Systematic investigation of CD spectra of aryl benzyl sulfoxides interpreted by means of TDDFT calculations

Article abstract of DOI:10.1021/jo902363c

(Chemical Equation Presented) The CD spectra of 13 crystalline aryl benzyl sulfoxides 1a-m with various substituents on the two aromatic rings were recorded in solution and in the solid state. Solution CD spectra were very homogeneous along the series, consisting in most cases of a couplet-like feature in the 200-300 nm region. The red-shifted component of the couplet, corresponding to the sulfoxide-centered n-π* transition, is always positive for (R) absolute configuration in accordance with Mislow?s rule. The presence of a strong electron-withdrawing substituent on the phenyl ring (nitro or ester group) alters the shape of the CD spectrum. CD calculations using the TDDFT method were run for eight representative compounds using DFT-optimized geometries. In all cases, calculated spectra were in very good agreement with experimental ones and allowed for rationalization of the diverse spectral behaviors. It is demonstrated thatTDDFT//DFTcalculations represent a reliable option for assigning the absolute configuration of this class of compounds whenever crystals suitable for X-ray are not available. Solid-state CD spectra recorded with the KCl pellet technique were in some cases in agreement with those in solution. However, in other cases new and strongCD signals appeared which were interpreted as being due to intermolecular couplings in the crystals.

Full text of DOI:10.1021/jo902363c

pubs.acs.org/joc
Systematic Investigation of CD Spectra of Aryl Benzyl Sulfoxides
Interpreted by Means of TDDFT Calculations
Gennaro Pescitelli,† Sebastiano Di Pietro,† Cosimo Cardellicchio,‡
Maria Annunziata M. Capozzi,§ and Lorenzo Di Bari*,†
†
ꢀ
Dipartimento di Chimica e Chimica Industriale, Universita di Pisa, via Risorgimento 35, 56126 Pisa, Italy,
‡CNR-ICCOM, Dipartimento di Chimica, via Orabona 4, 70126 Bari, Italy, and Dipartimento di Scienze
ꢀ
Agroambientali, Chimica e Difesa Vegetale, Universita di Foggia, Via Napoli 25, 71100 Foggia, Italy
§
ldb@dcci.unipi.it
Received November 5, 2009
The CD spectra of 13 crystalline aryl benzyl sulfoxides 1a-m with various substituents on the two
aromatic rings were recorded in solution and in the solid state. Solution CD spectra were very
homogeneous along the series, consisting in most cases of a couplet-like feature in the 200-300 nm
region. The red-shifted component of the couplet, corresponding to the sulfoxide-centered n-π*
transition, is always positive for (R) absolute configuration in accordance with Mislow’s rule. The
presence of a strong electron-withdrawing substituent on the phenyl ring (nitro or ester group) alters
the shape of the CD spectrum. CD calculations using the TDDFT method were run for eight
representative compounds using DFT-optimized geometries. In all cases, calculated spectra were in
very good agreement with experimental ones and allowed for rationalization of the diverse spectral
behaviors. It is demonstrated thatTDDFT//DFT calculationsrepresent a reliable option forassigning
the absolute configuration of this class of compounds whenever crystals suitable for X-ray are not
available. Solid-state CD spectra recorded with the KCl pellet technique were in some cases in
agreement with those in solution. However, in other cases new and strong CD signals appeared which
were interpreted as being due to intermolecular couplings in the crystals.
Introduction
asymmetric synthesis,3-5 because they are easily prepared
and induce a high level of enantioselection and the sulfinyl
moiety can be easily removed at the end of the synthetic
strategy.
Several methods have been proposed to establish the
absolute configuration of sulfoxides. Crystals suitable for a
satisfactory diffractometric X-ray analysis6 were not ob-
tained frequently, as result of the lack of the particular weak
interactions7 required to assembly the crystal building. On
the other hand, several configurational assignments were
A recent interest arose in the biological activity of chiral
sulfoxides.1 For example, omeprazole, a powerful proton
pump inhibitor, is nowadays one of the best-selling drugs.
Some alkyl sulfoxides inhibit alcohol dehydrogenases.2
Many chiral sulfoxides play important roles in metabolic
pathways as the first oxidation products of the correspond-
ing sulfides, including methionine and biotin.1 Further-
more, enantiopure sulfoxides have been widely used in
(1) Bentley, R. Chem. Soc. Rev. 2005, 34, 609–624.
(2) Cho, H.; Plapp, B. V. Biochemistry 1998, 37, 4482–4489.
(3) Fernandez, I.; Khiar, N. Chem. Rev. 2003, 103, 3651–3705.
(5) Pellissier, H. Tetrahedron 2006, 62, 5559–5601.
(6) Mikolajczyk, M.; Drabowicz, J. Top. Stereochem. 1982, 13, 333–468.
(7) Naso, F.; Cardellicchio, C.; Capozzi, M. A. M.; Capitelli, F.; Bertolasi,
V. New J. Chem. 2006, 30, 1782–1789.
~
(4) Carreno, M. C. Chem. Rev. 1995, 95, 1717–1760.
DOI: 10.1021/jo902363c Published on Web 01/25/2010
r
J. Org. Chem. 2010, 75, 1143–1154 1143
2010 American Chemical Society

Pescitelli et al.
JOCArticle
based on chiro-optical spectroscopies since the earlier work
of Mislow and co-workers,8 who demonstrated that electro-
nic circular dichroism (CD) and optical rotatory dispersion
(ORD) are convenient tools to assign the absolute configura-
tion of these molecules. The sulfinyl group is a moderately
strong chromophore, showing an absorption band around
215 nm with ε ≈ 2500 M-1 cm-1 in alcoholic solvents.9 This
so-called primary band8b is associated with a transition from
oxygen lonepairnO to π*-type SdO orbital.10 ThetwonO and
π* orbitals are heavily mixed with bonding and nonbonding
σ-type S-C orbitals, respectively. The tendency of sulfoxide-
centered orbitals to mix with nearby orbitals of proper
symmetry and energy is also evident in ethylene episulfoxide
(CH2)2SO described by Gottarelli et al.11 For this compound,
the virtual orbital involved in the first transition observed
around 210 nm is rather of a σ*S-O type, and such a transi-
tion acquires a σ-σ* character. A large orbital mixing is
also apparent for alkyl aryl sulfoxides. They show a single
band around 250 nm in cyclohexane that decomposes in
two bands in alcoholic solvents and acetonitrile, the first
SCHEME 1. Sulfoxides 1a-m
1
one (at 260-290 nm) being assignable to the phenyl Lb
These results were tentatively extended to phenyl derivatives
as a qualitative means of interpretation, because the coupled-
oscillator model requires that the two coupling chromo-
phores undergo negligible conjugation or orbital overlap.
This is not strictly the case for alkyl phenyl sulfoxides where a
favorable overlap is expected between the sulfur lone pair
and the adjacent π system, which is thought to dictate the
most stable molecular conformation.15
Rather surprisingly, no other systematic discussion of CD
spectra of sulfoxides has been reported since then in the
literature. In particular, time-dependent density functional
theory (TDDFT) calculations, which currently represent the
most popular approach to quantitative predictions of chir-
optical properties,17 were applied only to a few specific cases
and were found to correctly reproduce experimental CD
spectra or optical rotations and to allow safe configurational
assignments to be established.18-21
Now we report the CD spectra of 13 aryl benzyl sulfoxides
(1a-m, Scheme 1) with various substituents on the two
aromatic rings. Several sulfoxides with one benzyl- and one
aryl-type substituent have been described to show biological
activity.22 The availability of a relatively large number of
enantiopure sulfoxides originated from our previous inves-
tigation, which led to the important result that aryl benzyl
sulfoxides could be easily prepared in a high enantiomeric
purity(upto >98% ee) by a tert-butyl hydroperoxide (TBHP)
enantioselective oxidation of the corresponding sulfides in
the presence of a complex between titanium tetraisoprop-
oxide and commercially available (S,S)- or (R,R)-hydroben-
zoin (Scheme 2).23
transition and the second one (250-260 nm) to the sulfoxide-
centered n-π* transition,8b,12 whose π* orbital is however
largely delocalized on the phenyl ring.
In correspondence with the apparent absorption bands,
chiral nonracemic sulfoxides show intense CD bands or
Cotton effects (CE) that are diagnostic of their absolute
configuration, as first recognized by Mislow.8 In particular,
ORD spectra of alkyl aryl sulfoxides show a CE in the
235-255 nm region whose sign correlates with the config-
uration, a positive CE corresponding to the (R) enantiomer
and a negative CE to the (S). This empirical correlation has
been employed for configurational assignments of several
alkyl aryl sulfoxides,13 but it is not free from pitfalls and
limitations.14 Only recently, the first nonempirical interpre-
tations of CD spectra of this class of compounds have
appeared. In particular, Rosini and co-workers reported
the CD spectra of several alkyl aryl sulfoxides where the aryl
group is phenyl, p-substituted phenyl, 1- and 2-naphthyl, and
9-anthryl.15,16 In the case of naphthyl and anthryl deriva-
tives, CD spectra were well reproduced by coupled-oscillator
calculations with DeVoe method taking into account the
main aromatic transitions and the sulfoxide-centered transi-
tion, this latter described as σ-σ* according to Gottarelli.11
(8) (a) Mislow, K.; Ternay, A. L., Jr.; Melillo, J. T. J. Am. Chem. Soc.
1963, 85, 2329–2330. (b) Mislow, K.; Green, M. M.; Laur, P.; Melillo, J. T.;
Simmons, T.; Ternay, A. L., Jr. J. Am. Chem. Soc. 1965, 87, 1958–1976.
(9) Jaffe, H. H.; Orchin, M. Theory and Applications of Ultraviolet
Spectroscopy; Wiley: New York, 1962.
(10) Frolov, Y. L.; Sinegovskaya, L. M.; Gusarova, N. K.; Efremova, G.
G.; Trofimov, B. A. Russ. Chem. Bull. 1978, 27, 901–904.
(11) (a) Bendazzoli, G. L.; Palmieri, P.; Gottarelli, G.; Moretti, I.; Torre,
G. J. Am. Chem. Soc. 1976, 98, 2659–2660. (b) Moretti, I.; Torre, G.;
Gottarelli, G. Tetrahedron Lett. 1976, 711–714. (c) Bendazzoli, G. L.;
Palmieri, P.; Gottarelli, G.; Moretti, I.; Torre, G. Gazz. Chim. Ital. 1979,
109, 19–25.
This oxidation protocol is particularly efficient, because it
is insensitive to the presence of water, works at room
(17) (a)Crawford, T.D. Theor. Chem. Acc. 2006, 115, 227–245. (b)Diedrich,
C.; Grimme, S. J. Phys. Chem. A 2003, 107, 2524–2539. (c) Rappoport, D.;
Furche, F. Lect. Notes Phys. 2006, 706, 337–354. (d) Autschbach, J. Chirality
2010, in press; DOI: 10.1002/chir.20789.
(12) Mangini, A.; Pallotti, M.; Tiecco, M.; Dondoni, A.; Vivarelli, P. Int.
J. Sulfur Chem. A 1972, 2, 69–78.
(13) Donnoli, M. I.; Superchi, S.; Rosini, C. Minirev. Org. Chem. 2006, 3,
77–92.
(18) Stephens, P. J.; Devlin, F. J.; Cheeseman, J. R.; Frisch, J.; Rosini, C.
(14) Laur, P. H. Steric aspects of sulfur chemistry. In Sulfur in Organic
and Inorganic Chemistry; Senning A., Ed.; Marcel Dekker: New York, 1972;
Vol. 3, pp 91-274.
(15) Rosini, C.; Donnoli, M. I.; Superchi, S. Chem.;Eur. J. 2001, 7, 72–
79.
Org. Lett. 2002, 4, 4595–4598.
(19) Petrovic, A. G.; He, J.; Polavarapu, P. L.; Xiao, L. S.; Armstrong,
D. W. Org. Biomol. Chem. 2005, 3, 1977–1981.
(20) Casarini, D.; Lunazzi, L.; Mancinelli, M.; Mazzanti, A.; Scafato, P.
Chirality 2009, 21, 16–23.
(16) Donnoli, M. I.; Giorgio, E.; Superchi, S.; Rosini, C. Org. Biomol.
Chem. 2003, 1, 3444–3449.
(21) Wang, Y.; Raabe, G.; Repges, C.; Fleischhauer, J. Int. J. Quantum
Chem. 2003, 93, 265–270.
1144 J. Org. Chem. Vol. 75, No. 4, 2010

Pescitelli et al.
JOCArticle
SCHEME 2. Enantioselective Synthesis of Aryl Benzyl
Sulfoxides
Aryl benzyl sulfoxides 1b,c,f (Figure 1b,c) show CD
spectra similar to alkyl aryl sulfoxides,15 with minimal
differences due to the replacement of an alkyl with the
benzyl group. They consist, for (R) absolute configuration,
of a moderately intense positive band around 260 nm and a
negative one around 230 nm. A smaller positive band may
appear around 300 nm, sometimes as a shoulder on the long-
wavelength side of the stronger close band. Based on the
similarity with alkyl phenyl sulfoxides,15 the CE at 260 may
be assigned to the sulfoxide group primary transition, while
those at 225 and 300 nm to benzene 1La and 1Lb transitions,
respectively. The sulfoxide primary band has positive sign
for the (R)-configuration for all compounds and follows
Mislow’s empirical rule for alkyl aryl sulfoxides.8 With
respect to 4- and 3-methoxyphenyl compounds 1b and 1c,
the 2-methoxy isomer (R)-1d [2OMe/H] shows an additional
positive CD band around 280 nm, which has a correspond-
ing strong absorption band at 281 nm in CH3CN (ε ≈ 5600
M-1 cm-1; see Figure S2, Supporting Information). Pecu-
liar red-shifted spectra are shown by the nitrophenyl sulf-
oxide 1a [4NO2/H] and the ester 1e [2COOMe/H]
(Figure 1a), consisting of a series of bands of alternating
sign in the 200-350 nm region, the first of which (from the
right) is again positive for the (R) enantiomers.
temperature, is performed in n-hexane, and can be easily
scaled up.23 The (R)-sulfoxides were obtained by using the
(S,S)-hydrobenzoin as a ligand of titanium, whereas the
(S)-sulfoxides were obtained in the presence of the (R,R)-
hydrobenzoin. This stereochemical outcome and the high
enantioselectivity of the oxidation were also substantiated by
a theoretical study.23
In the present investigation, we recorded the CD spectra of
1a-m both in solution and in the solid state. Solid-state CD
spectra24 are useful tools for stereochemical investigations of
organic molecules, in particular for absolute configurational
assignments.25 We also report the results of TDDFT calcula-
tions of CD spectra for representative aryl benzyl sulfoxides,
using DFT-computed structures and available X-ray geo-
metries. Aim of the present contribution is to demonstrate
that TDDFT calculations are able to reproduce CD spectra
of aryl benzyl sulfoxides and can be employed as a useful tool
for the prediction and interpretation of CD of these and
related substrates.
The 4-bromophenyl derivatives (R)-1g-m with a substi-
tuted benzyl phenyl ring have CD spectra similar to that of
the parent compound (R)-1f [4Br/H] (Figure 1c-e). They are
consistent with that of (S)-benzyl phenyl sulfoxide described
by Rosini et al. (extrema at 257 nm, Δε = -30.4 M-1 cm-1
,
and 226 nm, þ48.1).15 This demonstrates that both the
p-bromo-substitution at the phenyl ring and the various
substituents on the benzyl phenyl ring affect the general
appearance of the CD spectrum only to a small extent. The
only compound in the series showing some difference is the
nitro derivative 1g [4Br/4NO2], whose positive CD band in
the 240-300 nm region is detectably split (Figure 1e).
Results
1. Experimental CD Spectra in Solution. Aryl benzyl
sulfoxides 1a-m (Scheme 1) are indicated in the follow-
ing text with the notation 1x [R0/R00], where R0 and R00 are
the substituents at the phenyl and the benzyl phenyl ring,
respectively, e.g., 1h [4Br/2NO2] is 40-bromophenyl (2-nitro-
phenyl)methyl sulfoxide. Figure 1 shows the CD spectra of
compounds (R)-1a-m measured in acetonitrile solutions.
The spectra for available (S) enantiomers of 1a and 1b,
shown in Figure S1 (Supporting Information), are the perfect
mirror images of the corresponding (R) antipodes.
2. Calculations of CD Spectra. 2.1. Choice of Calculation
Method: Model Compounds. Simulation of CD spectra of
1a-m in solution started from the generation of input
geometries and the choice of the most suitable functional
and basis set for DFT and TDDFT calculations. A series of
preliminary calculations were performed on methyl phenyl
(2), ethyl phenyl (3), and benzyl phenyl (4) sulfoxides
(Scheme 3). Due to the spectral homogeneity of alkyl aryl
sulfoxides discussed above, the choice of such model com-
pounds was considered appropriate. Conformational
searches with molecular mechanics methods led for each
compound to a single or a limited number of low-energy
minima that were optimized with DFT. B3LYP/6-31G(d)
optimized geometry for 2 was in very good agreement, in
terms of bond lengths and angles, with the structure calcu-
lated by Agranat and co-workers with the larger 6-311G(d,p)
(22) (a) Montgomery, J. I.; Toogood, P. L.; Hutchings, K. M.; Liu, J.;
Narasimhan, L.; Bradena, T.; Dermyer, M. R.; Kulynych, A. D.; Smith, Y. D.;
Warmus, J. S.;Taylor, C.Bioorg. Med. Chem. Lett. 2009, 19, 665–669.(b) Catto,
M.; Nicolotti, O.; Leonetti, F.; Carotti, A.; Favia, A. D.; Soto-Otero, R.;
ꢁ
ꢁ
Mendez-Alvarez, E.; Carotti, A. J. Med. Chem. 2006, 49, 4912–4925.
(c) Venkatesan, A. M.; Davis, J. M.; Grosu, G. T.; Baker, J.; Zask, A.; Levin,
J. I.; Ellingboe, J.; Skotnicki, J. S.; DiJoseph, J. F.; Sung, A.; Jin, G.; Xu, W.;
McCarthy, D. J.; Barone, D. J. Med. Chem. 2004, 47, 6255–6269. (d) Ng, H. P.;
May, K.; Bauman, J. G.; Ghannam, A.; Islam, I.; Liang, M.; Horuk, R.;
Hesselgesser, J.; Snider, R. M.; Perez, H. D.; Morrissey, M. M. J. Med. Chem.
1999, 42, 4680–4694. (e) Dannhardt, G.; Kiefer, W.; Nowe, U. Arch. Pharm.
1995, 328, 681–686. (f) Honda, Y.; Masuda, Y.; Yoshida, T.; Sato, F.;
Kurokawa, M.; Hosoki, K. Arzneim.-Forsch. 1995, 45, 1057–1060. (g) Nelson,
M. L.; Park, B. H.; Levy, S. B. J. Med. Chem. 1994, 37, 1355–1361. (h) Hornsby,
P. J.; Harris, S. E.; Aldern, K. A. Chem.-Biol. Interact. 1984, 51, 335–346.
(i) Fulton, A.; Norman, T. R.; Cheng, H.; Burrows, G. D. J. Affective Disord.
1982, 4, 261–269.
basis set.26 A torsional energy scan relative to C2 -C1 -S-O
dihedral (phenyl/SdO dihedral, dPh/SO) revealed the pre-
sence of a single minimum with dPh/SO between 0 and 10°
and a maximum at dPh/SO ≈ 100° destabilized by around
5 kcal mol-1. B3LYP reoptimization of the minimum led
to dPh/SO = þ3.0° with 6-31G(d) and þ7.1° with 6-311G(d,p)
basis set, reproducing Agranat’s results.26 Compounds 3
0
0
(23) (a) Naso, F.; Capozzi, M. A. M.; Bottoni, A.; Calvaresi, M.;
Bertolasi, V.; Capitelli, F.; Cardellicchio, C. Chem.;Eur. J. 2009, 15,
13417–13426. (b) Capozzi, M. A. M.; Cardellicchio, C.; Naso, F.; Rosito,
V. J. Org. Chem. 2002, 67, 7289–7294.
(24) (a) Kuroda, R.; Honma, T. Chirality 2000, 12, 269–277. (b) Kuroda,
R. Solid-state CD: application to inorganic and organic chemistry. In Circular
Dichroism: Principles and Applications, 2nd ed.; Berova, N., Nakanishi, K.,
Woody, R. W., Eds.; Wiley-VCH: New York, 2000; pp 159-184.
ꢁ
€
(25) Pescitelli, G.; Kurtan, T.; Florke, U.; Krohn, K. Chirality 2009, in
(26) Marom, H.; Biedermann, U.; Agranat, I. Chirality 2007, 19, 559–
press; DOI: 10.1002/chir.20795.
569.
J. Org. Chem. Vol. 75, No. 4, 2010 1145

Pescitelli et al.
JOCArticle
FIGURE 1. CD spectra of sulfoxides (R)-1a-m in acetonitrile.
SCHEME 3. Model Sulfoxides 2-5
small extent. In the minimum-energy structures of (R)-3 and
(R)-4, the phenyl/SdO dihedrals dPh/SO are again in the range
between þ3° and þ6°.
TDDFT calculations were run on DFT-optimized geome-
tries for compounds 2-4 with various functionals (BP86,28
B3LYP,29 PBE0,30 and BH&HLYP,31 in order of increasing
HF exchange amount) and basis sets (SVP, TZVP, and
aug-TZVP, in order of increasing size),32 and the resulting
CD spectra were compared with experimental literature data
(see Figure S3, Supporting Information, for (S)-3).15 For 3
and 4, the Boltzmann-weighted average CD spectrum was
evaluated at 300 K over three DFT (B3LYP/6-311þG(d,p)//
B3LYP/6-31G(d)) energy minima (a, gþ, and g-). Among
the functionals used, the best match between experimental
and calculated CD spectra, in terms of relative intensity
and position of the two major bands, was obtained with
BH&HLYP. This functional afforded a systematic frequency
overestimation of ∼3,000 cm-1 for all predicted transitions
(corresponding to a blue shift of 12 and 20 nm at 200 and
260 nm, respectively, e.g., for (S)-4, average calculated CD,
bandwidth σ = 2500 cm-1: 226 nm (Δε = -20 M-1 cm-1),
202 nm (þ25); experimental in CH3CN:15 247 nm (-14),
217 nm (þ26)). Both B3LYP and PBE0 overestimated the
energy separation between the two major bands, affording
a less satisfying agreement (see Figure S3 in Supporting
and 4 have an additional degree of freedom relative to
the rotamerism around the S-C1 bond. In both cases, three
rotamers were found by DFT calculations, which are labeled
a (anti), gþ (gauche plus), and g- (gauche minus), correspond-
0
00
ing to C1 -S-C1-C2 (or C1 ) dihedrals dPhS/C1≈ 180°, þ60°,
and -60°, respectively (see Table S1, Supporting Information).
For compound 3 with (R) configuration, a and gþ conformers
had similar population quite larger than that of conformer
g-. For compound (R)-4, conformer a was the most stable,
while the two g conformers had similar energies. Populations
evaluated at the B3LYP/6-311þG(d,p)//B3LYP/6-31G(d)
level were slightly different from those obtained with B3LYP/
6-31G(d), while the difference between B3LYP/6-311þG(d,p)//
B3LYP/6-31G(d) and full B3LYP/6-311þG(d,p) results
was minimal (Table S1, Supporting Information). Moreover,
inclusion of PCM (polarizable continuum solvation model)27
for acetonitrile in B3LYP/6-311þG(d,p) single point calcula-
tions affected the computed relative populations to a very
(28) (a) Becke, A. D. Phys. Rev. A 1988, 38, 3098–3100. (b) Perdew, J. P.
Phys. Rev. B 1986, 33, 8822–8824.
(29) Becke, A. D. J. Chem. Phys. 1993, 98, 5648–5652.
(30) Adamo, C.; Scuseria, G. E.; Barone, V. J. Chem. Phys. 1999, 111,
2889–2899.
(31) Becke, A. D. J. Chem. Phys. 1993, 98, 1372–1377. (Gaussian’03 and
Gaussian’09 add a B88-type28a gradient exchange correction to the original
BH&H scheme).
(32) Schafer, A.; Huber, C.; Ahlrichs, R. J. Chem. Phys. 1994, 100, 5829–
5835.
(27) Continuum Solvation Models in Chemical Physics: From Theory to
Applications; Tomasi, J., Mennucci, B., Cammi, R, Eds.; Wiley: Chichester,
2007.
1146 J. Org. Chem. Vol. 75, No. 4, 2010

Pescitelli et al.
JOCArticle
Information for 3). BP86 performed worst along the series. As
for the basis set, using BH&HLYP as functional the agree-
ment observed was similar for the three sets tested. The good
performance of the smallest basis set SVP is especially notice-
able. Use of small basis sets like 6-31G(d) has been suggested
as an expedient to save computational time viable in some
conditions.33 For the present compounds, however, the over-
all computational time using TZVP is still reasonable, and
therefore the BH&HLYP/TZVP combination was employed
in the following calculations and a frequency correction34 of
-3,000 cm-1 (þ20 nm at 260 nm) was applied in all cases. The
only exceptions concern methoxyphenyl derivatives 1b
[4OMe/H] and 1d [2OMe/H], for which B3LYP led to better
results than BH&HLYP (no frequency correction was needed
in these cases).
The same sequence described above of (a) conformational
search, (b) DFT geometry optimizations at the B3LYP/6-
31G(d) level, (c) single point calculations at the B3LYP/6-
311Gþ(d,p) level to estimate relative energies, (d) TDDFT
calculations with BH&HLYP/TZVP (or B3LYP/TZVP) on
all populated minima (including PCM for a few representa-
tive cases), and (e) averaging of CD spectra was applied to
several representative aryl benzyl sulfoxides among the series
1a-m. Diagrams and relevant data of DFT-calculated en-
ergy minima are summed in Tables S2-3 (Supporting
Information), and the results of TDDFT calculations are
shown in Figures 2, 3, 4, and 5 and S7.
FIGURE 2. CD spectra calculated for sulfoxides (R)-1f [4Br/H]
(a) and (R)-1i [4Br/4OMe] (b) with BH&HLYP/TZVP on B3LYP/
6-311Gþ(d,p)//B3LYP/6-31G(d) structures. Average spectra are
obtained by Boltzmann weighting at 300 K of the calculated spectra
for various energy minima (three for 1f, six for 1i, of which the
results for the lowest-energy four are shown; see Table S2 in
Supporting Information). Band-width σ = 2500 cm-1, frequency
2.2. Compound 1f [4Br/H]. The most straightforward case
is compound (R)-1f [4Br/H], which will be described first.
DFT calculations resulted in three minima (labeled as a, gþ,
and g-, see Table S2 in Supporting Information) relative to
the rotamerism around the S-CH2 bond, with a slight
preference for gþ (absolute minimum, 38% population at
300 K) and a (þ0.05 kcal mol-1, 33%) over the g- con-
former (þ0.20 kcal mol-1, 27%). The second possible degree
of conformational freedom, that is, the rotamerism around
the phenyl-CH2 bond, led to a single energy minimum with
the CH2-S bond nearly perpendicular to the benzyl phenyl
ring. The low-energy structures for 1f are shown in Figure S4
(Supporting Information). Calculated CD spectra are very
similar for a and gþ conformers in the low-energy region
and consist of a positive couplet-like feature between 210 and
300 nm (frequency corrected, Figure 2a). A monosignate and
weaker negative CD band is instead computed for g-
between 210 and 300 nm. As a consequence, the weighted
average spectrum is dominated by a and gþ contributions.
For (R)-1f it consists of a moderately intense positive cou-
plet-like feature with extrema at 260 and 220 nm (frequency
corrected), followed by two weaker CEs around 200 and 185
nm, respectively positive and negative. Overall, the average
calculated CD for (R)-1f [4Br/H] is in almost perfect agree-
ment with the experimental one (compare Figures 2a and 1c;
a direct comparison is shown in Figure 7b below). Introduc-
tion of a solvent model (PCM)27 for acetonitrile in both
energy and CD calculations did not improve the overall
result, actually, the agreement with the experimental spec-
trum was worse at high energies (see Supporting Informa-
tion, Figure S7c).
correction -3,000 cm-1
.
CD spectra recorded at variable temperatures from 262 to
348 K (-36 to þ50 °C) are in accordance with the conforma-
tional picture found by calculations, because increasing the
temperature led to a slight intensity decrease for the positive
CD band, while the negative one was relatively less affected
(Figure S5, Supporting Information).
The first positive CE (from the red) calculated for (R)-1f
essentially arises from a HOMO f LUMO transition.
Analysis of Kohn-Sham orbitals (MO) revealed that both
HOMO and LUMO are delocalized over a large molecular
portion (see Figure S6, Supporting Information). In parti-
cular, the HOMO (MO 74) may be thought to arise from the
combination of oxygen lone pair nO, a bonding σS-C1
between sulfur and methylene carbon, and bonding π orbi-
tals over the two aromatic rings, both of benzene e1g (π3)
type.9 Conversely, the LUMO (MO 75) consists of the two
aromatic π* orbitals (benzene e2u type, or π5) plus an
antibonding σ*S-C1 MO. Therefore, the first observed tran-
sition for 1f [4Br/H] corresponds to the sulfoxide primary
band described by Mislow. The situation for the negative CE
at 230 nm is less clear-cut, as it results from the superposition
of many different transitions, two of which are associated
with moderately intense negative rotational strengths. The
main contribution to the first of them, computed at 222 nm
(frequency corrected), is the transition to the LUMO from
the occupied MO 71, which may be described as a combi-
0
nation of oxygen lone pair nO, a bonding σS-C1 between
(33) Giorgio, E.; Tanaka, K.; Verotta, L.; Nakanishi, K.; Berova, N.;
Rosini, C. Chirality 2007, 19, 434–445.
(34) Bringmann, G.; Gulder, T. A. M.; Reichert, M.; Gulder, T. Chirality
2008, 20, 628–642.
sulfur and phenyl carbon, and a phenyl π orbital of e1g type
(Figure S6 in Supporting Information). The second transi-
tion concurring to the negative CD band is calculated at
J. Org. Chem. Vol. 75, No. 4, 2010 1147

Pescitelli et al.
JOCArticle
215 nm and it is instead the combination of several single-
electron promotions from MO 71 and MO 74 to various
virtual orbitals, with small CI (configuration interaction)
coefficients. As for the 1Lb-type aromatic transition, it
appears as one of the components of two transitions pre-
dicted around 245 nm (frequency corrected), both with
negligible oscillator and rotational strength, submerged by
the sulfoxide primary band. The fact that the present
TDDFT calculations seem to especially overestimate the
energy of this kind of transition (benzene 1Lb) has been
observed before,35 but it has no or very small effect on the
overall shape of the computed CD curves of aryl benzyl
sulfoxides, because of the inherent weakness of the asso-
ciated CD bands.
2.3. Methoxy Substituent on the Benzyl Phenyl Ring:
Compound 1i [4Br/4OMe]. The results of calculations on a
second 4-bromophenyl derivative, namely, (R)-1i [4Br/4OMe],
did not substantially deviate from those previously discussed.
The only difference consists in the additional degree of
freedom relative to the rotamerism of the methoxy group,
and thus six energy minima needed to be considered (Table S2
in Supporting Information). In this case too, the agreement
between the average calculated and experimental CD spectrum
was very good (Figures 1d and 2b, direct comparison shown in
Figure 7d below), which also allowed us to confirm (R)-(þ)
absolute configuration associated with positive CD at 260 nm.
In fact, compound 1i is the only one in the series whose
absolute configuration had not been assigned by X-ray analy-
sis23 or known before. Similarly to what found for 1f, inclusion
of PCM for acetonitrile in CD calculations did not affect much
the two major transitions, while some difference with respect to
the calculations in vacuo was observed at higher energies (see
Supporting Information, Figure S7f).
FIGURE 3. CD spectra calculated for sulfoxides 1g [4Br/4NO2]
(a) and 1h [4Br/2NO2] (b) with BH&HLYP/TZVP on B3LYP/6-
311Gþ(d,p)//B3LYP/6-31G(d) structures. Average spectra are ob-
tained by Boltzmann weighting at 300 K of the calculated spectra
for three energy minima (see Table S2 in Supporting Information).
The spectra labeled with (PCM) are the average obtained by
employing PCM solvation model for acetonitrile both in single
point (DFT) and TDDFT calculations (Table S3 and Figure S7d,e
in Supporting Information). Band-width σ = 2500 cm-1, frequency
correction -3,000 cm-1
.
2.4. Nitro Substituent on the Benzyl Phenyl Ring: Com-
pounds 1g [4Br/4NO2] and 1h [4Br/2NO2]. For the com-
pounds containing a nitro-substituted benzyl phenyl ring
such as 1g [4Br/4NO2] and 1h [4Br/2NO2], some differences
arose in terms of calculated geometries and spectra with
respect to the cases presented so far. The 4-nitro isomer
(R)-1g allowed only three rotamers a, gþ, and g- (Table S2
in Supporting Information); their relative energies (and
populations at 300 K) were þ0.09 kcal mol-1 (38%), 0
(absolute minimum, 45%) and þ0.57 kcal mol-1 (17%),
that is, the g- is quite less populated than for (R)-1f [4Br/H].
This is probably due to an unfavorable dipole-dipole align-
ment between the nitrobenzene and sulfinyl group. For the
2-nitro isomer (R)-1h, the a rotamer is by far the most
populated (87%), whereas the g- is not at all (overall
population <2%), perhaps due to repulsive electrostatic
interaction between nitro and sulfoxide groups in the g
rotamers. Average TDDFT-calculated CD spectra were
in both cases in good agreement with experimental ones
(Figures 1e and 3, a direct comparison for 1g is shown in
Figure 7c below), but some details need to be added. For 1g
[4Br/4NO2], the splitting of the first CD band between 240
and 350 nm was not reproduced by calculations in vacuo,
unless admitting a larger contribution from the g- isomer
than predicted by B3LYP/6-311þG(d,p)//B3LYP/6-31G(d)
(Figure 3a). The result was practically unchanged using
geometries optimized at B3LYP/6-311þG(d,p) level or in-
troducing PCM for acetonitrile in energy calculations (see
Supporting Information, Table S3). However, when PCM
was employed in TDDFT calculations too, all conformers
acquired a positive CD signals around 240 nm (frequency
corrected, Figure S7d, Supporting Information). As a result,
the weighted average spectrum showed two positive bands
above 230 nm similar to the experimental one (Figure 3a),
although the intensity of the short-wavelength component is
overestimated.
In the CD spectrum of (R)-1h [4Br/2NO2] three transitions
of alternating sign -/þ/- were predicted in the long-wave-
length region (>280 nm, frequency corrected, Figure 3b).
The middle one is associated with the largest absolute
rotational strength and is conceivably responsible for the
positive CE at 305 nm seen in the experimental spectrum
(Figure 1e). The assignment is supported by the fact that,
after running DFT (single point) and TDDFT calculations
with PCM, the average spectrum did display a more intense
positive band at 312 nm (frequency corrected, Figures 3b and
S7e, Supporting Information). The difference with respect to
the calculations in vacuo is mainly due to a switch in
population between the two gþ conformers (related to the
nitrophenyl ring flip and labeled as I and II, Table S2,
Supporting Information). Therefore, for both compounds
(R)-1g [4Br/4NO2] and 1h [4Br/2NO2] the inclusion of a
solvation model partially improved the agreement between
(35) (a) Pescitelli, G.; Di Bari, L.; Caporusso, A. M.; Salvadori, P.
Chirality 2008, 20, 393–399. (b) Mazzini, F.; Pescitelli, G.; Di Bari, L.;
Netscher, T.; Salvadori, P. Chirality 2009, 21, 35–43.
1148 J. Org. Chem. Vol. 75, No. 4, 2010

Pescitelli et al.
JOCArticle
FIGURE 5. CD spectra calculated for sulfoxides 1a [4NO2/H]
(a) and 1e [2COOMe/H] (b) with BH&HLYP/TZVP on B3LYP/
6-311Gþ(d,p)//B3LYP/6-31G(d) structures. Average spectra are
obtained by Boltzmann weighting at 300 K of the calculated spectra
for three (1a) or four (1e) energy minima (see Table S2, Supporting
Information). The spectrum labeled with PCM is the average
obtained by employing PCM solvation model for acetonitrile in
single point (DFT) and TDDFT calculations (Table S3 and Figure
FIGURE 4. CD spectra calculated for sulfoxides (R)-1b [4OMe/H]
(a) and (R)-1d [2OMe/H] (b) with B3LYP/TZVP on B3LYP/6-
311Gþ(d,p)//B3LYP/6-31G(d) structures. Average spectra are
obtained by Boltzmann weighting at 300 K of the calculated
spectra for various energy minima (three for 1d, six for 1b, of which
the results for the lowest-energy four are shown; see Table S2
in Supporting Information). Band-width σ = 2500 cm-1, no fre-
quency correction.
S7a in Supporting Information). Band-width σ = 2500 cm-1
frequency correction -3,000 cm-1
,
.
calculated and experimental CD spectra. However, it must be
stressed that the calculations in vacuo were already enough
accurate for, e.g., the hypothetical purpose of assigning abso-
lute configurations.
around S-CH2 (a, gþ and g- rotamers) and C-OMe bonds
(two rotamers indicated as I and II). In this case too, the
observed positive couplet-like feature between 200 and 300 nm
for the (R) configuration (Figure 1b) was correctly reproduced
by the average B3LYP/TZVP-calculated CD (Figure 4a). Use
of the PCM solvation model in both energy and CD calcula-
tions did not affect substantially the result (see Table S3 and
Figure S7b, Supporting Information). For the o-isomer (R)-1d
[2OMe/H], only one conformation is possible for the methoxy
group to avoid a steric contact with the sulfoxide one; more-
over, the g- rotamer is very disfavored with respect to a and
gþ due to the electrostatic repulsion between the two SdO and
OMe groups (Table S2, Supporting Information). The pecu-
liar shape of the CD spectrum of (R)-1d, showing two distinct
positive bands in the long wavelength region, is well repro-
duced by B3LYP/TZVP calculations (Figures 1b and 4b). For
both isomers 1b [4OMe/H] and 1d [2OMe/H], the first (from
the red) CE is due to the HOMO f LUMO transition
corresponding to the sulfoxide primary transition, with MO
characters similar to those analyzed above for 1f. The diffe-
rences observed in the remaining calculated CD bands seem to
be related to the fact that the perturbation exerted by the
methoxy substituent on the phenylsulfinyl ring is larger for the
p- than the o-isomer. In fact, in both cases the electron-
donating group destabilizes all the virtual orbitals localized
on the attached phenyl ring; however, this effect is weaker for
the o-isomer, and hence a less pronounced impact on the CD is
manifest.
The three calculated transitions above 280 nm for 1h
all involve the LUMO, which is an almost pure π* orbital
localized on the nitrobenzene ring. On the contrary, the
stronger positive CD calculated at 253 nm (frequency
corrected) mainly arises from the HOMO f LUMOþ1
transition. These latter MOs are both delocalized on the
whole molecule including the sulfinyl group and may be
thought to correspond, respectively, to the HOMO and
LUMO described above for 1f [4Br/H] and the associated
transition to the sulfoxide primary band. Therefore the
discrepancies between the CD spectrum of 1f and its
o-nitrobenzyl analogue 1h [4Br/2NO2] are mainly due to an
energy redistribution of MOs, while the sign and character
of the major CD band (sulfoxide primary band) remain
unchanged.
2.5. Methoxy Substituent on the Phenyl Ring: Compounds
1b [4OMe/H] and 1d [2OMe/H]. As far as sulfoxides with
substituents different from bromine on the phenyl ring are
concerned, compound 1b [4OMe/H] has a CD spectrum
similar to that of 1f [4Br/H], and both DFT and TDDFT
calculations are in agreement with the already described trend.
Six low-energy conformations were obtained through B3LYP/
6-31G(d) geometry optimizations (Table S2, Supporting In-
formation), which arise from a combination of the rotamerism
J. Org. Chem. Vol. 75, No. 4, 2010 1149

Pescitelli et al.
JOCArticle
2.6. Nitro or Carboxy Substituent on the Phenyl Ring:
Compounds 1a [4NO2/H] and 1e [2COOMe/H]. The singular
CD spectra of the nitro and methoxycarbonyl derivatives 1a
[4NO2/H] and 1e [2COOMe/H] were also well reproduced
by CD calculations (Figures 1a and 5, a direct comparison
for 1a is shown in Figure 7a below). The calculations
correctly predicted both the distinctive red shift with respect
to the other derivatives, and the series of CD bands with
alternating sign appearing in the 180-350 nm range
(frequency corrected). For 1a, DFT geometry optimiza-
tions resulted in the usual three rotamers a, gþ, and g-, the
latter being the most stable (Table S2, Supporting
Information), for which TDDFT computed quite different
CD spectra (Figure 5a). In the resulting average spectrum,
the third CD band (experimentally a positive maximum at
228 nm for (R)-1a) is partially submerged by flanking
negative bands and appears as a negative maximum around
225 nm (frequency corrected). The agreement with the
experimental spectrum improved after introducing PCM
for acetonitrile in DFT (single point) and TDDFT calcula-
tions (see Supporting Information, Table S3 and Figure
S7a). With respect to the calculations in vacuo, the relative
population of the a isomer increased substantially (from
29% to 42%). As a consequence, the weighted average CD
spectrum displayed a positive band at 215 nm (frequency
corrected) and acquired the same appearance of the experi-
mental spectrum with a series of five bands with alternating
sign (Figure 5a). The first band around 300 nm is due to the
HOMO f LUMO transition, but these two MOs have quite
different character with respect to 1f [4Br/H] and analogous
cases. In fact, the HOMO (MO 68) is a combination of nO,
FIGURE 6. Solid-state CD spectra measured as KCl pellets for
sulfoxides (R)-1c, 1d, 1g, and 1h.
Supporting Information). As a result, only two conforma-
tions, a(I) and gþ(I), contributed significantly to the average
calculated CD of (R)-1e [2COOMe/H] (Figure 5b), which is
in good agreement with the experimental one (Figure 1a).
3. CD Spectra in the Solid State. The absolute configura-
tion and the crystal structure of sulfoxides 1a, 1c-e, 1g,h and
1j-l had been investigated by X-ray analysis.23 In the present
work, we decided to measure CD spectra on microcrystalline
samples of these compounds with the technique of KCl
pellet24,25 (see Figures 6 and S9 in Supporting Information).
The observed solid-state CD spectra of compounds 1c
[3OMe/H] and 1d [2OMe/H] (Figure 6a), and of 1k [4Br/
2OMe], 1l [4Br/3Cl] and to some extent 1e [2COOMe/H]
(Figure S9a) were comparable with the spectra in solution.
On the other hand, the spectra of 1g [4Br/4NO2] and 1h
[4Br/2NO2] (Figure 6b) and of 1a [4NO2/H] and 1j [4Br/3MeO]
(Figure S9b) showed substantial differences with respect to the
corresponding solution spectra.
σ
S-C1, and π (e1g-type) orbitals, the latter localized over the
benzyl phenyl but not the nitrophenyl ring, whereas the
LUMO (MO 69) is a π* orbital localized on the nitrophenyl
ring (Figure S8, Supporting Information). The second
calculated CD band is largely due to a π-π* transition of
the nitrophenyl ring. The third transition is mainly a
promotion from the HOMO to MO 71, which consists of
two aromatic π* orbitals plus an antibonding σ*S-C1, thus
resembling the LUMO of 1f. Overall, the apparent red shift
observed for 1a with respect, e.g., to the bromine analogue
1f [4Br/H] is justified by the stabilizing effect exerted by the
electron-withdrawing nitro substituent on the orbitals of
the attached phenyl ring. This effect is stronger on the
virtual orbitals, first of all the LUMO, than on the HOMO
and following occupied orbitals.
Similar arguments may be applied to the discussion of the
CD spectrum of 1e [2COOMe/H] with an electron-with-
drawing carboxy group on the phenylsulfinyl ring. The first
calculated transition at 293 nm (frequency corrected) was
again mainly the HOMO f LUMO. The two MOs have
character similar to that of the corresponding MOs of 1a, the
HOMO being a combination of nO, σS-C1, and benzyl phenyl
π orbitals and the LUMO being mainly a π* orbital localized
over the carboxyphenyl ring. As in the case of 1d [2OMe/H]
discussed above, the g- rotamer of (R)-1e had much higher
DFT-computed energy than a and gþ (Table S2, Supporting
Information) because of the electrostatic repulsion between
the ester and the sulfinyl moiety. The ester group may assume
two different conformations, with either the CdO or the
C-OMe facing the sulfinyl group, the former (labeled as I)
being much more stable than the latter (II, Table S2,
As an example of the first class, the X-ray structure of com-
pound (R)-1d [2OMe/H] shows a conformation of gþ type
with respect to the S-C1 torsion, with a value of dPhS/C1
=
þ65.6°. DFT calculations afforded a very similar geometry for
the gþ rotamer. The CD spectrum calculated by TDDFT on
this geometry (Figure 4b) is in agreement with the solid-state
CD spectrum of (R)-1d (Figure 6a). In many situations when
the solid-state CD is determined by the absolute conformation
of a single, “isolated” molecule, a very good agreement has
been observed between experimental solid-state CD spectra
and those calculated with TDDFT using the X-ray geometries
as input structures.25
On the other hand, solution and solid-state CD spectra of
sulfoxides 1g [4Br/4NO2] and 1h [4Br/2NO2] are very differ-
ent (Figures 1e and 6b), because new bands appear in the
latter. The DFT-optimized geometry of 1g for the gþ
1150 J. Org. Chem. Vol. 75, No. 4, 2010

Pescitelli et al.
JOCArticle
FIGURE 7. Comparison between experimental and calculated CD spectra for sulfoxides (R)-1a,f,g,i. See captions of Figures 2, 3, and 5 for
details on calculated spectra.
rotamer is in very good agreement with the X-ray geometry,
showing a gþ conformation with dPhS/C1=þ51.8°. However,
the TDDFT-calculated CD for this geometry (Figure 3a) is
not in agreement with the solid-state CD of (R)-1g
(Figure 6b), because the negative bands at 270 and 370 nm
are not reproduced. Conversely, the X-ray geometry of 1h
shows an a conformation with dPhS/C1 = -166.9°, similar to
the DFT-optimized geometry for the a rotamer, with the
exception of an increased torsion angle of the nitro group
with respect to the phenyl ring (observed dihedral, -34°;
calculated, -25°). TDDFT-calculated CD for the DFT-
optimized and the X-ray geometry were very similar to each
other, but neither could reproduce the intense positive CD
band seen at 345 nm in the solid-state CD spectrum of (R)-1h
(Figure 6b). Probably, in the solid-state CD spectra of
compounds 1g, 1h, and others behaving similarly, effects
inherent to the crystalline state are present, such as inter-
molecular interactions of the exciton type between aryl
chromophores belonging to distinct molecules packed to-
gether in the crystal.24,25
d
Ph/SO dihedrals are below 12° (absolute values), with the only
exception of compound 1c [3OMe/H], in which it is 24°. In
our B3LYP/6-31G(d)-optimized geometries, the values of the
phenyl/SdO dihedrals dPh/SO are in the 0-10° range, with
values up to 15° observed only for a few hindered gauche
conformers (relative to the rotamerism around S-C1 bond).
0
Such a preferential conformation around the C1 -S bond
in alkyl or benzyl aryl sulfoxides means that the sulfur lone
pair is not perfectly perpendicular to the aromatic plane, an
arrangement that would imply dPh/SO ≈ 35°. A perfect
parallelism between the sulfur lone pair (in a sp3(S) atomic
orbital)15 and the aromatic π system (made up of 2p(C) AOs)
would clearly maximize their overlap. However, since the
extent of π-overlap between two equal p-type AOs follows
the cosine of the relative dihedral angle,9 in the present case
the amount of overlap loss may be roughly estimated around
15% at most. The subsequent destabilization is therefore
quite small and may be easily overcome by other factors such
as steric ones. In fact, in the arrangement with the sulfur lone
pair perpendicular to the aromatic plane, the C1 methylene
hydrogens would come too close to the ortho phenyl hydro-
0
gens (attached to C2 or C6 ). For example, on the torsional
0
Discussion
energy surface for methyl phenyl sulfoxide (2), for the point
0
Ground-State Properties. Geometry optimizationswith the
DFT method on model compounds 2-4 as well as on several
aryl benzyl sulfoxides belonging to the series 1 have resulted
in a small set of low-energy structures. An almost coplanar
arrangement between the SdO bond and the attached phenyl
ring is always present, as shown also by experimental7,23,36
and other computational investigations.19,26 In particular,
in X-ray structures of sulfoxides 1a, 1c-e, 1g,h, and 1j-l the
corresponding to dPh/SO = 35° the shortest C1-H to C2 -H
˚
0
(or C6 -H) distance reached 2-2.2 A values, that is, below
the sum of two van der Waals hydrogen radii.
The case of nitro-substituted compound 1a [4NO2/H]
deserves a further discussion. On the basis of the analysis
of UV spectra of aryl methyl sulfoxides, it has been suggested
that a nitro group in a position para to the sulfinyl may
enhance the electron-donating attitude of this latter.15 How-
ever, this effect is not manifest in our calculations. In fact,
analysis of Kohn-Sham orbitals, charge densities, and calcu-
lated geometries for 1a does not suggest any augmented
(36) Benassi, R.; Folli, U.; Iarossi, D.; Mucci, A.; Schenetti, L.; Taddei, F.
J. Chem. Soc., Perkin Trans. 2 1989, 517–521.
J. Org. Chem. Vol. 75, No. 4, 2010 1151

Pescitelli et al.
JOCArticle
charge donation from the sulfinyl or increased conjugation
with the phenyl ring with respect, e.g., to benzyl 4-bromophe-
nyl sulfoxide (1f). To deal with a smaller system we decided to
investigate methyl 4-nitrophenyl sulfoxide (5, Scheme 3), to be
compared with methyl phenyl sulfoxide (2). The absolute
minimum found by B3LYP/6-311G(d,p) calculations on 5
had dPh/SO = 5.2° and the distance between sulfur and phenyl
configurational assignment based on the comparison be-
tween calculated and experimental CD spectra. Moreover,
as discussed above, these discrepancies were alleviated by
introducing a PCM solvation model for acetonitrile in
energy and excited states calculations. The good perfor-
mance of the TDDFT//DFT calculation scheme employed
is noteworthy considering that the “calculated” spectrum is
in fact the weighted sum over three to six low-energy
structures, most of which with sizable population at the
working temperature. It can be concluded that relative
populations of various conformers and their component
CD spectra were both well predicted by the calculations,
even in the absence of a solvent model.
˚
0
carbon rS-C1 = 1.832 A; for 2, the two values were 7.1° and
˚
1.830 A, respectively. The charge density on the sulfur atom
predicted with the Mulliken scheme was 0.68 for 5 and 0.69
for 2; with the Natural Bond Orbitals scheme, the two values
were 1.18 for both molecules. These results confirmed that
the p-nitro group does not amplify the electron-donating
tendency of the sulfinyl group or the aromatic/sulfinyl con-
jugation. On the other hand, it is evident that the nitro group
stabilizes both the HOMO and the LUMO (the latter more
than the former) by comparing the MO energies of 2 and 5.
Moreover, it greatly influences the character of both HOMO
and LUMO in terms of localization on the aromatic
rings. This effect is the same observed for nitro-containing aryl
benzyl sulfoxides 1a [4NO2/H], 1g [4Br/4NO2], and 1h [4Br/
2NO2] and has consequences on their CD spectra (vide infra).
Excited-State Properties. CD spectra in acetonitrile solu-
tion have been recorded for 13 aryl benzyl sulfoxides
(R)-1a-m, most of which had not been previously reported
in the literature, with the exception of numerical CD data for
(S)-1a and 1b.15 As already underlined in the Results section,
the aspect of the CD spectrum is quite well preserved along
the whole series and consists of a couplet-like feature be-
tween 200 and 300 nm with the first (long wavelength)
branch being positive for (R) absolute configuration. CD
spectra deviating from the general trend were recorded
for compounds 1d [2OMe/H] and 1g [4Br/4NO2] (the first posi-
tive band is split) and for 1a [4NO2/H] and 1e [2COOMe/H]
(the whole spectra are sizably shifted to the red). In all cases
considered, the CD associated with the so-called sulfoxide
primary transition is positive for the (R) configuration, that
is, it is in keeping with Mislow’s empirical rule for alkyl aryl
sulfoxides.8 Such a consistency was not surprising because the
effect of replacing an alkyl group with the benzyl one was found
to be minimal, even when the aromatic ring carried auxochro-
mic substituents. On the other hand, molecular orbital analysis
(actually, Kohn-Sham orbitals were considered here)37 re-
vealed that the MOs involved in the relevant transitions are
delocalized on the benzyl phenyl ring too; therefore the fact that
a benzyl group does not appreciably affect spectral outcomes of
sulfoxides is less obvious than anticipated.
For the majority of (R)-sulfoxides considered, the CD
calculated for the anti and gaucheþ conformer (relative to
the rotamerism around S-C1 bond) were quite similar at low
energies. However, one would expect the CD spectra of aryl
benzyl sulfoxides to be at least partially affected by the
exciton coupling between the two aromatic rings, when in a
proper reciprocal arrangement.38 Considering for example
1
the benzene La transition, which is long-axis polarized in
p-disubstituted benzenes,9 the exciton coupling between two
such transitions in aryl benzyl sulfoxides is expected to be
much more effective for the gauche than for the anti con-
former. Clearly, the exciton coupling is not the dominant
mechanism of optical activity in the long wavelength region,
although it may contribute to the CD in the high energy
region for some substrates. For example, for 1f [4Br/H] the
calculated CD band around 190 nm (frequency corrected) is
oppositely signed for a and gþ rotamers (Figure 2a).
Our results suggest that the most reasonable mechanism of
optical activity to be considered for aryl benzyl sulfoxides (as
well as for their alkyl phenyl analogues)8b is the intrinsic
chirality of the chromophoric system. This latter must necessa-
rily include the whole molecule, because MO analysis demon-
strated that the two major CD bands are due to transitions
involving Kohn-Sham orbitals delocalized over the whole
molecule, including both aromatic rings. As reported in the
Introduction, CD spectra of alkyl aryl sulfoxides have been
interpreted through the coupled-oscillator model, in terms of
the exciton coupling between the sulfoxide primary transition
and the phenyl 1La transition.15 Such a treatment is only
allowed upon the assumption that the sulfoxide and the
benzene chromophores are isolated from each other, a situation
that occurs if there is severe steric hindrance to SdO/aryl
conjugation, as when the aromatic moiety is a 1-naphthyl or
a 9-phenanthryl group.16 The above findings demonstrate that
this is not generally true for all aryl sulfoxides; as a consequence,
a fundamental condition of the coupled-oscillator model,
namely, that the two (or more) interacting chromophores are
nonconjugated, is not valid for alkyl phenyl sulfoxides, though
the model may appear to provide a correct answer.15
CD spectra have been calculated with the TDDFT method
for the eight benzyl phenyl sulfoxides 1a-b and 1d-i, using
DFT-optimized geometries as input structures. In the very
large majority of cases, a very good agreement has been
observed between calculated and experimental spectra. A
direct comparison between experimental and calculated
spectra is shown in Figure 7 for four relevant compounds.
Minor discrepancies, consisting in the appearance (intensity)
of one band at intermediate energy, have been detected only
for compounds 1a [4NO2/H], 1g [4Br/4NO2], and 1h [4Br/
2NO2], but they would not preclude a correct absolute
Our results demonstrate that a full and fairly accurate
prediction of CD spectra of aryl benzyl sulfoxides and of
related compounds may be achieved by means of the
TDDFT//DFT calculation approach. Since the number of
populated conformers to be taken into account for these
substrates is rather small, the problems related with extensive
(37) (a) Stowasser, R.; Hoffmann, R. J. Am. Chem. Soc. 1999, 121, 3414–
ꢁ
3420. (b) Kar, T.; Angyan, J. G.; Sannigrahi, A. B. J. Phys. Chem. A 2000,
104, 9953–9963.
(38) Harada, N.; Nakanishi, K Circular Dichroic Spectroscopy-Exciton
Coupling in Organic Stereochemistry; University Science Books: Mill Valley,
CA, 1983.
ꢁ
1152 J. Org. Chem. Vol. 75, No. 4, 2010

Pescitelli et al.
JOCArticle
conformational analyses are not encountered,25 and the
overall computational cost is quite limited. Therefore, when
empirical predictions based on Mislow’s rules are judged
insufficient, a full quantum-mechanics CD calculation should
be considered as a valid practical option. In that respect, a
comparably efficient alternative to electronic CD is repre-
sented by vibrational CD (VCD) calculations.13,39
The apparently singular CD spectra of nitro and carboxy
derivatives 1a [4NO2/H], 1e [2COOMe/H], and to a minor
extent, 1h [4Br/2NO2] could be explained through the analysis
of the configurations (CI terms) contributing to the two major
transitionsand of the relativeMOs. Theelectron-withdrawing
group exerts a stabilization effect on the orbitals of the
attached ring and provokes a large redistribution of the
MOs energy scale and of the corresponding transitions, thus
altering the CD shape. Furthermore, it affects the virtual
orbitals (especially the LUMO) more than the occupied ones,
leading to the red shift observed, e.g., for 1a and 1e.
One of the aspects that our calculations did not reproduce
was the substantial hyper/batochromic effect apparent in the
absorption spectrum of o-methoxyphenyl isomer 1d, relative
to the sulfoxide primary band, with respect to the p-isomer
1b. However, the CD spectra of the two compounds were
both well predicted. The observed phenomenon is reminis-
cent of the ortho-effect seen in the absorption spectra of
2- versus 4-methoxy benzaldehyde and acetophenone,40
which must be related to the different mobility of the
methoxy group in the two isomers and cannot be reproduced
by calculations performed on single structures where the
methoxy group is always coplanar with the aryl ring.
Solid-state CD spectra have been recorded for compounds
1a, 1c-e, 1g,h, and 1j-l that were available as crystalline
samples characterized by X-ray analysis. Quite surprisingly,
although almost half of the compounds showed solid-state
CD spectra in accordance with solution ones, the second half
gave solid-state CD spectra apparently dominated by inter-
molecular interactions in the crystals. These effects are
difficult to be quantified,25 as they would require calculating
the CD on a more or less large ensemble of molecules as
found in a crystal cluster,41 which is beyond the scope of the
present paper. We tried to highlight a possible relation
between the spectral behavior and the crystal shape. The
above compounds crystallize either in the monoclinic P21 or
in orthorhombic P212121 space groups, both types permit-
ting in principle symmetry-allowed intracrystalline exciton
interactions to be established. Unfortunately, no clear cor-
relation was found between samples showing solid-state
intrinsic CD effects and their crystal class.
Solution CD spectra appear very homogeneous along the
series and show in turn close similarity to the general trend
observed for alkyl phenyl sulfoxides. A typical CD spectrum
consists in a couplet-like feature between 210 and 300 nm
with moderate intensity and positive sign of the long wave-
length branch for (R) configuration. Clear differences with
respect to typical spectra were only observed for compounds
with auxochromic substituents at the 2- and sometimes
4-position of the phenylsulfinyl ring, such as nitro, methoxy-
carbonyl, and to a minor extent, methoxy. However, in all
cases, analysis of transitions and relative MOs computed
with DFT demonstrated that the CD corresponding to the
so-called primary sulfoxide band is always positive for (R)
absolute configuration.
Microcrystalline solid-state CD spectra were similar to
solution ones for some sulfoxides, while for others they
showed very different shapes with new bands and/or bands
with altered intensity. These latter cases were interpreted as
due to intermolecular couplings between molecules closely
packed in the crystals. It must be concluded that for aryl
benzyl sulfoxides solid-state CD does not always reflect the
chirality of the isolated molecule and should be considered
with caution as a source of stereochemical information.
The most important result from our investigation is that
TDDFT calculations, run with hybrid functionals such as
BH&HLYP and B3LYP and a moderately large basis set
(TZVP) on DFT-computed geometries, are able to fully
reproduce CD spectra of aryl benzyl sulfoxides. Therefore,
as noted by other authors,20 TDDFT//DFT calculations
lend themselves as a reliable tool for interpreting the CD
spectra of these and other related compounds and for
determining their absolute stereochemistry.
Experimental Section
Enantiopure aryl benzyl sulfoxides 1a-m were prepared by
tert-butyl hydroperoxide enantioselective oxidation of the cor-
responding sulfides in the presence of catalytic amounts of the
complex between titanium tetraisopropoxide and (S,S)- or (R,
R)-hydrobenzoin according to previous work.23 The enantio-
meric purity of aryl benzyl sulfoxides 1a-m was measured by
HPLC after recrystallization and was in all cases g97%.23
Employment of the (S,S)-ligand yielded (R)-sulfoxides, as es-
tablished mainly by X-ray diffraction analysis (compounds 1a,
1c, 1d, 1e, 1f, 1g, 1h, 1j, 1k, 1l, 1m) and as supported by the
theoretical model derived for this type of oxidation.23 The
absolute configuration of compound (R)-(þ)-1b is known,15
and that of (R)-(þ)-1i is confirmed by the present results.
CD spectra were recorded with a Jasco J-710 spectropolari-
meter with the following measurement parameters: scan speed,
50 nm/min; bandwidth, 1 nm; response, 1 s; 4-8 accumulations.
Solution samples were prepared by dissolving the proper
amount of substrates in acetonitrile to afford ∼1 mM concen-
tration, which assures the prevalence of monomeric species.8b
Cells with 0.01-0.05 cm path lengths were employed to keep
absorbance below 1 au. Solid-state samples were prepared with
the KCl pellet technique as follows.25 Less than 1 mg of the
crystalline compound was mixed with ∼100 mg of oven-dried
KCl, finely ground and pressed at ∼10 ton under vacuum for 15
min to produce a glassy translucent disk. Several samples were
prepared for each compound to ensure reproducibility. The
pellet was mounted on a rotatable support placed as close as
possible to the detector. On each sample, several CD spectra
were measured upon rotation of the disk around the incident
axis direction and 180° flip around the vertical. These spectra
Conclusions
We have reported CD spectra of a series of aryl benzyl
sulfoxides with various substituents on the two rings, both in
acetonitrile solution and in the solid state as KCl pellets.
(39) Drabowicz, J.; Zajac, A.; Lyzwa, P.; Stephens, P. J.; Pan, J.-J.;
Devlin, F. J. Tetrahedron: Asymmetry 2008, 19, 288–294. and references
therein.
(40) (a) Dearden, J. C.; Forbes, W. F. Can. J. Chem. 1958, 36, 1362–1370.
(b) Dearden, J. C.; Forbes, W. F. Can. J. Chem. 1959, 37, 1305–1314.
ꢁ
ꢁ
ꢁ
ꢁ
(41) (a) Kerti, G.; Kurtan, T.; Borbas, A.; Szabo, Z. B.; Liptak, A.;
ꢁ
ꢁ
€
ꢁ
Szilagyi, L.; Illyes-Tunde, Z.; Benyei, A.; Antus, S.; Watanabe, M.;
Castiglioni, E.; Pescitelli, G.; Salvadori, P. Tetrahedron 2008, 64, 1676–1688.
(b) Bringmann, G.; Maksimenka, K.; Bruhn, T.; Reichert, M.; Harada, T.;
Kuroda, R. Tetrahedron 2009, 65, 5720–5728.
J. Org. Chem. Vol. 75, No. 4, 2010 1153

Pescitelli et al.
JOCArticle
were almost superimposable to each other, demonstrating the
absence of detectable spectral artifacts.25
functions taken from the most diffuse functions of aug-cc-
pVDZ set.43 Dipole-velocity rotational strengths were employed
to calculate CD spectra; however, the difference with dipole-
length values was negligible (<15%) for most transitions using
TZVP basis. CD spectra were generated by applying a Gaussian
band shape with 2000-2500 cm-1 half-height width. For
BH&HLYP/TZVP calculations, a frequency correction of
-3,000 cm-1 was applied as described in the text, correspond-
ing to þ12 nm and þ20 nm at 200 and 260 nm, respectively. For
each compound, all CD bands calculated above 190 nm had
energies below the estimated ionization potential threshold and
were mostly due to transitions involving virtual orbitals with
negative eigenvalues.44
Computational Section
Conformational searches with MMFF, geometry optimiza-
tions, torsional energy scans, and single-point calculations with
DFT (B3LYP/6-31G(d), 6-311G(d,p), and 6-311Gþ(d,p)) were
run with Spartan’06 and Spartan’09 (WaveFunction, Inc.,
Irvine, CA) using default parameters and convergence criteria.
DFT calculations including solvation model (IEF-PCM, with
parameters for acetonitrile) and TDDFT calculations were run
with Gaussian’03 and Gaussian’09 (Gaussian, Inc., Wallingford,
CT)42 using various functionals and basis sets as described
in the text. The number of transitions included in TDDFT
calculations was 16 for SVP basis set, 16-24 for TZVP, and
24 for aug-TZVP. This latter was constructed by augmenting
the TZVP basis set with a (1s1p1d/1s1p) set of primitive
Acknowledgment. Professor Francesco Naso and Professor
Carlo Rosini are gratefully thanked for useful discussions. This
work was financially supported by the MIUR, Rome (PRIN
Project: Stereoselezione in sintesi organica. Metodologie e
applicazioni; FIRB project: RBPR05NWWC) and by the
University of Bari.
(42) (a) Frisch, M. J. et al. Gaussian 03, Revision D.01; Gaussian, Inc.:
Wallingford, CT, 2004. (b) Frisch, M. J. et al. Gaussian 09, Revision A.02;
Gaussian, Inc.: Wallingford, CT, 2009. (Complete references given in Supporting
Information.)
Supporting Information Available: Experimental and com-
putational data, complete ref 42, and Cartesian coordinates for
all DFT-optimized structures. This material is available free of
charge via the Internet at http://pubs.acs.org.
(43) Kendall, R. A.; Dunning, T. H.; Harrison, R. J. J. Chem. Phys. 1992,
96, 6796–806.
(44) Casida, M. E.; Jamorski, C.; Casida, K. C.; Salahub, D. R. J. Chem.
Phys. 1998, 108, 4439–4449.
1154 J. Org. Chem. Vol. 75, No. 4, 2010