J. Phys. Chem. 1994, 98, 67 14-6718
6714
Theoretical Study of the Enolic Forms of Acetylacetone. How Strong Is the H-Bond? J. J. Dannenberg' and Raphael Rios Department of Chemistry, City University of New York-Hunter Avenue, New York, New York 10021
College and The Graduate School, 695 Park
Received: February 23, 1994; In Final Form: April 4, 1994"
Completely optimized ab initio calculations at the Hartree-Fock and second-order Mlaller-Plesset (MP2) levels using basis sets up to D95++** and semiempirical calculations using AM1, PM3, and S A M l are reported for acetylacetone and both the cis and trans isomers of its 1- and 2-enols. Comparison of the energies (MP2/D95** used for examples) of the various species indicates that the internal H-bond of the cyclic, conjugated cis-2-enol stabilizes that species by 12.0 kcal/mol, which is considerably more than for the cyclic 1-enol (6.0 kcal/mol). Furthermore, as the enhanced stability of the H-bond is more than for other resonance-stabilized H-bonds, the existence of aromatic character in this 6-T-electron system is supported. The non-H-bonded trans-Zen01 is more stable than the trans-1-enol by 5.0 kcal/mol, which we attribute to conjugation. The symmetric CZ, structure of the cis-2-enol is predicted to be 2.5 kcal/mol less stable than the unsymmetrical structure by both MP2/D95* * and MP2/D95++(d,p) before the zero-point vibrational energy correction and 0.2 kcal/mol less than the unsymmetrical structure after correction (using scaled HF/D95** frequencies and the MP2/D95++** energies). The calculated structures of the cis-2-enol agree reasonably well with that reported by Karle but are not consistent with that reported by Shibata. Both Hartree-Fock (because of the electron correlation error) and the three semiemperical methods are inadequate for this problem, as they all give artificially high energies for the CZ,structure. We suggest that this structure may still be overestimated a t the highest levels calculated due to incomplete correction for electron correlation which favors this structure.
Both malonaldehydeand acetylacetoneare postulated to have unusually strong H-bonds in their cyclic, conjugated enolic forms. Conjugated enols of this kind contain 6 ?r electrons within a ring and thus are potentially aromatic systems. Estimates of the strength of these bonds1 have varied widely (up to 26 kcal/mol in stabilization energy for malonaldehydez). Clearly,the amount of stabilization attributed to such bonds depends upon the nature of the system to which the cyclic enol is compared. Ideally, one should compare the enol to a model in which the H-bond is broken but the rest of the molecule is unaffected. In this paper we shall evaluate the stabilizing contributions that are attributable to H-bonding and conjugation, both on an individual basis and in concert. We chose acetylacetone rather than the more simple malonaldehyde,s as the former has the possibility of enolizing to form double bonds at either the 1(leading to a nonconjugated enol) or the 2-position. In addition, the cis and trans arrangements about the double bond will differ little in steric repulsions, as the OH will be exchanged with a CH3 rather than an H. The structure of the cis-2-enol has been somewhat controversial. On the basis of two separate gas-phase electron diffraction studies, Karle4 determined the enol to be symmetrical with a short O.-O distance (2.381 A) and the H-bond in the plane of the ring, while Shibatas reported an unsymmetrical enol with a longer 0-0 distance (2.512 A) and the H-bond out of the plane. The only known crystal structure of the enol form is unsymmetrical;6 however, it is part of a cocrystal, which could clearly have an unsymmetrical influence. On the other hand, structures of several 3-derivatives are symmetrical.7 Photoelectron spectroscopy, which has been used to characterize the gas-phase structure of malonaldehyde, gives more ambiguousresults for acetylacetone.8 An NMR study of deuterated acetylacetonesuggests an unsymmetrical enol exists in s o l ~ t i o n . ~ Molecular orbital calculations (both ab initio and semiempirical) are reported for acetylacetone (I), its 2-enols (cis and trans) (IIa,b), and its I-enols (cis and trans) (IIIa,b). Compari*Due to an editorial error, an incorrect abstract was published in Advance ACS Abstracts, May 1, 1994.
O y Y 0
I
IIa
IIh
IIC
M3vm2 cH3*0H0
IIIa
a 2
IIIb
son of the energies and structures of these isomers will enable us to evaluate the contributionsof conjugation,H-bondingand their cooperative effects. The calculated structures of the unsymmetrical and the symmetrical CZ,forms of the cis-Zen01(IIc) are compared with the experimentalreports in an attempt to elucidate the structure. Methods
Ab initio calculations were performed at both the H a r t r e e Fock (HF) and second-order Mlaller-Plesset (MP2) levels. The 6-31G*, 6-31G**, D95*, D95**, and D95++** basis setslowere used for both HF and MP2. MP2/D95++** calculations were performed on IIa, and IIc, only. MP2 calculations used the frozencore approximation. Semiempiricalcalculationsused the AM 1, I ' PM3,12 and SAMl 13 approximationsto molecular orbital theory. We used Gaussian 92 revision B14 to perform the ab initio calculations and both AMPAC versions 2.115 (AM1) and 4.516 (PM3 and SAM1) for the semiempirical studies. All geometries were completely optimizedI7 at each level of calculation. The stationary states (minima except for the CZ,
0022-365419412098-67 14%04.50/0 0 1994 American Chemical Society
The Journal of Physical Chemistry, Vol. 98, No. 27, 1994 6715
The Enolic Forms of Acetylacetone
TABLE 1: Relative Energies (kcal/mol) W-ma
W-IIa
In
Ib
IIa
IIb
IIh
W
IIc
IIc-IIa
IIb-IIa
0.0 0.8 0.0 0.0 0.1
0.0 0.8 0.2 0.2 0.0
2.6 0.0 6.4 0.1 0.9
13.7 11.1 13.9 11.1 11.4
17.8 15.1 17.4 14.4 11.0
22.7 20.1 22.1 19.2 15.8
12.5 8.6 12.6 8.3 6.5
10.0 8.6 6.2 8.2 5.6
11.1 11.1 7.5 11.0 10.5
4.9 5.0 4.7 4.8 4.8
9.0 9.1 8.2 8.1 4.4
20.1 20.1 15.7 19.1 15.0
0.0 1.5 0.0 1.1
1.0 2.4 1.3 2.2
13.4 12.3 13.8 12.5
17.1 16.0 16.6 15.2
23.1 21.9 22.6 21.2
9.7 9.6 8.8 8.7
22.0 21.9 21.0 21.2
12.2
11.2
17.2
4.2 3.1 3.7 2.5 2.5 -0.2
5.9 5.9 6.0 6.0
1.4
5.3 3.1 5.3 2.5 2.5 0.0
12.3 12.3 12.2 12.5
0.5
1.1 0.0 1.6 0.0 0.0 0.2
12.0
6.0
5.0
17.0
0.0 0.0 0.0
0.0 0.0 0.0
0.7 2.7 2.2
5.8 9.3 9.9
10.0 13.6 14.3
5.7 11.0 5.7
21.5 27.4 11.0
20.8 24.6 8.8
5.1 6.6 7.7
-4.3 -2.5 -8.6
-0.1 1.7 4.2
4.9 8.3 3.5
IIIb-IIb
HF 6-31G* 6-31G** D95' D95** ZP coF MP2 (Frozen-Core) 6-31G* 6-31G** D95* D95** D95++** ZP cor0
Semiempirical AM 1 PM3 SAM 1
"Corrected for zero-point vibrational error using a 0.85 scaling factor and HF/D95++**. by 1.1 kcal/mol, somewhat less than the reported 2.4 kcal/mol. c"YYo14 All three semiempirical methods incorrectly predict I to be the 6 2%
17.2
IIIb
6.0
IIIa
11.0
12.0
global minimum, although AM 1predicts little differencein energy between I and IIa. Two conformations of the keto form were obtained, In, which has a 140.4' dihedral angle between the planes of the carbonyls, and Ib, whose corresponding angle is 93.4'. Ib contains a weak C-H-0 interaction,**which renders it unsymmetrical. Comparing the cis- and trans-2-enols, IIa and IIb, provides an estimate of the H-bonding energy in the conjugated enols. The ab initio calculations predict this difference to be 7.5-13 kcal/ mol. Only the HF/D95* and HF/D95** calculations predict this difference to be less than 10 kcal/mol. Comparison of the energies of the trans-1-enol, IIIb, with that of the trans-2-eno1, IIb, provides an estimate of the contribution of conjugation to the stability of IIa. This difference is about 9 (8.7 for MP2/D95**) kcal/mol beforeand 5 kcal/molafter ZPVE for all ab initio calculations. This difference includes both the stabilization due to conjugation and that due to the augmented substitution of the double bond. The energies of the hydrocarbons IVa and IVb were calculated at the HF/D95** level to evaluate
I
1
CH3 CH3
Ila
d. b .H'
IIa'
IIC
Figwe 1. Schematic representation of the energy differences for the different structures of acctylacctone. Energies are taken from MP2/ D95** with zero-point vibrational correction (see Table 1).
transition state for H-transfer in IIc) were characterized by calculating the force constants for all HP and semiempirical calculations. Zero-point vibrational corrections were calculated for the H F optimizations. Results and Discussion
The energies of the various structures at differing levels of theory are collected in Table 1and displayed graphically in Figure 1. The various methods differ in finding either I or IIa to be most stable. Examination of the table shows that those basis sets containing a polarization function only on the heavy atoms (*) favor the keto form, I, while those with polarization functions on the hydrogens as well, favor the enol, IIa. The highest level of calculation (MP2/D95**) predicts IIa to be more stable than I
cH3 cH3
the stabilization due to augmented substitution. The energy difference is 1.9 before and 1.7 kcal/mol after ZPVE correction for the effect of increased substitution. Using this value as a correction, about 3 kcal/mol would be attributable to the effect of polarization upon conjugation. The difference in energy between the cis- (H-bonded) and trans-1-enols, IIIa and IIIb, provide an estimate of the H-bonding energy in the absence of conjugation or cooperativity. This value is consistently predicted to be 5-6 (6.0 for MP2/D95**) kcal/ mol by all ab initio methods, somewhat larger than that of water dimerI9 (3.6 kcal/mol). The difference in energy between the unconjugated, non-Hbonded enol, IIIb, and the conjugated, H-bonded end, IIa, is 17-23 (17.2 for MP2/D95** corrected for ZPVE) kcal/mol. If one combines the stabilization due to H-bonding (6.0 kcal/mol) with that of conjugation (5.0 kcal/mol), one estimates the combination of conjugation and H-bonding to be 1 1.Okcal/mol, some 6.2kcal/mol less than calculateddirectly (using thecorrected MP2/D95** energies). This value is significantly greater than that calculated for the cooperativity in acetic acid dimer (2.5-3 kcal/mol)*O or for infinite chains of H-bonded enols of 1,3-
6716 The Journal of Physical Chemistry, Vol. 98, No. 27, 1994 0 NH 0
Figure 2. Numbering scheme for Table 2.
cyclohexanedione(3.9 kcal/mol).21 Is thisincreasedcooperativity due to aromaticity, or can it be otherwise explained? Let us first examine the comparison with acetic acid dimer. Acetic acid dimer (V)has 8 T electrons and thus is not aromatic ‘ * 3 W y Y \
9
, \
V
(potentially antiaromatic). However, its formation also involves a repulsion between the hydroxyl H’s that is not present in the isolated monomers. It seems unlikely that the H-*H repulsions22 could account for the 5-6 kcal/mol difference in cooperativity between acetic acid dimer and IIa, particularly when one considers that acetic acid dimer cooperativity is shared between two H-bonds. The reported experimental and MP2/6-3 1G(d) geometries indicate the C=O and 0-H bonds to be longer and the C-0 and O--Odistances to be shorter than those for acetic acid dimer. The comparison with the estimate for infinite chains of the anti-enol of 1,3-cyclohexanedione(VI)is alsovery revealing. The
VI estimate was based upon ab initio calculations on the anti-enol form of 1,3-propanedione (malonaldehyde) constrained to the conformation it would have if it were part of a cyclohexane ring. The cooperativity for this system (3.9 kcal/mol) is greater than that for acetic acid dimer but less than that for IIa. Yet, one might reasonably expect the cooperativity in an infinite chain to be greater than that in a finite cyclic structure (for example, the cooperativity/molecule of six 1,3-~yclohexanedionesarranged in a ring is slightly (0.2 kcal/mol) less than that for the infinite chain). Once again the O.-O distance is less than the 2.561 A observed in crystals of 1,3-~yclohexanedione,23 which crystallizes as infinite chains of the anti-enol. We are tempted to attribute part of the cooperativity of IIa to aromaticity. The geometries of the various species (Table 2) follow trends analogous to those of the energetics. Thus, comparison of the C-C, C = C , C-0, and C=O bond lengths for IIb and IIIb indicates lengthening of the double bonds and shortening of the single bonds upon passing from the unconjugated IIIb to the conjugated IIb. These changes are greater than those found for the analogous hydrocarbons. For example, the internal C3-C4 bond decreases by 0.050A upon going from IIIb to IIb, while the analogous ((2243) bond decreases only 0.033 A upon going from IVa toIVb. Thedifferencecan beattributed to thepolar character of the carbonyl group coupled with the polarizability of the conjugated system. Comparison of the bond lengths of IIIa and IIIb provide a means of estimating the effect of H-bonding alone upon the
Dannenberg and Rios geometry. The C=O bond is slightly lengthened while the C - 0 bond is slightly shortened, but both effects are less that those for conjugation discussed above. Formation of thecyclicconjugated structure IIa involves greater structural changes than expected from theindividual contributions of the effects of conjugation and H-bonding, in parallel with the energetics previously discussed. Thus, the C - 0 bond (1.258 A) isO.021 and0.018 Alonger, theC-Obond (1.337A) 0.035and 0.033 A shorter, the C = C bond (1.376 A) 0.016 and 0.027 A longer, and the C-C bond (1.453 A) 0.025 and 0.075A shorter than those for ID and IIIa, respectively at MP2/D95++**. Comparison of the O-.H distances for IIa (1.625 A) and IIIa (1.955 A) indicates a significant shortening upon conjugation. The O-.H distance for IIIa is similar to that for water dimer?‘ while that for IIa can only be compared with those of crystals with extended cooperativity such as 1,3-~yclohexanedione. If IIa be aromatic, then the transition state for intramolecular H-transfer (being more symmetrical) would arguably be more aromatic. As mentioned above, experimental evidence has not been consistent concerning the possibility that the CZ,structure with the H equally bonded to each of the O’s, IIc, is the ground or a transition state. Earlier MO calculations on this problem have been restricted to RHF and semiempirica125calculations, both of which consistently overestimatethe energy of IIc compared to the present MP2 calculations. The highest level calculations used for all structures in this study (MP2/D95**) predict a difference of only 2.5 kcal/mol. As this was the lowest barrier among all the calculational methods, we reexamined only IIa and IIc using completely optimized MP2/D95++** calculations.The energy difference remained at 2.5 kcal/mol. This difference is likely to disappear after correction for zero-point vibrational energy (ZPVE), which favors IIc by 2.9 kcal/mol at the HF/ D95** level. The experimental evidence is confusing. The gas-phase electron diffraction study by Karle was performed at 100-1 10 OC. He determined that the sample contained 66 f 5% enol, the rest being the keto form. He did not consider the possibility of more than one conformation for either tautomer. Shibata assumed the sample to be 100% enol at 21 OC on the basis of a reported AH of 2.4 kcal/mo126(other reports range from 1.9 to 4.1 kcal/ mol).27 Others have reported 65% (liquid, 100 0C)28and 90% (gas, 100 OC).29 Karle’s reported structure is in good agreement with the best optimized structures reported here for IIc. In particular, the six atoms forming the ring are planar and the 0-0distances agree within 0.017 A. The 0-H-O angle was reported to be nearly linear in comparison to the calculated 159O. The calculated conformations of the methyl groups in both IIc and the dione agree with Karle’s values. For IIc each methyl group has a C-H bond in the molecular plane that eclipses the C-0 bonds, while in the dione structures the C-H bonds in the molecular plane are anti to the C = O bonds. Karle specifically tested the significanceof the rotational conformations of the methyl groups by evaluating the fit to the data upon rotation of the methyl groups. On the other hand, Shibata’s structure differs from the best calculated for IIa in several ways. His HOCC dihedral is 26’ (versusOo calculated by all methods). His O*-O distance is 0.034 A shorter than the best calculation reported here. His methyl groups have a C-H bond 30’ out of the molecular plane, while our best geometry has one C-H bond eclipsed with the C - 0 bond and another anti to the C = O bond. The last difference could result from an averaging of the two equivalent structures for IIa. To test the possibility that another minimum might exist that corresponds to Shibata’s structure, we recalculated IIa with the dihedral angle constrained to be 26O at the MP2/D95** level. The energy is 3.3 kcal/mol higher than for the completely
The Enolic Forms of Acetylacetone
The Journal of Physical Chemistry, Vol. 98, No.27, 1994 6717
TABLE 2: Geometrical Parameters (A and deg)' C I ~ Z c&3
C&4
C445
CrO
C4-0
0-H
6-H
0-0
GH-0
h HF/6-31G* HF/6-3 lG** HF/D9 5 * HF/D95** MP2/D95* MP2/D95**
1.508 1.507 1.511 1.509 1.513 1.511
1.528 1.527 1.529 1.529 1.532 1.531
HF/6-31G* HF/6-3 1G** HF/D95* HF/D95** MP2/6-31G* MP2/6-3 1G** MP2/D95* MP2/D95**
1.509 1.508 1.511 1.510 1SO7 1SO6 1.512 1.511
1.523 1.523 1.525 1.525 1.522 1.521 1.527 1.525
1.523 1.522 1.525 1.525 1.525 1.525 1.532 1.531
1.511 1.510 1.513 1.512 1.510 1SO9 1.515 1.514
1.191 1.191 1.195 1.195 1.227 1.227 1.231 1.230
1.192 1.192 1.194 1.195 1.228 1.227 1.213 1.230
1.496 1.495 1.499 1.498 1.493 1.493 1.499 1.497 1.497 1.493
1.348 1.348 1.353 1.353 1.367 1.367 1.376 1.376 1.376 1.382
1.458 1.458 1.464 1.462 1.448 1.446 1.456 1.453 1.452 1.430
1.510 1.510 1.512 1.511 1.509 1.508 1.513 1.511 1.512 1.525
1.318 1.317 1.320 1.319 1.338 1.335 1.340 1.337 1.338 1.319
1.213 1.213 1.215 1.216 1.253 1.254 1.257 1.258 1.259 1.243
0.962 0.958 0.961 0.960 1.003 0.999 1.003 1.003 1.004 1.049
1.499 1.498 1.503 1SO2 1.495 1.494 1.501 1.499
1.335 1.335 1.339 1.339 1.353 1.353 1.361 1.360
1.474 1.474 1.480 1.480 1.470 1.469 1.478 1.478
1.515 1.514 1.517 1.516 1.516 1.515 1.521 1.519
1.345 1.344 1.348 1.347 1.370 1.370 1.373 1.372
1.201 1.201 1.204 1.204 1.237 1.237 1.240 1.240
0.947 0.943 0.947 0.944 0.974 0.965 0.974 0.967
1.500 1.499 1.502 1.501 1.499 1.498 1.504 1.502 1.502 1A97
1.398 1.398 1.404 1.404 1.403 1.402 1.411 1.410 1.410 1.416
1.262 1.262 1.264 1.264 1.292 1.291 1.295 1.294 1.295 1.315
1.190 1.185 1.188 1.185 1.214 1.201 1.213 1.200 1.201
1.319 1.319 1.324 1.324 1.340 1.340 1.350 1.349
1.510 1.510 1.513 1.512 1.511 1.510 1.516 1.515
1.522 1.521 1.524 1.523 1.520 1.520 1.526 1.525
1.508 1.507 1.510 1SO9 1.504 1.503 1.509 1.508
1.352 1.350 1.353 1.352 1.369 1.368 1.372 1.370
1.197 1.197 1.200 1.200 1.234 1.233 1.237 1.237
0.952 0.948 0.951 0.949 0.981 0.973 0.98 1 0.975
1.319 1.319 1.324 1.324 1.340 1.340 1.349 1.348
1.505 1.505 1.510 1.509 1.499 1.498 1.507 1S O 5
1.523 1.522 1.524 1.523 1.524 1.524 1.529 1.528
1.514 1.513 1.516 1.514 1.513 1.512 1.517 1.516
1.351 1.350 1.354 1.354 1.372 1.370 1.376 1.375
1.191 1.191 1.194 1.194 1.226 1.225 1.229 1.229
0.947 0.942 0.947 0.944 0.973 0.966 0.974 0.967
1.326 1.326
1.523 1.522
1.514 1.514
1.327 1.327
1.329 1.328
1.490 1.489
1.332 1.332
1.512 1.511
1.192 1.192 1.195 1.195 1.232 1.231
3.965 3.973 3.977 4.046 4.046
Ib 3.557 3.569 3.562 3.569 3.974 3.593 3.555 3.559
IIa HF/6-31G* HF/6-31GS* HF/D95* HF/D95** MP2/6-3 1G* MP2/6-3 1G** MP2/D95* MP2/D95** MP2/D95++** exptlb
Ilb HF/6-31G* HF/6-31G** HF/D95* HF/D95** MP2/6-31G* MP2/6-3 1G** MP2/D95* MP2/D95** IIC HF/6-31G* HF/6-31G** HF/D95* HF/D95** MP2/6-3 1G* MP2/6-3 1G** MP2/D95* MP2/D95** MP2/D95++** exptlC IIIa HF/6-31G* HF/6-31G** HF/D95* HF/D95** MP2/6-3 1G* MP2/6-31G** MP2/D95* MP2/D95**
IIIb HF/6-31G* HF/6-31G** HF/D95* HF/D95** MP2/6-31G* MP2/6-3 1G** MP2/D95 * MP2/D95**
IVa HF/D95* HF/D95 * *
IM HF/95* HF/95** a Atom
numbering follows the IUPAC convention. Shibata et ala4 Karle et al.3
1.813 1.802 1.824 1.792 1.688 1.646 1.691 1.625 1.626 1.626
2.002 2.003 2.026 2.010 1.93 1 1.924 2.003 1.955
2.638 2.627 2.647 2.622 2.594 2.558 2.596 2.547 2.549 2.512
142.0 142.3 141.9 142.8 148.0 149.7 148.0 150.6 150.6 137.0
2.328 2.320 2.325 2.320 2.381 2.361 2.378 2.359 2.363 2.381
156.0 156.4 158.1 156.5 157.2 158.5 157.2 159.0 159.1 175.0
2.775 2.772 2.791 2.778 2.768 2.757 2.817 2.779
136.9 137.0 136.2 136.8 141.6 142.1 139.0 140.7
Dannenberg and Rios
6718 The Journal of Physical Chemistry, Vol. 98, No. 27, 199'4
optimized IIa. The O.-O distance also increases to 2.598 A. If one averages these values, assuming a 2:l eno1:keto ratio and Karle's values for the O--O distances and dihedral angles in the enol and keto structures, one obtains 2.510 A for 0-0. This value is reasonably close to that reported by Shibata. Thus, much of the differences in the two reports might derive from Shibata's assumption that his sample was 100%enol. The 2.4 kcal/mol enthalpy difference might be significantlyovercome at the experimental temperatures by A( T U ) , which will surely favor the acyclic keto tautomer. For example, a 5 eu AM at 400 K would lower the keto free energy by 2 kcal/mol. One should note, however, that the measured 0-0 distance for IIa should be less than that calculated from the potential energy surfacedue to the extreme anharmonicity expected for the vibration that connects the two equivalent structures. Nevertheless, the calculations are totally inconsistent with the reported 26O HOCC dihedral angle. Photoelectron spectra of gas-phase acetylacetone have been interpreted as consistent with an unsymmetrical enol, although the evidence is less convincing than for malonaldehyde. Clark reported the photoelectronspectra of thecarbon 1sorbitals, which he compared to single-point RHF calculations on hypothetical structures. He was able to correlate the spectra with his calculations. The 1s orbital energies from our calculations agree qualitatively with those reported by Clark.'b However, his correlation with experimental ionizations substantially depends upon the hole relaxation energies, which presumably depend upon the geometries used for IIa and IIc, which are different in the two studies. In any case, the high frequency of the exciting radiation (much less than a vibration) suggests that a double-minimum potential surface might be consistent with these spectra even if the zero-point vibration was above the symmetrical barrier. The calculated enthalpy difference between the keto and enol forms is somewhat less than the experimental result. The difference becomes even less after consideration of the ZPVE correction. Theenergies of Table 1indicate that both polarization functions on hydrogen and inclusion of correlation energy (MP2) lower the energy of the enol, IIa, with respect to I. These are the same factors that lower the energy of the symmetric IIc with respect to IIa. If the calculations are not yet converged with respect to the enthalpy difference between I and IIa, they may not yet be converged with respect to that between IIa and IIc. This implies that the relative energy of IIc might still be overestimated. Since the O.-O separation is smallest in IIc, somewhat larger in IIa, and larger still for I, one might reasonably expect the electron correlation error to be in the order IIc > IIa > I, in accord with the above argument. Furthermore, the calculated geometries for the keto conformers, In and Ib, differ from that reported by Karle, particularly with respect to dihedral angle and 0--0 separation. We located two minima, while he assumed only one keto form. Nevertheless, since the 0-0 distances of both In and Ib (4.046 and 3.559 A) are significantly longer than that reported by Karle (2.767 A), they could not average to his structure. The facts that the 0-0 separation is calculated to be too large and the enol to be too energetic are again consistent with the suggestion that incomplete correction for energy correlation might be responsible. The comparisons between the experimental and theoretical results lead us to suggest that the symmetric form, IIc, is likely to lie below the zero-point vibration and may even be a minimum on the surface. The only evidence that requires a double-well potential is the photoelectron studies, which are somewhat ambiguous.
Conclusion The present calculations indicate that the H-bond stabilizes the cis-2-enol of acetylacetone by 12.5 kcal/mol, which is about 6-8 kcal/mol more than expected for a simple OH+-Ointeraction. The enhanced stabilization is significantly greater than that
previously calculated for acetic acid dimer and infinite chains of enols of other 1,3-diones, which also benefit from resonance interactions. Since the cyclic conjugated enolic structure contains 6 r electrons, this added stability might be attributed to aromatic character. If this phenomenon be shown to be general by further studies, current molecular modeling will need to be revised accordingly. The best calculations predict the conjugated enol to have a double-well potential with respect to internal proton (or H atom) transfer with a barrier of only 2.5 kcal/mol. This barrier is (a) likely to be below the zero-point vibrational level and (b) probably still overestimated by the current calculations. These results are in good agreement with the gas-phase electron diffraction study of Karle but in clear disagreement with that of Shibata.
Acknowledgment. This work was supported in part by grants from the PSC-BHE, NIH/MBRS (GM08176), New York City Alliance, NSF, and IBM Corp. References and Notes (1) For a useful discussion, see: Emsley, J. Structure and Bonding, Springer-Verlag: Berlin, 1984; p 148. (2) Isaacson, A. D.; Morokuma, K. J . Am. Chem. Soc. 1975,97,4453. (3) For ab initio studies of malonaldehyde, see: Frisch, M. J.; Scheiner, A. C.; Schaefer, H. F., III; Binkley, J. S.J. Chem. Phys. 1985, 82, 4194. Binkley, J. S.; Frisch, M. J.; Schaefer, H. F., I11 Chem. Phys. Lett. 1986,126, 1. (4) Lowry, A. H.; D'Antonio, C. G. P.; Karle, J.J. Am. Chem.Soc. 1971, 93, 6399. (5) Iijima, K.; Ohnogi, A.; Shibata, S . J . Mol. Struct. 1987, 156, 111. (6) Camerman, A.; Mastropaolo, D.; Camerman, N. J . Am. Chem. Soc. 1983, 105, 1584. (7) Emsley, J.; Ma, L. Y.Y.;Bates, P. A,; Motevalli, M.; Hursthouse, M. B. J. Chem. Soc., Perkin Trans. 2 1989, 527-33. (8) (a) Hush, N. S.;Livett, M. K.; Peel, J. B.; Willett, G. D. Aust. J . Chem. 1987,40, 599. (b) Clark, D. T.; Harrison, A. J. Electron Spectrosc. Relat. Phenom. 1981, 23, 39. (c) Brown, R. S.;Tse, A,; Nakashima, T.; Haddon, R. C. J. Am. Chem. Soc. 1979,101,3157. (d) Brown, R. S.J. Am. Chem. Soc. 1977, 99, 5497. (9) Egan, W.; Gunnanson, G.; Bull, T. E.; Forstn, S.J. Am. Chem. Soc. 1976, 99,4568. (10) Hehre, W. J.; Ditchfield, R.; Pople, J. A. J. Chem. Phys. 1972, 56, 2257. Hariharan,P. C.;Pople, J. A. Theor. Chim.Acta 1973,28,213. Gordon, M. S. Chem. Phys. Lett. 1980, 76, 163. Dunning, T. H.; Hay, P. J. Mod. Theor. Chem. 1976, 1-28. Clark, T.; Chandrasekhar, J.; Spitznagel, G. W.; Schleyer, P. v. R. J . Compur. Chem. 1983, 4, 294. Frisch, M.J.; Pople, J. A.; Binkley, J. S . J. Chem. Phys. 1984,80, 3265. (11) Dewar, M. J. S.;Zoebisch, E. G.; Healy, E. F.; Stewart, J. J. P. J. Am. Chem. Soc. 1985,107, 3902. (12) Stewart, J. J. P. J . Comput. Chem. 1989, 10, 209. (13) Dewar, M. J. S.;Jie, C.; Yu,J. Tetrahedron 1993, 49, 5003. (14) Gaussian, Inc., Pittsburgh, PA. (15) Graciously furnished by M. J. S.Dewar and E. Healy. (16) SemiChem, Shawnee, KS. (17) Maximum force
