Ab initio calculation of optical rotatory dispersion (ORD) curves: A simple and reliable approach to the assignment of the molecular absolute configuration

Article abstract of DOI:10.1021/ja046875l

In this paper, both Hartree-Fock (HF) and density functional theory (DFT) methods have been used to make ab initio calculations of the optical rotatory power of selected molecules at several wavelengths; that is, part of the optical rotatory dispersion (ORD) curve has been predicted. This approach constitutes a new, simple, and reliable method to assign the molecular absolute configuration, at least for rigid molecules such as those studied in the present work. In fact, in this way, it is possible to overcome the difficulties connected to some relevant cases, in particular that of (-)-β-pinene, for which even a very high-level (DFT/ B3LYP/6-311++G(2d,2p)) calculation affords the wrong sign of the optical rotation at 633 nm. On the contrary, the predicted ORD curve, even using small basis sets, reproduces (below 400 nm) the experimental trend well, allowing for the correct configurational assignment. This result clearly shows that to have a reliable configurational assignment the comparison between experimental and predicted rotation values must be carried out at different wavelengths and not at a single frequency. The reason for this is that working at wavelengths approaching the absorption maximum the [α]_λ values become larger and their prediction becomes more reliable. Coupling the use of an inexpensive instrument (a polarimeter working at a few wavelengths) with the use of a DFT-calculation package can also allow the experimental organic chemist to arrive, quickly and reliably, at the assignment of the molecular absolute configuration.

Full text of DOI:10.1021/ja046875l

Published on Web 09/17/2004
Ab Initio Calculation of Optical Rotatory Dispersion (ORD)
Curves: A Simple and Reliable Approach to the Assignment
of the Molecular Absolute Configuration
†
‡
‡
,†
Egidio Giorgio, Rosario G. Viglione, Riccardo Zanasi, and Carlo Rosini*
Contribution from Dipartimento di Chimica, UniVersit a` degli Studi della Basilicata,
Via N. Sauro 85, 85100 Potenza, Italy, and Dipartimento di Chimica,
UniVersit a` degli Studi di Salerno, Via S. Allende, 84081 Baronissi, Salerno, Italy
Received May 27, 2004; E-mail: rosini@unibas.it
Abstract: In this paper, both Hartree-Fock (HF) and density functional theory (DFT) methods have been
used to make ab initio calculations of the optical rotatory power of selected molecules at several wavelengths;
that is, part of the optical rotatory dispersion (ORD) curve has been predicted. This approach constitutes
a new, simple, and reliable method to assign the molecular absolute configuration, at least for rigid molecules
such as those studied in the present work. In fact, in this way, it is possible to overcome the difficulties
connected to some relevant cases, in particular that of (-)-â-pinene, for which even a very high-level (DFT/
B3LYP/6-311++G(2d,2p)) calculation affords the wrong sign of the optical rotation at 633 nm. On the
contrary, the predicted ORD curve, even using small basis sets, reproduces (below 400 nm) the experimental
trend well, allowing for the correct configurational assignment. This result clearly shows that to have a
reliable configurational assignment the comparison between experimental and predicted rotation values
must be carried out at different wavelengths and not at a single frequency. The reason for this is that
λ
working at wavelengths approaching the absorption maximum the [R] values become larger and their
prediction becomes more reliable. Coupling the use of an inexpensive instrument (a polarimeter working
at a few wavelengths) with the use of a DFT-calculation package can also allow the experimental organic
chemist to arrive, quickly and reliably, at the assignment of the molecular absolute configuration.
Introduction
through the calculation of the optical parameter â, which is
Thanks to recent progress in computational methods, the ab
initio prediction of optical rotation (OR) is now possible,
directly connected to the trace of the frequency-dependent
electric dipole-magnetic dipole polarizability tensor G′, that is
1
-7
and therefore, the theoretical assignment of the molecular
absolute configuration (AC) could be carried out, taking into
account that, for instance, the OR at the sodium D line, that is,
the most common parameter to label optically active compounds,
can be reliably calculated. According to the general theory,
the OR is obtained as the specific rotation [R]λ, for each angular
frequency ω ) 2πν ) 2πc/λ ) 2πc νj of the incident radiation,
-4
2
2
1
.34229 × 10 â νj (n + 2)
[R] )
λ
3
M
8
-10
1
â ) - Tr[G′(ω)]
3ω
4
π
ω
G′ (ω) ) -
J(〈0| µˆ |j〉〈j| mˆ |0〉)
Râ
∑
R
â
†
Universit a` degli Studi della Basilicata.
Universit a` degli Studi di Salerno.
2
2
h
j*0
‡
ω
j
- ω
(
1) (a) Polavarapu, P. L. Mol. Phys. 1997, 91, 551. (b) Polavarapu, P. L.
Tetrahedron: Asymmetry 1997, 8, 3397. (c) Polavarapu, P. L.; Chakraborty,
D. K. J. Am. Chem. Soc. 1998, 120, 6160. (d) Polavarapu, P. L.; Zhao, C.
Chem. Phys. Lett. 1998, 296, 105. (e) Polavarapu, P. L.; Chakraborty, D.
K. Chem. Phys. 1999, 240, 1. (f) Polavarapu, P. L.; Zhao, C. J. Am. Chem.
Soc. 1999, 121, 246. (g) Polavarapu, P. L.; Chakraborty, D. K.; Ruud, K.
Chem. Phys. Lett. 2000, 319, 595. (h) Polavarapu, P. L. Chirality 2002,
3
-1
where the specific rotation is in unit of deg[dm(g/cm )] , â is
in bohr , the radiation wavenumber is in cm ; n is the refractive
index of the medium, M is the molar mass in g/mol; ωj is the
4
-1
1
4, 768. (i) Polavarapu, P. L. Angew. Chem., Int. Ed. 2002, 41, 4544. (l)
Polavarapu, P. L.; Petrovic, A.; Wang, F. Chirality 2003, 15, 143.
2) (a) Kondru, R. K.; Lim, S.; Wipf, P.; Beratan, D. N. Chirality 1997, 9,
(3) (a) Cheeseman, J. R.; Frisch, M. J.; Devlin, F. J.; Stephens, P. J. J. Phys.
Chem. A 2000, 104, 1039. (b) Stephens, P. J.; Devlin, F. J.; Cheeseman, J.
R.; Frisch, M. J.; Mennucci, B.; Tomasi, J. Tetrahedron: Asymmetry 2000,
11, 2443. (c) Stephens, P. J.; Devlin, F. J.; Cheeseman, J. R.; Frisch, M. J.
J. Phys. Chem. A 2001, 105, 5356. (d) Stephens, P. J.; Devlin, F. J.;
Cheeseman, J. R.; Frisch, M. J. Chirality 2002, 14, 288. (e) Mennucci, B.;
Tomasi, J.; Cammi, J. R.; Cheeseman, J. R.; Frisch, M. J.; Devlin, F. J.;
Gabriel, S.; Stephens, P. J. J. Phys. Chem. A 2002, 106, 6102. (f) Stephens,
P. J.; Devlin, F. J.; Cheeseman, J. R.; Frisch, M. J.; Rosini, C. Org. Lett.
2002, 4, 4595. (g) Stephens, P. J.; Devlin, F. J.; Cheeseman, J. R.; Frisch,
M. J.; Bortolini, O.; Besse, P. Chirality 2002, 14, 288. (h) Stephens, P. J.;
Devlin, F. J.; Cheeseman, J. R.; Frisch, M. J.; Bortolini, O.; Besse, P.
Chirality 2003, 15, S57. (i) Stephens, P. J.; McCann, D. M.; Butkus, E.;
Stoncius, S.; Cheesman, J. R.; Frisch, M. J. J. Org. Chem. 2004, 69, 1948.
(
4
1
2
1
69. (b) Kondru, R. K.; Wipf, P.; Beratan, D. N. J. Am. Chem. Soc. 1998,
20, 2204. (c) Kondru, R. K.; Wipf, P.; Beratan, D. N. Science 1998,
82, 2247. (d) Kondru, R. K.; Wipf, P.; Beratan, D. N. J. Phys. Chem. A
999, 103, 6603. (e) Kondru, R. K.; Chen, C. H.; Curran, D. P.; Beratan,
D. N.; Wipf, P. Tetrahedron: Asymmetry 1999, 10, 4143. (f) Kondru, R.
K.; Beratan, D. N.; Friestad, G. K.; Smith, A. B., III; Wipf, P. Org. Lett.
2
000, 2, 1509. (g) Ribe, S.; Kondru, R. K.; Beratan, D. N.; Wipf, P. J. Am.
Chem. Soc. 2000, 122, 4608. (h) Specht K. M.; Nam, J.; Ho, D. M.; Berova,
N.; Kondru, R. K.; Beratan, D.; Wipf, P.; Pascal, R. A.; Kahne, D. J. Am.
Chem. Soc. 2001, 123, 8961. (i) Perry, T. L.; Dickerson, A.; Khan, A. A.;
Kondru, R. K.; Beratan, D.; Wipf, P.; Kelly, M.; Hamann, M. T.
Tetrahedron 2001, 57, 1483.
12968
9
J. AM. CHEM. SOC. 2004, 126, 12968-12976
10.1021/ja046875l CCC: $27.50 © 2004 American Chemical Society

Ab Initio Calculation of ORD
A R T I C L E S
transition frequency from ground state |0〉 to excited state |j〉, µˆ
and mˆ are the electric and magnetic dipole operators, respec-
tively.
It is recognized by some authors3a,7a,7d that at the sodium D
method to assign the molecular absolute configuration. However,
the HF/small basis set approach does not work properly12 for a
simple molecule that fulfils the criterion reported above, that
is, (+)-camphor, because the wrong sign of OR is predicted by
a HF/small basis set treatment. In addition, a recent paper6c by
Vaccaro et al. pointed out that even a very accurate treatment
can fail: this is the case of (-)-â-pinene, for which a DFT/
B3LYP/6-311++G(2d,2p) calculation affords +21 (at 633 nm),
but the experimental value, in the gas phase, is -17. These facts
throw some doubts on the reliability of these calculations.
Therefore, in this paper, we approached the problems related
to the OR calculation of these and other molecules to understand
when we can trust the ab initio prediction of the OR to solve
one of the most important problems of organic stereochemistry,
that is, the assignment of the absolute configuration.13 In
summary, the present paper is organized as follows. First, we
show that a reliable calculation of the optical rotary power at
the sodium D line (i.e., far from the absorption region of the
most common organic molecules) is a difficult task, taking into
account the smallness of the â value (the parameter that
line â is, in general, a small quantity because two of the diagonal
components of G′ almost cancel the third one. As a consequence,
even small changes in the electronic distribution may produce
rather large contributions to the OR computed value. Such
changes may be induced by several factors, for example, the
choice of basis set, electron correlation, solvent effect, equili-
brium geometry, vibrational contributions, and the effect of dis-
persion passing from a static to a dynamic approach. All of
these effects have been deeply explored and found to be very
significant and also sometimes cooperate to give fortuitous
cancellation of errors.1
a,3a-c,4b,4c,7a,7b
Currently, there seems to
be a general agreement on the computational requirements need-
ed for reliable OR predictions,1
g,3c,4a,7b
that is, the use of a dy-
namic method together with a proper treatment of the electron
correlation and the use of large basis sets containing diffuse
functions. Within these respects, the recently introduced cavity
ring-down polarimetry (CRDP)6 for probing, with unprecedent-
ed sensitivity, the circular birefringence and circular dichroism
in the gas phase marks the beginning of a new era of OR
measurements and possibly of absolute configuration assign-
ments, as almost homogeneous comparison between experi-
mental and theoretical results can now be done, for example,
a,7b
determines the sign and magnitude of [R] ). One way to have
λ
larger (and more easily predictable) â values is to approach the
absorption region of the molecule under investigation: this
clearly suggests that the calculation of the optical rotation at
several wavelengths (instead of at a single frequency) and the
comparison with the corresponding experimental data certainly
constitute a safer way to assign the absolute configuration. This
conclusion is not new: Sjoberg and co-workers showed14
experimentally, as early as 1955, that a comparison of chiroptical
data made on several wavelengths affords a more reliable answer
than a comparison at a single wavelength, as far as the
configurational assignment is concerned. However, such crite-
rion did not find a systematic application in the field of AC
assignment by the ab initio calculation of OR: only the AC of
11
the case of (S)-propylene oxide. Furthermore, we have recently
12
shown that even HF/small basis set (for example, a HF/6-31G*)
calculations can be employed for a reliable calculation of OR
(at least in the cases of highly unsaturated and/or aromatic mole-
cules possessing low-lying Cotton effects, determining the OR
at the sodium D line in sign and order of magnitude). The numer-
ical agreement between predicted and experimental values is
1
g,3c,4a,7b
poorer with respect to the above-quoted calculations,
(
-)-2,8,9-trihydroxy-3,4-dihydro-2H-anthracen-1-one has been
but the sign and order of magnitude of OR are correctly repro-
duced. In addition, this simplified treatment can also be applied
to the case of large molecules, which cannot be dealt with using
extended basis sets, but these are the real target of interest for
experimental chemists. All of these considerations clearly
strengthen the expectation that such calculations, affording a
correct value of the optical rotatory power, constitute a reliable
2
h
assigned just doing the calculation at three different wave-
lengths. So, in the second section of this paper, for the first
time, a systematic use of the prediction of ORD curves and
their comparison with experimental data for AC assignments
will be made. To this end, we shall not only employ the DFT/
B3LYP method, which, following Stephens and co-workers,3c
has to be considered as the method of choice but also describe
the results of the corresponding HF computations to try to
understand the reasons of its failure. As it will be clear later,
we shall calculate the ORD curves near to but not at the
frequency of the resonance, following a protocol first introduced
(
4) (a) Grimme, S. Chem. Phys. Lett. 2001, 339, 380. (b) Grimme, S.; Furche,
F.; Ahlrichs, R. Chem. Phys. Lett. 2002, 361, 321. (c) Grimme, S.;
Bahlmann, H.; Haufe, G. Chirality 2002, 14, 793. (d) De Meijere, A.;
Khlebnikov, A. F.; Kozhushkov, S. I.; Kostikov, R. R.; Schreiner, P. R.;
Wittkopp, A.; Rinderspacher, C.; Menzel, H.; Yufit, D. S.; Howard, J. A.
K. Chem.sEur. J. 2002, 8, 828.
1f
(5) Autschbach, J.; Patchkovski, S.; Ziegler, T.; Van Gisbergen, S. J. A.;
Baerends, E. J. J. Chem. Phys. 2002, 117, 581.
by Polavarapu, which assumes the infinite-lifetime approxima-
tion. The calculation of the ORD in the resonant frequency
(6) (a) Muller, T.; Wiberg, K. B.; Vaccaro, P. H. J. Phys. Chem. A 2000, 104,
7
f
5
959. (b) Muller, T.; Wiberg, K. B.; Vaccaro, P. H.; Cheeseman, J. R.;
region has been presented by Ruud et al. only this year.
Frisch, M. J. J. Opt. Soc. Am. B 2002, 19, 125. (c) Wiberg, K. B.; Vaccaro,
P. H.; Cheeseman, J. R. J. Am. Chem. Soc. 2003, 125, 1888. (d) Wiberg,
K. B.; Wang, Y.; Vaccaro, P. H.; Cheeseman, J. R.; Trucks, G.; Frisch, M.
J. J. Phys. Chem. A 2004, 108, 32.
Results and Discussion
Smallness of â. In this section, we will start showing that
the accurate [R]λ calculation is really a difficult task when
carried out at wavelengths far from the absorption region,
whereas the OR values predicted at wavelengths that approach
(
7) (a) Ruud, K.; Taylor, P. R.; Astrand, P. Chem. Phys. Lett. 2001, 337, 217.
(
b) Ruud, K.; Helgaker, T. Chem. Phys. Lett. 2002, 352, 533. (c) Ruud,
K.; A° strand, P.-O.; Taylor, P. R. J. Comput. Methods Sci. Eng. 2003, 3, 7.
(d) Ruud, K.; Stephens, P. J.; Devlin, F. J.; Taylor, P. R.; Cheeseman, J.
R.; Frisch, M. J. Chem. Phys. Lett. 2003, 373, 606. (e) Pecul, M.; Ruud,
K.; Rizzo, A.; Helgaker, T. J. Phys. Chem. A 2004, 108, 4269. (f) Norman,
P.; Ruud, K.; Helgaker, T. J. Chem. Phys. A 2004, 120, 5027.
8) Rosenfeld, L. Z. Phys. 1928, 52, 161.
(13) Two very recent examples of assignment of absolute configuration by the
(
(
D
ab initio calculation of [R] are: (a) Takeshi, K.; Miyako, F.; Toshiyuki,
9) Condon, E. U. ReV. Mod. Phys. 1937, 9, 432.
H.; Tominari, C.; Junko, N.; Yukio, O.; Satoshi, H. Chem. Pharm. Bull.
2003, 51, 20. (b) Vogensen, S. B.; Greenwood, J. R.; Varmin, A. R.; Brehm,
L.; Pickering, D. S.; Nielsen, B.; Liljefors, T.; Clausen, R. P.; Johansen,
T. N.; Krogsgaard-Larsen, P. Org. Biomol. Chem. 2004, 2, 206.
(14) Djerassi, C. Optical Rotatory Dispersion: Application to Organic Chem-
istry; McGraw-Hill: New York, 1960; p 236.
(
(
10) Buckingham, A. D. AdV. Chem. Phys. 1967, 12, 107.
11) Giorgio, E.; Viglione, R. G.; Zanasi, R.; Rosini, C. Chem. Phys. Lett. 2003,
3
76, 452.
(
12) Giorgio, E.; Minichino, C.; Viglione, R. G.; Zanasi, R.; Rosini, C. J. Org.
Chem. 2003, 68, 5186.
J. AM. CHEM. SOC.
9
VOL. 126, NO. 40, 2004 12969

A R T I C L E S
Table 1.
Giorgio et al.
molecule
method
ω-1G
′
xx
ω-1G
′
yy
ω-1G
′
zz
â
[R]
D
H2O2 (æ ) 120°)
HF
B3LYP
HF
B3LYP
HF
B3LYP
HF
B3LYP
HF
B3LYP
-1.138
-1.201
-0.754
-0.820
5.949
5.850
0.408
0.240
0.198
0.387
0.891
0.960
1.447
0.016
-0.049
0.014
18.4
-55.3
(
(
(
(
+)-propylene oxide
-)-dimethylallene
+)-camphor
-0.736
-0.812
-3.259
-4.687
2.925
4.891
91.855
116.468
9.5
1.587
0.015
10.2
-2.066
-0.438
-3.450
-5.847
-31.037
-37.053
-0.208
-0.241
0.039
0.239
-0.043
-2.324
-118.2
-137.0
9.8
60.6
-6.7
-358.9
-)-Troeger’s base
-60.688
-72.442
4
π
1
the absorption region are more likely to be reliable. Therefore,
the calculation of OR at several wavelengths that approach the
absorption region could afford a configurational assignment that
is safer than that obtained through a prediction at a single
frequency. Usually, in reporting OR calculations, little attention
is devoted to the diagonal component of the G′ (see refs 3a
and 7d for a few exceptions), and almost exclusively [R]D results
are given. At the sodium D line, quite frequently, two of the G′
diagonal components almost cancel the third one, and as a
consequence, error propagation may produce a serious effect
ending with a quite large relative error on â. To illustrate this
point we compare in Table 1 the HF and DFT/B3LYP diagonal
components and the average value of the ω G′ tensor,
computed at 589.3 nm, adopting the fairly good Sadlej polar-
izability consistent basis set,15 for some test molecules. Obvi-
ously, neither of the two methods provides exact solutions, but
some indications about the effect of changing the electron
distribution can be clearly seen.
As a rule, for all cases, two diagonal tensor components are
positive (or negative), almost cancelling the third one, giving a
relatively small â. Going from HF to DFT/B3LYP, the following
can be observed: (i) a sign change of â, despite a substantial
agreement between tensor components, as in the simple case of
H2O2; (ii) a good coincidence of the results, as in propylene
oxide; and (iii) a fortuitous agreement of the predicted optical
rotations because a large discrepancy between some of the tensor
components exists, as in the case of (-)-dimethylallene. For
â(0) )
J(〈0| µˆ |j〉‚〈j| mˆ |0〉)
∑
2
3
h
j*0
ωj
Now, assuming an approximation introduced in the past to
compute several kinds of second-order molecular properties,16
which consists of choosing an average excitation energy ∆ω
for all states, the last expression can be rewritten as
4
π 1
â(0) ≈
J(〈0| µˆ |j〉‚〈j| mˆ |0〉)
∑
j*0
2
3
h
∆ω
The sum on the rhs of the above equation is vanishing because
8
,9
of Kuhn’s sum rule for the rotatory strengths. Of course, â is
not totally vanishing because the approximation of the average
excitation energy is a quite crude one. However, this justifies
the smallness of â, especially for the static limit, that is, far
from resonance. Hence, one way to obtain more reliable â
values, that is, smaller relative errors, would be to get far away
from the static limit and, even better, making ω approach a
transition frequency, for example the first one. Obviously, the
smaller the denominator in the equation giving â is, the larger
the absolute value of all three diagonal components of G′ will
be; simultaneously, the latter will not have the tendency to cancel
out when summed together, and then the magnitude of â will
increase as observed. This suggests computing the OR dispersion
curves, or at least the OR at several wavelengths, to appreciate
the trend of the results, which is opposite for two enantiomers.
To this end, the adopted method of calculation should provide,
at least, the correct sign of the rotational strength associated
with the first Cotton effect. This is a less difficult task than the
OR calculation at a single wavelength because even the HF
method usually gives the correct answer. However, it should
be mentioned that the HF transition frequencies are very often
-
1
(
+)-camphor, the ratio between HF and DFT/B3LYP tensor
components is as large as 1.7, whereas the ratio between the
OR predictions is much larger (>6). The case of (-)-Troeger’s
base is quite instructive: the agreement between HF and DFT/
B3LYP tensor components is within 22% on average, whereas
the DFT/B3LYP optical rotation is more than 50 times larger
than the HF value.
1
1,17
largely shifted toward high energy
and, as a consequence,
the HF OR dispersion curves are blue shifted by an amount
that can be as large as 100 nm even when adopting large basis
sets. However, the approach discussed in the following is general
and can be employed at any level of theory; in particular, small
basis set DFT/B3LYP OR dispersion curves could be very useful
in the case of large organic molecules. So, in the following
discussion, we shall use this approach to carry out ab initio
calculations of OR at single frequencies and of ORD curves
for some selected molecules.
Taking into account the previous theoretical works on OR
determination,3 there is nothing new about the comparison
between HF and DFT/B3LYP results; here, we would remark
that to obtain accurate OR predictions time-dependent correlated
methods and high-quality (at least aug-cc-pVDZ) basis sets are
required mainly because of the smallness of â, due to the near
cancellation of the G′ diagonal components. This fact is so
general that it deserves to be justified in some way. From the
definitions of â and G′ one has
,4,7
Synthesis, Measurements, and Calculations. The test
molecules chosen (Table 2) are all rigid systems (to avoid
4
π
1
â(ω) )
J(〈0| µˆ |j〉‚〈j| mˆ |0〉)
(1)
∑
2
2
(15) (a) Sadlej, A. J. Collect. Czech. Chem. Commun. 1988, 53, 1995. (b) Sadlej,
A. J. Theor. Chim. Acta 1991, 79, 123.
3h
j*0
ω - ω
j
(
16) (a) Karplus, K.; Das, T. P. J. Chem. Phys. 1961, 34, 1683. (b) Pople, J. A.
J. Chem. Phys. 1962, 37, 53.
and within the static limit (ω ) 0)
(17) Diedrich, C.; Grimme, S. J. Phys. Chem. A 2003, 107, 2524.
12970 J. AM. CHEM. SOC.
9
VOL. 126, NO. 40, 2004

Ab Initio Calculation of ORD
A R T I C L E S
Table 2.
Figure 1. Experimental (s) and predicted (b, HF/6-31G*; 2, B3LYP/6-
31G*) ORD curves for 1a. The experimental ORD curve has been measured
in hexane.
synthesis of (-)-3b has been carried out by Wittig olefination
1
8
starting from commercial (+)-3a. For the synthesis of (-)-
b and (-)-2b, we used the same procedure starting, respectively,
1
from commercial (-)-1a and (-)-2a. All the [R]D values
reported in Table 2 have been measured in hexane solution at
c ≈ 1 g/100 mL, which is the concentration mostly used by
experimental organic chemists (in the case of (-)-5 only the
concentration is 0.6 g/100 mL): the solvent choice has been
made taking into account that hexane is the best solvent3c to
compare experimental values and the calculated results for the
isolated molecule. For the same reason, all of the ORD spectra
have also been measured in hexane solution. Our analysis starts
considering verbenone 1a and corresponding diene 1b.
It is noteworthy that to the best of our knowledge 1b is
reported for the first time in an optically active form. We have
chosen these two molecules on the basis of the reasoning that
they are both rigid, have the same chiral skeleton, and differ
only in the nature of chromophoric system (R,â-unsaturated
ketone in 1a and a conjugated diene in 1b). In addition, for
both of these compounds, the optical rotatory power at 589.3
nm is determined in sign and order of magnitude by low-lying
1
2
Cotton effects (i.e., they belong to case a)), so for them, an
HF/small basis set treatment should afford the correct value of
OR. Table 2 clearly shows that the optical rotatory power is
correctly calculated in sign and order of magnitude also in the
case of 1b, which possesses a low OR value. The calculated
ORD curves (HF and DFT/B3LYP, small basis set, together
with the experimental ones in hexane) of 1a and 1b are reported
in Figures 1 and 2, respectively. The experimental ORD curve
of 1a presents a clearly negative Cotton effect centered at ∼350
nm, whereas for 1b, only a negative plain curve is measured
down to 250 nm. It is quite easy to note that the experimental
trend is correctly reproduced by the calculations. However, an
important aspect must be pointed out: the HF result gives an
ORD curve that is substantially blue shifted with respect to the
experimental one, whereas the DFT/B3LYP result is much
better. This it is quite evident for 1a. The behavior of the HF
curve is in agreement with the intrinsic features of this
uncorrelated method, which usually provides electronic excita-
difficulties coming from the conformational flexibility) having
known absolute configuration. Furthermore, it is interesting to
note that compounds 1a-4a are ketones, and 1b-4b are the
corresponding olefins. So, we have pair of molecules (a and b)
where different chromophores are inserted in the same chiral
backbone. In addition, in the case of non-comercially available
molecules (such as 1b, 2b, and 3b), they have been obtained
from the corresponding ketone (1a, 2a, and 3a). The
17
tion energies at shorter wavelengths. Moreover, it should be
pointed out that both time-dependent HF and DFT methods
adopted here are not suitable for making predictions in cor-
(18) Money, T.; Palme, M. H. Tetrahedron: Asymmetry 1993, 4, 2363.
J. AM. CHEM. SOC.
9
VOL. 126, NO. 40, 2004 12971

A R T I C L E S
Giorgio et al.
Figure 2. Experimental (s) and predicted (b, HF/6-31G*; 2, B3LYP/6-
Figure 5. Experimental (s) and predicted (b, HF/6-31G*; 2, B3LYP/6-
31G*) ORD curves for 1b. The experimental ORD curve has been measured
31G*) ORD curves for 3a. The experimental ORD curve has been measured
in hexane.
in hexane.
Figure 3. Experimental (s) and predicted (b, HF/6-31G*; 2, B3LYP/6-
31G*) ORD curves for 2a. The experimental ORD curve has been measured
Figure 6. Experimental (s) and predicted (b, HF/6-31G*; 2, B3LYP/6-
31G*) ORD curves for 3b. The experimental ORD curve has been measured
in hexane.
in hexane.
basis set level, both using the HF and DFT/B3LYP methods.
Again, the Cotton effect of 2a at 300 nm is very well reproduced
by DFT/B3LYP calculation, whereas the HF method, even if it
gives the correct negative Cotton effect, provides a considerable
blue shift (Figure 3). It is also noteworthy that the experimental
ORD curve of 2b is very well reproduced both at the DFT/
B3LYP and HF level (Figure 4).
Also in the case of 3b, everything is working very well: OR
at 589 nm (Table 2), HF/ORD and DFT/ORD as reported in
Figure 6. This result is not surprising, taking into account that
3
b is a case a) molecule as well. The case of 3a is really
interesting; in fact, as we discussed in the Introduction, only a
DFT approach (even with a small basis set) provides3c the
correct OR value at 589 nm: these results are collected in the
Table 2. Of course, the ORD curve calculated within a DFT
scheme (Figure 5) reproduces very well the experimental data.
However, even the HF/small basis set ORD curve gives a trend
very similar to the experimental one, the only difference being
that at wavelengths longer than 409 nm the predicted and the
experimental curve are opposite in sign, in agreement with the
discussion about the smallness of â far from the resonance,
previously given. This means that the calculation of the optical
rotation in a range of wavelengths, that is, a part of an ORD
curve, when compared to the experimental one, constitutes a
safer tool for configurational assignment. At this stage some
comments are needed: first of all, a calculation using a small
basis set guarantees a fast answer, so it is certainly more
convenient to repeat such a calculation several times to obtain
Figure 4. Experimental (s) and predicted (b, HF/6-31G*; 2, B3LYP/6-
31G*) ORD curves for 2b. The experimental ORD curve has been measured
in hexane.
respondence to an electronic resonance; in fact, the calculation
of the ORD in the resonant frequency region has been made
7f
possible by Ruud et al. only this year. Therefore, we limit our
discussion to only the portion of ORD curve that approaches
the resonance.
The same comments, about rotatory power and ORD calcula-
tions, can be made in the case of 2a and 2b, that is, a saturated
ketone and a simple olefin, which again, belong to case a but
have much simpler chromophores and lower OR values than
1
a. The values of OR are collected in Table 2, and the ORD
curves are reported in Figures 3 and 4, respectively. The data
in Table 2 clearly show that the [R]D values of 2a and 2b are
correctly reproduced (sign and order of magnitude) at a small
12972 J. AM. CHEM. SOC.
9
VOL. 126, NO. 40, 2004

Ab Initio Calculation of ORD
A R T I C L E S
at least a short ORD curve rather than to do a prediction at a
single frequency but using an extended basis set. In fact,
considering that the computational complexity of the available
methods is proportional at least to the 4th power of the basis
set size, it is easy to see that sixteen 6-31G* basis set
calculations take the same amount of time as a single aug-cc-
pVDZ basis set calculation, with the former basis set being
roughly half the size of the latter. In addition, the computer
time is not the only parameter to take into account; that is, the
memory requirement could be so high, even for a medium-size
organic molecule adopting a large basis set, to make the
calculation not feasible at all, at least on a desktop PC. Second,
the ORD HF/small basis set calculation on (+)-camphor affords
the right answer because we penetrate in the n-π* Cotton effect,
and clearly in this range, the OR values are determined by this
CD band only; in other words even (+)-camphor becomes a
true case a) molecule. This could suggest the real trick to do
reliable OR calculations: using wavelengths that are sufficiently
near to a Cotton effect so that the sum in the Rosenfeld eq 1 is
dominated by a single term and the problem of the smallness
of â is overcome, see the discussion about the smallness of the
â parameter given previously. This is particularly true in this
case, where we are dealing with a valence-shell Cotton effect
Figure 7. Experimental (s) and predicted (b, HF/6-31G*; 2, B3LYP/6-
31G*) ORD curves for 4a. The experimental ORD curve has been measured
in hexane.
(
6
because of an n-π* transition), and thus, even the use of a
-31G* basis set is sufficient (the calculated rotational strength
for the n-π* transition of 3a at HF/6-31G* level is +1.8 ×
-40
1
×
0
(erg esu cm/Gauss) versus an experimental value of +2.4
10- ). Third, now we can understand why the OR calculation
40
at 589.3 nm is wrong with the HF/6-31G* approach: the use
of this basis set, although it is sufficient to give the right sign
of the 290 nm CD band, places this Cotton effect at about 250
nm; that is, it is blue shifted, with respect to the experimental
CD band, about 40 nm. This means that its positive contribution
to the [R]D value can be overcome by a negative contribution
coming from shorter wavelength CD bands. By contrast, the
DFT method places this Cotton effect in the right position on
the wavelength scale, and therefore now its contribution to [R]D
cannot be overcome by negative contributions coming from
higher energy CD bands. In addition, this basis set may introduce
tremendous errors in reproducing the higher-energy Cotton
effects, so the sum in the Rosenfeld eq 1 provides a wrong result.
In other words, at 589.3 nm, â is a very small number, its
magnitude is smaller than the computation uncertainty, and its
sign could be wrong, which actually happens. On the contrary,
near 290 nm, the G′ tensor is determined predominantly by a
strong contribution from the n-π* transition, and because the
â value is a large number, its magnitude is larger than its
computation uncertainty, and the sign is safely predicted. The
cases of 4a, (+)-nopinone, a saturated ketone, and corresponding
alkene 4b, (-)-â-pinene, are even more illustrative. In the case
of 4a, the experimental rotation is a small number (for this
reason this molecule constitutes a difficult problem), and the
numbers produced by HF/6-31G* calculations are even smaller
Figure 8. Experimental (s) and predicted (b, HF/6-31G*; 2, B3LYP/6-
31G*; (, B3LYP/Sadlej) ORD curves for 4b. The experimental ORD curve
has been measured in hexane.
considering that London and non-London calculations afford
predictions that are opposite in sign. Most importantly, even
the DFT/B3LYP/6-311++G(2d,2p) calculation affords the
wrong answer (at λ ) 633 nm, the calculated value is +21 and
the experimental value is -17); note that here, we are comparing
the theoretical prediction with an experimental value measured
in gas phase, hence no solvent effects are present. We have
calculated the ORD curve of 4b at DFT/B3LYP/Sadlej level
obtaining a curve which reproduces very well the experimental
ORD trend (Figure 8). We used the DFT/B3LYP method and
11
the Sadlej basis set because we reported that this approach
works very well for the OR calculation. It is interesting to note
that at 589.3 nm this method also gives the wrong result, and
only at wavelengths shorter than 400 nm, the sign of OR is
correctly reproduced, pointing out again that the calculation of
an ORD spectrum affords the right configurational answer. It
is even more interesting and important to note that even an HF
or DFT/B3LYP calculation using a small basis set (6-31G*)
reproduces (Figure 8) the experimental shape (at least below
350 nm), confirming the power and reliability of ORD calcula-
tion for configurational assignments. It is important, at this stage,
to note that 4b (and 5, vide infra) in contrast to the previous
1a-4a does not belong to the above-defined case a) molecules.12
In fact, the CD spectrum of (-)-â-pinene 4b in the gas phase
(Table 2), so a clear answer cannot be given. The DFT
calculations are certainly better. However, the prediction (both
at the HF and DFT level) of the ORD curves eliminates any
doubt (Figure 7). For 4b, (Table 2) we have that the predicted
B3LYP values both at the small (6-31G*) and large (Sadlej)
basis set level afforded the wrong sign. The correct (in sign)
result obtained at the HF/6-31G* level is certainly fortuitous,
1
9
shows, between 220 and 170 nm, a sequence of positive/
negative bands, whereas in the hexane solution, only the positive
Cotton effect is measurable. As a consequence, the contribution
of this Cotton effect to the OR at 589 nm is +170 versus an
J. AM. CHEM. SOC.
9
VOL. 126, NO. 40, 2004 12973

A R T I C L E S
Giorgio et al.
in the near-UV region) compound, by means of ab initio
calculation of the optical rotatory power. The first step is
measuring the experimental ORD curve in a suitable range of
wavelengths and the second step is calculating the ORD curve
in the same range of frequencies by means of the DFT/B3LYP/
6
-31G* method. This approach will couple the better accuracy
of the DFT method (with respect to the HF one) in terms of
wavelength position of the individual Cotton effects, allowing
a simple comparison with the experimental data, with the speed
of the calculation, even if it is repeated several times.
Conclusions
The main result of this investigation is that the calculation
of the OR repeated at different wavelengths (prediction of the
ORD curve) constitutes a reliable method for the assignment
of the molecular absolute configuration. One could argue that
this approach strongly depends on the measurement of the ORD
curve, which nowadays is not so easy to obtain because modern
chiroptical spectroscopy is mainly based on circular dichroism.
Figure 9. Experimental (s) and predicted b, HF/6-31G*; 2, B3LYP/6-
31G*) ORD curves for 5. The experimental ORD curve has been measured
in hexane.
experimental value of -15, indicating clearly that higher Cotton
effects determine the sign of [R]D. Thus, the case of this
molecule is deeply different from that of 2b and 3b, for which
a very good agreement between the experimental and calculated
values is found. In other words, although 2b and 3b belong to
class a) (see above) and for them even a simplified treatment
gives the correct answer, it could be possible to predict a lot of
difficulties for the OR calculation at 589 nm in the case of (-)-
â-pinene (4b) because [R]D is determined (sign and magnitude)
by shorter-wavelength CD bands.
20
However, it must be noticed that modern polarimeters allow
us to make OR measurements at several wavelengths and thus,
the availability of OR values at some different wavelengths
could be a simple alternative to the measurement of a continuous
ORD spectrum. Moving at shorter wavelengths, we go nearer
to absorption regions, which may guarantee that we are
approaching a single Cotton effect with the consequent reduction
of the contributions due to the other Cotton effects in the
Rosenfeld sum; this leads to a larger â, and, hence, a most
reliable predicted OR value. Another criticism could be that
one could attempt the AC assignment directly predicting the
CD spectrum itself. A CD calculation requires the introduction
of a shape factor to compare the overall shape of the experi-
mental spectrum with the predicted one; often, this comparison
in not so easy because a large number of near-in-frequency and
opposite-in-sign transitions may derive. Therefore, such an
approach can be safely used only when the CD spectrum
presents a well-defined, isolated Cotton effect, so the problem
of the shape of the spectrum can be avoided; a representative
case is that of the saturated ketones, which have a low-energy
An additional interesting comment can be made concerning
(
-)-Troeger’s base 5. For this compound, the OR computation
by the DFT/B3LYP method gives satisfactory results (see Table
), whereas the HF calculation (even when using quite large
2
basis sets, such as aug-cc-pVDZ) always produces the wrong
sign. For this reason, Stephens et al. pointed out3 that we can
trust only in OR calculations carried out by the DFT/B3LYP
method. In a previous paper, we suggested12 that because (-)-
Troeger base presents a very intense, positive, low-lying CD
band that provides a strong (+200), positive contribution to the
optical rotation at 589.3 nm, it is absolutely required to have a
correct description of the higher-energy Cotton effects, so a
method that is more accurate than HF is needed. In Figure 9,
we report the experimental ORD curve (hexane) together with
those calculated by the DFT/B3LYP/6-31G* method: clearly
a satisfactory reproduction of the experimental data is provided,
at least down to 350 nm. At shorter wavelengths, the sequence
c
3i
2
90-nm band well separated from other Cotton effects. In the
present approach, we have simply to compare a few numbers
experimental and predicted OR, at different wavelengths),
(
which, in addition, at least in the cases similar to those described
in this paper, are obtained in a reasonable time of calculation.
One could also argue that in this paper we have treated only
rigid molecules. However, in principle, the same treatment can
be extended also to flexible molecules. Clearly, an accurate
conformational analysis is an absolute prerequisite and this step
can cost some more computational effort, even if no fundamental
reasons prevent its application to these systems. To conclude,
by coupling the use of an inexpensive instrument (a polarimeter
working at a few wavelengths) with the use of an ab initio DFT
calculation package, the experimental organic chemist will
arrive, quickly and reliably, at the assignment of the molecular
absolute configuration.
(-, +, -) of the Cotton effects is correctly reproduced, even if
a blue shift of about 50 nm (of the predicted curve with respect
to the experimental one) is observed. It is noteworthy that a
simple HF/6-31G* computation reproduces the experimental
ORD trend only below 300 nm. Here, the blue shift (of the
calculated vs experimental curve) is more than 80 nm. The above
results clearly show the limitations of ORD protocol at Hartree-
Fock level, and it seems that a DFT/B3LYP calculation at
several wavelengths could be the method of choice to reach a
reliable AC assignment, also within a small basis set scheme,
which is necessary to deal with medium-large size molecules,
as the biological active ones, that is, the largest part of molecules
having practical interest. We are now able to propose a simple
and reliable protocol to assign the molecular absolute config-
uration of an organic unsaturated and/or aromatic (i.e., absorbing
(20) It is to be noticed that nowadays polarimeters working at a few discrete
wavelengths (880, 633, 577, 546, 435, 405, 365, 334, 325, 313, 302, 296,
2
80, and 253 nm) are commercially available at a contained (15 000 euro
ca.) price. Clearly, with such instrumentation, the problem posed by the
(19) Drake, A. F.; Mason, S. F. Tetrahedron 1977, 33, 937.
(-)-â-pinene molecule can be easily solved.
1
2974 J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004
9

Ab Initio Calculation of ORD
A R T I C L E S
5
7300-U (poly(dimethylsiloxane) (PDMS) phase) column. THF was
Computational Methods
freshly distilled prior its use on sodium benzophenone ketyl and was
stored under nitrogen atmosphere. n-Butyllithium 2.5 M in hexane, (-)-
fenchone (2a), (+)-camphor (3a), (+)-nopinone (4a), (-)-â-pinene (4b)
All calculations have been carried out on a simple PC endowed with
a single PentiumIV 2.2-GHz processor. All geometries have been fully
optimized at DFT/B3LYP/6-31G* level using the Gaussian 98 package.
All of the geometries are real minima; no imaginary frequencies were
found.
(Aldrich products) and (-)-verbenone (1a) and (-)-Troeger’s base (5)
(Fluka products) were used as purchased. Methyltriphenylphosphonium
bromide (Aldrich) was dried under vacuum for 3 h before its use.
All the OR calculations have been carried out by means of time-
21
(-)-4-Methylverbenene (1b). To a stirred solution of methyltriph-
enylphosphonium bromide (11.43 g, 32.02 mmol) in 80 mL of
anhydrous THF was added 14 mL of n-butyllithium (2.5 M in hexane)
dropwise under a nitrogen atmosphere. The solution was warmed at
dependent HF and DFT methods as available within Gaussian,
2
2
23
Dalton, and Turbomole packages. In particular, (i) all the OR
calculations at HF/6-31G* level have been carried out using London
orbitals (which ensure the origin independency of the results) as
implemented in the Dalton 1.2.1 package; (ii) all the OR calculations
at DFT/B3LYP/6-31G* level have been carried out using London
orbitals as implemented in the Gaussian 03 package; (iii) the OR
calculation at the DFT/B3LYP/Sadlej level has been carried out with
Turbomole 5.6 package. A reviewer noticed that the B3LYP functional
implemented in Turbomole is different from that of Gaussian03.
However, the differences between the functionals for properties in
general are small, although there are large differences in absolute
energies. We remark that Turbomole has been used in this work only
in the case of the DFT/B3LYP/Sadlej calculation of 4b, because
Gaussian03 gave convergence problems.
5
0 °C and stirred for 2h, obtaining a red coloration. At this point, to
the solution was added (-)-verbenone (1a) (3.02 g, 20.1 mmol)
dropwise in 20 mL of anhydrous THF. The obtained solution was stirred
under reflux for 24 h, cooled at room temperature, and then about half
of the volume of the solvent was removed under reduced pressure, and
5
0 mL of petroleum ether was added. The organic layer was washed
successively by 2 × 30 mL of water and 2 × 30 mL of brine. The
organic layer was then dried over anhydrous Na SO , filtered, and
2
4
evaporated at reduced pressure. Chromatography on silica gel (eluent:
petroleum ether) of the crude residue gave, after evaporation of solvent
at reduced pressure, the crude product. After fractional distillation, we
recovered 1.85 g (62% yield) of pure (-)-1b as a colorless liquid. [R]
)
20
D
Experimental Section
1
- 40 (c ) 1; hexane). H NMR (500 MHz, CDCl ): δ 0.84 (s, 3H);
3
1
13
General Procedures. H NMR and C NMR spectra were recorded
1
6
.37 (s, 3H); 1.47(d, 1H, J ) 9.0 Hz); 1.81 (s, 3H); 2.13 (t, 1H, J )
1
13
3
in CDCl on Varian-Inova 500 ( H 500-MHz and C 125-MHz) or
.0 Hz); 2.53 (m, 1H); 2.61(t, 1H, J ) 5.5 Hz); 4.59 (s, 2H); 5.80
1
Bruker Aspect 300 ( H 300-MHz) spectrometers. UV and CD spectra
were recorded in hexane solution on a JASCO J-600 spectropolarimeter.
Optical rotations were measured with a JASCO DIP-370 digital
polarimeter. ORD curves were recorded in hexane solution on a JASCO
J-810 spectropolarimeter equipped for ORD measurements. Column
chromatography was carried out with silica gel Merck 60 (80-230
mesh). Gas chromatographic analyses were carried out on a GC/MS
Hewlett-Packard 5080 series II, MS detector HP 5971, with a Supelco
13
(s,1H). C NMR (125 MHz, CDCl
3
): δ 22.0, 23.1, 26.4, 35.8, 43.8,
+
4
1
8.4, 51.5, 104.6, 120.8, 148.7, 150.4 MS (EI): m/z 148 (M , 24),
33 (42), 106 (36), 105 (100), 103 (7), 79 (13), 77 (15). Anal. Calcd
for C11
H16: C, 89.12; H, 10.88. Found: C, 89.70; H, 10.30.
(-)-Methylenefenchone (2b). To a stirred solution of methyltriph-
enylphosphonium bromide (15.98 g, 44.73 mmol) in 80 mL of
anhydrous THF was added 19 mL of n-butyllithium (2.5 M in hexane)
dropwise under nitrogen atmosphere. The solution was warmed at 50
°
C and stirred for 2 h, obtaining a red coloration. At this point, to the
(
21) Helgaker, T.; Jensen, H. J. Aa.; Joergensen, P.; Olsen, J.; Ruud, K.; Aagren,
H.; Auer, A. A.; Bak, K. L.; Bakken, V.; Christiansen, O.; Coriani, S.;
Dahle, P.; Dalskov, E. K.; Enevoldsen, T.; Fernandez, B.; Haettig, C.; Hald,
K.; Halkier, A.; Heiberg, H.; Hettema, H.; Jonsson, D.; Kirpekar, S.;
Kobayashi, R.; Koch, H.; Mikkelsen, K. V.; Norman, P.; Packer, M. J.;
Pedersen, T. B.; Ruden, T. A.; Sanchez, A.; Saue, T.; Sauer, S. P. A.;
Schimmelpfennig, B.; Sylvester-Hvid, K. O.; Taylor, P. R.; Vahtras, O.
Dalton, release 1.2; 2001.
solution was added (-)-fenchone (2a) (4.27 g, 28.07 mmol) dropwise
in 20 mL of anhydrous THF. The obtained solution was stirred under
reflux for 30 h, cooled at room temperature, and then about half of the
volume of the solvent was removed under reduced pressure, and 50
mL of petroleum ether was added. The organic layer was washed
successively by 2 × 30 mL of water and 2 × 30 mL of brine. The
(
22) (a) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.;
Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A.
D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi,
M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.;
Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.;
Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.;
Cioslowski, J.; Ortiz, J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz,
P.; Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-
Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe,
M.; Gill, P. M. W.; Johnson, B. G.; Chen, W.; Wong, M. W.; Andres, J.
L.; Head-Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian 98, Gaussian,
Inc.: Pittsburgh, PA, 1998. (b)Frisch, M. J.; Trucks, G. W.; Schlegel, H.
B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A.,
Jr.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.;
Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.;
Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda,
R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai,
H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Adamo,
C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A.
J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.;
Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich,
S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A.
D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A.
G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.;
Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham,
M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.;
Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian
2 4
organic layer was then dried over anhydrous Na SO , filtered, and
evaporated under reduced pressure. Chromatography on silica gel
(eluent: petroleum ether) of the residue gave, after evaporation of
solvent at reduced pressure, the crude product. After fractional
distillation we recovered 2.78 g (66% yield) of pure (-)-2b as a
colorless liquid. [R]
MHz, CDCl ): δ 1.05 (s, 3H); 1.08 (s, 3H); 1.20 (s, 3H); 1.22 (d, 2H);
1.45 (m, 3H); 1.68 (t, 1H); 1.84 (s, 1H); 4.57 (s, 1H); 4.61 (s,1H). MS
20
1
D
) - 68 (c ) 1.08; hexane). H NMR (300
3
(EI): m/z 150 (M
+
, 28), 135 (10), 121 (14), 107 (100), 91 (20), 79
18: C, 87.93; H, 12.07.
(20), 67 (9), 41 (12). Anal. Calcd for C11
H
Found: C, 87.51; H, 12.49.
(-)-Methylenecamphor (3b). To a stirred solution of methyltriph-
enylphosphonium bromide, (15.95 g, 44.65 mmol) in 80 mL of
anhydrous THF was added 19 mL of n-butyllithium (2.5 M in hexane)
dropwise under nitrogen atmosphere. The solution was warmed at 50
°
C and stirred for 2h, obtaining a red coloration. At this point, to the
solution was added (+)-camphor (3a) (4.25 g, 27.8 mmol) dropwise
in 20 mL of anhydrous THF. The obtained solution was stirred under
reflux for 24 h, cooled at room temperature, and then about half of the
volume of the solvent was removed under reduced pressure, and 50
mL of petroleum ether was added. The organic layer was washed
successively by 2 × 30 mL of water and 2 × 30 mL of brine. The
0
3, Gaussian, Inc.: Pittsburgh, PA, 2003.
(
23) Ahlrichs, R.; Bar, M.; Baron, H.-P.; Bauernschmitt, R.; Bocker, S.; Ehrig,
M.; Eichkorn, K.; Elliott, S.; Furche, F.; Haase, F.; Haser, M.; Horn, H.;
Hattig, C.; Huber, C.; Huniar, U.; Kattannek, M.; Kohn, A.; Kolmes, C.;
Kollwitz, M.; May, K.; Ochsenfeld, C.; O¨ hm, H.; Schafer, A.; Schneider,
U.; Treutler, O.; Arnim, M. v.; Weigend, F.; Weis, P.; Weiss, H.
TURBOMOLE, version 5.6; Universitat Karlsruhe: Karlsruhe, Germany,
2 4
organic layer was then dried over anhydrous Na SO , filtered, and
evaporated at atmospheric pressure (warning, the product 3b sublimes
easily). Finally, chromatography on silica gel (eluent: pentane) of the
2
002.
J. AM. CHEM. SOC.
9
VOL. 126, NO. 40, 2004 12975

A R T I C L E S
Giorgio et al.
residue yielded, after evaporation of pentane at atmospheric pressure,
is gratefully acknowledged. We warmly thank Professor Gio-
vanni Natile, Dipartimento Farmaco-Chimico, Universit a` di Bari,
Italy, for allowing us to use the Jasco 810 spectropolarimeter
and record the ORD spectra. Particular thanks are also given to
Professor Philip J. Sthephens, Department of Chemistry,
University of Southern California for helpful discussions and
to Professor Kenneth Ruud, Department of Chemistry, Univer-
sity of Tromsoe, Norway, for a preprint of ref 7f.
2
0
2
.90 g (68% yield) of 3b as a colorless solid. [R]
D
) - 36 (c ) 0.96;
1
hexane). H NMR (300 MHz, CDCl
3
): δ 0.78 (s, 3H); 0.91 (s, 3H);
0
4
.94 (s, 3H); 1.25 (m, 2H); 1.65 (t, 1H); 1.78 (m, 2H); 1.87 (d, 1H);
.63 (s, 1H); 4.67 (s,1H). MS (EI): m/z 150 (M , 28), 135 (59), 121
+
(30), 107 (100), 93 (61), 79 (68), 41 (21). Anal. Calcd for C11H18: C,
8
7.93; H, 12.07. Found: C, 88.10; H, 11.90.
Acknowledgment. Financial support from Universit a` della
Basilicata, Universit a` di Salerno, and MIUR-COFIN (Roma)
JA046875L
12976 J. AM. CHEM. SOC.
9
VOL. 126, NO. 40, 2004