J . Phys. Chem 1986. YO. 1582-1585
1582
Experimental and Theoretical Study of [C,H,O]+* Cations. Evidence for the Existence of Stable [CH=CH-CO]+* Ions in the Gas Phase Guy Bouchoux, Yannik Hoppilliard,* Jean-Pierre Flament, Laboratoire de SynthPse Organique, Ecole Polytechnique. 91 I28 Palaiseau Cedex, France
Johan K. Terlouw, and F. van der Valk Analytical Chemistry Laboratory, University of Utrecht, 3522 AD Utrecht, The Netherlands (Received: October 15, 1985)
Ab initio molecular orbital calculations using polarization basis sets and incorporating valence-electron correlation have been used to determine structures and relative energies of isomers in the C3H20+.potential energy surface. In addition to the high-energy keto/enol species [HC=C-CHO]+./ [HC=C-C-OH]+. and the experimentally characterized propadienone and cyclopropenone radical cations, an isomer which has no neutral stable counterpart, [CH=CH-C=O]+., is reported to be a stable species. Experimental evidence is presented that this ion is generated by the gas-phase dissociative ionization of maleic anhydride and cyclopentene-3,5-dione. Both the calculations and the experimental observations show that its heat of formation is close to that of the most stable C3H20+.isomer, [propadienone]+..
Introduction The structure and stability of neutral C3H20molecules has been the subject of some recent theoretical investigation^.'-^ In contrast, [C3H20]+.ions have not been studied in great detail. The He I photoelectron spectrum of cyclopropenone has been discussed by Harshbarger et a1.: who also reported its electron impact mass spectrum. More recently, the ionization energy of pyrolytically generated methyleneketene was measured and its ionic heat of formation was e~tablished.~ We report here experimental and theoretical evidence for the existence of a not yet identified [C,H20]+. isomer, [CH= CH-CO]'., as a stable species in the gas phase. The experimental data come from mass spectrometric measurements and they will be presented first. The second part of the paper deals with a molecular orbital study of the new [C3H20]+-ion, [2]+., and its most important isomers: r
[21+'
C11+' CCH~C-C-OHl+' 14 1
>+.
C31"
CCH~C-CHOI+' El+'
Experimental Results Collisional Activation Spectra. Collisional activation (CA) mass spectrometry is now a well-established method for the structure analysis of charged and neutral species.61' This technique was used in order to characterize the [C3H20]+.ions coming from (i) direct ionization of methyleneketene (reaction a) and (ii) dissociation of ionized maleic anhydride and cyclopentene-3,5dione (reaction b). ( I ) Komornicki, A.: Dykstra, C. E.: Vincent, M. A,; Radom, L. J . Am. Chem. SOC.1981, 103, 1652. (2) Farnell, L.; Radom, L. Chem. Phys. Lett. 1982,91, 383; 1983,99, 516. (3) Greenberg, A.; Tomkins, R . P. T.; Dobrovolny, M.; Liebman, J . F. J . Am. Chem. Soc. 1983. 105. 6855. (4) Harshbarger, W. R.: Kuebler, N. A,; Robin, M.R. J . Chem. Phyf.. 1974, 60, 345. ( 5 ) Terlouw, J. K.; Holmes, J. L.; Losing, F. P. Can. J . Chenz. 1983, 61,
1722. (6) Levsen, K.; Schwarz, H. Mass. Spectrom. Reo. 1983, 2, 7 7 . ( 7 ) Holmes, J. L. Org. Mass Spectrom. 1985, 20, 169.
x =CHI .o
The measurements were made using the Utrecht University ZAB-2F mass spectrometer using helium as target gas as described in ref 8. Neutral methyleneketene (1) was pyrolytically generated from acrylic trifluoroacetic anhydride (reaction a).'5 Figure 1 shows the CA spectra of the m/z 54 [C3H20]+.ions, [A]'. and [B]'.. Both spectra are dominated by peaks at m / z 53 and m/z 26 which are also present in the unimolecular dissociation spectra. These two peaks can readily be explained by the formation of [CHCCO]' and [CH=CH]+. ions. The two dissociation processes may well need the same amount of energy as is indicated by the calculated heats of formation of the corresponding final states: AHfo([CHCCO]+)5+ AHfo(H.) = 1197 kJ.mol-' AHfo([CH=CHlf.)
+ AHfo(CO) = 1215 kJ-mo1-I
However, differences are observed between the CA spectra of ions [A]'. and [B]'. in the m/z 12-14 and m / z 40-41 regions, i.e., for those dissociations which only involve one carbon atom. These peaks are expected to give information about the localization of the two C-H bonds in the stable [A]'. and [B]'. ions. It appears that [A]'. upon collision yields m/z 14 (CH,) and m/z 40 (loss of CH,); thus, ions [A]'. should contain a CH2 group. For ion [B]'., in contrast, there are peaks at m / z 13 (CH) and m / z 41 (loss of CH), which points to the presence of a CH group. The peak intensity ratio ( m / z 4l)/(m/z 40) is equal to 0.125 for [A]'. and 6.0 for [B]'.. It seems reasonable to attribute structure [ I ] + . (ionized methyleneketene) to ion [A]'. not only because of its mode of generation but also because [A]'. is the only [C,H,O]'. ion containing a CH2 group. It is less obvious to assign a structure to ion [B]'.: if its generation (reaction b) occurs without rearrangement in the precursor ion, then [B]'. would be [ 2 ] + ([CH=CH-CO]+.). However, it would be the cyclic analogue (8) Terlouw, J. K.; Burgers, P. C.: Hommes. H. Org. Mass Spectrom. 1979. 1 4 . 387.
0022-3654/86/20901582$01.50/0 0 1986 American Chemical Society
Stable [CH=CH-CO]'.
The Journal of Physical Chemistry, Vol. 90, No. 8, 1986 1583
Ions in the Gas Phase
TABLE I: Heats of Formation of Neutral and Ionized C3H20 (1, 3) 154
A 3
138'
9.47d
1052
From Benson's additivity method9 with (CO)d(ketenes) = -74 kJ. mol-' (ref 5). bValues derived on the basis of isodesmic reactions employing 4-31G and 6-31G*basis set calculations (ref 3). 'Ionization energy determined with energy selected electrons (ref 5). Adiabatic IE value obtained by photoelectron spectroscopy (ref 4).
31
II
54
I lS3
CH=CH-C=O
26
ionization energy (IE) of the precursor molecule (IE = 11.10 eV") and the AH,' of the C 0 2 lost (AHfo(C0), = -393 kJ-mol-I 12). The estimation of AH?(."([B]'.) can be compared to the relevant thermochemical data of structures [1]+. and [3]+.. Table I summarizes these values also obtained from combination of the AHf'(neutra1, C3H20) with their known ionization energies. It appears that AHHf[B]'.) is very close to AH?( [l]'.) but less than AHfo([3]+.) by ca. 100 kJ-mol-'. Consequently, the structure of ion [B]'. is certainly not [3]+. on the basis of the thermodynamic data, whereas it cannot be [1]+. on the basis of the CA spectra. Three structures remain to be. investigated: [2]+., [4]+., and [5]+-. No thermochemical data are available for these, and we have therefore used a molecular orbital approach. Molecular Orbital Calculations The five [C3H20]'. ions [l]'. to [5]+.have been investigated. Complete geometry optimizations were made by a standard gradient procedure using the 3-21G basis set of atomic orbitals in the restricted Hartree-Fock (RHF) appr~xirnation.'~Single point energy calculations were performed using 4-31G* and 631G** basis sets for the five isomeric ions (Table 11). Finally, electron correlation was taken into account by means of configuration interaction (CI) calculations using the CIPSI algorithm as described in ref 15; the results are presented in Table 111, and the geometries (bond lengths in angstroms, bond angles in degrees) are displayed in the text. Zon [Z 1'. ([CH,=C=C=O]'-). Neutral propanedienone (1) H loto
100.0
Figure 1.
o& *! / /
111.0
[3]+. if cyclization yields a gain in stability. The two structures [2]+. and [3]+- support the presence of peaks m / z 13 and m / z 41 in their CA spectra. Thermochemistry. The experimentally determined thermochemical data of the system C3H20-[C3H20]+-provide further information for the structure elucidation of ion [B]+-. If AHfo([B]+-) is available, it can be compared to known AHf'([C,H,O]+.) values obtained from experiment or theoretical calculations. This section deals with the experimental results, whereas specific molecular orbital calculations are presented in the next part. Measurement of the appearance energy (AE) of ion [B]'. was performed using energy selected electrons." With cyclopentene-3,5-dione as the precursor molecule a value for AE( [B]'.) = 11.8 f 0.2 eV is obtained, the large uncertainty arising from the weak gradual onset in the appearance energy curve. In order to obtain the heat of formation of ion [B]'., we have to combine the preceding AE value with AH? of the neutral precursor. For cyclopentene-3,5-dione a AH: value of -232 kJ.mol-I is calculated by the additivity method of B e n s ~ n . ~ .This ' ~ value, combined with AHfo(CH2CO)= -48 kJ-mol-',12 gives AHfo([B]+.) = 954 A 20 kJ-mol-]. Note that maleic anhydride cannot be expected to yield a threshold value for AHfo([B]+.) because of the high
(1 1) Levin, R. D.; Lias, S. G. Natl. Stand. R e j Data Ser. (US., Narl. Bur. Stand.) 1982, NSRDS-NBS 71. (12) Pedley, J. B.; Rylance, J. Computer Analyzed Thermochemical Data; Organo and Organometallic Compounds. University of Sussex, 1977. (13) Burgers, P. C.; Holmes, J. L.; Lossing, F. P.; Mommers, A. A,; Povel, F. R.; Terlouw, J. K.Can J . Chem. 1982, 60, 2246. (14) The MONSTERGAUSS Set Of programs was used for optimum geometry search, and the single point calculations are presented in Table I1 (M. Petersen and R. Pokier, Chemistry Department, Toronto, Canada, 198 1). (15) Huron, B.; Malrieu, J. P.; Rancurel, P. J . Chem. Phys. 1973, 32,
(9) Benson, S. W. 'Thermochemical Kinetics", 2nd ed.;Wiley: New York, 1976. (10) For cyclopentene-3,5-dione, the increment for CO-(C,J(C) was assigned the same value as CO-(C,)(O), Le., -140 kJmol-'; the ring strain energy was taken to be equal to that of maleic anhydride, Le., 15 kJmol-'.
(16) (a) Bunker, P. R.; Landsberg, B. M.; Winnewisser, B. P. J. Mol. Spectrosc. 1979, 74, 9. (b) Brown, R. D.; Champion, R.; Elmes, P. S.; Godfrey, P. D. J . A m . Chem. SOC.1985, 107, 4109. (17) (a) McKee, M. L.; Lipscomb, W. N. J . A m . Chem. SOC.1981, 103, 406. (b) Nobes, R. H.; Bouma, W. 3.; Radom, L. Chem. Phys. Lett. 1982, 89, 497.
la
H
"
0- 1.111
1.m.
- p.ol,
"v:
1.1s.
1.011
lo''
H
H
I .117
12'.1
1.01s
P
/
1b'+
was theoretically investigated by Radom et a1.I.' Their configuration interaction calculations' predict a planar bent structure in agreement with experiment.I6 However, the energy of the linear C, structure is only slightly above that of the planar bent molecule (the difference is less than 1 kJ.mol-' at the MP3/6-31G* level).'
5945.
Bouchoux et al.
1584 The Journal of Physical Chemistry, Vol. 90, No. 8, 1986
TABLE 11: Total (hartree) and Relative (in Parentheses, kJmol-') Energies of [C3H20]+.Ions Calculated for 3-21G Equilibrium Geometries ion structure 3-21G 4-31G* 6-31G** -189.053658 (14) -189.248633 (17) [la]'. [CH++C=O]+. -188.181 592 (10) -188.181 622 (IO) -189.054897 (14) -1 89.249 948 ( I 4) [lb]'. -189.255 118 (0) -189.059040 (0) [2a]+. [ CH=CH=C=O] +* -188.185301 (0) -189.252686 (6) -188.182314 (8) -189.056631 (6) [2b]+. [CH=CH-C=O]+*
[31+. [4a]+. [4h]+* [51+.
-188.151 632 -188.138687 -188.135720 -188.094651
[CHEC-C-OH]+* [CHEC-CHO]'.
TABLE 111: Total (hartree) and Relative (in Parentheses, kJmol-') Energies of [C3H20]+-Ions Calculated for 3-21G Equilibrium Geometries after Configuration InteractionY CI/6-31G**//3-21Gb ion structure CI/4-31G*//3-21G la]+lh]+. 2a] 2b]+* 31 4a] 4h]+. 51'. +a
+ -
[CH,=C=C=O]+*
-189.578 079 -189.577511 -189.569 592 -189.565 180 -189.533 SI0 -189.515 346 -189.512105 -189.483 298
[CH=CH-C=O]+* [CH=CH-C=O]+* [CH=C-C-OH]+* [CH=C-CHO]+*
(0) (1) (22) (34) (1 17) (165) (173) (249)
2 0 21 33 1 I4 145 154 24 1
(88) (122) (130) (238)
-189.040903 -188.998507 -188.994 198 -188.977931
(48) (159) (170) (213)
-189.237711 -189.201 952 -189.197216 -1 89.176 801
(46) (140) (152) (206)
configuration [2b]+-(Tables I1 and 111). It follows from a comparison of the energies for 111'. and [2]+that the two species have a similar thermodynamic stability. Introduction of the correlation effect leads to a greater stabilization for [1]+. than for [2]+. (Table 111).
-
Ion [3]+. ([CH=CH-C=O]+.). symmetry group:
This ion falls into the CZc
0-
Configuration interaction calculation a t the 4-3 lG* level using the variational perturbation method CIPSI (ref 15). Relative energy estimated according to the additive procedure of ref 17: AE(CII631G**) N AE(CI/4-31G*) AE(6-31G**) - AE(4-31G*).
+
For ion [1]+-two minima are found at the S C F level by using the 3-21G basis set. They correspond to the two above-mentioned geometries: CZl,([la]'.) and planar bent ([lb]'.). The calculated 3-21G and 4-31G* total energies of [la]'. and [lb]'. are the same. Ion [lb]'. becomes slightly more stable (3 kJ.mol-I) in a 6-3 1G** calculation, but this difference is reduced to 2 kJ.mo1-I after inclusion of the electron correlation by the additivity method (Table 111). Clearly the two structures, which are unlikely to be experimentally distinguishable, have comparable energies. Ion [ 2 ] +([CH=CH-C=O]+.). . No information is available about the neutral counterpart of ion [2]+.. Geometry optimization
As expected for a three-membered ring compound, its energy, relative to that of the linear isomers, is more sensitive to the choice of the basis set and to the correlation effect. The difference in energy between [lb]'. and [3]+. is 78 kJ-mol-I with the 3-21G basis (Table 11), but it becomes 114 kJ.mol-' at our highest level of calculation (CI/6-31G**, Table 111). The latter difference is in acceptable agreement with the difference in heats of formation derived from Table I, 77 kJ.mol-I. Zon [ 4 ] + .([CH=C-C-OH]+.). Two conformations of this ion, [4a]+.and [4b]+.,were examined. At the CI/6-31G** level
4a"
H 129.0
11(.0
H
2b
18kB
140.1
I
8.081
i'.
H 4b"
0
'3 'c
180.8
2b"
H
of this neutral biradical leads to conformations 2a and 2b with the 3-21G basis set. Their total energies are -188.420 561 and -188.413635 hartree, respectively, and thus 2a is more stable than 2b by 18 kJ-mol-I. We have optimized neutral methyleneketene in its C2, conformation with the same 3-21G basis. The total energy of this molecule, la, is -1 88.480069 hartree; consequently, 2a is less stable than l a by 156 kJ.mol-I. Two structures, [2a]+.and [Zb]'., were obtained after geometry optimization of the cationic system. At all levels of theory, the cis isomer [2a]+. is predicted to be lower in energy than the trans
[4a]+.is predicted to be lower in energy than [4b]+-by 9 kJ.mol-' (Table 111). These ethynylhydroxycarbene ions are higher in energy than the three preceding [C3H20]+.ions but lower than the carbonyl isomer [5]+. by ca. 100 kJ.mol-I. This situation differs from that encountered for the saturated analogues: the ionized carbenes [R-C-OH]'. are less stable than [R-CHO]+. (R = H, CH3) by some 50 k J . m ~ l - ~ . ' * . ~ ~ Zon [ 5 ] + ([CH=C-CHO]'.). . This structure is the highest in energy among the five isomers investigated: ( 1 8) For experimental determination of AHfo([R-C-0H]+.) see: (a) Terlouw, J. K.; Wezenberg, J.; Burgers, P. C.; Holmes, J. L. J . Chem. SOC., Chem. Commun. 1983, 1121. (b) Burgers, P. C.; Mommers, A. A,; Holmes, J. L. J . Am. Chem. SOC.1983, 105, 5976. (19) For theoretical ab initio investigations see: (a) Bouma, W. J.; Macleod, J. K.; Radom, L. Inf. J . Muss Specfrom. Ion Phys. 1980, 33, 87. (b) Osamura, Y . ;Gcddard, J. D.; Schaefer 111, H. J . Chem. Phys. 1981, 74, 617. (c) Bouchoux, G.; Flament, J. P.; Hoppilliard, Y. I n r . J . Muss Spectrom. Ion Processes 1984, 57, 179. (d) Apeloig, Y.; Karni, M.; Ciommer, B.; Depke, G.; Frenking, G.;Meyn, S.; Schmidt, J.; Schwarz, H. Int. J . Muss Specfrom. Ion Processes 1984, 59, 21.
J . Phys. Chem. 1986, 90, 1585-1589 0;
5 .+
\
n
The energy of [SI+.relative to [lb] is 241 kJ-mol-I with the 6-31G** basis set and after estimation of the electron correlation effect.
Conclusion The main conclusion of the MO investigation is that the [C3H20]+.ions, [1]+. and [2]+., are very close in energy. The three other isomers, [3]+-, [4]+., and [5]+., are sufficiently high in energy to be rejected as a candidate for the structure of ions [B]'.. Combining these results with the experimental findings of the preceding section, we conclude that ionized maleic anhydride and cyclopentene-3,5-dione eliminate C 0 2 and CH2=C=0, respectively, without any rearrangement to produce ions of structure [CH=CH-C=O]+. ([2]+-). Thus, ion [2]+. is a stable species with a heat of formation much lower (by ca. 90 kJ.mol-I; see Table
1585
111) than that of the related closed-ring species, [cyclopropenone]+.. It is likely that a similar situation obtains for the methyl-substituted homologues, [methylcyclopropenone]+. vs. [CH3-C=CHC=O]+.. The latter ion has been proposed to be a key intermediate in the decomposition of ionized furan,20and we propose that it is formed as a stable [C4H40]+.product ion in the dissociative ionization of itaconic and citraconic anhydride and the methyl ester of allenecarboxylic acid. The product ions generated in these dissociations, to which the structure [CH3CH=C= C=O]+. has incorrectly been as~igned,'~ show collisional activation characteristics very close to that of [methylcyclopropenone]+. but have an enthalpy of formation which is significantly 10wer.~
Acknowledgment. We are grateful to Dr. F. P. Lossing for the measurement of the appearance energy of [C3H20]+.ions from cyclopentene- 3,5 -dione. (20) Bouchoux, G.; Dagaut, J.; Fillaud, J.; Burgers, P. C.; Terlouw, J. K. N o w . J. Chim. 1985, 9, 25. (21) Maeda, K.; Semeluk, G. P.; Lossing, F. P. Int. J . Muss Spectrom. Zon Phys. 1968, 1, 395. Lossing, F. P.; Traeger, J. C. Int. J . Mass Spectrom. Ion Phys. 1976, 19, 9.
LASER CHEMISTRY, MOLECULAR DYNAMICS, AND ENERGY TRANSFER Small System Size Artifacts in the Molecular Dynamics Simulation of Homogeneous Crystal Nucleation in Supercooled Atomic Liquids J. Dana Honeycutt and Hans C. Andemen* Department of Chemistry, Stanford University, Stanford, California 94305 (Received: June 24, 1985; In Final Form: December 4, 1985)
We have devised a method for determining the time of critical nucleus formation in molecular dynamics simulations of homogeneous nucleation in a supercooled liquid. We have studied nucleation in systems of 500 and 1300 Lennard-Jones particles with reduced density 0.95 at a reduced temperature of 0.45. The size of the critical nucleus and the time to formation of a critical nucleus display an anomalous dependence on the size of the simulated system, indicating that the small system size and periodic boundary conditions are producing artifacts in the crystallization process.
Introduction In 1976 Mandell, McTague, and Rahman' first reported observation of homogeneous nucleation of a crystalline phase in molecular dynamics (MD) simulations of a supercooled Lennard-Jones liquid. Since then, several MD studies of homogeneous nucleation for various model systems have been r e p o ~ t e d . ~Some -~ ( I ) M. J. Mandell, J. P. McTague, and A. Rahman, J. Chem. Phys., 64, 3699 (1976). (2)'M. J: Mandell, J. P. McTague, and A. Rahman, J. Chem. Phys., 66, 3070 (1977). (3) M. Tanemura, Y. Hiwatari, H. Matsuda, T. Ogawa, N. Ogita, and A. Ueda, Prog. Theor. Phys., 58, 1079 (1977). (4) C. S . Hsu and A. Rahman, J. Chem. Phys., 70, 5234 (1979). ( 5 ) C. S. Hsu and A. Rahman, J . Chem. Phys., 71, 4974 (1979). (6) R. D. Mountain and P. K. Basu, J . Chem. Phys., 78, 7318 (1983). (7) R. D. Mountain and A. C. Brown, J . Chem. Phys., 80, 2730 (1984). ..
0022-3654/86/2090-1585$01 .50/0
investigators have concluded that the periodic boundary conditions (PBC) used in these simulations may interfere with the process of nucleation and growth for systems of up to 500 particle^.^.^ Others have concluded or seem to have assumed that their systems, which range in size from 432 to 4000 particles, were free of PBC artifacts or other small system size Clark,Io in a study of amorphous Lennard-Jones systems, reported that a 2 16-particle system nucleated more readily than an 864-particle system. We reported in a previous letter" that the catastrophic crystal growth observed in molecular dynamics simulations of highly (8) J. N. Cape, J. L. Finney, and L. V. Woodcock, J . Chem. Phys., 75, 2366 (1981). (9) A. C. Brown and R. D. Mountain, J . Chem. Phys., 80, 1263 (1984). (IO) J. H. R. Clarke,J. Chem. Soc., Furuduy Trans. 2, 75, 1371 (1979). (11) J. D. Honeycutt and H. C. Andersen, Chem. Phys. Letr., 108, 535 (1984).
0 1986 American Chemical Society