An ab initio and DFT study of the fragmentation and isomerisation of MeP(O)(OMe)+

Article

Reference

An ab initio and DFT study of the fragmentation and isomerisation of MeP(O)(OMe)+

BELL, A. J., et al.

Abstract

The fragmentation behaviour of the ion MeP(O)OMe+ has been investigated using quantum mechanical calculations at the B3LYP and MP2 levels to support experiments made with an Ion Trap Mass Spectrometer. Two mechanisms for the loss of CH2O are found, one involving a 1,3-H migration to phosphorus and the other a 1,2-methyl migration to give P(OMe)2+

followed by a 1,3-H migration. In each case an ion-dipole complex is formed that rapidly dissociates to yield CH2O. The relative importance of each route has been previously determined experimentally via isotopic labelling experiments, and the theoretical results are found to be consistent with these experimental results. The mechanisms suggested in the earlier work involving a 1,4 H migration to O are shown to be energetically unfavourable.

BELL, A. J., et al . An ab initio and DFT study of the fragmentation and isomerisation of MeP(O)(OMe)+. Physical Chemistry Chemical Physics , 2004, vol. 6, no. 6, p. 1213-1218

DOI : 10.1039/b315944b

Available at:

http://archive-ouverte.unige.ch/unige:3654

Disclaimer: layout of this document may differ from the published version.

An ab initio and DFT study of the fragmentation and isomerisation of MeP(O)(OMe)

A. J. Bell,^aA. Citra,^bJ. M. Dyke,^bF. Ferrante,^cL. Gagliardi^cand P. Watts*^a

a Dstl, Porton Down, Salisbury, Wiltshire, UK SP4 0JQ. E-mail: pwatts.rsg@ukgateway.net

b Department of Chemistry, Southampton University, Southampton, UK SO7 1BJ

c Dipartimento di Chimica Fisica ‘‘ F. Accascina ’’, Universita` degli Studi di Palermo, 90128 Palermo, Italy

Received 9th December 2003, Accepted 22nd January 2004 F|rst published as an Advance Article on the web 17th February 2004

The fragmentation behaviour of the ion MeP(O)OMe^þhas been investigated using quantum mechanical calculations at the B3LYP and MP2 levels to support experiments made with an Ion Trap Mass Spectrometer.

Two mechanisms for the loss of CH₂O are found, one involving a 1,3-H migration to phosphorus and the other a 1,2-methyl migration to give P(OMe)2þfollowed by a 1,3-H migration. In each case an ion-dipole complex is formed that rapidly dissociates to yield CH₂O. The relative importance of each route has been previously determined experimentallyviaisotopic labelling experiments, and the theoretical results are found to be consistent with these experimental results. The mechanisms suggested in the earlier work involving a 1,4 H migration to O are shown to be energetically unfavourable.

Introduction

In an initial investigation into the electrospray ionisation ion trap mass spectrometry of simple organophosphates esters,¹ we reported that the collision induced fragmentation of proto- nated dimethyl methylphosphonate, I, resulted in the loss of methanol and the formation of the CH3P(O)OCH3þion, II, which in turn fragmented under appropriate collision induced dissociation (CID) conditions to lose CH₂O (see Fig. 1).

It was observed that fragmentation of the deuterated ion CD3P(O)OCH3þ formed from an isotopomer of I, Ia (see below) produces a mixture of CH₂O and CD₂O, indicating that a scrambling of the methyl groups is involved during frag- mentation. Scrambling did not however occur in II prior to fragmentation as only unlabelled methanol was eliminated fromIIIformed from the isotopomerIa. More detailed studies of this scrambling involving the isotopomersIa,Ib andIc

were made and mechanistic schemes were proposed but were not investigated further.²In the present work the scrambling of the methyl groups on elimination of formaldehyde fromII has been investigated using quantum mechanical calculations at the B3LYP and MP2 levels in an attempt to understand the mechanism(s) of fragmentation of II. The structure of

one of the postulated intermediate ions, MeOPOMe^þhas been studied previously at the B3LYP level with a smaller basis set than that used in the present work.³Mass spectrometric^4,5and computational⁵ studies of phosphonate ions have been reported but as these were concerned with odd-electron ions it is difficult to relate the results of these studies to the present work although superficially they appear similar.

Computational details

All electronic structure calculations were performed using the Gaussian 98 programme.⁶ DFT(B3LYP) and MP2 calcula- tions have been performed with a 6-311þG (2d,2p) basis set for all atoms. The approach adopted in these calculations can be described as follows: geometries were initially optimized at the DFT/B3LYP level with a 6-31G** basis set. Further geometry optimizations and frequency calculations, reported in this work, have been performed at the B3LYP/6- 311þG(2d,2p) and MP2/6-311þG(2d,2p) levels. Structures were optimized with standard convergence criteria and vibra- tional frequencies were calculated in all cases to characterize the stationary points and provide zero point energies (ZPE).

All total energy comparisons include ZPE corrections.

Transition states have been located with the synchronous transit-guided quasi-Newton (QST2) method. In the DFT calculations, a grid of 99 radial points and 590 angular points, together with the Becke weighting scheme of integration, have been used.

All the ions considered in this work have an even number of electrons and both closed shell singlet and open-shell triplet state calculations were performed. In all cases, the triplet states are much higher in energy ( > 100 kJ mol¹) (see also ref. 3) and are not considered to be important in understanding the chemistry of this reaction system.

Results

The results of calculations for all relevant, non-isotopically labelled isomers and transition states are summarised in Table 1 and Figs. 12 and 13, later.

Fig. 1

PCCP

www.rsc.org/pccp

R E S E A R C H P A P E R

DOI:10.1039/b315944b

MeP(O)OMe⁺(II)

The geometries of II were optimised from different starting points but all lead to approximately planar structures with methyl groups eithertrans,IIt, orcis,IIc, to each other, (see Fig. 2) the former being 27.0 kJ mol¹and 27.1 kJ mol¹lower in energy at the B3LYP and MP2 levels respectively. This is most likely due to steric interactions between the methyl groups in the cis isomer that are not present in thetransiso- mer. Whilst the barrier for interconversion between IIt and IIc has not been investigated it is expected to be considerably smaller than those associated with the fragmentation ofIIor its isomerisation to VI. The cis minimum energy structure for II differs from the structure given in an earlier study³ at the B3LYP/6-31G* level because a larger basis set has been used in the present work.

P(OMe)2+

(VI)

In the earlier study,²this ion (not shown in Fig. 1 but see Figs.

10 and 11, later) was suggested as the intermediate structure in the scrambling of the methyl groups. Two stable conforma- tions ofVIshown in Fig. 3 have been found in this study; a trans–trans,VItt, form and a lower symmetry trans-cis form, VItc, that allows a hydrogen bonding interaction between a hydrogen in one methoxy group with the oxygen atom in the other methoxy group. The latter is found to be more stable

by 10.4 kJ mol¹and 10.8 kJ mol¹at the B3LYP and MP2 levels respectively. Again the barrier to their interconversion has not been determined but will be much smaller than those associated with their fragmentation.

Methyl shift in the transition II!VI

IIt is more stable thanVItc by 6.1 kJ mol¹and 28.3 kJ mol¹ andIIc is more stable thanVItt by 10.6 kJ mol¹and 12.1 kJ mol¹at the B3LYP and MP2 levels. The energy of the transi- tion state for the 1,2-methyl shift reactionIIt!VItc is 220.9 kJ mol¹and 245.0 kJ mol¹above that ofIIt, and the transi- tion state forIIc!VItt 202.8 kJ mol¹ and 229.6 kJ mol¹ above that ofIIc using B3LYP and MP2 levels. The structures of the transition states are shown in Fig. 4.

The non-observation in the experimental work of the reverse reactionVI!II, as demonstrated by the lack of scrambling in II, is interesting. The slightly lower values of the transition state for the reaction VI to X (see later) compared to that between II and VI (22 and 5 kJ mol¹ at the B3LYP and MP2 levels respectively) is sufficient to favour the forward reaction to the exclusion of the reverse reaction.

Fragmentation of MeP(O)OMe⁺, II

Direct loss of CH2O from II was suggested in the earlier experimental study²to occur in one stepviaa cyclic transition state in which a hydrogen atom on the methoxy ligand under- goes a 1,4-H migration to the oxo oxygen with a concurrent breaking of the P–O bond to release formaldehyde. The pre- sent calculations indicate that the fragmentation ofIIby this mechanism (if it occurred) proceeds through a stable ion–

dipole complex CH₂O.(MeP(OH))^þ(VII) which subsequently loses CH2O to formIV.

Two minimum energy conformations of VII that differ in the orientation of the O–H bond were found, differing in energy by 1.7 and 1.9 kJ mol¹at the B3LYP and MP2 levels respectively. These are shown in Fig. 5.

The lowest energy form of VIIwas found to be less stable than IIt by 22.1 and 44.8 kJ mol¹ with the transition state for IIt!VII (shown in Fig. 6) higher in energy than IIt by 312.1 and 311.4 kJ mol¹at the B3LYP and MP2 levels respectively.

A second possible reaction pathway involves a 1,3-H migra- tion by a methoxy hydrogen atom to phosphorus to give a complex isomeric with VII, CH₂O.(MePH(O)H)^þ (VIII) see Fig. 7.

VIII was found to be 49.1 kJ mol¹ and 56.8 kJ mol¹ higher in energy thanIIt at the B3LYP and MP2 levels respec- tively, and so is less stable thanVII.

Fig. 2 Structures ofIIt andIIc.

Fig. 3 VItc andVItt.

Fig. 4 Transition states forIIt!VItc andIIc!VItt.

Table 1 Energetics (in kJ mol¹) of three reaction paths for the fragmentation ofIIat B3LYP and MP2 levels of theory. All values are referred to the energy ofIIt; ZPV corrected energies are enclosed between parenthesis

B3LYP/

6-311þG(2d,2p)

MP2/

6-311þG(2d,2p) Reaction:IIt!VII!IVþFormaldehyde

TS 328.9 (312.1) 329.1 (311.4)

VII

Structure A 28.9 (23.8) 51.3 (46.7)

Structure B 26.2 (22.1) 48.5 (44.8)

IVþFormaldehyde 158.2 (139.3) 191.5 (172.4) Reaction:II!VI!X!XIIþFormaldehyde

TSIIt!VItc 228.0 (220.9) 251.2 (245.0)

VItc 2.8 (6.1) 24.5 (28.4)

IIc 27.4 (27.1) 27.3 (27.1)

TSIIc!VItt 237.6 (230.3) 263.0 (256.7)

VItt 13.4 (16.5) 35.4 (39.2)

TSVItc!Xt 208.5 (197.7) 249.6 (239.9)

Xt 125.2 (119.1) 153.8 (148.7)

TSVItt!Xc 211.2 (200.3) 256.1 (247.0)

Xc 121.2 (114.8) 148.9 (143.7)

XIItþFormaldehyde 263.9 (245.3) 299.3 (281.1) XIIcþFormaldehyde 264.2 (245.1) 300.5 (281.8) ReactionIIt!VIII!XIþFormaldehyde

TS 222.4 (210.5) 244.2 (232.5)

VIII 55.6 (49.1) 62.9 (56.8)

XIþFormaldehyde 233.0 (209.4) 247.6 (233.2)

The transition state forIIt!VIII(see Fig. 8) is 210.5 and 232.5 kJ mol¹aboveIIt at the B3LYP and MP2 levels. This is much lower in energy than the transition state forIIt!VII and thus the 1,3-H migration to P rather than the 1,4-H migra- tion to O will be the preferred mechanism although it has greater steric constraints. This is despite the considerably lower energy of the final products viathe 1,4-H migration,IVand CH2O, 139.3 and 172.4 kJ mol¹ higher than IIt at the B3LYP and MP2 levels compared with the final productsvia the 1,3-H migration,XIand CH₂O, 209.4 and 233.2 kJ mol¹ higher thanIIt at the B3LYP and MP2 levels.

Fragmentation of P(OMe)₂^þVI

In the earlier experimental work²it was also suggested that the dimethoxyphosphenium ion P(OMe)₂^þ (VI) fragments via a concerted 1,4-H migration and loss of CH₂O. The present cal- culations show that this is a two step process, analogous to that for MeP(O)OMe^þ leading to VII, resulting in an ion- dipole complex MeOH.(POCH₂)^þ(IX). This complex would be expected to fragment to lose MeOH, and not CH2O. IX is 259.5 and 321.6 kJ mol¹ higher in energy than MeP(O)OMe^þ, and the transition state betweenVItc!IXlies

315.7 kJ mol¹aboveVItc at the MP2 level. This would dis- sociate into methanol and the ion PO^þCH2 with an energy 506.5 kJ mol¹aboveIIt at the MP2 level. This is by far the most endoergic route and it is not surprising that it is not observed. As with II an alternative mechanism involving a 1,3-H migration to P is possible, leading to the intermediate CH2O.(HP(O)Me)^þ (X). Calculations show that only the hydrogens on the left hand carbon in VItc shown in Fig. 3 can migrate to phosphorus. This structure has been optimised and two conformations (see Fig. 9) were found to differ in energy by 4.3 and 4.0 kJ mol¹at the B3LYP and MP2 levels respectively.

The lowest energy conformation (on the left of Fig. 9) is higher in energy thanIIt by 114.8 and 143.7 kJ mol¹at the B3LYP and MP2 levels respectively.

The transition state for the reactionVItc!Xis calculated to be 197.7 and 239.9 kJ mol¹at the B3LYP and MP2 levels respectively. The results of these calculations are consistent with reactionVItc!IXnot being observed and withVItc!X being the preferred pathway for the fragmentation of VI (either tc or tt). It should be remembered that sufficient energy is present inVIto make the interconversion betweenVItc and VItt rapid thus making (in the absence of isotopic effects) the methyl groups equivalent. The importance of this will become apparent when the fragmentation of the isotopomers is discussed.

The combined energy of the fragments CH2OþXIIis 130.3 and 138.1 kJ mol¹higher than that ofXat the B3LYP and MP2 levels respectively.

Discussion

The reaction scheme originally proposed² for the fragmenta- tion of MeP(O)OMe^þII is shown in Fig. 10 with k1and k2

being comparable and rate limitingi.e.much slower than k3

and k₄.

The reaction pathways considered in detail in the present work are shown in Fig. 11 with the relative energies shown dia- grammatically in Figs. 12 and 13 at the B3LYP and MP2 levels respectively. As was seen in the Results section, formation of IVa is not energetically feasible due to a large barrier and an isomer,XI, is formed instead. Similarly structuresIVc/d were shown not to be formed as their immediate precursor(s) would have eliminated MeOH not CH2O and an isomer ofXwould have formed instead.

At this point it is useful to ask: are the results of the present calculations consistent with the experimental results of the earlier studies?

Although the mechanisms of formaldehyde elimination and the structures of the product ions (with the exception ofIV, see Fig. 11) are different from those originally proposed,² the kinetic scheme is the same if it is assumed that the breakdown of the ion–dipole complexes occurs rapidly. Whilst the energetics suggest that this rapid breakdown rather than rever- sion back to the precursor ion is marginal in the case ofVIII, dissociating toXIand CH₂O, and unlikely in the case of X, dissociating toXII and CH2O, three additional factors need to be taken into consideration. The first is the neglect so far of entropic effects. Using the computed entropies obtained Fig. 6 Transition state forIIt!VII.

Fig. 7 Structure of ion–dipole complexVIII.

Fig. 8 Transition state forIIt!VIII.

Fig. 9 Conformations of the ion–dipole complexX.

Fig. 5 Structures of the ion–dipole complexVII.

from the quantum mechanical calculations listed in Table 2 and assuming a temperature of 1200 K (see later), an entropic contribution, TNS, of 172 kJ mol¹ is calculated for the dissociation of X. This not only makes the dissociation step more favourable than the reverse step but makes the dissocia- tion ofXexoergonic and thusVIwill go straight toXIIand CH₂O. Similar considerations apply to the formation and dis- sociation ofVIII. The second is that whilst the energetics of the dissociation process allow it to be reversible, it is in practice irreversible as the background concentration of formaldehyde is effectively zero. The third factor to be taken into consideration

is that all of the ionic species with the exceptions ofIV,XI, and XII are maintained at the same suprathermal energy in the ITMS whilst the excitation voltage is applied (as it is during the reaction period) but IV,XI, and XII, once formed, are rapidly cooled to ambient temperature by collisions with the helium bath gas. This leaves the stepsIIt!VIandIIt!VIII rate limiting.

In the earlier work²it was proposed that for the unlabelled II or its ¹³C labeled isotopomer that k1/k2¼0.5 (and k3/ k4¼1.0). As at first approximation ca. 90% fragmentation ofIIoccurs in 30 ms (even at the lowest fragmentation energies Fig. 11 Presently proposed scheme for the fragmentation ofII.

Fig. 12 Energetics in kJ mol¹ for the reactions shown in Fig. 11 calculated at the B3LYP level referenced to IIt (see earlier Figures for structures).

Fig. 13 Energetics in kJ mol¹for the reactions shown in Fig. 11 calculated at the MP2 level referenced toIIt (see earlier Figures for structures).

Fig. 10 Original scheme²for the fragmentation ofII. The superscripts on the methyl groups serve to allow the various isotopic labels to be referenced. The rate coefficients will be discussed later.

used) giving an overall rate (i.e. k₁þk₂) of at least 80 s¹, and since the ions being fragmented are only ‘ on resonance ’i.e.at suprathermal energies for a fraction of this time, the rates are likely to be much greater than this. Taking the energetics for the transition states for the stepsIIt!VIandIIt!VIIIfrom Table 1 and the associated entropies from Table 2, the rates of fragmentation can be calculated from the usual transition state equations A¼(ekBT/h)exp(NS*/R) and k¼Aexp(E*/

RT). It is found that in order to obtain rates80 s¹a tem- perature of1200 K must be assumed. This is of course the internal temperature of the excited species and not that of the ion trap. Similar high internal temperatures of ions in an ion trap have been proposed previously from modelling stu- dies.⁷ The ratio k₁/k₂ resulting from such calculations is 0.1 and not the 0.5 suggested by the experimental data. If however the entropy of activation for the TS IIt!VI is assumed to be16.8 kJ mol¹K¹rather than3.8 kJ mol¹ K¹calculated from the data in Table 2 then the expected ratio ofk1/k2is obtained. This value of16.8 kJ mol¹K¹is more in accord with the other transition states and is not unreason- able given that only the single reference method with a modest basis set has been used. A multireference method, such as CASSCF/MRCI, with a larger basis set would certainly be more appropriate to calculate the energy and entropy change fromIIt to the TSIIt!VIand this is probably the major rea- son for the inconsistency with the experimental results. Also it should be borne in mind that in calculatingNSfor this process only the harmonic oscillator vibrational frequencies have been used and this is expected to be poor for low frequency vibrational modes.

When the deuterium labelled isotopomers ofI, and therefore ofII, were investigated a significant kinetic isotope effect was observed, although, through lack of temporal data, it was only manifested in the product ratios i.e. the fraction of ¹Me retained. The energetics for the transition states for the steps IIt!VI,IIt!VIIIandVI!Xcalculated at the B3LYP level for the three isotopomers together with the normal isotope containing species are given in Table 3. As was expected the

13C containing isotopomer shows a negligible isotope effect.

Using the B3LYP activation energies and entropies (that for the TS IIt!VI was adjusted from 1.4 kJ mol¹ K¹ to 17.1 kJ mol¹K¹to give the correct ratio ofk₁/k₂as for the rate calculations described above) the fraction of ¹Me

retained in the product ions is calculated for the three isotopo- mers and given in Table 4.

The calculated and observed data are not in quantitative agreement but that the same trend is shown is very satisfactory given that there are aspects of the experimental results that are ill-understood.

Conclusions

A study of the organophosphate ion MeP(O)OMe^þhas been carried out by quantum mechanical calculations in order to understand its fragmentation behaviour in an ion trap mass spectrometer. The ion fragments to lose CH₂O and isotopic labelling experiments have showed previously that two com- peting mechanisms are involved, one of which must involve a scrambling of the methyl groups. This was proposed in the paper describing the original experimental work² to occur through the dimethoxyphosphenium ion P(OMe)2þ formed from MeP(O)OMe^þviaa 1,2-methyl migration. 1,4-H migra- tions in both MeP(O)OMe^þ and P(OMe)₂^þ were proposed to explain the elimination of formaldehyde.

All isomers have been studied using electronic structure calculations at the B3LYP and MP2 levels and fragmentation pathways from each were considered. In each case barriers to the originally proposed 1,4-H migration steps leading to loss of CH₂O are found to be much higher in energy than 1,3-H migrations in which a methoxy hydrogen atom migrates to the phosphorus atom. These produce the ion–dipole complexes CH₂O.(MeP(O)H)^þand CH₂O.(MePOH)^þwhich then dissoci- ate to give the experimentally observed products. These cal- culations have, therefore, led to a much greater understanding of the mechanisms of fragmentation and isomerisation of the ion MeP(O)OMe^þobserved in an ion trap mass spectrometer.

Acknowledgements

Support from the EC ‘‘ Reactive Intermediates ’’ Research Training Network and the Leverhulne Trust and from the Ministero dell’Istruzione, dell’Universita` e della Ricera, MIUR is gratefully acknowledged. We thank Centro Universitario di Calcolo, Palermo, for the computer time on the Linux cluster.

The authors are grateful to acknowledge the valuable contri- butions made by Peter Lewis in the early stages of this work before his untimely death.

Table 2 Computed entropies (J mol¹K¹) of ionic species and some transition states considered in this work

Comp B3LYP/6-311þG(2d,2p) MP2/6-311þG(2d,2p)

IIt 353.6 352.0

IIc 350.7 346.9

VIIa 346.9 342.3

VIIb 339.0 335.7

TS IIt!VII 324.4 323.1

IV 287.2 298.0

VItc 351.1 356.6

VItt 347.4 345.3

TS IIt!VItc 355.3 348.2

TS IIc!VItt 357.4 348.2

Xt 354.5 351.5

Xc 354.9 352.4

TS VItc!Xt 335.7 334.8

TS VItt!Xc 340.7 334.8

XIIt 277.1 277.1

XIIc 277.6 277.1

VIII 333.6 331.1

TS IIt!VIII 320.2 318.5

XI 280.9 280.5

CH2O 218.2 218.2

Table 3 Energetics (kJ mol¹) of the two main reaction paths for unlabelledII and its three isotopomers calculated at the B3LYP/6- 311þG(2d,2p) level corrected for ZPV. All energies are referred to the energy of theIIt for the particular isotopomer

Normal isotopes ¹³C CD3P POCD3

TS IIt!VItc 220.9 220.6 220.5 220.6

TS VItc!Xt 197.7 197.4 199.4 197.5

TS IIt!VIII 210.5 210.1 210.0 213.5

Table 4 Comparison of the calculated and experimentally observed fraction of¹Me retained in the product ion in the fragmentation of

1MePOMe^þ,I

Position of label in isotopomer ¹³CH3P POCD3 CD3P Calculated retention of¹Me 0.64 0.58 0.68 Mean experimentally observed

retention of¹Me

0.66 0.57 0.78

References

1 A. J. Bell, D. Despeyroux, J. Murrell and P. Watts,Int. J. Mass.

Spectrom. Ion Proc., 1997,165/166, 5333.

2 J. D. Barr, A. J. Bell, D. O. Konn, J. Murrell, C. M. Timperley, M. J. Waters and P. Watts,Phys. Chem. Chem. Phys., 2002,4, 2200.

3 K. K. Thoen, L. Gao, T. D. Ranatunga, P. Vainiotalo and H. I.

Kenttamaa,J. Org. Chem., 1997,62, 8702.

4 J. R. Holtzclaw and J. R. Wyatt, Org. Mass Spectrosc., 1988, 23, 261.

5 L. N. Heydorn, Yun Ling, G. de Oliveira, J. M. L. Martin, C.

Lifshitz and J. K. Terlow, Z. Phys. Chem., 2001, 215, 141;

L. N. Heydorn, C. Y. Wong, R. Srinivas and J. K. Terlow,Int.

J. Mass Spectrom., 2003,225, 11; L. N. Heydorn, P. C. Burgers, P. J. A. Ruttink and J. K. Terlow,Int. J. Mass Spectrom., 2003, 228, 759.

6 M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A.

Robb, J. R. Cheeseman, V. G. Zakrzewski, J. A. Montgomery, R. E. Stratmann, J. C. Burant, S. Dapprich, J. M. Millam, A. D.

Daniels, K. N. Kudin, M. C. Strain, O. Farkas, J. Tomasi, V.

Barone, M. Cossi, R. Cammi, B. Mennucci, C. Pomelli, C. Adamo, S. Cliffoird, J. Ochterski, G. A. Petersson, P. Y. Ayala, Q. Cui, K.

Morokuma, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B.

Foresman, J. Cioslowsli, J. V. Ortiz, B. B. Stefanov, G. Liu, A.

Liashenko, P. Piskorz, L. Komaromi, R. Gomperts, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng, A.

Nanayakkara, C. Gonzalez, M. Challacombem, P. M. Gill, B. G.

Johnson, W. Chen, M. W. Wong, J. L. Andres, M. Head-Gordon, E. S. Replogle and J. A. Pople, Gaussian 98, Gaussian Inc., Pittsburgh, PA, 1988.

7 D. E. Goeringer and S. A. McLuckey,J. Chem. Phys., 1996,104, 2214.