Barriers to stereoinversion of N-aryl-1,3,2-benzodithiazole 1-oxides studied by dynamic enantioselective liquid chromatography

Article abstract of DOI:10.1021/jo981885o

Free energies of activation for the enantiomerization of a series of racemic N-aryl-1,3,2-benzodithiazole 1-oxides have been determined by dynamic high-performance liquid chromatography (DHPLC) on a chiral stationary phase. From a comparison of experiment

Full text of DOI:10.1021/jo981885o

J . Org. Chem. 1999, 64, 1483-1486
1483
Ba r r ier s to Ster eoin ver sion of N-Ar yl-1,3,2-ben zod ith ia zole
1-Oxid es Stu d ied by Dyn a m ic En a n tioselective Liqu id
Ch r om a togr a p h y
J oakim Oxelbark and Stig Allenmark*
Department of Chemistry, Go¨teborg University, SE-41296 Go¨teborg, Sweden
Received September 16, 1998
Free energies of activation for the enantiomerization of a series of racemic N-aryl-1,3,2-
benzodithiazole 1-oxides have been determined by dynamic high-performance liquid chromatography
(DHPLC) on a chiral stationary phase. From a comparison of experimental and computer-simulated
chromatograms, the barriers to stereoinversion at sulfur were found to be around 80 kJ /mol and
relatively insensitive to effects from substituents in the N-aryl group. Throughout the series the
(+)-forms (436 nm) were found to be of (S)-configuration.
Ch a r t 1
In tr od u ction
In our continuing investigation of the configurational
stability and chiroptical properties of N-substituted 1,3,2-
benzodithiazole 1-oxides, we have found that the N-
benzyl-substituted compound (1a ) exhibits a remarkably
low barrier to stereoinversion at sulfur of about 95 kJ /
mol.1 Changes in the ring system were found to raise this
barrier quite dramatically. The low stereointegrity was
first observed by Klein et al.,2 who synthesized com-
pounds of this type for studies of their antifungal activity.
Because sulfoxides have been shown to enantiomerize via
several different pathways, we have been interested in
determining the mechanism of this stereoinversion. Usu-
ally, sulfoxides enantiomerize by a pyramidal inversion
mechanism, the barrier being about 160 kJ /mol.3 2,5-Di-
tert-octylthiophene sulfoxide, however, the most stereo-
labile (although achiral) sulfoxide prepared, is known to
undergo pyramidal inversion, with a barrier of only 60
kJ /mol.4 This process has been suggested to be facilitated
by aromatic stabilization of the pseudoplanar transition
state. Aryl sulfinamides5 and benzylic sulfoxides,6 on the
other hand, have been shown to enantiomerize by a
radical homolysis mechanism, and allylic sulfenates7 via
a reversible sigmatropic rearrangement.
them suitable for studies by dynamic HPLC (DHPLC), a
technique8 similar to dynamic NMR (DNMR) although
applicable within a different time scale. The shape of the
chromatogram of the resolved racemate is dependent on
the interconversion rate of the enantiomers. At a certain
temperature the racemate will collapse into a single
broadened peak due to the fast enantiomerization. This
process is possible to computer simulate, and by com-
parison with experimental data, the rate constants of the
interconversion process can be determined. DHPLC has
been shown to produce results in good agreement with
those obtained by independent techniques,9 except for the
rare case when the stationary phase shows catalytic
activity.10 The rapid enantiomerization of compounds
1b-e makes isolation of the enantiomers very difficult,
and DHPLC thus proved to be a simple and effective way
to determine their barriers to inversion. In this work
DNMR was excluded because of the high temperatures
needed and the limited stability of the analytes.
To gain further insight into this problem, a series of
N-phenyl-substituted 1,3,2-benzodithiazole 1-oxides (1b-
d ), Chart 1, was synthesized. These compounds were
thought to permit an extended charge delocalization in
the transition state, or to stabilize a possible diradical,
thus increasing the enantiomerization rate. The low
barrier to stereoinversion of these compounds makes
Resu lts a n d Discu ssion
The barriers to stereoinversion of 1b-e were deter-
(1) (a) Allenmark, S.; Oxelbark, J . Enantiomer 1996, 1, 13-22. (b)
Allenmark, S.; Oxelbark, J . Chirality 1997, 9, 638-642.
(2) Klein, L. L.; Yeung, C. M.; Weissing, D. E.; Lartey, P. A.; Tanaka,
K.; Plattner, J . J .; Mulford, D. J . J . Med. Chem. 1994, 37, 572-578.
(3) Rayner, D. R.; Gordon, A. J .; Mislow, K. J . Am. Chem. Soc. 1968,
90, 4854-4860.
mined to around 80 kJ /mol using DHPLC as shown in
(8) (a) Bu¨rkle, W.; Karfunkel, H.; Schurig, V. J . Chromatogr. 1984,
288, 1-14. (b) J ung, M.; Schurig, V. J . Am. Chem. Soc. 1992, 114, 529-
534. (c) Veciana, J .; Crespo, M. I. Angew. Chem., Int. Ed. Engl. 1991,
30, 74-76. (d) Eiglsperger, A.; Kastner, F.; Mannschreck, A. J . Mol.
Struct. 1985, 126, 421-432. (e) Mannschreck, A.; Andert, D.; Eigl-
sperger, A.; Gmahl, E.; Buchner, H. Chromatographia 1988, 25, 182-
188. (f) Mannschreck, A.; Zinner, H.; Pustet, N. Chimia 1989, 43, 165-
166.
(4) (a) Mock, W. L. J . Am. Chem. Soc. 1970, 92, 7610-7612. (b)
J enks, W. S.; Matsunaga, N.; Gordon, M. J . Org. Chem. 1996, 61,
1275-1283.
(5) Booms, R. E.; Cram, D. J . J . Am. Chem. Soc. 1972, 94, 5438-
5446.
(6) Miller, E. G.; Rayner, D. R.; Thomas, H. T.; Mislow, K. J . Am.
Chem. Soc. 1968, 90, 4861-4868.
(7) Bickart, P.; Carson, F. W.; J acobus, J .; Miller, E. G.; Mislow, K.
J . Am. Chem. Soc. 1968, 90, 4869-4876.
(9) Gasparrini, F.; Lunazzi, L.; Misiti, D.; Villani, C. Acc. Chem. Res.
1995, 28, 163-170.
(10) Schurig, V.; Leyrer, U. Tetrahedron: Asymmetry 1990, 1, 865-
868.
10.1021/jo981885o CCC: $18.00 © 1999 American Chemical Society
Published on Web 02/13/1999

1484 J . Org. Chem., Vol. 64, No. 5, 1999
Oxelbark and Allenmark
Ta ble 2. Selected Ch r om a togr a p h ic Da ta a
Ta ble 1. Resu lts of DHP LC Sim u la tion s
compd
∆Gq a/kJ mol-1
temp/Kb
runsc
range/Kd
compd
k1′
k2′
Rb
elutionc
abs configd
app
1b
1c
1d
1e
81.0
79.5
81.1
80.2
274.4
263.8
256.5
261.6
10
7
3
11
11
4
1b
1c
1d
1e
1.58
1.73
1.77
2.12
3.85
2.69
2.00
2.81
2.44
1.56
1.13
1.33
(-) (+)
(-) (+)
(+) (-)
(+) (-)
(R)
(R)
(S)
(S)
12
13
a
a
Gibb’s free energy calculated from the mean of kfapp and kr
.
Column: (3S,4R) Whelk-O1 (200 × 4.6 mm). Flow rate: 1.0
app
b Mean temperature. c Number of simulated chromatograms. Tem-
perature range of the experimental chromatograms.
mL/min. Mobile phase: 45% hexane in dichloromethane with 1%
d
methanol added. Temperature: -18 °C. R ) k2′/k1′. c Elution
b
d
order determined at 436 nm. Absolute configuration of the first
eluted enantiomer.
F igu r e 1. Experimental (‚‚‚) and simulated (s) chromato-
grams of 1b at -10 °C (Whelk-O1, 45% hexane and 1%
methanol in dichloromethane, 0.5 mL/min). Parameters
used: tM ) 5.2 min; N ) 3800.
F igu r e 2. CD spectra of (1S,3S)-2a and the late-eluting
enantiomers11 of 2b-e in acetonitrile (arbitrary concentrations
used).
Sch em e 1
mers and of (1S,3S)-2a ,1a from which it is evident that
they all have the same absolute configuration. This
implies that compounds (+)-1b-e are all of the (S)-
configuration. The previously reported1a (R)-configuration
of (+)-1a refers to the optical rotation at 302 nm, which
is opposite in sign to that obtained at 436 nm. In the
present work all optical rotations were measured at 436
nm.
The bis(oxide)s obtained were composed of 80-95% of
the meso diastereomers. Therefore, only small amounts
of the enantiomers were prepared, except for 2d , of which
NMR and mass spectra of the respective diastereomers
were recorded. Since the oxidation state of the ring
system affects the position of the CD band at the longest
wavelength quite significantly1a and the spectra of 2d and
the other isolated bis(oxide)s in the series (2b,c,e) show
great similarities, there is no reason to doubt the identity
of the latter, even though the amounts obtained were not
sufficient for NMR analysis.
The low separation factor of 1d is well explained by
the reversal of the elution order. The chromatographic
retention is influenced by the substituents in such a way
that k′R increases with the π-accepting ability of the aryl
substituent, which is in accordance with what has been
found1b for the different analogues of 1a .
Table 1. The effect from the substituents on the phenyl
ring on the enantiomerization rate is clearly within the
error limit of the method. No trend can be distinguished
in the ∆Gqappvalues given in Table 1. Extrapolation of the
data to 260 K results in ∆Gq
values which should
app(260)
be comparable. This extrapolation does not change the
relative order of the ∆Gqappvalues. The fit of the simulated
chromatograms differs slightly among the different ana-
lytes but is generally quite satisfactory, as can be seen
in Figure 1. Best fits were obtained for 1b,c because they
exhibit somewhat better peak shapes, i.e., tailing is less
pronounced. A general observation is that the height of
the second enantiomer peak in the simulated chromato-
grams is slightly lower than expected. This might be
caused by the different number of theoretical plates found
for the first and second enantiomers in the experimental
chromatograms. Despite this slight mismatch, however,
the experimental and simulated half-height widths seem
to be almost identical. The equilibria8a involved in the
overall enantiomerization process are shown in Scheme
1.
Chromatographic data for 1b-e are given in Table 2.
Polarimetric detection at 436 nm of the eluted enantio-
mers indicated a reversal of elution order in the series.
The reversed elution order was confirmed by oxidizing
enantiomerically enriched 1b-e. It was found that
oxidation of (+)-1b-e consistently yielded an excess of
the last eluted enantiomer11 of 2b-e. Figure 2 shows the
CD spectra in acetonitrile of these last eluted enantio-
Exp er im en ta l Section
Dyn a m ic HP LC. Low-temperature HPLC12 was carried out
using a 5µ (3S,4R) Whelk-O1 sorbent13 (Regis Technologies,
Inc., Morton Grove, IL) in a 200 × 4.6 mm column immersed
in a thermostated cooling bath. The column was connected to
the injector by a steel capillary of ca. 1 m length, also immersed
in the cooling bath. When the temperature was changed, the
(11) The column used was a 200 × 4.6 mm Kromasil CHI-DMB
(EKA Chemicals Co., Bohus, Sweden). The mobile phase was composed
of 15% ethyl acetate in cyclohexane, except for 2e, for which neat tBME
was used. Ethyl acetate was added to the solution injected, to increase
the solubility.

Barriers to Stereoinversion of Benzodithiazole Oxides
J . Org. Chem., Vol. 64, No. 5, 1999 1485
system was left to equilibrate thermally for 20 min prior to
chromatography. The mobile phase was composed of 45%
n-hexane and 1% methanol in dichloromethane. UV detection
was carried out at 240 nm.
Sim u la tion of Ch r om a togr a m s. Simulations of the ex-
perimental chromatograms were performed with a slightly
modified14 version of the SIMUL15 program, which is based
on the discontinuous plate model.16 Plate number, retention
times, and the void time of the column are determined
experimentally. These figures together with rate constants for
the stationary phase has been determined to be up to 5 kJ /
mol higher than in the mobile phase.14b This study was
performed using the Whelk-O1 phase and atropisomeric naph-
thamides as analytes. The result can be rationalized by the
assumption that in the analyte/selector complex the transition
state cannot be reached unless a hydrogen bond is partially
broken.
Assign m en t of Absolu te Con figu r a tion . The enantio-
mers of the monoxides 1b-d were collected using enantiose-
lective HPLC at -20 °C and immediately poured into a cooled
(-20 °C) and dried solution of 1 equiv mCPBA in dichloro-
methane. After 20 min at this temperature the reaction
mixture was worked up as fast as possible to prevent hydroly-
sis. Drying and solvent evaporation gave the bis(oxide)s 2b-
e. Each of these were analyzed by enantioselective chroma-
tography11 with respect to the location of the predominating
enantiomer of the trans isomer. The enantiomer present in
excess was isolated and its CD spectrum recorded. The CD
spectra of these enantiomers of 2b-e were then compared with
the spectrum of (1S,3S)-2a in acetonitrile. Compound 2d was
prepared on a larger scale, allowing for NMR and MS analysis
of the meso and racemic form.
2-Ar yl-1,3,2-ben zod ith ia zoles: Gen er a l P r oced u r e. A
solution of the appropriate aniline (8 mmol) and triethylamine
(16 mmol, 1.62 g) in 20 mL of dry THF was added dropwise to
a cooled (-70 °C) and stirred solution of benzene-1,2-disulfenyl
chloride (8 mmol, 1.69 g) in 120 mL of dry THF under nitrogen.
The solution was left to thaw slowly (a few hours) and then
kept overnight at room temperature. The solution was filtered
and evaporated under reduced pressure to yield a black, tarry
oil. Filtration through silica as a chloroform solution was
followed by flash chromatography with 30% chloroform in
hexane as the eluent.
2-(3,4,5-Tr im eth oxyp h en yl)-1,3,2-ben zod ith ia zole: 1H
NMR (400 MHz, CDCl3) δ 3.75 (s, 3H), 3.80 (s, 6H), 6.62 (s,
2H), 7.15-7.20 (m, 2H), 7.36-7.41 (m, 2H); 13C NMR (101
MHz, CDCl3, 1H-decoupled) δ 56.02, 60.82, 97.83, 120.50,
126.49, 135.13, 138.04, 150.39, 152.69.
the reaction in the mobile and stationary phases, kmob and kstat
,
are used as input parameters. Rate constants are then changed
until the simulated and experimental chromatograms show
the best possible correspondence.
Determination of the plate number for the simulation turned
out to be the crucial step in determining the reaction barrier.
A change of (500 plates causes only minor differences in peak
width but is associated with a change in plateau height
corresponding to a change in rate of up to (10%. This should
be compared with the error in rate arising from the uncertainty
in fitting the height of the plateau, which is less than (5%.
Use of the correct experimental plate number, i.e., eq 1,
would produce too narrow peaks. This is due to the fact that
2
N ) 5.54(tR2/wh
)
(1)
(2)
2
N ) 5.54(tR(tR - tM)/wh
)
the discontinuous plate model does not take analyte diffusion
into account. Therefore, as suggested in the program user’s
manual,15 the plate number used for the simulation was
determined according to eq 2, where tR is the retention time
of the analyte, tM is the void time, and wh is the half-height
peak width.
Still, this determination is to some extent subjective, since
the tailing and fronting experimental chromatograms are
different from the Gaussian-shaped simulated ones, and what
is considered a best fit is somewhat arbitrary.
2-(4-Meth oxyp h en yl)-1,3,2-ben zod ith ia zole: 1H NMR
(400 MHz, CDCl3) δ 3.72 (s, 3H), 6.70-6.74 (m, 2H), 7.14-
7.19 (m, 2H), 7.22-7.27 (m, 2H), 7.35-7.40 (m, 2H); 13C NMR
(101 MHz, CDCl3, 1H-decoupled) δ 55.43, 113.65, 120.15,
121.92, 126.37, 137.96, 148.16, 157.03.
The strategy was to determine the plate number of a few
chromatograms with a plateau of about 10-40% of the peak
height and to calculate a mean. This plate number was then
used for the simulation of chromatograms with the same
analyte and flow rate but at different temperatures. To permit
a precise determination of the rate constants, the plateau must
exceed 10% of the peak height. In principle, it would also be
possible to determine the activation parameters of the reaction.
In this particular case, however, the temperature range
spanning the coalescence of the peaks is too narrow.
Rate constants in the stationary phase were arbitrarily set
to 1/10 of the rate constants in the mobile phase. The shape of
the simulated chromatogram is determined by the apparent
rate constants of the respective directions (kfapp and krapp). kfapp
2-P h en yl-1,3,2-ben zod ith ia zole: 1H NMR (400 MHz,
CDCl3) δ 6.94-7.00 (m, 1H), 7.03-7.08 (m, 2H), 7.13-7.19 (m,
2H), 7.26-7.30 (m, 2H), 7.33-7.37 (m, 2H); 13C NMR (101
1
MHz, CDCl3, H-decoupled) δ 120.16, 120.34, 124.69, 126.50,
128.68, 138.19, 154.18.
2-(4-Nitr op h en yl)-1,3,2-ben zod ith ia zole: 1H NMR (400
MHz, CDCl3) δ 7.18-7.22 (m, 2H), 7.40-7.44 (m, 2H), 7.52-
7.56 (m, 2H), 8.08-8.13 (m, 2H); 13C NMR (101 MHz, CDCl3,
1H-decoupled) δ 119.21, 120.46, 124.35, 126.82, 131.50, 143.65,
159.12.
and kr
are nonequivalent due to the presence of the chiral
app
2-Ar yl-1,3,2-b en zod it h ia zole S-Oxid es. The oxidations
selector and can be expressed as the kmob and kstat mean values
weighted by the residence time in each phase.8c Because there
is an infinite number of pairs of kstat and kmob yielding the same
were carried out as described previously.1b
2-(3,4,5-Tr im eth oxyp h en yl)-1,3,2-ben zod ith ia zole 1-ox-
id e (1b): 1H NMR (400 MHz, CDCl3) δ 3.85 (s, 3H) 3.87 (s,
6H), 6.74 (s, 2H), 7.36-7.43 (m, 1H), 7.53-7.57 (m, 2H), 7.86-
7.90 (m, 1H); 13C NMR (101 MHz, CDCl3, 1H-decoupled) δ
56.23, 60.90, 101.84, 119.39, 125.73, 126.15, 131.43, 136.49,
136.89, 141.50, 141.91, 153.58; MS (FAB) m/e 337 (M), 338
(M + 1).
k
k
app, they can either be set equal or to a fixed ratio, unless
mob is known from independent experiments. If, however, kmob
is found by an independent technique, kstat can be obtained
from the kapp values determined. To set kstat lower than kmob is
justified by the fact that the barrier to enantiomerization in
2-(4-Meth oxyph en yl)-1,3,2-ben zodith iazole 1-oxide (1c):
1H NMR (400 MHz, CDCl3) δ 3.80 (s, 3H) 6.86-6.91 (m, 2H),
7.32-7.38 (m, 1H), 7.39-7.44 (m, 2H), 7.49-7.53 (m, 2H),
7.83-7.87 (m, 1H); 13C NMR (101 MHz, CDCl3, 1H-decoupled)
δ 55.51, 114.56, 119.27, 125.65, 126.00, 126.81, 131.24, 133.25,
142.13, 142.21, 158.76; MS (FAB) m/e 277 (M), 278 (M + 1).
2-P h en yl-1,3,2-ben zod ith ia zole 1-oxid e (1d ): 1H NMR
(400 MHz, CDCl3) δ 7.06-7.12 (m, 1H), 7.20-7.28 (m, 3H),
7.32-7.44 (m, 4H), 7.71-7.75 (m, 1H); 13C NMR (101 MHz,
CDCl3, 1H-decoupled) δ 119.62, 123.12, 125.89, 126.28, 126.38,
129.76, 131.62, 141.17, 141.55, 141.97; MS (FAB) m/e 247 (M),
248 (M + 1).
(12) Temperatures down to -80 °C have been used successfully.
See: (a) Wolf, C.; Pirkle, W. H.; Welch, C. J .; Hochmut, D. H.; Ko¨nig,
W. A.; Chee, G.-L.; Charlton, J . L. J . Org. Chem. 1997, 62, 5208-5210.
(b) Villani, C.; Pirkle, W. H. Tetrahedron: Asymmetry 1995, 6, 27-
30.
(13) (a) Formerly wrongly designated as (R,R). (b) Pirkle, W. H.;
Brice, L. J .; Widlanski, T. S.; Roestamadji, J . Tetrahedron: Asymmetry
1996, 7, 2173-2176.
(14) (a) Only user interface modifications. (b) Gasparrini, F.; Misiti,
D.; Pierini, M.; Villani, C. Tetrahedron: Asymmetry 1997, 8, 2069-
2073.
(15) (a) QCPE Program No. 620. (b) J ung, M. QCPE Bull. 1992, 12,
52.
(16) See ref 8b and references therein.

1486 J . Org. Chem., Vol. 64, No. 5, 1999
Oxelbark and Allenmark
2-(4-Nitr op h en yl)-1,3,2-ben zod ith ia zole 1-oxid e (1e):
1H NMR (400 MHz, CDCl3) δ 7.42-7.48 (m, 1H), 7.54-7.66
(m, 4H), 7.92-7.96 (m, 1H), 8.26-8.31 (m, 2H); 13C NMR (101
(101 MHz, CDCl3, 1H-decoupled) δ 126.66, 128, 23, 129.06,
130.01, 133.24, 136.55, 146.10.
1
MHz, CDCl3, H-decoupled) δ 119.71, 119.89, 125.61, 126.13,
Ack n ow led gm en t . We thank Prof. F. Gasparrini
and Prof. D. Misiti for placing their simulation program
and computer facilities at our disposal. Special thanks
are due to Dr. C. Villani for the kind help during the
stay of J .O. in Rome. This work was supported by a
grant (K-AA/KU 02508-321) from the Swedish Natural
Science Research Council (NFR). A travel grant (to J .O.)
within the Swedish-Italian exchange program of NFR
is also gratefully acknowledged.
126.74, 132.32, 140.06, 140.86, 143.94, 147.17; MS (FAB) m/e
292 (M), 293 (M + 1).
tr a n s-2-P h en yl-1,3,2-ben zodith iazole 1,3-dioxide (tr a n s-
2d ): 1H NMR (400 MHz, CDCl3) δ 7.45-7.53 (m, 3H) 7.61-
7.67 (m, 2H), 7.82-7.87 (m, 2H), 7.96-8.01 (m, 2H); 13C NMR
(101 MHz, CDCl3, 1H-decoupled) δ 126.52, 129.38, 129.44,
129.94, 133.74, 135.13, 146.01.
cis-2-P h en yl-1,3,2-b en zod it h ia zole 1,3-d ioxid e (cis-
2d ): 1H NMR (400 MHz, CDCl3) δ 7.43-7.53 (m, 3H) 7.55-
7.67 (m, 2H), 7.82-7.87 (m, 2H), 8.00-8.06 (m, 2H); 13C NMR
J O981885O