THERMOCHEMISTRY AND KINETICS

The RDA reaction is usually one of several parallel fragmentations of a molecular ion. The relative abundance of RDA fragments thus depends on the pertinent critical energy (6) compared with critical energies of competing fragmentations. So far, thermochemical data, ionization energies (IE), appearance energies (AE), and the heats of formation derived therefrom, have been reported for a few simplest systems. Budzikiewicz, Brauman, and Djerassi (4) discussed the thermochemistry of the RDA reaction in ionized cyclohexene 1, making use of IE and AE data then available (7,8). Reexaminations (9,lO) gave different values for IE(1) and AE(C,H,+') (Table I). According to Winters and Collins (lo), the energy lies near the thermochemical required to produce C,H,+' and C,H, from 1+' threshold given by AHAC,H,+') + C,H,) (Fig. 1).This implies that the DA between C,H,+' and C,H, has a negligible activation energy in contrast to RDA of the neutral molecules ( E , = 115 kJ/mol) (14). Nevertheless, DA be1+*

1+.

Scheme B

RDA REACTION IN MASS SPECTROMETRY

87

Table I. Ionization and appearance energies of C,H,,+* and C,H,+*. IE (C~HIO+') AE (CaH,") (eV) (eV) References 9.2 9.57 8.92 8.94 8.95 8.75

11.2 11.91 10.67

-

4'7,s 9 10 11 12 13

tween ionized butadiene and ethylene was found not to proceed under ion cyclotron resonance (ICR) conditions (15). The time dependence of the RDA reaction and competing processes has been studied with (3,3,6,6-D,)cyclohexene (2) by the field ionization kinetics (FIK) method (16). The ordinary EI mass spectrum of 2 displays C,(H,D),+' species in a nonstochastic distribution as a result of hydrogen migrations within the molecular ion. The time-resolved study has shown that the hydrogen migration competed with the RDA process even at the shortest lifetime of the parent ions amenable to FIK measurements (ca. lo-" s). The stochastic distribution of the C,(H,D),+' fragments was achieved after ca. lCb9s (16). The FIK results also showed that the relative abundance of C,H,+' fragments, formed from unlabeled cyclohexene, had a maximum at ca. 3 x 1@l1 s and rapidly faded toward longer lifetimes of the parent ions. Although the absolute values of the FIK rate constants have been subject to criticism (17), qualitatively they agree well with other findings (18). The absence of the kinetic shift (19) in the AE measurements (10) is consistent with a very steep slope of the log k versus internal energy curve so that there are few ions I+'

E (kJ /mol)

coordinate

Figure 1. The schematic energy profile for the RDA reaction in ionized cyclohexene.

TURECEK AND HANUS

88

capable of decomposing via RDA reaction with low ( 4 0 6s-') rate constants. Reports on RDA decomposition of metastable 1+* seem to be contradictory: Winters and Collins (10) observed no metastable ions whereas an intense signal was noted in another study (20). Ions 1+* produced by charge exchange gave a very weak metastable ion signal corresponding to the RDA reaction (18). A reexamination of the EI data (unpublished results from the authors' laboratory) has shown a very small fraction of metastable 1+' decomposing unimolecularly in the first field-free region ([m*]/ to 1C5torr resulted [C,H,+*] = 6 x 10-6). Increasing the pressure from in a 50-fold increase in the signal intensity due to collision-induced decomposition (CID) of stable I+'. The RDA in ionized norbornene (3) has been studied in detail (21,22). In contrast to the behavior of cyclohexene, where the lowest-energy decomposition corresponds to loss of methyl, the RDA is the most favorable fragmentation of 3 + *(22). AE measurements, combined with deuterium labeling, showed that the RDA products correspond to threshold-energy (cyclopentadiene)+' and ethylene (22). A fairly strong metastable-ion signal was observed for the RDA reaction in this case (22). The release of kinetic energy (T,) in RDA loss of ethylene from 3+', 4+', and 5 + * was correlated with the activation energy of the Diels-Alder reaction of the neutral dienes with maleic anhydride in solution (23). Since the kinetic energy release during fragmentation could result from the activation energy of the reverse process, it was suggested that the gas-phase data might be useful for estimation of energy parameters of solution reactions (23). Nevertheless, the kinetic energy release in metastable 3 + * is rather low and with respect to the thermochemical measurements (22) it appears that it may be largely due to a nonfixed excess energy (22a).

3 T,(eV)

0.056

4

0.098

5

0.15

The molecular ion of limonene (6) decomposes via RDA to isoprene ion and molecule (24). Since the isoprene subunits are not equivalent in the parent ion, the C5H,+' and/or C5H, fragments can originate from the ring or side-chain parts of the molecular ion. Labeling studies (24,25) disclosed that the ionized fragment C5H8+'comes evenly from both parts of the molecular ion. The appearance energy measurements gave AE(C,H,+') = 11.6 eV with some uncertainty in pinpointing the actual threshold (24). Taking the reported ionization energy [IE = 8.3 eV (24)] and the calculated AH, of 6 (3.4 kJ/mol)(26), we obtain an energy profile (Fig. 2) in which the critical energy for the RDA considerably exceeds the thermochemical threshold. Neither metastable nor collisionally activated ions 6+' have been found to fragment via RDA reaction (24). This was attributed to the insufficiency of the small


89

6

coordinate

Figure 2. The schematic energy profile for the RDA reaction in ionized limonene.

amount of energy deposited by collisions to induce RDA fragmentation (24). In the light of the observed migration of hydrogen about the ring (24), it also seems possible that the long-lived molecular ions of 6 (stable or metastable) do not retain the original position of the ring double bond and the RDA fragmentation then proceeds from rearranged ions, giving products other than isoprene. Fragment ions from cyclohexene-l-o1(7),prepared from 2-ethylcyclohexanone by loss of ethylene, have been reported to decompose via RDA reaction (27).Unimolecular decompositions of metastable 7+' give a moderate fraction of RDA products C,H,O+' (2.8% of the sum of all fragments, C F + ) whereas in the CID spectra C,H,O+' is one of the major ions. RDA decomposition of metastable or collisionally activated 7+' was not preceded by hydrogen migrations, as established by deuterium labeling (Scheme C) (27). This is probably due to an exceptional stability of 7+', compared with other doublebond positional isomers, that sets a high energy barrier to hydrogen migration. The presence of the RDA in this case served as one of the arguments in assigning structure to 7+' (27). The formation of butadiene and (1,1,4,4-D4)butadieneions from 4-vinyl(3,3,6,6-D4)cyclohexene(8) has been investigated recently (28). The appearance energies of C,H,+' and C,H,D,+' have been found to be equal within

q 7 + - 3D&l+*

DO

D

0

7

Scheme C

?'+.

TURECEK AND HANUS D O 1,3-shift

D

OD

I

I+-

/CiD

CID

Scheme D

experimental error (AE = 11.07eV), the critical energy for the RDA reaction exceeding the calculated thermochemical threshold by 45 kJ/mol. No metastable molecular ions decomposing to C,H, +',C,H,D, or C,H,D, could be detected in 8+*. The CID of stable molecular ions gave RDA fragments, albeit in very low abundance. Of these, the C,H,D,+* ion was about twice as abundant as C,H,+' and C,H,D,+'; this may be attributed to RDA decomposition of a very small fraction of the original and singly isomerized parent ions (Scheme D). Bicyclo[4.3.0]nona-3,7-dienes(9, 10) give abundant C5H,+' ions due to RDA fragmentation. Although the EI mass spectra of the cis- and transannulated isomers differ in the relative abundance of C5H6+', the experimental critical energies are equal and correspond to formation of threshold-energy products (Fig. 3). In contrast to the situation in ion-source fragmentation, the major fraction of metastable 9+' and lo+' does not decompose via RDA reaction. The mass-analyzed kinetic energy (MIKE)spectra

+.,

+

El

(kJImoi)

I

I

\

coordinate

Figure 3. The schematic energy profile for the RDA reaction in 9+'and

lo+*.


91

display C,H,+' fragments of extremely low abundance. On the contrary, the CID spectra of 9+' and lo+' show abundant C,H,+' fragments which, moreover, originate from the intact cyclopentene part of the parent ion, as confirmed by deuterium labeling (Turefek, F.; H a n d , V.; Pancii., J.; Gaumann, T.; Stahl, D., to be published).

9

The above examples indicate that RDA decomposition of I+', 6+', 8+', 9+', and lo+' is a fast, high-frequency-factor reaction which is able to compete with rearrangements as long as rapidly decomposing ions (T 4 1P s) are involved. In this respect, the RDA reaction resembles simple-cleavage fragmentations rather than bond-forming rearrangements (30). This is illustrated by internal energy dependence of the relative abundances of (M Br) + and (M - C,H,Br) +' ions from 5-bromonorbornenes 11and 12 (Scheme E) (31). Although the rates of bromine loss from 11+'and 12+' depend on configuration, the abundance ratios [(M - Br)+]/[(M - C,H,BR)+'] show the same trend, the former ions being preferred at low electron energies. [M - Br])'fM

- C2H3Br]+'

30eV

13eV

0.59

11

0.056

0.091

Br

Scheme E

If there are competing rearrangements with critical energies lower than that of the RDA reaction, the branching ratio will be sensitive to frequency factors and thermochemical considerations will be of little value. This may be demonstrated by comparing the thermochemical data for RDA decompositions in 9 +' and the related cis-bicyclo[4.3.0]non-3-ene (13). If the reverse

-

13

activation energy for the RDA reaction in 13+' are neglected, the critical + C,H, and 13+'--f C a , ' + C f i energies of decompositions 9 + ' C& +' are calculated to differ by 9 kJ/mol (unpublished results from the authors' +

TURECEK AND HANUS

92

1+.

1+.

I

RDA

Scheme F

laboratory). In spite of this small difference, the relative abundances of the RDA products differ considerably; given the uncertainties in estimating the AHf values, the discrepancy impairs the value of thermochemistry-based predictions of the reaction course even with related systems. On the other hand, the AH, values can be useful if the competing fragmentations are of a similar type, as has been documented recently (32). Substituted dihydropyrans (Scheme F) decompose mainly via RDA reaction and loss of the C6 substituent. With ethoxy derivative 14 the critical energies of these competing fragmentations differ by 9 kJ/mol and the reactions occur to a similar extent (32). With the acetyl derivative 15, RDA fragmentation is disfavored by 79 kJ/mol and the loss of aceteyl prevails. Thermochemical considerations can be a valuable guide for predicting the charge localization in fragment ions formed by RDA decomposition ( 3 3 3 ) . Provided the reaction proceeds from the same electronic state of the reactants and via the same mechanism to form diene and ene fragments, the charge site will be dictated by the relative ionization energies of the products (Fig. 4).As pointed out by Audier (34), this generalization is valid as long as the reverse activation energies are comparable for the competing channels so as not to overwhelm the differences in the ionization energies. 111. MECHANISM

The peculiarities encountered with the mass spectral RDA have stimulated numerous investigations of the mechanism. Attention has been focused largely on answering the question of whether the RDA reaction proceeds in a concerted or a stepwise manner. This problem is difficult to attack by mass spectrometric means, for the possible intermediates in a two-step mechanism are transient isomers of the reactants, while the products are identical for both mechanisms. The experimental approaches have therefore used specially designed model compounds and relied upon some general reactivity principle such as conservation of orbital symmetry (3541), charge localization (4244), or differences in the behavior of odd- and even-electron ions (4547).


93

energ

-7A+

€1i

B+'

Eg- EA

= IE(B) - IE(A)

A B+'

Figure 4. The hypothetic reaction profile for RDA decompositions leading to A+' or B+*.

A. Orbital symmetry considerations

Woodward-Hoffman rules (48) represent a well-established tool for predicting the course of concerted reactions in organic chemistry and attempts have been made to extend the use of the orbital symmetry correlations to ion decompositions in the gas phase (35,49-51). RDA decomposition of a labeled 4-vinylcyclohexene was reported to produce charged butadiene fragments originating from the ring and side-chain parts of the molecular ion with an unequal frequency (52). This unusual finding was explained by Dougherty (35) to be a simple result of orbital correlations. With groundstate reactant ions (35), abstraction of an electron from the cyclohexene v(CC) orbital followed by a concerted RDA reaction would lead to products having the electron vacancy in the symmetrical x1orbital of butadiene, which corresponds to an excited state of the C,H,+' ion. On the other hand, correlation of the vinylic v(C-C) orbital is not symmetry-restricted, so the parent ions ionized at the vinyl group may produce butadiene ion and molecule in the ground state. The latter channel would thus require a lower critical energy and should proceed preferentially (35). A reexamination of the experimental data (25) has shown, however, that the asymmetry in formation of charged diene fragments was very low in 6,8, and 1,4-dimethyl4-vinylcyclohexene (16) and even absent with tricyclic compounds 17 and 18. The strong metastable-ion signal for the RDA reaction, claimed as evidence for decomposition from the ground state, has not been reproduced (25,53). The photoelectron spectrum of 8 (unpublished results from the authors' laboratory) reveals that the ionization potential of the vinylic r(CC) orbital (9.56 eV) is higher than that of the cyclohexene v(C-C) (8.91 eV). Hence 8+' ionized in the vinyl group does not have ground-state electron configuration.

/o- m m 16

17

18

94

TURECEK AND HANUS

Figure 5. The formal correlation diagram for the RDA reaction in ionized cyclohexene.

It should be noted (cf. also ref. 51) that a conventional correlation diagram (48) gives an incorrect prediction of the charge site even in the simplest case of RDA in ionized 1 (Fig. 5). According to such a formally constructed diagram, the lowest-energy path would lead to formation of neutral butadiene and charged ethylene, contrary to experiment. The orbital symmetry rules use a minimum set of relevant orbitals that undergo first-order changes along the reaction coordinate (54). For closed-shell systems this approach has been proved to be quite adequate from both a qualitative (48) and a quantitative (55) point of view. With odd-electron ions there are numerous examples of rule violations, e.g., [2 + 21 (56-59) and [6 + 21 (60-63)retrogressions (Scheme G). The course of RDA reaction in ionized cyclohexene indicates that the conventional description within the frame of n(C-C) and o(C-C) orbitals may not be fully adequate with odd-electron ions. Pancii. and Turerek (28) have applied recently the topological molecular orbital (TMO-CI) method (55) to investigation of reaction coordinates of the RDA reaction in 8+'. Calculations using the conventional basis of orbitals yielded unrealistic predictions for the ion radical 8+', reversing the relative stability of the reactants and products and giving an absolute energy minimum along the reaction coordinate. Meaningful results have been obtained after having included the allylic n(CH,) orbitals into the basis set and the calculations then correctly reproduced the experimental thermochemical data (28). The choice of the .rr(CH,) orbitals was motivated by considerable stabilization of cyclohexene ion (IE = 8.92 eV) compared with ethylene (IE = 10.50 eV) (64).Figure 6 shows a three-dimensional reaction profile for the RDA reaction in 8+'. The concerted reactions with supra-supra or supra-antara geometry (diagonals connecting the reactants and products) correspond to pure energy maxima in the transition state and are therefore forbidden. Nevertheless, the barrier to the supra-supra process is lower than for the supra-antara one so that, at least formally, the Woodward-Hoffmann rules are preserved. The lowest-energy pathway is stepwise and includes sequential cleavage of the C-3-C-4 and C-5-C-6 bonds. The second step overcomes a barrier which sets the energy excess over the thermo-


95

20

23

YS'

Scheme G

chemical threshold in the products. The mechanistic routes are altered in the first excited state of 8+' (Fig. 7), in which the RDA reaction is predicted to proceed via a concerted, supra-supra mechanism. Since the first excited state of 8+' arises by ionization from the vinylic r(C-C) orbital (vide supra), it follows that the removal of the electron vacancy farther from the cyclohexene ring strongly favors the concerted, supra-supra mechanism. Including the n(CH,) orbitals into the "active frame" (54) of relevant wave functions enables one to assess the effect of substituents in allylic positions. On substitution, the hyperconjugative effect of allylic bonds will be weakened, which would lower the energy barrier to the concerted RDA process. Hence, the effective mechanism of the reaction in a given system should depend on substitution and, moreover, the reaction may proceed independently by both mechanisms if the critical energies become comparable. The stepwise course of the RDA reaction in 8+' also follows from a recent M I N D O / ~study (29).

TURECEK AND HANUS

96

ENERGY RI1) STATE

4-V1NYLCYCLOHEXENE:D conrotatory

___)