Decomposition Modes of Dioxirane, Methyldioxirane and Ž

31 July 1998

Chemical Physics Letters 292Ž. 1998 97±109

Decomposition modes of dioxirane, methyldioxirane and dimethyldioxirane Ð a CCSDž/ T , MR-AQCC and DFT investigation

Dieter Cremer a,), Elfi Kraka a, Peter G. Szalay b a Department of Theoretical Chemistry, UniÕersity of Goteborg,ÈÊ Kemigarden 3, S-41296, Goteborg, È Sweden b Department of Theoretical Chemistry, EotÈÈÕos Lorand Â ±UniÕersity, P.O. Box 32, H-1518, Budapest, Hungary Received 7 April 1998; accepted 8 June 1998

Abstract

Formation and decomposition of dioxirane Ž.2a , methyldioxirane Ž.2b and dimethyldioxirane Ž.2c in the gas phase were investigated by carrying out CCSDŽ. T , MR-AQCC and B3LYP calculations with the 6-31G Ž d, p . , 6-311qG Ž 3df, 3pd . and cc-VTZ2Pqf, d basis sets. The inclusion of f functions in the basis set was essential to determine the heat of formation o y r D Hf Ž.298 of carbonyl oxide Ž.1a and 2a to be 27.0 and 0.3 kcal mol, respectively. With the latter value, we calculate the same ring strain energy for cyclopropane, oxirane and 2a. Molecule 2a decomposes at 298 K with an activation enthalpy of 18 kcalrmol to methylenebisŽ.Ž. oxy 3a , which is calculated to be 1.2 kcalrmol less stable than 2a, in contrast to previous investigations. Two methyl substituents increase the ring opening barrier to 23 kcalrmol and, thereby guarantee the kinetic stability of 2c. The biradicals 3 decompose with barriers smaller than 4 kcalrmol to esters and therefore will be difficult to intercept in dioxirane decomposition reactions. q 1998 Elsevier Science B.V. All rights reserved.

1. Introduction oxygen reactions, the reactions of singlet oxygen with diazo compounds, alkyl radical±ozone reac- Dioxiranes and their acyclic isomers carbonyl ox- tions, the matrix photooxidation of oxygen±alkene ides play an important role in many oxidation pro- mixtures and as intermediates in reactions of car- cesseswx 1±4 . The former are ;20 kcalrmol more bonyl compounds with atomic oxygenwx 1±4 . Curci stable than the latterwx 5 and, accordingly, it was et al.wx 10 postulated dioxiranes as oxidants in the possible to synthesize and investigate the parent caroate±ketone system and Murray and Jeyaraman dioxiranewx 6 and several substituted dioxiranes w 1± wx11 based their synthesis of dialkyldioxiranes on this 3,7±9x . Both isomers are probably generated as de- system. Apart from this, dioxiranes may play a role composition products of primary ozonides in the in the chemistry of the polluted atmospherewx 12,13 ozonolysis of alkeneswx 1±5 . Furthermore, they are and in enzymatic processeswx 14 . produced or supposed to be produced in carbene± Due to the importance of dioxiranes in many oxidation processes there is a large literature on ) Corresponding author. Fax: q46-31-7722933. these compounds covering different aspects of their

Scheme 1.

chemistry viewed both from the point of experimen- be formed by direct decomposition of 3 Ž.Scheme 1 . tal and theoretical investigationswx 1±5 . In particular, Our investigation will be based on coupled cluster quantum chemical investigations added much to the Ž.CC , multi-reference averaged quadratic CC Ž MR- understanding of the properties of dioxiranes and AQCC.Ž. and density functional theory DFT calcula- their chemistrywx 5,9,15,16 . Nevertheless, it is still tions. The primary goal of our work will be to unclear why certain dioxiranes are stable, but others explain the increase in stability when substituting are not. Murray and Jeyaramanwx 11 succeeded in dioxiranes by methylŽ. alkyl groups and to predict synthesizing dimethyldioxirane, which can be kept which products can be expected when generating for several days at room temperature, while dioxi- dioxiranes. rane itself has only been observed as a labile species in the gas-phase ozonolysis of ethenewx 6 . Mass spectrometry investigations have found that dioxi- 2. Computational methods wx rane decomposes to CO22 , H , CO and H 2 O 4 . However, in matrix isolation experiments esters as For the parent systems 1a to 7a, CCSDŽ.Ž T CCSD dioxirane rearrangement products were observed with perturbative inclusion of tripleŽ. T excitations . rather than the decomposition productswx 1,7,9 . wx17 were carried out using Dunning's correlation We will investigate in this Letter the rearrange- corrected cc-VTZ2Pqf, d basis, which is composed ment and decomposition modes of dioxirane and its of aŽ. 10s 5p 2dwx 4s 3p 2d contraction augmented by a methyl substituted derivatives. For this purpose, we set of spherical f functions for the heavy atoms and a investigate the rearrangement of carbonyl oxide Ž.1a , set of spherical d-functions for H atoms ŽŽ.a C: syn- and anti-methylcarbonyl oxide Ž1b-syn and 1b- 1.097, 0.318Ž. d , 0.761 Ž. f ; a ŽO . : 2.314, 0.645 Ž. d , anti.Ž. and dimethyl carbonyl oxide 1c to the corre- 1.428Ž. f ; a ŽH . : 1.407, 0.388 Ž p . , 1.057 Ž d ..wx 18 and sponding dioxiranes 2a, 2b and 2c, respectively; ring has proven to lead to reliable results in the case of opening of the latter to the methylenebisŽ. oxy biradi- dioxiraneswx 15,16 . The ring opening of 2 to form 3 cals 3a, 3b and 3c; rearrangement of the biradicals 3 represents a multi-reference problem involving two to formic acid or esters Ž.4a, 4b and 4c ; rotation of 4 configurations of the same symmetry. As a result, to the less stable conformation 5; decomposition of MR-AQCCwx 19 calculations were performed em- q s q wx either 4 or 5 to CO RO222Ž.7; for 7a,R O HO ploying a 6-311 GŽ. 3df, 3pd basis set 20 . MR- q s or R 22CO Ž.6; for 6a,R 22H where 6 may also AQCC is essentially a modified MR-CI procedure, D. Cremer et al.rChemical Physics Letters 292() 1998 97±109 99 which contrary to MR-CI leads to nearly size-exten- Since the extension of the CCSDŽ. T and MR- sive resultswx 21 . According to test calculations wx 22 , AQCC calculations to the methyl substituted systems potential energy surfaces obtained by MR-AQCC are 2b and 2c was not feasible because of computational parallel to those of full CI calculations and, there- limitations, we checked various methods as to fore, MR-AQCC provides reliable energy differ- whether they can provide a reasonable, but computa- ences. tionally much less expensive description of the reac- The reference space for the MR-AQCC of 2, tions of Scheme 1. It turned out that density func- TS2-3 and 3 was carefully chosen to give a balanced tional theoryŽ. DFT with Becke's three-parameter description at all three geometries. The highest occu- functional B3LYPwx 23 provided the best description. pied MOs and the LUMO of 2 are DFT covers significant correlation effects in an un- specified way and has proven to be useful in the s p 6a11 ± Ž.OO ; 2b ± ŽOO bond . ; description of typical single-determinant problems 1a ±pŽ.Ž.OO antibond, CO nonbond ; 4b ±s) OO . but also some multi-reference problems, in which 22one configuration is still dominating the wavefunc- The ground stateŽ. GS of 2 is defined by the tion. For biradicals such as 3, unrestricted DFT 2 2 2 electron configuration . . .Ž. 6a112Ž. 2bŽ. 1a Ž.UDFT can provide reasonable descriptions, in par- 1 Ž A,41 p.. Ring opening to 3 formally involves a ticular if the symmetry of the molecule is broken rotation of the two 2p p electron pairs at the O Ž.broken symmetry UDFT as was shown in the case wx atoms into the ring plane so that both the 6a1 ±sŽ.OO of metal bonding for transition metal complexes 24 . and the 4b2 ±s)Ž.OO become doubly occupied and, Broken symmetry UDFT calculations of singlet bi- consequently, s-bonding between the O atoms is no radicals lead to a mixing of singlet and tripletŽ and longer possible. At the equilibrium OO distance of 3 higher.Ž. state s .

Ž.close to 2 AÊ the two p MOs 1a21 and 2b are of Mixing might be assessed by calculating the ex- comparable energy and, therefore, both the pectation value of the operator S 2 for the Kohn± 2 2 2 ...Ž. 6a122Ž. 1aŽ. 4b and the Sham single determinant wavefunction. However, 2 2 2 2 . . .Ž. 6a112Ž. 2bŽ. 4b electron configuration con- ²S : is of somewhat dubious value in DFT since the 1 tribute to the A,21 p GS of 3. The analysis of the Kohn±Sham wavefunction corresponds to the un- 1 ring opening process reveals that the A,21 p of 3 physical situation of non-interacting electrons. Calcu- corresponds to a low-lying excited state of dioxirane lation of ²S 2 : requires knowledge of the pair den- ™ ™ obtained by 2b124b or 1a 224b double excita- sity, which is not known. Therefore, we refrained 1 1 tion and that the A,411p and A,2p state undergo from improving the UDFT results by simple summa- an avoided crossing situation close to TS2-3. Hence, tion rules or spin-projection techniques applied for a consistent description of 2,TS2-3 and 3 requires ab initio wavefunctions and instead based the direct the inclusion of orbitals 6a112 , 2b , 1a and 4b 2 into use of the UDFT results on the fact that the electron the reference space, which in the present work was density is less sensitive to spin contamination effects extended by adding the unoccupied orbitals 7a1 and than the wavefunction. This was confirmed by the 5b2 to have some more flexibility in the MR-AQCC observation that the properties of 3 calculated at the calculations. UB3LYP level are in good agreement with MR- All possible excitations of the 6 active electrons AQCC results. among the 6 active orbitals were considered, thus Explorative B3LYP calculations were carried out leading to a 6=6 CASŽ. complete active space . with the 6-31GŽ. d, p basis setwx 25 , while for the After freezing the three core orbitals, all S and D more accurate calculations Pople et al.'s 6-311q excitations out of these reference functions were GŽ. 3df, 3pd basiswx 20 was used, which is slightly included in the wavefunction. The orbitals were opti- larger than the cc-VTZ2Pqf, d basis of the CC mized at the multiconfiguration SCFŽ. MCSCF level calculations. For all molecules and transition states using the same 6=6 CAS wavefunction. For the considered, B3LYP vibrational frequencies at the 6-311qGŽ. 3df, 3pd basis set used, the dimension of optimized geometries were determined to character- the CI became larger than 10 million. ize the stationary points. Zero-point energyŽ. ZPE 100 D. Cremer et al.rChemical Physics Letters 292() 1998 97±109 D. Cremer et al.rChemical Physics Letters 292() 1998 97±109 101

Fig. 1. Geometries of 1,TS1-2 and 2 Ž.see Scheme 1 . Distances in A,Ê angles in degree, dipole moments m in debye. Numbers in normal print refer to B3LYPr6-31GŽ. d, p results, numbers in italics to B3LYPr6-311qG Ž 3df, 3dp . results and numbers in bold to CCSDŽ. T rwx4s 3p 2d 1f r 3s 2p 1d results. Black balls denote the positions of O atoms, large white balls those of C atoms and small white balls those of H atoms. and thermal corrections were used to obtain reaction differ from the CC results on average just by 2.3 and activation enthalpies at 298 K. Utilizing experi- kcalrmolŽ 6-31G Ž d, p .. and 1.7 kcalrmol Ž 6-311q o mental heats of formation D Hf Ž.298 for suitable GŽ. 3df, 2pd , respectively, which is comparable to the reference compoundswx 26,27 , heats of formation for magnitude of the vibrational corrections. all molecules shown in Scheme 1 were determined. Calculations were carried out with COLOGNE96 3.1. Formation and decomposition of dioxirane wx28 , ACES II wx 29 , COLUMBUS wx 30 and GAUSS- IAN94wx 31 . CCSDŽ. T calculations predict dioxirane 2a to be 27.3 kcalrmol more stable than carbonyl oxide 1a o r ŽDDHf Ž.298 , Table 1 . , which is 3 kcal mol higher 3. Results and discussion than predicted by the best CCSDŽ. T calculation carried out previouslywx 5 . Recently, it was shown that Calculated geometries are shown in Figs. 1±3, the geometry of 2a, in particular the OO bond length, while the corresponding energies, ZPE values, reac- can only be calculated accurately when CCSDŽ. T o tionŽ. activation enthalpies DDHf Ž.298 and heats of calculations are carried out with a TZ2P basis set o wx formation D Hf Ž.298 are listed in Table 1 Ž parent that includes f-type polarization functions 15,16 . system.Ž and Table 2 methyl substituted systems . . The necessity of f-functions for predicting the cor- Details of the MR-AQCC calculations are given in rect geometry is confirmed by our calculationsŽ Fig. Table 3. In Table 4, methyl stabilization effects are 1. and, in addition, it turns out to be relevant when compared. determining the stability of 2a. Therefore, we recal- o o Heats of formation D Hf Ž.298 were calculated culated the D Hf Ž.298 of 1a and 2a with the help of y r o using the experimental value of 4a Ž 90.6 kcal mol the CCSDŽ. T DDHf Ž.298 between 12 Ž.and 4 wx o s 27. as a suitable reference and comparing with the Ž.Table 1 . The values thus obtained ŽD Hf Ž.298 q y r y r experimental values for CO22H Ž 94.1 kcal mol 27.0 and 0.3 kcal mol, Table 1. are 3 and 6.3 wx q y r wx r 27.Ž and CO HO2 84.2 kcal mol 27. . Devia- kcal mol below the best previous valuesŽ 30.2 and tions were smallest for the CCSDŽ. T rMR-AQCC 6.0 kcalrmolwx 5. , which clearly demonstrates the valuesŽ. 0.1 and 1.9 kcalrmol, Table 1 so that the necessity of using f-type polarization functions. following discussion is based, if not otherwise noted, The revised values describe the dioxirane ring as o on differences in enthalpies, DDHf Ž.298 calculated having the same conventional strain energyŽ. CSE at the CC level of theory. However, B3LYP results y26.4 kcalrmolŽ obtained from the homodesmotic 102 D. Cremer et al.rChemical Physics Letters 292() 1998 97±109 D. Cremer et al.rChemical Physics Letters 292() 1998 97±109 103

Fig. 2. Geometries of TS2-3, 3,TS3-4 and 4 Ž.see Scheme 1 . See caption of Fig. 1.

q q ™ q wx reaction: 2 HOOH 2CH32 OH HOCH OH with earlier results 5 . Once 2a is formed in the o y y ™ r 2CH3f OOH and D H Ž.298 values of 0.3, 32.6, reaction 1a 2a, it possesses 46.3 kcal mol of y48.0, y93.5 and y30.9 kcalrmolwx 26,27. as excess enthalpy, which is sufficient to break the OO cyclopropane Žy26.5 kcalrmolwx 32,33. or oxirane bondŽ. OO bond length in TS2-3a: 2 A,Ê Fig. 2 and to Žy27.6 kcalrmolwx 32,33. . Since cyclotrioxolane open 2a to the biradical 3a since the activation possesses a higher CSEŽ 33 kcalrmolwx 31. , it is enthalpy for this process is just 18 kcalrmol at the reasonable to assume that special electronic effects MR-AQCC level of theory. UB3LYP predicts values balance the strain of the three cyclic molecules cy- between 19.6 and 21.7 kcalrmol, which is surpris- clopropane, oxirane and 2a and that the CH bonds in ingly good in view of the fact that the wavefunction the cyclic molecule take over one of these balancing for both the TS2-3 and the biradical 3a are domi- functions. In related work, we found that the adia- nated by two configurations, one corresponding to 1 batic CH bond stretching frequencies and force con- the A1 , 4p and one to theŽ with regard to 2a doubly wx 1 stants 35 , which provide a reliable measure for the Ž.D excited . A1 , 2p state as shown in Table 3. CH bond strength, decrease in the order cyclo- UB3LYP includes the D excited configuration at the propane, oxirane, 2a, which is parallel to the en- price of a triplet contamination. It seems that con- hanced possibility of anomeric delocalization of the trary to an orbital-based method a density-based O lone pair electrons into the s ) Ž.CH orbitals and method is less sensitive to the triplet contamination, the larger electron withdrawing ability of the C atom which indicates that it is not justified to use in this in 2a relative to that of the C atoms in oxirane or case an orbital-based method such as spin-projection cyclopropane. to get a non-contaminated state. B3LYP describes The CCSDŽ. T activation enthalpy for isomeriza- both isomerization to and opening of 2a with reason- tion of 1a is 19 kcalrmolŽ. Table 1 in agreement able accuracy, which is sufficient to investigate the 104 D. Cremer et al.rChemical Physics Letters 292() 1998 97±109

Fig. 3. Geometries of TS3-6,TS4-5,TS4-7,TS5-6 and 5 Ž.see Scheme 1 . See caption of Fig. 1. corresponding processes of the methyl substituted likely that both molecules possess comparable stabil- o y r molecules. ities ŽD Hf Ž.298 for 3a: 0.9; for 2a: 0.3 kcal mol, In view of the magnitude of the calculated ring Table 1. . Ring opening is accompanied by a opening barrier, there is little chance to isolate 2a at strengthening of the CO bondsŽ shortening from room temperature in the gas phase. Even in the 1.385 to 1.329 AÊ . and the loss of a weak OO bond solution phase where there is the possibility of en- Ž.lengthening from 1.51 to 2.34 A,Ê Figs. 1 and 2 . The ergy dissipation, a barrier of 18 kcalrmol is too dipole moments of the two molecules are 2.5 D Ž.2a small to guarantee kinetic stability under normal and 2.4 D Ž.3a , which means that formation of 3a conditions, which explains why 2a could only be should have little dependence on the dielectric con- observed at low temperatureswx 6 . stant of the environment. The reactions of the biradical 3a should proceed 3.2. Decomposition of methylenebis() oxy rapidly via TS3-4 Ž.ŽH-migration or TS3-6 fragmen- q tation into H 22CO.Ž. , since the CCSD T activation o The biradical 3a is just 1.2 ŽDDHf Ž.298 , MR- enthalpies for these processes are just 1.8 and 2.4 AQCC.Ž. to 3.2 B3LYP kcalrmol less stable than kcalrmolŽ. Table 1 , i.e. even at low temperatures 2a. We note that most previous calculations underes- there is little chance of ever detecting the singlet timated the stability of 3a, suggesting relative ener- biradical 3a experimentally. Both formic acid 4a and r wx q r gies 11±12 kcal mol 4,36,37 . However in view of CO22H Ž.6a are more than 90 kcal mol Ž Table 1 . the electronic structures of 2a and 3a, it is more below 3a so that the back formation of the biradical D. Cremer et al.rChemical Physics Letters 292() 1998 97±109 105 ) 82.1 84.2 y y 3-6 7 . wx TS ) 3.3 2.2 12.8 3.4 4.6 20.7 2.3 4.3 11.3 9.1 2.3 17.8 7.9 2.4 8.5 8.8 0.2 10.0 3s2p1d , respectively. 94.2 8.0 94.1 y y y y y y r y y 5-6 6 21.0 wx TS y Values from experiment 26,27 . 86.3 ) y Ž. 4-7 5 21.0 G 3df, 3pd geometries were used. CC enthalpies were TS y q 189.54402 hartree 4s3p2d1f y 6-311 4-5 79.0 r TS y 298 values, see text. Ž. o ) f H D 93.7 12.9 69.5 4.0 68.8 90.3 14.1 73.5 5.1 68.6 91.1 13.0 73.9 4.5 71.3 89.3 12.8 69.0 4.9 62.7 90.3 11.6 69.6 4.3 5.3 92.9 11.5 65.2 3.8 62.9 90.6 90.6 c y y y y y y y Ž. Ž.Ž. mol obtained from reaction energies, zero-point energies ZPEs unscaled and G 3df, 3pd , and r a q 3-4 4 TS 298 kcal Ž.Ž . Ž . o f H DD while in all other cases B3LYP bb bb 189.84163 6-311 bb 2 2-3 3 y TS , and 0.3 17.7 0.9 2.7 1-2 21.9 21.9 5.1 3.1 24.1 23.2 6.3 3.9 28.0 20.3 4.6 2.8 23.4 21.7 3.2 2.8 27.3 18.0 1.2 1.8 21.2 19.6 1.7 2.1 y y y y y y y ,TS 1 1-2 2 mol. Heats of reaction TS 12244455 34 112232 r 189.76222 6-31G d, p , y 0 19.0 0 20.7 are 19.6 18.6 20.4 18.1 16.8 15.8 21.2 19.9 16.3 20.9 15.0 13.7 14.8 16.5 27.0 46.0 189.65736 21.8 189.57992 21.5 189.35431 20.1 4 y 1 Ž. G 3df, 3pd ZPEs and temperature corrections. For the calculation of Ž. Ž. q r r ry Ž. Ž. Ž. ry r r n Ž . Ž Ž .. Ž Ž .. Ž basis set 6-31G d, p 19.6 18.7 20.4 19.0 17.1 16.1 21.3 20.0 16.5 21.1 15.1 13.7 14.9 16.6 6-31G d, p 6-31G d, p 0 20.4 Ž.Ž . 6-311 G 3df, 3pd G 3df, 3pd G 3df, 3pd x x x r r ry r r r r G 3df, 3pd results see also Table 3 . r Ž. Ž. q q q q 4s3p 2d1f 4s3p2d1f 4s3p2d1f 6-311 B3LYP 6-311 B3LYP 3s2p1d 3s2p1d w structure 3s2p1d w CCSD T CCSD T 6-311 B3LYP w 6-311 r 298 B3LYP Ž. Ž. o 298 CCSD T f Ž. Ž. H o f H MR-AQCC The B3LYP and CCSD T energies of Absolute energies in hartree, relative energies in kcal ZPE B3LYP Energy B3LYP DD obtained with B3LYP b Table 1 Energetics of dioxirane formation and decompositionQuantity according to the mechanism ofEnergy Scheme 1 Method with regard to a c temperature corrections. CC geometries were obtained for D 106 D. Cremer et al.rChemical Physics Letters 292() 1998 97±109 ) ) ) 103.3 98.5 85.0 y y y Ž. Ž. Ž. 98.8 74.8 73.3 73.5 89.0 74.5 88.0 88.0 73.3 73.5 87.1 101.8 y y y y y y y y y y y y Ž. Ž. Ž. Ž. Ž. Ž. Ž. Ž. 5.8 2.8 3-4 4 11.0 y y y TS 298 values are obtained as described in the text. Ž. o f 14.2 H y D 2-3 3 2.2 a y TS from Table 1. 3 mol obtained from reaction energies, zero-point energies ZPEs unscaled and r 19.8 25.0 14.7 3.8 23.619.5 23.113.8 13.1 8.4 3.2 Me 4.0 Me 6.8 25.3 21.024.1 23.6 23.6 11.2 11.220.5 2.7 22.2 H 4.324.4 Me 27.423.4 22.2 9.0 23.1 1.0 7.0 H 11.1 1.0 3.2 4.0 , and 2 y y y y y y y y y y y , 1-2 298 kcal Ž.Ž . Ž . o f ,TS H 1 1-2 2 11 22 3 2 TS DD the migrating group is indicated in parentheses. 3-4 formation and decomposition Scheme 1 1.9 18.1 268.24593 22.6 228.91609228.91120 25.0 18.1 y 2c y 1 y y mol. Heats of reaction r wx Me, synMe, anti 37.4 37.3 36.5 36.2 38.1 36.8 35.6 33.8 H 38.9 35.4 Me 39.0 Me, anti 0Me, syn 16.9 10.6 33.0 Me, syn Me, anti Me, synMe, syn 0Me, anti 0Me, anti 23.8 0 13.6 22.4 15.5 29.1 Me;Me 54.9 53.9 55.4 53.4 53.5 53.1 Me 56.4 Me;Me 0 21.4 Me;Me Me;Me Me;Me 0 20.0 from Ref. 26,27 . For TS b: b: c: b: c: b: b: b: c: c: c: b: b: b: b: Molecule ) and dimethyldioxirane 2b Ž. Ž. Ž . n r r r r Ž. Ž. Ž. 298 values with 6-31G d, p 6-31G d, p basis set 6-31G d, p structure Ž.Ž . o f H D 298 B3LYP 298 Best Ž. Ž. o o 298 f f Ž. H H o f H Absolute energies in hartree, relative energies in kcal ZPE B3LYP D Energy B3LYP DD DD Table 2 Energetics of methyldioxirane QuantityEnergy Method with regard to a Experimental temperature corrections. Best values are obtained by using CC corrections for D. Cremer et al.rChemical Physics Letters 292() 1998 97±109 107 is impossible. On the other hand, the excess energy Methyl group stabilization effects can be mea- of 4 in the gas phase is more than sufficient for an sured by the formal reactions 5 internal rotation of 4a into the HÐOÐC O anti HyXyHqCH CH ™ CH yXyHqCH , form 5a Žactivation enthalpy 11.6 kcalrmol, Table 33 3 4 1 1.Ž and subsequent decomposition into 6a activation Ž. r . y y q enthalpy 65.3 kcal mol, Table 1 . Alternatively, 4a H X CH 333CH CH can directly decompose to COqHOŽ7a, Scheme 2 ™ CH yXyCH qCH , 2 1. via TS4-7 since the corresponding activation en- 334 Ž. thalpyŽ. 70 kcalrmol, Table 1 is lower than the where XsCOO. The largest stabilization is found value of 76.9 kcalrmol needed for decomposition of for 1c Ž.y22.9 kcalrmol, Table 4 since the effect of r 4a via TS4-5a and TS5-6a into 6. Both CO22 , H , a syn-positioned Me group leads to 13 kcal mol of a r CO and H2 O were found in the gas-phase ozonolysis second Me group to 10 kcal mol, where both hyper- of ethenewx 5 , which should proceed via intermediate conjugative and inductive stabilization of the COO carbonyl oxide to these products following the mech- group play a role. As a result of additional stabilizing anism sketched in Scheme 1. On the basis of our through-space interactions between the CH3 group calculations we can conclude that biradical 3a is and the terminal O atom, syn-1b is more stable than kinetically highly unstable and even more difficult to anti-1b by 3 kcalrmol as was already pointed out in detect than its two isomers 2a and 1a. On the other early investigations on carbonyl oxides by Cremer hand, it is a fact that 2 is significantly stabilized by wx34,38 . This is also the reason why syn-1b has a methyl groups and therefore, we have to investigate higher isomerization barrierŽ 22.4 kcalrmol, Table in the following whether this is also true for 3. 2.Ž than 1a while anti-1b has a lower one 15.5 kcalrmol.Ž and that of 1c 20 kcalrmol, Table 2 . is 3.3. Methyl substituted carbonyl oxides, dioxiranes comparable to 1a. From 1 to 4, the stabilization and methylenebis() oxy biradicals effect of two methyl groups decreases from 22.9 Ž.1 to 3.2 kcalrmol Ž.4, Table 4 because the role of the All energy and enthalpy data for the methyl sub- substituents changes. For 2c, steric repulsion be- stituted systems are based on B3LYPr6-31GŽ. d, p tween the Me groups leads to a widening of the calculationsŽ. Table 2 , which were in some cases external ring angleŽŽ. HCH 2a : 115.9; HCC Ž.2b : improved by CCSDŽ. T corrections found for the 116.6; CCCŽ.2c : 118.08, Fig. 1 . and, thereby, to a parent systemŽ. Table 1 . Heats of formation slight decrease of the internal OCO angle and a o D Hf Ž.298 for 1b and 1c were derived from formal stiffening of the OO bond. These effects are also o reactionsŽ. 1 and Ž.Ž 2 see below . and the D Hf Ž.298 found for TS2-3 and biradicals 3 and are the reason value for 1a Ž.see Table 1 . For all other molecules, for an increase in the ring-opening barrierŽ TS2-3b: o o r D HffŽ.298 was derived from the best DDH Ž.298 22.2, TS2-3c: 23.1 kcal mol, Table 2. and a lower o values of Table 2 using the D Hf Ž.298 of 1 as a stability of the methyl substituted biradicals 33bŽ : reference. Comparison with the experimental 7.0, 3c: 11.1 kcalrmol above 2, Table 2. . o D Hf Ž.298 of 4 Žstarred values in Table 2 . suggests Formation of biradical 3 requires an opening of o errors of calculated D Hf Ž.298 values lower than 2 the OCO angle to keep the pair±pair repulsion be- kcalrmol. tween the in-plane electron lone pairs at the O atoms

Table 3 MR-AQCCr6-311qGŽ. 3df, 3pd results obtained from a 6=6 CAS referencea

Molecule MCSCF MR-AQCC Main configuration c12Second configuration c y y 222 222y 2a 188.736406 189.337355 . . .Ž.Ž.Ž. 6a112 2b 1a 0.908 . . . Ž.Ž.Ž. 2b 122 1a 4b 0.165 y y 222 222y TS2-3a 188.767303 189.304992 . . .Ž.Ž.Ž. 6a112 2b 1a 0.843 . . . Ž.Ž.Ž. 2b 122 1a 4b 0.378 y y 222 222 3a 188.725062 189.330069 . . .Ž.Ž.Ž. 6a122 1a 4b 0.790 . . . Ž.Ž.Ž. 6a 112 2b 4b 0.436 a Energies in hartree. For the notation of the MOs see text. The coefficients of the MR-AQCC expansion are denoted as c12and c . 108 D. Cremer et al.rChemical Physics Letters 292() 1998 97±109

Table 4 )21 kcalrmol. caused by the OO locking effect of Stabilization by methyl groups as determined by reactionsŽ. 1 and the substituents. The parent compound is kinetically Ž.2 a unstable at room temperatureŽ activation enthalpy 18 Structure ReactionŽ. 1 Reaction Ž. 2 Sum kcalrmol. and opens to biradical 3a, which rear- 1 y13.5 Ž.y10.4 y9.5 y22.9 ranges with rather low barriers to formic acid, CO y y y y 2 TS1-2 10.0 Ž.13.8 11.9 21.9 qH , and COqH O. For the alkyl substituted 2 y10.4 y8.3 y18.7 22 TS2-3 y10.0 y7.8 y17.8 biradicals 3, ester rearrangement is the preferred 3 y5.5 y4.7 y10.2 decomposition mode, which proceeds with barriers TS3-4 y6.7 y3.7 y10.4 lower than 4 kcalrmol. Accordingly, there is little 4 y9.1 5.9 y3.2 chance to intercept biradical 3 in the decomposition a reaction of 2. Of the three R CO isomers consid- B3LYPr6-31GŽ. d, p values in kcalrmol referring to the syn 22 forms if two forms are possible. Values in parentheses: stabiliza- ered in this work, dioxiranes are the only ones which tion energy for anti form. Energies for methane and ethane are: can be both thermodynamically and kinetically stable y40.52401 and y79.83874 hartree, respectively. B3LYPr6-311 and, therefore, intercepted experimentally. The im- qGŽ. 3df, 3pd values change slightly as, e.g. for 1: y14.1, y10.2, proved D H oŽ.298 values for 2 obtained in this work y r f 24.3 kcal mol. reveal that the dioxirane ring is less strained than previously thought. small. However, OCO widening implies a reduction Acknowledgements of the external angleŽŽ. HCH 2a : 116.7; HCH Ž.3a : 8 . 111.5 , Figs. 1 and 2 , which is hindered if steric Preliminary calculations by S. Andersson are ac- repulsion between the substituents becomes larger. knowledged. This work was supported at GoteborgÈ Kinetic stability of a compound at room temperature University by the Swedish Natural Science Research requires that all activation enthalpies for decomposi- CouncilŽ. NFR . PGS acknowledges financial support tion or rearrangement reactions are higher than 21 from FKFP, Grant No. 0145r97. Calculations were r kcal mol, which means that dioxirane 2c is kineti- done on a CRAY C90 of the Nationellt Superdator- cally stable at room temperature since the two methyl centrumŽ. NSC , Linkoping,È Sweden. DC and EK groups lock the OO atoms in a bond. This effect thank the NSC for a generous allotment of computer should be more pronounced with the increasing steric time. bulk of the substituents, as was verified by the fact that bismesityldioxirane forms stable crystals at room temperaturewx 9 . References Biradicals 3 decompose via H or Me group migra- tionŽ. ester rearrangement , where the former process wx1 W. Sander, Angew. Chem., Int. Ed. Engl. 29Ž. 1989 344. is considerably faster as is suggested by the activa- wx2 R.W. Murray, Molecular structure and energetics, in: J.F. tion enthalpies calculated for H and Me migration in Liebman, A. GreenbergŽ. Eds. , Unconventional Chemical 3b Ž.1 vs. 4 kcalrmol, Table 2 . The decomposition Bonding, vol. 6, VCH Publishers, New York, 1988, p. 311. of 3b or 3c via TS3-6 is a high-energy process, wx3 R.W. Murray, Chem. Rev. 89Ž. 1989 1187. wx which we did not study further since it cannot com- 4 S.A. Kafafi, R.I. Martinez, J.T. Herron, Molecular structure and energetics, in: J.F. Liebman, A. GreenbergŽ. Eds. , Un- pete with the ester rearrangement. It is interesting to conventional Chemical Bonding, vol. 6, VCH Publishers, note that Me group migration leads to an increase of New York, 1988, p. 283. the dipole moment from 2.7 to 3.5 DŽ. Fig. 2 , which wx5 D. Cremer, J. Gauss, E. Kraka, J.F. Stanton, R.J. Bartlett, suggests that the ester rearrangement is slightly Chem. Phys. Lett. 209Ž. 1993 547. wx favoured in solvents with large dielectric constants. 6 R.D. Suenram, F.J. Lovas, J. Am. Chem. Soc. 100Ž. 1978 5117. In conclusion, we state that dialkyl substituted wx7 G. Bucher, W. Sander, Chem. Ber. 12Ž. 1992 1851. dioxiranes can be synthesized in solution because of wx8 A. Russo, D.D. DesMarteau, Angew. Chem., Int. Ed. Engl. their increased kinetic stabilityŽ activation enthalpies 32Ž. 1993 905. D. Cremer et al.rChemical Physics Letters 292() 1998 97±109 109

wx9 W. Sander, K. Schroeder, S. Muthusamy, A. Kirschfeld, W. wx28 E. Kraka, J. Gauss, J. Grafenstein,È F. Reichel, L. Olsson, Z. Kappert, R. Boese, E. Kraka, C.P. Sosa, D. Cremer, J. Am. Konkoli, Z. He, D. Cremer, COLOGNE 96, Goteborg,È 1996. Chem. Soc. 119Ž. 1997 7265. wx29 J.F. Stanton, J. Gauss, J.D. Watts, W.J. Lauderdale, R.J. wx10 R. Curci, M. Fiorentino, L. Troisi, J.O. Edwards, R.H. Pater, Bartlett, ACES II, Quantum Theory Project, University of J. Org. Chem. 45Ž. 1980 4758. Florida, Gainesville, FL, 1992. wx11 R.W. Murray, R. Jeyaraman, J. Org. Chem. 50Ž. 1985 2847. wx30 R. Shepard, I. Shavitt, R.M. Pitzer, D.C. Comeau, M. Pep- wx12 R. Gutbrod, E. Kraka, R.N. Schindler, D. Cremer, J. Am. per, H. Lischka, P.G. Szalay, R. Ahlrichs, F.B. Brown, J.-G. Chem. Soc. 119Ž. 1997 7330. Zhao, Int. J. Quantum Chem. S 22Ž. 1988 149. wx13 R. Gutbrod, R.N. Schindler, E. Kraka, D. Cremer, Chem. wx31 M.J. Frisch, G.W. Trucks, H.B. Schlegel, P.M.W. Gill, B.G. Phys. Lett. 252Ž. 1996 221. Johnson, M.A. Robb, J.R. Cheeseman, T. Keith, G.A. Peters- wx14 T.A. Dix, S. Benkovic, Acc. Chem. Res. 21Ž. 1988 101. son, J.A. Montgomery, K. Raghavachari, M.A. Al-Laham, wx15 S.-J. Kim, H.F. Schaefer III, E. Kraka, D. Cremer, Mol. V.G. Zakrzewski, J.V. Ortiz, J.B. Foresman, C.Y. Peng, P.Y. Phys. 88Ž. 1996 93. Ayala, W. Chen, M.W. Wong, J.L. Andres, E.S. Replogle, R. wx16 E. Kraka, Z. Konkoli, D. Cremer, J. Fowler, H.F. Schaefer Gomperts, R.L. Martin, D.J. Fox, J.S. Binkley, D.J. DeFrees, III, J. Am. Chem. Soc. 118Ž. 1996 10595. J. Baker, J.P. Stewart, M. Head-Gordon, C. Gonzalez, J.A. wx17 K. Raghavachari, G.W. Trucks, J.A. Pople, M. Head-Gordon, Pople, GAUSSIAN 94, Revision B.3, Gaussian, Inc., Pitts- Chem. Phys. Lett. 157Ž. 1989 479. burgh, PA, 1995. wx18 T.H. Dunning Jr., J. Chem. Phys. 90Ž. 1989 1007. wx32 D. Cremer, E. Kraka, in: J.F. Liebman, A. GreenbergŽ. Eds. , wx19 P.G. Szalay, R.J. Bartlett, Chem. Phys. Lett. 214Ž. 1993 481. Molecular Structure and Energetics, Structure and Reactivity, wx20 R. Krishnan, M.J. Frisch, J.A. Pople, Chem. Phys. 72Ž. 1980 vol. 7, VCH Publishers, Deerfield Beach, FL, 1988, p. 65. 4244. wx33 D. Cremer, E. Kraka, K.J. Szabo, in: Z. RappoportŽ. Ed. , The wx21 P.G. Szalay, R.J. Bartlett, J. Chem. Phys. 103Ž. 1995 3600. Chemistry of Functional Groups, The Chemistry of the Cy- wx22 P. G. Szalay, in: R.J. BartlettŽ. Ed. , Modern Ideas in Cou- clopropyl Group, vol. 2, Wiley, New York, 1995, p. 43. pled-Cluster Methods, World Scientific, Singapore, 1997, p. wx34 D. Cremer, J. Am. Chem. Soc. 101Ž. 1979 7199. 81. wx35 D. Cremer, J.A. Larsson, E. Kraka, Theoretical and computa- wx23 A.D. Becke, J. Chem. Phys. 98Ž. 1993 5648. tional chemistry, in: C. ParkanyiŽ. Ed. , Theoretical Organic wx24 L. Noodleman, D.A. Case, Adv. Inorg. Chem. 38Ž. 1992 423. Chemistry, vol. 5, Elsevier, Amsterdam, 1998, p. 259. wx25 P.C. Hariharan, J.A. Pople, Chem. Phys. Lett. 66Ž. 1972 217. wx36 R.D. Bach, J.L. Andres, A.L. Owensby, H.B. Schlegel, wx26 J.D. Cox, G. Pilcher, Thermochemistry of Organic and J.J.W. McDouall, J. Am. Chem. Soc. 114Ž. 1992 7207. Organometallic Compounds, Academic Press, London, 1970. wx37 M. Cantos, M. Merchan,ÂÂ F. Tomas-Vert, B.O. Roos, Chem. wx27 J.B. Pedley, R.D. Naylor, S.P. Kirby, Thermochemical Data Phys. Lett. 229Ž. 1994 181. of Organic Compounds, 2nd edn., Chapman and Hall, New wx38 D. Cremer, E. Kraka, M.L. McKee, T.P. Radhakrishnan, York, 1986. Chem. Phys. Lett. 187Ž. 1991 491.