y axis. As is common in CNDO/S calculations on thiones,5 the
energy of the n→π* transition is vastly underestimated.
The splitting of the nϪ→π* and nϩ→π* transitions is pre-
dicted to be 12 600 cmϪ1, much larger than for 1,2-dithiones
(calculated 1300–2400 cmϪ1, experimental value 1450 cmϪ1)5
or 1,3-dithiones (calculated 400–1000 cmϪ1, no splitting
observed29). If we assign band D to the nϩ→π* transition, the
experimental splitting in 3c is 15 900 cmϪ1. Evidently, the results
of the calculations are in qualitative agreement with this
assignment.
Calculation of the contributions from the coupled oscillator
mechanism is straightforward.
For these calculations, it is necessary to know the geometry
of 3c in solution. In a recent work,34 we have studied the struc-
ture of N,NЈ-bis[(S)-(1-phenylethyl)]thiourea (4), which has
some structural similarity to 3c. As for all secondary alkyl
groups attached to planar frameworks, the two main conform-
ational types A and B (Scheme 2) need to be considered.35–37 It
Theoretical calculation of CD spectra
Calculation of CD spectra by the Schellman method21 com-
bines the Condon–Altar–Eyring (CAE),30 the magnetic–electric
coupling31,32 and the coupled oscillator33 mechanisms. In the
CAE mechanism one n→π* and one π→π* transition in the
same planar chromophore are “mixed” by chirally placed static
charges outside the plane of the chromophore. The application
of the program to 3c is complicated by the fact that we need to
consider two n→π* and two π→π* transitions in the dithiocarb-
amate anion chromophore, while the program permits only one
n→π* and one π→π* transition in each chromophore with
the CAE mechanism and only one n→π* transition in the
magnetic–electric coupling mechanism. In the dithiocarbamate
anion, the n→π* transitions originate in the antisymmetric
(ψnϪ) and symmetric (ψnϩ) combinations of the sulfur lone pair
orbitals (ψnA and ψnB, Fig. 1) and the treatment has to be modi-
fied. The rotational strengths of one n→π* transition can be
expressed by eqn. (2),32 where V, the first order perturbation
energy, is the energy of interaction between the transition quad-
rupole related to the n→π* transition and the chirally placed
static charges, qi. νππ* and νnπ* are the respective transition ener-
gies, and mnπ*A and mnπ*B are the magnetic transition
Scheme 2
was shown that the conformer of 4 corresponding to B is
strongly disfavoured by steric and coulombic repulsion between
the aromatic ring and the sulfur atom.34,37 The negative charge
in 3c must increase this repulsion both by making the electron
cloud around the sulfur atom more diffuse and by augmenting
the coulombic repulsion between the sulfur atom and the
negatively charged aromatic carbon atoms. Consequently, only
conformer A needs to be considered. However, force-field calcu-
lations on 4 indicated the existence of several close-lying energy
minima, forming what was called a “conformational family”,
differing in the value of the Hα–C–N–C dihedral angle (α). The
globe energy minimum was found at α = Ϫ35Њ.
AM1 geometry optimizations of the N-(1-phenylethyl)-
dithiocarbamate anion, starting with α = 0Њ (A) and 180Њ (B) led
to energy minima with α = Ϫ9.6 and Ϫ164.7Њ respectively, the
former being 10.2 kJ molϪ1 lower in energy.
Calculations by the Schellman method were performed for α
values in steps of 10Њ from α = ϩ40Њ to α = Ϫ40Њ, which should
include all feasible values for α in the A family. One calculation
was performed for α = 180Њ, corresponding to the B conform-
ation (Table 4). The phenyl ring was oriented with respect to
the Cα–N bond approximately as in the Za part of bis[(S)-
(1-phenylethyl)]thiourea, with the Cortho–Cipso–Cα–N dihedral
angles = Ϫ60 and ϩ120Њ. Very similar angles were found in the
AM1 optimized geometries.
Rn→π* = ϪVؒµππ*ؒmnπ*/(νππ* Ϫ νnπ*) = ϪRπ→π*
ψn = (2)Ϫ0.5(ψnA ψnB)
(2)
(3)
V = <ψn|Σqi/ri|ψπ* > =
(2)Ϫ0.5(<ψnA|Σqi/riЈ|ψπ*> <ψnB|Σqi/riЉ|ψπ*>) (4)
mnπ* = mnπ*A mnπ*B
(5)
moments corresponding to the respective transitions from ψnA
and ψnB to the π* orbital. The latter moments are vectors
oriented along the respectives S–C bonds. Since the lobes of π1*
on the sulfur atoms are in the same phase (Fig. 1), the indi-
vidual magnetic moments have the same direction along the C–
S bonds in the nϪ→π* transition and opposite directions in the
nϩ→π1* transitions, and the resultant magnetic transition
moment is y-oriented in the former and x-oriented in the latter
case. Since the π1→π1* transition is x-polarized and the π2→π1*
is y-polarized (Table 1), the scalar product µππ*ؒmnπ*, and thus
also Rn→π*, vanishes for the nϪ→π1–π1*→π1* and nϩ→π1*–
π2→π1* interactions.
Comparison of the experimental rotational strengths with
the calculated ones shows good agreement in the range
Ϫ40Њ ≤ α ≤ Ϫ20Њ except for the transition assigned as nϩ→π1*
(D). Positive rotational strengths for this transition are pre-
dicted only for α = ϩ40 and 180Њ, but for these conformations
the wrong signs are predicted for transitions A, B and 1Bb, and
1
at 180Њ also for C and Lb. The good agreement in sign and
order of magnitude for six transitions in the range
Ϫ40Њ ≤ α ≤ Ϫ20Њ gives good support for the assumption that the
molecules preferentially reside in this range.
Different explanations may be advanced for the incorrectly
predicted sign for band D. One may be that the assumption of
strictly x and y polarized π→π* transitions is an oversimplifi-
cation. Homoconjugation with the phenyl ring may change the
polarization directions and e.g. allow interaction between
an electrically nearly y-polarized π2→π1* and a magnetically
x-polarized nϩ→π1* transition. In a model calculation combin-
ing these two transitions with the CAE mechanism, the electric
moment was rotated 15 degrees from the (S2)C–N direction in
the direction of the substituent (Scheme 3). However, the
resultant rotational strengths were only 4% of those required
and with the wrong signs. Apparently, this is not a feasible
explanation.
However, the magnetic–electric coupling31,32 also contributes
to Rn→π*. An expression similar to eqn. (2) has been given for
the contribution from one n→π*–π→π* interaction.32 In this
case, µππ* is an electric transition moment in a different, chirally
placed chromphore (an “external” chromophore), in 3c the
phenyl ring, and the perturbation energy, V, is the energy of
interaction between µππ* and the n→π* transition quadrupole.
Similar contributions arise from all interactions of π→π* tran-
sitions in “external” chromophores with the n→π* transitions
in the “central” chromophore.
Using the Schellman method with appropriate combinations
of transitions, it is possible to calculate the terms from the
nϪ→π2* and nϩ→π1* transitions arising from the CAE and
mµ mechanisms, which contribute to the rotational strengths.
Another explanation for the incorrect prediction of the sign
for band D may be that the corresponding transition has not
been correctly assigned, i.e. it is not the nϩ→π1* transition.
830
J. Chem. Soc., Perkin Trans. 2, 1999, 827–832