ULTRAVIOLET SPECTRA OF POLVOLEFINS
July 20, 1964
2811
ORGANIC A N D BIOLOGICAL CHEMISTRY [CONTRIBUTION FROM THE
DEPARTMEKT OF
Organic Quantum Chemistry. BY
XORMAN
CHEMISTRY, WAYNE STATE UNIVERSITY,
DETROIT2,
MICHIGAN]
VII. Calculation of the Near-Ultraviolet Spectra of Polyolefins',* L. ALLINGERA N D
XfARY
ANN M I L L E R 3
RECEIVED JANUARY 29, 1964
T h e r-electronic spectra have been calculated for a number of dienes and trienes, conjugated, unconjugated, and cross-conjugated, utilizing the basic method of Pariser and Parr but with modification of the numerical values of certain of the required integrals. Butadiene was studied as the s-cis and s-trens conformers, and for various rotational arrangements intermediate between these extremes. I t was usually found possible t o calculate the wave lengths of the observable transitions to within 5-10 mp, and to obtain reasonable estimates of the oscillator strengths
Introduction The ultraviolet spectra of organic compounds have been of considerable use to organic chemists from the point of view of structural d e t e r m i n a t i ~ n . ~There are available empirical rules by which the spectrum of a compound which belongs to one of certain selected classes can be predicted with a high degree of accuracy. Woodward's rules5 offer a very practical method for the prediction of the absorption maximum in the ultraviolet spectrum of a conjugated polyene, as long as it belongs to a rather select, almost planar, noncross-conjugated system containing bond angles close to normal. There are any number of interesting compounds available which do not possess all of the appropriate characteristics, and for which the simple rules are inapplicable. We have therefore considered the very practical problem of the prediction of the spectra of olefins in general : conjugated, unconjugated, or cross-conjugated, and possessing any number of double bonds in any spatial orientation. There is available a theoretical method for the calculation of such spectra, which was developed b y Pariser and Parr,67and which was applied by them to a few simple systems. The quantum mechanical calculation of the electronic properties of even a simple polyene would be prohibitively laborious unless various approximations were made. The Pariser-Parr method, which has been relatively accurate in the limited area to which i t has been applied, makes two kinds of approximations. One is the wholesale neglect of large numbers of integrals which enter into the calculation on the basis that they are quite small. The second approximation, which in part compensates for errors introduced by the first, is to use empirical rather than theoretical values for certain specified integrals. The Pariser-Parr method has not been extensively applied to polyenes p r e v i o ~ s l y , ~and - ~ ~there is considerable uncertainty as to what kind of accuracy is to be (1) Paper V I : r . L. Allinger, L. A . Friebel-g. R . B. Hermann, and M. A. Miller, J . A n t . C h e m . S O L . 86, , 1171 (1963). ( 2 ) This research was supported in part b y Grants G-10346 and G19981 from the r a t i o n a l Science Foundation. ( 3 ) Predoctot-al Sational Institutes of Health Fellow, Ilivixion of General Medical Sciences, 1962-1963. (4) H. H . JaffC and M. Orchin, "Theory and Applications of Ultraviolet Spectroscopy," John Wiley and Sons, Inc., N e w York, N.Y . , 1962. ( 5 ) L. F. Fieser and M . Fieser, "Steroids," Reinhold Publishing Corp., New York, N . Y , 1969, p 16. (6) R Parisel- and R . G. Parr, J . Chrm P h y s . , 21, 466, 767 (1953). (7) R . Parisel-, ibid , 24, 250 (1956). ( 8 ) P Schiess and A . Pullman, J c h i m . p h y s . , S S , 101 (1956). 19) C. F Wilcox, J r . , S Winstein, and W . G. McMillan, J . A m . Chem. SOL., sa, 51.50 (1960). (10) R . B. Het-mann, J . 0 1 8 . Chem., 2 1 , 441 (1962):
expected from the calculations. We have therefore selected a number of examples of olefins for which experimental data are available for comparison, and which contain all of the different' kinds of features in which one is interested. In this paper we will discuss only the olefinic portion of the system; the effect of substituents such as methyl groups or alkoxy1 groups will be deferred to a later paper. The compounds for which calculations have actually been made in the present work are butadiene (11) (as a function of the rotational angle about the C-C single bond), 1,3,5-hexatriene (111) (in all of its planar conformations), bicycloheptadiene (or norbornadiene) (IV), bicyclo [2.2.2]-2,5,7-0~tatriene (V), tetrahydroacepentylene (VII), and symcis,cis,cis- 1,4 7-cyclononatriene (VI). ~
m 5
3
V
Iv
5
3
VI
6
6
2
VI1
Mathematical Method T h e *-electron approximation is used throughout the present work, including for the nonplanar systems in which its validity is much less certain than for the planar systems. This approximation is necessary a t this time if the problems are t o be held t o manageable size, and its validity will ultimately be found in the agreement between experimental and calculated quantities. Only the *-electrons are treated explicitly; the uelectrons together with the nuclei form the "core" or the field in which the *-electrons move. Thus, for a molecule with n T electrons, moving in the field of the core, the r-electron Hamiltonian operator is written n
=
C Hcore(i) i - 1
+
'/2
C
; : (e*/rij)
(1)
i'gj
The n-electron system wave functions are built from normalized antisymmetrized product functions, expressed as Slater determinants. T h e molecular orbitals (MO) are taken t o be ortho-
2812
NORMAN L. ALLINGER AND MARYANN MILLER
normal combinations of the atomic orbitals (AO) involved in the system being treated. 4i
(2)
= Ccipxp P
Simple LCAO molecular orbitals (Appendix 11) have been used in the present work. An electronic state is then given by a linear combination of the appropriate configurations
9 = Aiai
+ Azaz + . . . + A*@* + .
and the energy E of the state
E
Q
, ,
(3)
is found from eq. 4.
= J9X9dL'
(4)
In matrix notation, the energy is given by the diagonal elements of the diagonal matrix E , where
AHA-'
=
E
A is a unitary matrix and H is the Hamiltonian matrix with individual matrix elements
Xmn= J@mX@ndV The integrals X,,
(5)
Vol. 86
Evaluation of Atomic Integrals and Molecular Orbitals .--The integrals over atomic orbitals to be evaluated are the resoi1:mce and the Couintegrals pp4, the Coulomb attraction integrals e,,, lomb repulsion integrals ypq. The atomic orbitals are the notleless Slater AO'sI3; for the Slater orbital exponent j-, the empirical value14 TC = 1.59 was used rather than the theoretical value ( 1.625). While it is possible to calculate from the theoretical forrnulaG numerical values for the resonance integrals, such a calculation requires that approximations be made concerning the repulsion integrals contained in this formula. Thus there is considerable doubt as to the validity of the "theoretical" value obtained. Furthermore, the theoretical value is a function of the environment of the atom pair considered, and the calculation is tedious. We have therefore followed Pariser and Parr in assuming that the resonance integrals can be taken as constant between a given pair of atoms a t a fixed distance, and independent of their environment. We have accepted the numerical value of B = -2.40% e.v. a t 1.397 b. in benzene, as previously described.' For atoms which are not directly bound together, the theoretical formula indicates t h a t p falls off very rapidly with distance, and we have accepted the exponential relationship used previously ( 15).l In principle the values obtained from this formula
are expressed in terms of core integrals
and electronic repulsion integrals
(mklnl) = These molecular integrals are expressed in terms of atomic integrals using eq. 2
should be increased slightly to account for uu overlap if the orbitals are not parallel. For the molecules studied here the deviation from planarity is not great enough really to warrant attenipting such a refinement for atoms not bound together. For atoms which are bound together, the exponential variation (15) makes @ far too dependent on distance. Since it was found necessary to reduce empirically the one- and two-center repulsion integrals considerably below the theoretical value to account for electron correlation, and since the theoretical expression for p contains these integrals,6 a similar sort of reduction might be expected here. We have therefore elected to use the Mulliken relationship" for @ between bound atoms (16) where IV is an appropriate
where
The core integrals are then obtained as follows, assuming forma neglect of differential overlap
atomic valence state ionization potential, S is the overlap integral, and A is an empirical constant dependent on the type of overlap (1.00 for S,,, and 0.77 for Sn,,).For the present work, since we are considering only double bonds between carbon atoms and using benzene as our standard, this reduces to C
Employing the ZDO approximation, the electronic repulsion integrals ( i j l k l ) are obtamed as linear combinations of integrals over the atomic orbitals of the form ( p p l q q ) where
Coulomb Attraction Integrals.-In terms of atomic orbitals the theoretical expression for the one-center Coulomb attraction integral CY is The key t o the success of the method now depends on the empirical evaluation of up, Bpq, and ypq, and this evaluation is discussed below. Since we are using Hiickel type" rather than selfconsistent-field molecular orbitals, I t is possible to improve the orbitals by configuration interaction among the ground and singly-excited configurations. The configuration interaction treatment has been limited to, and includes all of, the singly excited configurations, together with the ground configuration. The intensities of the spectral transitions were calculated according to Mulliken and Rieke,12simplified by the zero differential overlap approximation. For transitions between nondegenerate states described by wave functions $k and IL1, the oscillator strength f is
f
=
1.085 X 10"w
C Q i z
i
=
(13)
x,y,s
where w is the frequency of the transition in cm.-l and
where i is the algebraic sum of the i's for the several electrons. (11) E. Huckel, Z . Physik, 70. 204 ( 1 9 3 1 ) . (12) R . S. Mulliken and C . A. Kieke, Re@. P r o w . P h y s . , 8 , 231 (1'241).
aLl =
w,- PcZ a KPP144) + ( 4 : P P ) l - C ( r : P P )
where ( p p l q q ) are the Coulomb repulsion integrals yp4,and ( 4 : p p ) and ( 7 : p p ) are Coulomb penetration integrals between xr,and neutral atoms q and 7 . These latter terms are small [ ( q : p p )E 0 . 1 for ~ p~ and ~ p neighbors, much less for p , q nonneighbors] , and have little effect on the calculated spectrum of an all carbon system. Neglecting penetration integrals, the CY'S are obtained from cyp
(e.v.) = -11.22 -
Cypq
(19)
P f q
Coulomb Repulsion Integrals.-These integrals are available directly from tables,16 and if the x-orbitals are not parallel, they should be factored into x , G , and u components. At distances (13) J. C. Slater, Phys. Re?., 36, 57 (1930). (14) R. G. P a r r and B. L. Crawford, J r , , J . Chem. P h y s . , 16, 526 (1'348) (15) (a) R . S. Mulliken, J . A m . Chem. Soc., 79, 4183 (1450); ( b ) J Phys Chem., 16, 295 (1952). (16) R. C. Sahni and J. W. Cooley, "Tables for Two-center Two-electron Coulomb Integrals." S a t i o n a l Aeronautics and Space Administration Technical X o t e D-146-11, Washington, D. C . , 1'360.
ULTRAVIOLET SPECTRA OF POLVOLEFINS
July 20, 1964
of less than 2.408 A. i t is necessary to reduce the theoretical values empirically in order t o account for electron correlation. This was done earlier following the method suggested by Schiess and Pullman.* The equation obtained1 was
ypq (e.v.) = 11.080 - 3.923277
+
0.69786r2 (7 5 2.408 A)
(20)
No attempt has been made to factor repulsion integrals at these close distances since the empirical curve is so arbitrary as it is. Computer Programs.-The calculations were done on an IBM i o i o computer using programs written in Fortran I1 (Full Fort r a n ) . The program sequence begins with the diagonalization of the matrix given in Appendix 11. T h e resulting eigenvectors, arranged in order of decreasing eigenvalue, are the MO coefficients tip. These, together with the matrices r and H,,,,, are the input data for a series of programs which calculate in order the quantities in eq. 8, 21-28,’
vi,=
VO
+
I k
-
ri
+
Jik
C(2Jfi-
f
Kik
2Jfk
f # i
-
-
Jii
Kfi
-
+
Kfk)
(24)
The subscript f in (23) and (24) refers to orbitals which are doubly occupied in the ground state. (VOis corrected for internuclear repulsion energy by the third summation in eq. 23.)17s
{Val vi,)
= 2””Iik
i
f (iilik)
+
The electronic integrals are evaluated as above, e.g.,
( v i k l v j k ) = -1ij
+ 2(ikljk) - (ijli;)- (zj’lkk) (ijljj) -
(vikl vjl) ( V i k l Vil}
= Ikl
=
Ci,j[2(iAij)
ff
- (&j)l
2(ik/jl) - ( i j / k l )
+ ( i k l d ) f (iilkl) +
c [2(.flkl) - (.fkIfl)l
f f k,l,i
(26) (27)
(28)
The configuration energies and configuration interaction matrix elements are then arranged to give the Hamiltonian matrix H, which is diagonalized to give the state energies (eigenvalues) and the configuration coefficients A a l k (eigenvectors). The oscillator strengths (eq. 13) are evaluated according t o the method given by Pariser’ from the state energies, configuration coefficients, molecular orbitals, and nuclear coordinates.
Discussion The calculated benzene spectrum’ was adjusted empirically to fit the observed spectrum, so the agreement between the two is quite good. Since there are a number of empirical adjustments which could have been made which would have given a satisfactory calculated benzene spectrum, there is no guarantee t h a t it will be possible to fit results for other compounds of interest using the same methods and numerical quantities as used for benzene. There are, moreover, additional
2813
factors peculiar to the different compounds which are not encountered in the benzene system, and which may be taken into account in varying ways. The inclusion of resonance integrals only between neighbors, for example, is not a clear cut concept in compounds such as norbornadiene. Thus for the sake of selfconsistency , we decided to omit resonance integrals between nonneighbors in the conjugated systems, but to include them (evaluated by the exponential formula) for those nonconjugated systems where the internuclear distances are between the benzene 1,3- and 1,4-distances. T h e method of evaluating data described above appears to us to be the most logical, b u t other possibilities were also examined : for example, for trans-butadiene the exponential formula was used to evaluate 312 and 323, with the other 3’s being set equal to zero. The results were quite unsatisfactory. When the Mulliken relationship was used for all resonance integrals (BIZ, 3 2 3 , 313, and o l d ) , the value for the low energy transition was satisfactory, b u t the second transition was in poor agreement with experiment. All of the pertinent numerical information is summarized in Appendices I and 11. The results of the calculation for butadiene as a function of the dihedral angle about the 2,3-bond are presented in Table I. Certain features of interest from the calculations can be seen in this table, as follows: the cis isomer should show absorption at longer wave length (230 mp) than the trans (212 mk)) and this absorption should be only about half as intense as that of the trans. As the dihedral angle a t the 2,3-bond changes from 0 to 90’ to 180’ the wave length of the N + V1 absorption drops from 230 to 182 and then rises to 212 mp. At the same time the oscillator strength goes from 0.5 to 0 to 1. The qualitative features calculated here are in good agreement with those observed from a study of more complicated systems containing butadiene structures with various 2,3-dihedral angles. T A B L EI CALCULATED TRANSITION ENERGIES AND OSCILLATOR STRENGTHS FOR BUTADIENE (11) AS A FUNCTION OF THE DIHEDRAL ANGLE Dihedral angle 8 , deg
0 30 60 85 90 95 120 150 180
-------N
-,VI-
f 230 0 465 210 463 201 392 183 089 182 000 183 091 581 195 204 908 212 997
A,
mr
-N A , mr
-
-
Transitions----------Vs--N f A, rnr
182 0 044 182 040 183 015 182 001 182 000 181 001 180 019 017 179 179 000
Va-
f
161 0 703 164 664 167 571 170 429 170 394 171 358 172 171 171 031 170 000
--N
-
A mr
V,-
f
133 lo-‘ 137 0 021 154 159 170 601 170 725 169 673 150 391 134 292 129 292
A study of the ground state of butadiene ( 8 = 0, 90, and 180’) by the SCF method was carried out recentlyl’b and complements the present work. The methods and numerical values used in the two studies were quite different, yet conclusions from the point of overlap between them (the variation of the energy of the T system with 8) are similar, namely that the s-cis- and (17) (a) R . G. Parr and R. Pariser, J. Chem. P k y s . , 28, 711 (1955); (b) I. Fischer-Hjalmars, paper presented a t the Winter Institute in Quantum
Chemistry and Solid State Physics, Sanibel Island, F l a . , Jan., 1962. T h e authors are indebted t o Dr. Fischer-Hjalmars for a preprint of this paper,
NORMANL. ALLINGERA N D MARYANNMILLER
2814
s-trans systems have similar r-energies, while the perpendicular form is some 10 kcal./mole higher. Recently the synthesis of 2,3-di-t-butyl-l,%butadiene (VIII) was reported.18 Due to the steric requirements of the t-butyl groups, it was expected that this com-
Vol. 86
I1 and compared with the experimental values for these compounds. I t is noted that the calculated wave lengths for the cisoid dienes tend to be rather consistently about 12 mp lower (0.24 e.v. higher) than those observed, with the exception of cyclopentadiene
t-Bu S
+
TABLE I1 VI TRANSITION &'AYE LENGTHS O F COSJUGATED DIESES AS A FUSCTION OF THE DIHEDRAL XXGLE E,za
VI11
Amp26
Compound
deg
(calcd )
s-trans-Butadiene l,3-Cyclopentadiene 1,3-Cyclohexadiene 1,3-Cycloheptadiene 1,3-Cyclooctadiene 1,3-Cyclononadiene, nonrigid, various possibilities from 65-110'
180 0 17 52 64 65 90 110
21227 250 245 230 216 216 202 211
A w (exptl )
209 -Iz7 242 256 5 248 229 219 5
Ref
21
28 pound would exist in the twisted form, corresponding 29 to 6 equals 90'. Nuclear magnetic resonance studies 29 confirmed this conformation. In the vapor phase, VI11 30 was reported to show a shoulder a t 209 mp and the 31 major peak a t 185 mp in good agreement with our calculated value of 152 mp for 90'-11. (The shoulder a t 209 was attributed to characteristic absorption of The transitions from the ground state to the two tripsubstituted ethylenes.l9 Alternatively, it might be due let states of lowest energy were also calculated for cisto the presence of a small amount of a transoid conand trans-I1 and for I I I a , and are compared with obformation .) served values3z in Table 111. The agreement here is The calculated oscillator strength for trans-butadiene rather poor, but since the empirically evaluated quantiappears to be too large by about a factor of 3, b u t the ties used in the calculation were obtained from transiratios of the oscillator strengths as a function of ditions to singlet states, this is not surprising.g3 Hamhedral angle appear to be about right. Since the oscilmond's work34 on the dimerization of dienes in the lator strength is quite a sensitive function of the electriplet state strongly suggests that the lowest triplet tron distribution,*O i t is expected that there may be apstate should be lower for s-cis-butadiene than i t is for preciable quantitative errors here, and the agreement the s-trans conformer, although cyclohexadiene itself is regarded as satisfactory. The agreement between is not an ideal model for the cis conformer. The enerthe calculated and experimentalz1values for the s-trans isomer is excellent, both for the N -+ VIand N -+ Vz gies calculated here are the same for the two conformtransitions: N + VI, calcd. 212, exptl. 209.4; N Vz, ers, but these values are of limited accuracy for reasons mentioned. calcd. 175, exptl. 175.4. Since butadiene actually exists in the s-trans conformation, the spectrum of the TABLE I11 s-cis conformer is not known. (There is experimental CALCULATED AND OBSERVED TRIPLET TRANSITION ENERGIES evidencez2which indicates that 3-7yo of cis-I1 may be Energy, e.v.--present a t room temperature, b u t no ultraviolet absorpTransition Calcd. Obsd. System tion has been attributed to it.) cis-I1 K -+ T I 2.26 3 0" While the electronic spectrum of butadiene itself for S + Tz 3.87 .. 2.22 3.22 trans-I1 S +TI 0 # 180' is not known, there are a number of substiS + T2 3.82 3.84 tuted cisoid dienes available for comparison purposes. 1.84 2.58 IIIa S + Ti The N -+ VI transition energies have been reported S + T2 3.18 3.22 for cyclopentadiene, cyclohexadiene, cycloheptadiene, This value is estimated from t h e value of cyclohexadiene cyclooctadiene, and cyclononadiene. The spectra were (2.85 e.v. (ref. 3 2 ) ) which must be corrected for nonplanarity measured in various solvents, and such values are gena n d for substitution. erally found to be about 10 mp higher than in the vapor 1,3-frans,S-Hexatriene (2-s-cis; 4-s-trans) .- Both phase.z3 From an examination of scale models i t is posthe l i q ~ i d ~ 5and 3 ~ vapor ~ p h a ~ e ~spectra ' , ~ ~ of this comsible to make a rather good estimate of the dihedral pound have been reported. Fine structure is observed angle between the olefinic linkages for each of these compounds except the last.24 For each of these angles (25) Measured fron Dreiding models, accurate to 3 ~ 5 ' . (26) These values are calculated for solution in hexane (see text) then, one can estimate by interpolation of the data in (27) Vapor phase. Table I where the butadiene linkage should absorb. (28) Present work (in hexane solvent): ref. 2.5 gives 238.5 mp. The presence of the two methylene groups on the ends (29 S. F. Mason, Quart. Rev. (London), 16, 287 (1961). (30) A. C , Cope, M. R. Kinter, and R . T. Keller, J . A m . Chem. Soc., 76, of the butadiene system will raise the value 10 my 2757 (1954). above that which would be observed in the butadiene it(31) R . W . Fawcett and J . 0. Harris, J . Cliem S O L . ,2669 (1954). self, and another 10 mp is added as a solvent correction. (32) D. F. Evans, ibid., 1735 (1960). (33) T h e empirical reduction of the value of the one- and two-center The calculated values thus obtained are listed in Table repulsion integrals a t short distances, for example, is a way of including the
-
7--
(18) H. Wynherg, A . I>eGroot, and D W. navies, Telrohedron Letters. 1083 (1963) (19) R . S. Bel-ry, J . Chem. P h y s , 38, 1934 (1963). ( 2 0 ) C. M Mosei-, J . Chem S O C ,34.55 (1954). f 2 l ) I,. C . Jones, Jr., and I,. W. Taylor, A n d Chem , 27, 228 (1955). (22) W B. Smith and J. 1,. Massingill, J . A m . Chem. SOL.,83, 4301 (11161) (2.3j Vapor phase spectra f o r some of the dienes are available: cyclohexadiene, 2.iO m p (ref. 3 7 ) , cyclopentadiene, 285 mp; and cyclooctadiene, 2 2 2 m p (rei. 28) ( 2 4 ) E. A . Braude. Chem. I n d . (London), 1557 (1954).
effect of electron correlations in the calculation. This correlation should be different for the triplet states, however. (34) G. S. Hammond and R. S. H . Liu, J A m . Chem Soc , 86, 477 (1963). (35) J. M. Shackelford, W. A . Michalowicz. and I,. H . Schwartzman, J . Org. Chem., 27, 1631 (1962). (36) J . C. H . H w a , P. L. d e Benneville, and H. J. Sims, J . A m . Chem S o c . , 82, 2537 (1960). (37) H . Schuler, E. 1,utz. and G. Arnold, Speclrochim. Acla, 17, 1043 (1961). (38) W. C . Price and A . D . Walsb, Proc. R o y . SOL.(London), A186, 182 (1946).
ULTRAVIOLET SPECTRA OF POLVOLEFINS
July 20, 1964
2S15
TABLE IV AXD OBSERVED SPECTRA OF CONJCGATED TRIENES CALCULATED
Compound
--N A, rn,u
IIla
244
1.461
201
0.000
182
0 062
IIIb
241
1.275
205
0.021
187
0.062
IIIc
254
0,844
199
0.070
184
0.205
IIId
266
0,473
215
0.010
192
0.689
355 32 285 13,500 IIIe 264 285 32 245(230) 9,100 IX 244 243 203(188) 12,100 chromophore calculated from the observed value.
0.714 0.342
203 213
0.000 0.000
198 205
0 000 0.862
IIIa X XI IIIb Ergosta-4,6,8(14),22-tetraene Cholesta-2,4,6-triene Cholesta-3,5,7-triene Tachysterol
Ref.
A, mpn
36
41,000 33,000 15,700 15,700 26,400 29 300 12,680 18,300
241.2 175 256 269 284 255 283 306 315 281 280
35 39 39 36 40 40 40 42 41
241.2 175 241 234 239 240 238 271 270 241 240
324 265
32 32b
284 230
L , ,
3
ma
*
p 5
6
4
IIIb
IIIe
X,mr
v3---
f
....
~
IIIc
2$5
1
IIId
f
....
-
, Fel-nholt, K g l . S o l s k p V i d r n s k a b Selskabs S k r i f l e r . No. 3, 1 (1'338) (.54) (a) A Almenningen, 0. Bastiansen, and M. Traetteberg. A d a C h m .%and., 12, 1 2 2 1 (1058); (b) L). J Mal-ais, S Sheppard, and B P. Stoicheff, T e f r a h e d r o n , 17, 163 (1962)
July 20, 1964
Where the geometry is not accurately known, standard bond lengths and angles were used.
trices of the form -012
Bond lengths of 11, C=C-C
1IIb-e
=
- P32
120'
r12 = 1.337 A.
V
I
VI
=
. . . . . . . . . -PIN
0. . . . . . . . . . . .
.-@3N
...................................... -PN2 -PN3., . . . . . . .0
( -ON1
r18 = 1.501 A.55 r23
-p13.
- p 4 3 . . . . . . . . . . -Pzx
Bond lengths and angles of I1
IIIa
2819
ACYLCOBALT CARBONYLS ADDEDTO ACETYLENES
2.375 A. (measured from Dreiding models)
rZ3= 1.337 A.
C=C-C
=
120'
r17 = 1.501 A.
C-C-C
=
109'28'
VI1
Average measured Dreiding model distances were used. The coordinates were calculated from various groups of measured distances until consistency between calculated and measured values was achieved
IX
Bond lengths of 11, C=C-C
=
120'
Appendix I1
(55) D. R. Lide, Jr., Telrahedron, 17, 125 (1962). (56) These d a t a for I have been published previously (ref. 1). For t h e other compounds treated in this paper t h e nuclear coordinates, atomic a n d molecular integrals, configuration energies, and configuration interaction elements are available from t h e authors.
RESEARCH CENTER
1 I
1
where P is the corresponding number in Hcore. For butadiene (11), CY was calculated from eq. 18 with t h e Coulomb penetration integrals (g : p p ) evaluated according to Parr and Crawford5' and ( r : p p ) neglected; eq.. 19 was used for the other compounds. The spectroscopic value of 11,080 e.v.j8 was used for the onecenter two-electron Coulomb repulsion integral -yllj and - 11.22 e.v. for the ionization potential Wc. Acknowledgment.-The authors are indebted t o Drs. D. D. Ebbing, R. B. Hermann, and H . E. Simmons for helpful discussion of many theoretical problems encountered during the course of this work, and to Mr. B. Gorden and Mr. R. A. Ford for the ultraviolet spectral measurements reported herein, and for the required synthetic work.
Numerical Data.-Selected numerical data for the calculated spectra are given in Tables VI1 and VIII.66 Molecular orbitals were obtained by diagonalizing ma-
[CONTRIBUTION FROM THE
I
O F THE
(57) R. G. Parr and B. L. Crawford, Jr., 3 . Chem. P h y s . , 16, 1049 (1948). These integrals are evaluated using hydrogenic orbitals. Slater orbitals would be more consistent since they are used f o r t h e other integrals calculated, but t h e differences are small. Later calculations of the S -+ V I energy for c i s - and frans-11, neglecting penetration integrals, including ( q : p p ) for Slater orbitals with ( r : p p ) zero, and including both ( q : p p ) a n d ( r : p p ) , gave results constant t o within 3 mp for both isomers. (58) H . A. Skinner and H . 0. Pritchard, T r a n s . F a r a d a y Soc., 49, 1254
(1953).
HERCULES POWDER
C O . , WILMINCTON
99,
DEL.]
The Addition of Acylcobalt Carbonyls to Acetylenes. ~-(2,4)-(Alkeno-4-lactonyl)cobalt Tricarbonyl Derivatives and Their Conversion to 2,4-Pentadieno-4-lactones BY RICHARD F. HECK RECEIVED FEBRUARY 5, 1964 Acylcobalt tetracarbonyls react with substituted acetylenes t o form T-(2,4)-(alkeno-4-lactonyl)cobalttricarbonyl derivatives. These complexes react with triphenylphosphine, with evolution of carbon monoxide, t o form mono(tripheny1phosphine) derivatives. The acetylcobalt tetracarbonyl-3-hexyne-triphenylphosphine complex was isolated and characterized. Reactions of the r-lactonyl complexes were investigated, The most significant finding was a new method of synthesizing 2,4-pentadieno-4-lactones b y the reaction of T-lactonyl complexes with dicyclohexylethylamine.
Introduction Alkyl- and acylcobalt carbonyls react readily with many compounds having unshared electron pairs. and phosphites3 form acylcobalt tricarbony1 monophosphines or monophosphites. The alkyland acylcobalt carbonyls can react with olefins to form ketones4 and with conjugated dienes to form l-acylmethyl-a-allylcobalt t r i c a r b o n y l ~ . ~The present paper is concerned with the reaction of acetylenes with acylcobalt carbonyls. Many reactions of acetylenes with transition metal carbonyls and their derivatives are known, b u t there ap(1) R. F. Heck and D. S. Breslow, 3 . A m . Chem. Soc., 84, 2499 (1962) R. F. Heck, i b i d . , 86, 6.51 (1963). (3) R. F. Heck, ibid., 85, 1220 (1983). ( 4 ) R.F. Heck, i b i d . , 85, 3118 (1983). ( 5 ) R. F.Heck, ibid., 86, 3381 (1963). (2)
pear to be only a few examples of the reaction of alkyl(or acyl)-metal carbonyls with acetylenes. Coffield reported t h a t methylmanganese pentacarbonyl reacted with acetylene a t 150' under pressure to produce a-dihydropentalenylnianganese tricarbonyl.6 The fate of the methyl group was not determined. Zeiss has investigated the reactions of various aryl- and alkylchromium compounds and other transition metal alkyls with acetylenes and found t h a t aromatic trimers are formed, sometimes accompanied by products containing two acetylene units and an alkyl7 or aryl group from the metal.8 Diethylnickel and diphenylacetylene yielded tetraphenylcyclohexadiene as well as hexa(6) T. H . Coffield, K . G Ihrman and W . Burns, i b i d , Sa, 1 2 5 1 (1980). (7) M . Tsutsui and H. Zeiss, i b i d . , 81,6090 (1959). ( 8 ) W. Herwig, W. Metlesics, and H . Zeiss, i b i d . , 81, 6203 (1959).
