The generalized anomeric effect in the 1,3-thiazolidines: Evidence for both sulphur and nitrogen as electron donors. Crystal structures of various N-acylthiazolidines including mercury(II) complexes. Possible relevance to penicillin action

Article abstract of DOI:10.1016/j.molstruc.2006.10.034

Evidence for the generalized anomeric effect (GAE) in the N-acyl-1,3-thiazolidines, an important structural motif in the penicillins, was sought in the crystal structures of N-(4-nitrobenzoyl)-1,3-thiazolidine and its (2:1) complex with mercuric chloride,

Full text of DOI:10.1016/j.molstruc.2006.10.034

Journal of Molecular Structure 837 (2007) 118–131
www.elsevier.com/locate/molstruc
The generalized anomeric effect in the 1,3-thiazolidines: Evidence
for both sulphur and nitrogen as electron donors. Crystal structures
of various N-acylthiazolidines including mercury(II) complexes.
Possible relevance to penicillin action
a,
b
a
b,
*
Sosale Chandrasekhar *, Deepak Chopra , Kovuru Gopalaiah , Tayur N. Guru Row
a
Department of Organic Chemistry, Indian Institute of Science, Bangalore 560 012, India
Solid State and Structural Chemistry Unit, Indian Institute of Science, Bangalore 560 012, India
b
Received 5 May 2006; received in revised form 2 October 2006; accepted 7 October 2006
Available online 13 December 2006
Abstract
Evidence for the generalized anomeric effect (GAE) in the N-acyl-1,3-thiazolidines, an important structural motif in the penicillins,
was sought in the crystal structures of N-(4-nitrobenzoyl)-1,3-thiazolidine and its (2:1) complex with mercuric chloride, N-acetyl-2-phen-
yl-1,3-thiazolidine, and the (2:1) complex of N-benzoyl-1,3-thiazolidine with mercuric bromide. An inverse relationship was generally
observed between the C2–N and C2–S bond lengths of the thiazolidine ring, supporting the existence of the GAE. (Maximal bond length
˚
˚
changes were ꢀ0.04 A for C2–N3, S1–C2, and ꢀ0.08 A for N3–C6.) Comparison with N-acylpyrrolidines and tetrahydrothiophenes indi-
cates that both the nitrogen-to-sulphur and sulphur-to-nitrogen GAE’s operate simultaneously in the 1,3-thiazolidines, the former being
dominant. (This is analogous to the normal and exo-anomeric effects in pyranoses, and also leads to an interesting application of Bald-
win’s rules.) The nitrogen-to-sulphur GAE is generally enhanced in the mercury(II) complexes (presumably via coordination at the sul-
phur); a ‘competition’ between the GAE and the amide resonance of the N-acyl moiety is apparent. There is evidence for a ‘push–pull’
charge transfer between the thiazolidine moieties in the mercury(II) complexes, and for a ‘back-donation’ of charge from the bromine
atoms to the thiazolidine moieties in the HgBr2 complex. (The sulphur atom appears to be sp2 hybridised in the mercury(II) complexes,
possibly for stereoelectronic reasons.) These results are apparently relevant to the mode of action of the penicillins.
ꢀ 2006 Elsevier B.V. All rights reserved.
Keywords: Amide resonance; Anomeric effect; Baldwin’s rules; Crystal structure; b-Lactam; Mercury(II); Penicillin; Stereoelectronic; Thiazolidine; X-ray
diffraction
1. Introduction
both theoretical and practical reasons. Theoretically,
because of the possibility of two opposing modes for
the GAE: the sulphur as the donor with the nitrogen as
the acceptor, or vice versa; and practically, because of
the presence of the N-acyl-1,3-thiazolidine moiety in the
penicillin antibiotics [4,5]. (The role of this moiety – if
any – in penicillin action has not been clarified.)
The present structural study was thus undertaken in the
hope that bond length changes would provide evidence for
the sought after GAE – an approach that was very success-
ful in pyranose systems [1]. There is also interesting
precedent, crystal structures of well over a hundred
The ‘generalized anomeric effect’ (GAE) is a stereoelec-
tronic term which refers to the preference for an antiperi-
planar orientation of a carbon-heteroatom bond relative
to an electron pair on another geminally situated hetero-
atom [1–3]. The possible existence of the GAE in the
N-acyl-1,3-thiazolidines (I, Scheme 1) is interesting for
*
Corresponding authors. Tel.: +91 80 2293 2689; fax: +91 80 2360 0529.
E-mail addresses: sosale@orgchem.iisc.ernet. in, ssctng@sscu.iisc.ernet.
in (S. Chandrasekhar).
0022-2860/$ - see front matter ꢀ 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.molstruc.2006.10.034

S. Chandrasekhar et al. / Journal of Molecular Structure 837 (2007) 118–131
119
R
R
O
O
O
O
1-5
X
R'
R
R
O
N
N
1
2
3
4
5
-
-
H
Ph
H
H
H
p-(NO2)-C6H5-
Me
p-(NO2)-C6H5-
Ph
Ph
N
R'
S
S
HgCl2
HgBr2
-
S
I
Ia
X0.5
R
6
1-5
O
R
R
O
N
3
Scheme 2. N-Acyl-1,3-thiazolidine derivatives prepared and studied in
this work, via X-ray diffraction determination of the crystal structures
(except 5). The indicated stoichiometry of 0.5 for ‘X’ above implies that
two molecules of the thiazolidine bind to one of the mercuric halide in the
case of 3 and 4 (cf. Schemes 4 and 5 below for the full structures).
4
5
2
N
N
S
1
S
S
I
Ib
I
R
R
O
ray diffraction analysis. Four analogs were so studied (1–
4, Scheme 2), two being mercury(II) complexes (3 and 4)
as these were expected to indicate how electrophilic coordi-
nation at the sulphur atom would modulate the GAE.
In the N-acyl-1,3-thiazolidines the GAE would act
alongside the amide resonance of the N-acyl moiety: either
in concert (sulphur as donor) or in opposition (nitrogen as
donor). These two GAE modes may be conveniently repre-
sented by the no-bond canonical forms Ia and Ib, respec-
tively, Ic then representing amide resonance (cf. Scheme
1). The above two modes may be termed the ‘sulphur-to-ni-
trogen’ and ‘nitrogen-to-sulphur’ GAE’s, respectively. The
anomeric effect (ꢀ2 kcals/mole) [1–3] is, of course, much
weaker than amide resonance (ꢀ20 kcals/mole) [9]; howev-
er, it is conceivable that electrophilic coordination at the
sulphur centre would enhance the ‘nitrogen-to-sulphur’
GAE in I (cf. Id and Ie), thereby opposing and weakening
the amide resonance.
N
N
S
S
I
Ic
R
R
R
O
O
O
N
E+
N
N
S
S
S
E
E
I
Ie
Id
Scheme 1. The possible modes of the generalized anomeric effect (GAE)
and competing charge transfers in N-acyl-1,3-thiazolidines I, represented
by the canonical forms Ia–Ie: sulphur-to-nitrogen GAE (Ia), nitrogen-to-
sulphur GAE (Ib), amide resonance (Ic), attack of electrophile (‘E’) at
sulphur (Id and Ie). The boxed structure shows the numbering scheme
employed for I.
Other interesting questions pertain to whether or not the
amide nitrogen atom in I would be pyramidalised by the
nitrogen-to-sulphur GAE, and to the conformation of the
thiazolidine ring. In fact, the above modes of charge trans-
fer could possibly be of importance in the mode of action
of the penicillins [5], as will be discussed later.
The principles of stereoelectronic theory, particularly
as apply to the ‘antiperiplanar lone pair hypothesis’
(ALPH) have been the subject of numerous reviews [1–
3], and will not be elaborated upon herein. A basic famil-
iarity with these principles is assumed in this paper, par-
ticularly the idea of the overlap of electron lone pairs
with adjacent and suitably aligned antibonding (r*) orbi-
tals. (Thus, in the nitrogen-to-sulphur GAE the lone pair
of electrons on the nitrogen atom would overlap with the
C2–S r* orbital; in the sulphur-to-nitrogen GAE a lone
pair on the sulphur atom would overlap with the C2–N
r* orbital.)
1,3-thiazolidine derivatives having been reported (although
no N-acyl derivatives, apparently). A random sampling of
these reports [6–8] indicates a general inverse relationship
between the C–N and C–S bond lengths of the thiazolidine
moiety (although with exceptions, cf. Table 1). This is sig-
nificant as analogous trends in the pyranoses have been
considered to evidence the anomeric effect [1]. Of particular
interest in the present study was the effect of the N-acyl
moiety on these trends.
Simple, readily prepared and crystalline N-acyl-1,3-
thiazolidine analogs were thus sought and subjected to X-
Table 1
˚
Typical values of the relevant anomeric bond lengths (in A) in substituted
1,3-thiazolidine derivatives possibly evidencing the GAEa
Derivative
C2–N3
S1–C2
Tricyclic penamb
4-Carboxylic acidc
Octahydrobenzothiazoleb
4-Carboxylic acidc
2-Phenyld
1.474 (5)
1.463 (7)
1.441 (4)
1.484 (7)
1.454 (2)
1.810 (4)
1.852 (8)
1.868 (3)
1.869 (6)
1.882 (2)
2. Experimental
a
From a random sampling of reported crystal structure data (standard
deviations in parenthesis).
2.1. General remarks
b
Ref. [6].
Ref. [7].
c
Organic compounds were prepared and characterized
with the following instruments: JASCO FT/IR 410 IR
d
Ref. [8].

120
S. Chandrasekhar et al. / Journal of Molecular Structure 837 (2007) 118–131
Spectrometer, JEOL JNM-LA 300 FT NMR System (1H
at 300 MHz and 13C at 75 MHz), Micromass Q-TOF
AMPS MAX 10/6A Mass spectrometer. NMR spectra
were recorded in CDCl3 solution with tetramethylsilane
as internal standard (unless stated otherwise), and IR spec-
tra as stated. The X-ray diffraction studies are described
separately below.
1,3-Thiazolidine and its 2-phenyl derivative were pre-
pared from 2-aminoethanethiol and the appropriate alde-
hyde by reported procedures [10,11], and N-acylated to
obtain 1, 2 and 5 as described below [12] (cf. Scheme 2);
these were converted to their mercury(II) complexes 3
and 5, respectively, [13]. Interestingly, all these amide deriv-
atives exhibited hindered rotation around the amide C–N
double bond and gave rise to twin resonances in the
NMR. (These were generally broad multiplets, ‘br m’, as
indicated below in the NMR data; note that the total inte-
gration value of the twin peaks is given.) Variable temper-
ature NMR studies in two cases (1 and 5) were also
performed as described further below.
22.7 and 23.1 (CO-Me), 28.9 and 30.4 (C5), 50.2 and 50.7
(C4), 64.0 and 65.0 (C2), 125.1 and 125.7 (phenyl C4),
127.5 and 128.0 (phenyl C2, C6), 128.3 and 128.8 (phenyl
C3, C5), 141.4 and 141.7 (phenyl C1), 168.3 and 169.1
(C@O); m/e 207 (M+), 179 (100%), 164, 118; HRMS: calcd.
for C11H13NOSNa (M + Na) 230.0616 (Found 230.0621).
N-Benzoyl-1,3-thiazolidine (5) was obtained from the
thiazolidine and benzoyl chloride in 95% yield as a color-
less liquid. mmax/cmÀ1 (thin film)1634, 1576, 1402; dH 3.01
(2H, br m, C5–H), 3.75 and 3.96 (2H, 2 · br m, C4–H),
4.49 and 4.73 (2H, 2 · br m, C2–H), 7.33–7.51 (5H, m,
Ar-H); dC 30.4 (C5), 48.0 (C4), 51.3 (C2), 126.9 (phenyl
C2, C6), 128.1 (phenyl C3, C5), 130.1 (phenyl C4), 135.9
(phenyl C1), 169.3 (C@O); m/e 193 (M+), 165, 105, 83
(100%); HRMS: calcd. for C10H11NOSNa (M + Na)
216.0459 (Found 216.0463).
2.3. Preparation of the mercury(II) complexes 3 and 4
These were prepared from 1 and 5 by a modification of a
reported procedure [13]. Thus, a stirred solution of the
nitrobenzoylthiazolidine 1 (3.0 mmol) in ethanol (15 ml)
was treated with a warm ethanolic solution of HgCl2 (0.2
M, 15 ml), when a white precipitate resulted. The mixture
was allowed to cool and filtered to collect the precipitate,
which was washed with ethanol and dried (Na2SO4) in vac-
uo. The solid was recrystallised from acetone to obtain the
pure complex 3 as colourless crystals (2.0 mmol, 66%); mp
110–114 ꢁC; mmax/cmÀ1 (KBr) 1650, 1634, 1596, 1514, 1416,
1345, 844; dH (acetone-d6) 2.93 and 3.12 (2H, 2 · br m, C5–
H), 3.81 and 3.93 (2H, 2 · br m, C4–H), 4.59 and 4.72 (2H,
2 · br m, C2–H), 7.85 (2H, d, J = 8.4 Hz, Ar-H ortho to
C@O), 8.34 (2H, d, J = 8.4 Hz, Ar-H ortho to NO2); dC
48.5 and 49.2 (C5), 51.3 and 51.4 (C4), 52.1 and 52.2
(C2), 124.4 (Caryl ortho to C@O), 129.4 (Caryl ortho to
NO2), 143.7 (Caryl at C@O), 149.5 (Caryl at NO2), 167.7
(C@O).
2.2. Preparation of the N-acyl-1,3-thiazolidines (general
procedure)
A stirred solution of the thiazolidine (5.0 mmol) in dry
CH2Cl2 (6.0 ml), at 0 ꢁC and under nitrogen, was treated
with a solution of the acid chloride (or the anhydride)
(5.2 mmol) in dry CH2Cl2 (6.0 ml) over 0.25 h. The reac-
tion mixture was maintained at 0 ꢁC for an additional
0.5 h and allowed to warm to 25ꢁC. The mixture was then
worked up by diluting with CH2Cl2 and washing with sat-
urated NaHCO3 solution (10 ml) and water (2 · 20 ml).
The organics were dried (Na2SO4) and distilled in vacuo
to remove solvent. The resulting crude residue was purified
by column chromatography over silica gel to obtain the
pure product.
Thus, N-(4-nitrobenzoyl)-1,3-thiazolidine (1) was
obtained from thiazolidine and 4-nitrobenzoyl chloride in
92% yield as a pale yellow solid; mp 71–74 ꢁC; mmax
(cmÀ1) (CHCl3) 1637, 1599, 1521, 1418; dH 3.05 and 3.14
(2H, 2· br m, C5–H), 3.75 and 4.02 (2H, 2· br m, C4–
H), 4.47 and 4.77 (2H, 2· br m, C2–H), 7.73 (2H, d,
J = 8.4 Hz, Ar-H ortho to C@O), 8.29 (2H, d, J = 8.4 Hz,
Ar-H ortho to NO2); dC 29.5 and 30.8 (C5), 48.0 and 48.2
(C4), 50.7 and 51.3 (C2), 123.4 (Caryl ortho to C@O),
127.9 (Caryl ortho to NO2), 141.6 and 142.0 (Caryl at C@O
in the two amide rotamers), 148.3 (Caryl at NO2), 166.8
(C@O); m/e 238 (M+), 210, 150 (100%), 104; HRMS: calcd.
for C10H10N2O3SNa (M + Na) 261.0310 (Found
261.0315).
The mercuric bromide complex 4 was similarly prepared
from N-benzoylthiazolidine (5) and mercuric bromide as a
colourless crystalline solid in 59% yield; mp 140–143 ꢁC;
m
max/cmÀ1 (KBr) 1626, 1575, 1393; dH (acetone-d6) 2.91
and 3.08 (2H, 2 · br m, C5–H), 3.86 (2H, br m, C4–H),
4.64 (2H, br m, C2–H), 7.46–7.58 (5H, m, Ar-H); dC 45.7
and 47.2 (C5), 51.28 and 51.32 (C4), 57.0 and 57.5 (C2),
128.2 (Caryl ortho to C@O), 129.1 (Caryl ortho to NO2),
131.0 (Caryl at C@O), 137.5 (Caryl at NO2), 169.8 (C@O).
2.4. Variable temperature NMR studies of 1 and 5: hindered
rotation around the amide bond [9]
N-Acetyl-2-phenyl-1,3-thiazolidine (2) was obtained
from 2-phenylthiazolidine and acetic anhydride in 89%
yield as a colourless solid; mp 65–68 ꢁC (reported [11]
64.3–64.8 ꢁC); mmax/cmÀ1 (CHCl3) 1650, 1401, 1352, 726;
dH 1.96 and 2.19 (3H, 2 · s, –CO-Me), 3.04–3.17 (2H, m,
C5–H), 3.90–3.99 and 4.25–4.31 (2H, 2 · m, C4–H), 6.01
and 6.50 (1H, 2 · s, C2–H), 7.22–7.39 (5H, m, Ar-H); dC
The N-acylthiazolidines 1–5 exhibited hindered rotation
around the amide C–N bond in the NMR, a pair of signals
being seen for a particular proton or carbon atom in most
1
cases as indicated above. Variable temperature H NMR
studies were performed in the case of the nitrobenzoyl
derivative 1 and the benzoyl derivative 5, and these showed
that the twin resonances for each proton of the thiazolidine

S. Chandrasekhar et al. / Journal of Molecular Structure 837 (2007) 118–131
121
Table 2
ring were considerably broadened in both cases, but much
˚
Relevant bond distances (in A) in the 1,3-thiazolidine derivatives 1–4
determined by X-ray diffraction in this studya
more so in the case of 5, at room temperature. Further-
more, at +50 ꢁC the resonances were sharper in the case
of 5 than in the case of 1. These apparently indicate that
the interconversion of the amide rotamers in 1 is slower
than in 5. Thus, the partial double bond of the amide C–
N moiety has greater double bond character in 1 than in
5, presumably a consequence of the greater electron with-
drawing ability of the 4-nitrophenyl group relative to a
phenyl group. (Both 1 and 5 exhibited very sharp resonanc-
es at À78 ꢁC, indicating that interconversion of the amide
rotamers has almost completely ceased.)
Compound
N3–C6
C6@O
C2–N3
C2–S1
1
1.337(8)
1.354(3)
1.359(4)
1.349(4)
1.355(1)
1.277(2)
1.223(7)
1.224(3)
1.223(4)
1.228(4)
1.218(2)
1.239(2)
1.477(8)
1.458(3)
1.437(5)
1.468(4)
1.441(2)
1.460(2)
1.785(8)
1.832(2)
1.832(5)
1.807(4)
1.787(3)
1.793(3)
2
3a
3b
4a
4b
a
The 2:1 mercury(II) complexes 3 and 4 provide two separate sets of
data for the two different thiazolidine moieties (indicated by ‘a’ and ‘b’, cf.
Schemes 2, 4 and 5); standard deviations are given in parenthesis.
2.5. X-ray diffraction studies
Table 3
˚
Typical carbon-heteroatom bond distances (in A) reported for crystalline
Single crystal X-ray diffraction studies of the N-acyl-
thiazolidine derivatives 1–4 prepared and characterized as
above, were performed on a Bruker AXS SMART APEX
CCD diffractometer using graphite monochromatic Mo
Ka radiation. The data was collected at room temperature
using an Oxford N2 cryosystem. The data was collected
using the package SMART and the data reduction was car-
ried out using SAINTPLUS. An empirical absorption cor-
rection was performed with SADABS. Reduced intensities
were analyzed using XPREP for the space group determi-
nation and merging. All the structures were solved using
SIR92 and refined using SHELXL present in the WinGx
(Version 1.64.05) program suite. All the hydrogen atoms
for these structures were located from the difference Fouri-
er map and refined accordingly. ORTEP diagrams of all
the compounds were generated by ORTEP32. The geomet-
ric calculations were done by PARST95. Full details of
these studies, including the relevant references, have been
deposited in the Cambridge Crystallographic Database
[14].
derivatives of tetrahydrothiophene (C–S) and N-acylpyrrolidine (C–N and
N–CO), as determined by X-ray diffraction
Compound
C–S
C–N
N–CO
Tetrahydrothiophenea
N-Acylpyrrolidineb
1.84
–
–
1.49
1.35
a
Ref. [15].
Ref. [16].
b
one of the points is excluded in most cases (cf. Figs. 5–8
in the ‘Appendix’.) The maximal bond length changes in
˚
˚
˚
1–4 are ꢀ0.04 A for C2–N3 and S1–C2, >0.08 A for N3–
C6 and ꢀ0.017 A for C6@O (cf. Table 2).
The site of complexation in the mercuric complexes 3
and 4 is the sulphur atom in both the cases studied (as
expected). Both 3 and 4 were composed of two molecules
of thiazolidine and one of mercury halide. Also, 3 was
dimeric and pentacoordinated, whereas 4 was monomeric
and tetracoordinated, at the mercury atom. (The difference
in coordination numbers is possibly due to the greater elec-
trophilicity of mercuric chloride relative to mercuric
bromide.)
The relevant crystallographic data of 1–4 have been col-
lected and tabulated in the ‘Appendix’ section, along with a
brief note on the lattice packing characteristics.
In general – although not invariably – the complexes 3
and 4 apparently evidence an increase in the nitrogen-to-
sulphur GAE relative to 1 and 2. Furthermore, both 3
and 4 furnish differing sets of data for the two thiazolidine
ligands in each of them (denoted as ‘a’ and ‘b’). According-
ly, the effect of the complexation on the structural features
of the thiazolidine ring are complicated in an interesting
manner.
The amide resonance of the N3-acyl moiety apparently
increases with increasing electron withdrawal at the acyl
carbonyl atom (C6), as indicated by the N3–C6 distance.
This length is expected to be inversely related to the extent
of amide resonance, and is thus at (nearly) a minimum in
the p-nitrophenyl derivative 3. (An exception is the mercu-
ric bromide complex 4, as discussed further below.)
There is no discernible pyramidalisation of the
ring nitrogen atom in any of the derivatives studied;
however, the thiazolidine ring is puckered although not
uniformly. (The annular bond angles at the heteroatoms
are apparently normal and generally show little variation.)
3. Results and discussion
3.1. Crystal structures of the N-acyl-1,3-thiazolidines 1–4:
general trends
Crystalline 1–4 were prepared and studied by X-ray dif-
fraction [14] as described above. The relevant bond length
data are summarized in Table 2. Two important trends are
noteworthy. Firstly, both the endocyclic C2–S and C2–N
bonds are shortened relative to the corresponding car-
bon-heteroatom bonds in tetrahydrothiophene [15] and
pyrrolidine [16], respectively (cf. Table 3). Secondly, there
is a general inverse relationship between the endocyclic
C2–S and C2–N lengths, as also between the endocyclic
C2–N and exocyclic C6–N lengths. Intriguingly, the C@O
bond lengths are only marginally affected.
These trends are not without exception, although plots
of the data in Table 2 afford approximate linear fits if

122
S. Chandrasekhar et al. / Journal of Molecular Structure 837 (2007) 118–131
Another interesting trend is that almost all the derivatives
1–4 crystallise with the amide C6@O bond in the antiperi-
planar orientation relative to the endocyclic C2–N bond
(with only one exception).
The above trends and structural features are discussed in
detail below.
In fact, the rather low pKa of 6.3 reported for 1,3-thiaz-
olidine [21] indicates a relatively stabilized conjugate base
form, possibly due to electron withdrawal by the sulphur
atom via the nitrogen-to-sulphur GAE. (The magnitude
of the pKa apparently rules out a simple inductive effect).
3.3. Inverse relationship between C–S and C–N lengths
3.2. Relative shortening of the (endocyclic) S1–C2 and C2–N
bonds: competing 5-endo-trig processes
In principle, an observed inverse relationship between
the C2–S and C2–N lengths may arise from either of the
two modes of the GAE, involving either the sulphur or
the nitrogen as the donor centre. However, in the light of
the above discussion the latter mode is presumably the
dominant one.
Crystallographic studies on numerous derivatives of
both tetrahydrothiophene [15] and N-acylpyrrolidine [16]
have been reported, and typical carbon-heteroatom bond
lengths are collected in Table 3. Thus, the C–S lengths
˚
are generally >1.80 A, and the C–N lengths generally
The best inverse correlation is apparently that between
the endocyclic C2–N and exocyclic N3–C6 lengths, the only
exception being an unusually short exocyclic value of
˚
>1.47 A. These compare interestingly with the values deter-
mined in this study for 1–4 (Table 2). Thus, the S1–C2
˚
˚
length in the thiazolidines can be as low as 1.7849 A, and
1.2768 A in 4b (Table 2). The correlation between the endo-
˚
the C2–N3 length as low as 1.4371 A. (Note that electron-
cyclic C2–N and C2–S lengths is beset with several excep-
tions, although these may well arise from the disparate
substitution pattern in the overall molecule. Thus the cor-
relation between the uncomplexed 1 and 2 is good, with a
decrease in the endocyclic C2–N length being accompanied
by an increase in the endocyclic C2–S length on going from
1 to 2. This (expectedly) follows a decrease in electron with-
drawal at the amide carbonyl group, the substituent therein
changing from p-nitrophenyl to methyl. Although the role
of the C2 phenyl substituent in 2 is unclear, the overall
trend indicates an enhanced nitrogen-to-sulphur GAE (in
2 relative to 1).
Another good correlation is that between the p-nitro-
benzoyl derivative 1 and its mercuric chloride complex 3.
The shortening of the endocyclic C2–N length and the cor-
responding lengthening of the C2–S in 3, indicate an
enhanced nitrogen-to-sulphur GAE, most likely due to
electrophilic coordination by mercury(II) at the sulphur
atom (in both 3a and 3b).
Both the above two cases involving the pairs 1/2 and 1/3,
demonstrate electronic effects at the termini of the N–C–S
system in 1,3-thiazolidine. These are also accompanied by
corresponding changes in the exocyclic amide N3–C6
lengths, which correlate inversely with the endocyclic C2–
N lengths and directly with the C2–S lengths. This clearly
establishes a ‘competition’ between amide resonance and
the GAE, the electron pair at the nitrogen centre being
effectively partitioned between the amide and C–N–S sys-
tems. [Similar trends are discernible in 2-phenyl-1,3-thiaz-
olidine (cf. Table 1) [8] and its N-acetyl derivative 2 (this
study, cf. Table 2).]
withdrawal by the N-acyl moieties in 1–4 would mute the
nitrogen-to-sulphur GAE, so the above changes should
be greater in principle.)
A possible explanation for these trends is the simulta-
neous operation of both the modes of the GAE, i.e., involv-
ing both the nitrogen and the sulphur centres as donors (cf.
Scheme 1). (A similar effect is the operation of both exo
and endo anomeric effects in the pyranose sugars [1,2]).
The extent of the bond shortening in 1–4 indicates that
the anomeric interaction involving the nitrogen centre as
the donor is the stronger one [17]. This is intriguing not
only because sulphur is regarded as a better donor than
nitrogen [2,18], but also because the nitrogen centre bears
an electron-withdrawing acyl group.
However, this could be related to the fact that 1,3-thiaz-
olidines undergo ring-opening via cleavage of the C2–S
bond (Eq. (1), Scheme 3) rather than of the C2–N bond
(Eq. (2), Scheme 3) [4,19]. This is explicable on the basis
of Baldwin’s rules [20]: Eq. (1) represents a favoured 5-
endo-trig process involving a second row element (sulphur),
but Eq. (2) a disfavoured 5-endo-trig process involving a
first row element (nitrogen). These arguments imply that
of the two no-bond canonical forms Ia and Ib, the former
would be higher in energy, and would thus contribute less
towards the overall charge transfer in the molecule.
H
N
N
(-H+)
(eqn. 1)
S
S
3.4. The Hg(II) complexes: the ‘push–pull’ mechanism in 3
and 4
R
N
R
N
(eqn. 2)
The data of the two mercury(II) complexes 3 and 4 (each
involving two thiazolidine units) show two interesting
trends, discussed below. (Tetrahydrothiophene reportedly
reacts with mercuric chloride via nucleophilic displacement
S
S
Scheme 3. Two possible modes for the ring opening of 1,3-thiazolidine:
both are 5-endo-trig processes, but Eq. (1) is favoured over Eq. (2).

S. Chandrasekhar et al. / Journal of Molecular Structure 837 (2007) 118–131
123
of a chloride ion rather than forming a complex [22]; this
indicates that tetrahydrothiophene is more nucleophilic
than 1 and possibly 5, presumably because of the electro-
negative N-acyl moiety in 1 and 5.)
relatively in 4a and 4b. (The bond distances also indicate
that the ‘push–pull’ effect is greater in 3 than in 4.)
The trends involving C2–N3 vanish if 3b is compared
with 4a or 4b, possibly because 3b possesses an unusually
low nitrogen-to-sulphur GAE, and an enhanced sulphur-
to-nitrogen GAE (because of the ‘push–pull’ effect). These
seem to overcome the difference in electrophilicity men-
tioned above.
The first trend is that in the mercuric chloride complex 3
the endocyclic C2–N and C2–S lengths in the two thiazolidine
moieties correlate inversely (cf. Table 2), as do the endocyclic
C2–N and exocyclic C6–N lengths. Clearly, the above argu-
ments proposed for the pairs 1/2 and 1/3 apply here too.
Interestingly, the inverse trends in 3 apparently imply a
‘push–pull’ mechanism, in which charge is transferred from
one thiazolidine ligand through the mercury atom to the
other thiazolidine ligand (cf. the no-bond form IV in
Scheme 4). This indicates that the nitrogen-to-sulphur
GAE dominates in thiazolidine ligand 3a, and the sul-
phur-to-nitrogen GAE in ligand 3b (Scheme 4). There are
also corresponding changes in the N-acyl moieties (amide
resonance is enhanced in 3b relative to 3a).
3.5. The Hg(II) complexes: the ‘back-donation’ mechanism
in 4
Not only does the C2–N distance decrease on going from
3b to 4a or 4b, but so too does the C2–S distance! A similar
trend involves 4a and 4b (but N3–C6, C6@O and C2–N3
behaving as expected).
These trends apparently imply an enhanced GAE in both
the modes, i.e., with both the nitrogen and sulphur centres
acting as donors, and which increases in the series 3b–4b–
4a. A possible explanation for this could be that there is a
back-donation of electrons from the sulphur atom to the
C2 centre, which progressively increases in the above series.
Considering first the set 4a and 4b, such a back donation
would overlay an enhanced sulphur-to-nitrogen GAE on
the nitrogen-to-sulphur GAE in ring 4a (cf. no-bond forms
V and VI, Scheme 5). It is possible that an electron lone
pair is donated by a bromine atom to the sulphur atom
via the mercury atom (cf. V).
Similar trends also exist in the mercuric bromide complex
4 (although with an interesting exception). Thus, in the two
thiazolidine moieties 4a and 4b, the exocyclic C6–N length is
inversely related to both the C6@O and C2–N3 lengths.
Intriguingly, however, these changes are not matched by
the C2–S lengths, which relate directly with the C2–N
lengths. This indicates an enhanced GAE in both the modes
in 4a, which is discussed in the next section. Also, the
˚
minimal N3–C6 length (1.2768 A) and the matching maxi-
˚
mal C6@O length (1.2390 A) in 4b evidence the maximum
level of amide resonance in this set of derivatives.
A similar effect (with Cl) may also involve 2 and 3a. A
measure of the above back-donation apparently also exists
in the uncomplexed 1 and 2; in these both the C2–S and
C2–N bonds are shortened relative to tetrahydrothiophene
and pyrrolidine (as mentioned). The effect is apparently
maximal in 4 because of the back-donation from the bro-
mine atom.
The second trend (mentioned above) is the increase in
the endocyclic C2–N3 length (Table 2) on going from the
mercuric chloride complex 3 to the mercuric bromide com-
plex 4, although this is only relative to 3a; the exocyclic
˚
amide C6–N length is drastically shortened (1.2768 A as
mentioned) in 4b; the C2–S length is also inversely related
to the C2–N length in this set (i.e., 3a vs. either 4a or 4b).
The likely explanation is that the lower electrophilicity of
mercuric bromide decreases the nitrogen-to-sulphur GAE
The C2–N3 and C2–S distances in 4a and 4b (generally)
`
show the usual inverse relationship vis-a-vis 1, 2 and 3a, as
do the N3–C6 and C6@O distances. The drastically lowered
˚
N3–C6 distance (1.2768 A) in 4b indicates the most
enhanced level of amide resonance in this set of com-
pounds; this is matched by the maximal C@O distance
(1.2390 A) observed in this set, but (intriguingly) not by
Ar
˚
Ar
S
O
O
`
the C2–S length vis-a-vis 4a.
N
b
N
b
Thus, the amide resonance is apparently not supported by
the sulphur-to-nitrogen GAE. Intriguingly, the above maxi-
mal amide resonance (in 4b) corresponds with electrophilic
coordination at the sulphur atom! These trends possibly
indicate a through-bond component to the sulphur-to-nitro-
gen GAE, which affects the S1–C2 and C2–N3 distances
minimally, yet enhancing amide resonance.
S
Cl
Cl
Hg
Hg
Cl
Cl
S
S
a
N
a
N
O
O
Ar
Ar
(Ar = p-NO2-C6H4-)
3.6. The apparent immutability of the amide C@O bond
IV
3
The amide carbonyl bond length apparently displays
Scheme 4. The ‘push–pull’ mechanism for charge transfer in the HgCl2
complex 3, in which the nitrogen-to-sulphur GAE operates in one ring
(marked ‘a’) and the sulphur-to-nitrogen GAE in the other (marked ‘b’).
(These correspond with 3a and 3b, respectively in Tables 2,5,6,8 and 9.)
only marginal changes (except in 4b). The changes in the
˚
C@O length are <0.016 A, and generally an order of mag-
nitude lower than those of the other lengths. In fact, an

124
S. Chandrasekhar et al. / Journal of Molecular Structure 837 (2007) 118–131
Ph
R
R
Ph
O
O
O
O
N
a
N
N
N
a
R'
R'
S
S
S
S
Br
Br
Hg
Hg
Br
Br
If
1-4
S
S
b
b
N
N
R
S
R
O
O
O
O
Ph
Ph
N
N
R'
R'
4
V
S
Ig
1-4
Ph
O
Ph
Scheme 6. The contribution of the canonical form If to the resonance of
the amide group would be small if R were an electronegative group (as in 1
and 3), but that of the no-bond form Ig would be considerable. (These
explain the relative immutability of the C@O distance.)
O
N
a
N
a
S
S
Br
Hg
Br
Br
Hg
3.7. Correlation with relevant spectroscopic characteristics
S
Br
b
S
N
b
The salient features of the spectral data of 1–4 are col-
lected in Table 4, and offer several interesting correlations
with the crystallographic data (cf. Section 2 and Table 2).
The IR values for the amide C@O group correlate quite
well with the N3–C6 bond lengths. The mercuric chloride
complex 3 shows two IR bands for C@O, which accords
with the fact that the two thiazolidine moieties are differ-
ent: the band at 1650 cmÀ1 may be assigned to 3a and
the band at 1634 to 3b on the basis of the ‘push–pull’ char-
ge transfer discussed at length above (cf. Scheme 4).
However, the fact that the mercuric bromide complex 4
shows only one IR band for C@O is at odds with the C@O
bond length data, which are far more different in 4a and 4b
(relative to 3a and 3b). On the other hand, the very low val-
ue for the C@O IR band in 4 (1626 cmÀ1) correlates very
N
O
Ph
O
Ph
VI
4
Scheme 5. The ‘back-donation’ mechanism in the HgBr2 complex 4,
leading to an enhanced sulphur-to-nitrogen GAE via electron donation
from the bromine atoms as shown in the no-bond form V. This is
superimposed on the ‘push–pull’ mechanism depicted in VI (as in 3, cf.
Scheme 4). Thus, both the nitrogen-to-sulphur and sulphur-to-nitrogen
GAE’s operate in ring ‘a’, but only the latter in ring ‘b’ (these correspond
to 4a and 4b, respectively in Tables 2, 5, 6, 8 and 9).
inverse relationship with the neighbouring N3–C6 bond is
to be expected on the basis of amide resonance (cf. canon-
ical form Ic, Scheme 1); although this relationship exists in
general, there are exceptions (e.g., on going from 1 to 2
both C@O and N3–C6 increase).
˚
well with the drastically low N3–C6 length (1.2768 A) and
˚
the maximal C@O length (1.2390 A) observed in 4b.
The above IR spectra were determined for the solids, so
the crystal lattice was likely unperturbed. Also, the twin
(C@O) bands in the IR spectrum of 3 are unlikely to be
due to the rotational isomerism of the amide moiety, which
was observed in solution (by NMR, vide infra) but not in
the crystal (cf. Section 3.11). (The above ambiguity in the
case of 4 remains unresolved.)
The NMR data also offer useful correlations, although
the data is complicated by the rotational isomerism around
the partial double bond (N3–C6) of the amide group. The
shift values (both 1H and 13C) at C2 of the thiazolidine ring
indicate increasing positive charge in the series 1-5-3, which
accords both with the expected increase in the nitrogen-to-
sulphur GAE and the S1–C2 lengths of 1 and 3 (Table 2).
(The C2 shift values of 2 and 4 are difficult to correlate with
the others.)
A possible reason may be the contribution of the canon-
ical form If (Scheme 6), which would be muted by an elec-
tron withdrawing group at C6. This would explain the
lower C@O length in 1 relative to 2 despite the shorter
N3–C6 length in 1, as also the nearly unchanged C@O
length in 1 and its complex 3a. An electron withdrawing
substituent at C6 could also raise the C@O bond order
via overlap of an oxygen lone pair with the r* antibonding
orbital at C6 to the substituent (cf. canonical form Ig).
Therefore, the C@O bond is apparently subject to multi-
ple (and compensatory) electronic effects, and so displays
little change in geometry. However, the data in Table 2 is
biased by the presence of the strongly electron withdrawing
p-nitrophenyl group in two of the cases (1 and 3). Thus,
considerable changes in the C@O length are observed in
the other cases, those in the mercuric bromide complex 4
being clearly substantial.
The C5–H resonances shift marginally upfield in 3 and 4
relative to 1 and 3, respectively, although the 13C resonance

S. Chandrasekhar et al. / Journal of Molecular Structure 837 (2007) 118–131
125
Table 4
Salient spectroscopic characteristics of the N-acylthiazolidines 1–5 (cf. Sections 2.2 and 3.7)a
Thiazolidine derivative 1–5
mmax (cmÀ1) (C@O)
dH (dC)
C2
C4
C5
C6 (dC)
N-(4-Nitrobenzoyl) (1)
N-Acetyl-2-phenyl (2)
HgCl2 Complex (3)
HgBr2 Complex (4)
N-Benzoyl (5)
1637 (CHCl3)
1650 (CHCl3)
1650, 1634 (KBr)
1626 (KBr)
4.47, 4.77 (50.7, 51.3)
6.01, 6.50 (64.0, 65.0)
4.59, 4.72 (52.1, 52.2)
4.64b (57.0, 57.5)
3.75, 4.02 (48.0, 48.2)
3.95, 4.28 (50.2, 50.7)
3.81, 3.93 (51.3, 51.4)
3.86b (51.3, 51.3)
3.05, 3.14 (29.5, 30.8)
3.04–3.17b (28.9, 30.4)
2.93, 3.12 (48.5, 49.2)
2.91, 3.08 (45.7, 47.2)
3.01b (30.4)
166.8
168.3, 169.1
167.7
169.8
169.3
1634 (KBr)
4.49, 4.73 (51.3)
3.75, 3.96 (48.0)
a
Shown are the IR carbonyl stretching frequency of the N-acyl group (column 2), and the NMR chemical shifts (both 1H and 13C, the latter
parenthesized) for the ring positions of the thiazolidine moiety (columns 3–5); the 13C shift for the carbonyl carbon atom (C6) is also given (column 6); the
twin NMR resonances arise from rotational isomerism around the amide partial double bond.
b
Broad.
shifts downfield in 3 relative to 1. (Possibly, the 13C shifts are
more sensitive to through-bond charge transfer.) The 13C
NMR shifts of the amide carbonyl group (C6) seem to vary
erratically (rather like the C@O lengths discussed above).
However, the observed downfield shift with an increase in
the nitrogen-to-sulphur GAE (1 vs. 3 and 4), as also (possi-
bly) with a decrease in amide resonance (1 vs. 2 and 5), are
generally supported by the bond length data (Table 2).
show marginal, if any, pyramidalisation at N3. Although h
is practically 360ꢁ in all the cases, a shows some variation,
being as low as À161.6ꢁ in 3a. This indicates some pyrami-
dalisation, which interestingly correlates with an enhanced
nitrogen-to-sulphur GAE in 3a.
However, the general lack of pyramidalisation in 1–4
most likely implies that the nitrogen-to-sulphur GAE
essentially involves the amide p system in its entirety.
Apparently, this is energetically preferable to pyramidalis-
ing the amide nitrogen centre, so pyramidalisation is not a
reliable measure of the perturbation of amide resonance.
3.8. Absence of pyramidalisation at N3
An important question is whether the competition
between amide resonance and the GAE in 1–4 leads to
the pyramidalisation of the amidic N3 centre [23]. (Pyrami-
dalisation has been considered in the penicillins as a conse-
quence of strain [24].)
There are three measures of pyramidalisation at an
amide nitrogen centre [23,24]: the distance of the nitrogen
atom from the plane formed by its three substituents (‘h
value’); the angle of deviation of the amide system from
planarity (a); and the sum of the bond angles around the
nitrogen atom (h). The extent of pyramidalisation is pro-
portional to h, a and the deviation of h from 360ꢁ.
Of these a and h are perhaps easier to employ. Their val-
ues in 1–4 are listed in Table 5. (The a has been estimated
from the two torsion angles around the amide N3–C6 bond,
involving the flanking endocyclic C2 and C4 atoms.) These
3.9. Ring puckering
The 1,3-thiazolidine ring in 1–4 is expectedly puckered
(cf. Figs. 1–4 and Scheme 7). The ring torsion angles (cf.
Table 6) indicate that the generally preferred conformation
is the twisted envelope [25]: this is particularly so in the
uncomplexed 1 and 2, in which the C5 atom occupies the
‘flap’ position as depicted in Ih (Scheme 7). The thiazoli-
dine rings in the complexes 3 and 4 are less puckered, being
nearly planar in 4. The groups in the ‘flap’ position are dif-
ferent in 3a and 3b, as shown in Ii and Ij, respectively.
The envelope conformations of 1–3 are also supported
by a Cremer and Pople analysis [26] of the crystallographic
data, the magnitudes of u(2) (the phase angle) and Q(T)
(total puckering amplitude) being listed in Table 7. For
an envelope conformation of a five membered ring, u(2)
should be an integral multiple (k) of 36; [for a half-chair,
u(2) = 36k + 18]. The data in Table 7 indicates the
Table 5
Possible pyramidalisation of the amide nitrogen atom (N3) in 1–4, as
measured by the sum of the angles around N3 (h) and the torsion angle
around the N3–CO amide bond (a)a
Compound
h (ꢁ)
a (ꢁ)b
1
359.3
359.7
357.8
357.3
358.9
359.3
0.79; 169.1
0.57; 174.0
2.06; 161.6
6.78; 169.0
0.70; 167.0
1.91; 167.9
2
3a
3b
4a
4b
a
The values in parenthesis in the case of 3 and 4 refer to the second
thiazolidine ring in the 2:1 complexes (cf. Schemes 2, 4 and 5).
Fig. 1. ORTEP diagram of the asymmetric unit cell of the N-(4-
nitrobenzoyl)-1,3-thiazolidine 1 at 50% ellipsoidal probability. (Crystallo-
graphic numbering is shown.)
b
Signs omitted; the two values refer to the ‘cisoid’ and ‘transoid’ torsions
(ꢀ0ꢁ and ꢀ180ꢁ, respectively).

126
S. Chandrasekhar et al. / Journal of Molecular Structure 837 (2007) 118–131
O
Ar
HgCl2
S
S
N
N
R
S
Ar-OC
Ar
N
Ar
O
HgCl2
Ij
Ih
Ii
3
Ii/Ij
Ih
R
Ar
3a Ii
3b Ij
p-NO2-C6H4
p-NO2-C6H4
1
2
H
Ph
p-NO2-C6H4
Me
Scheme 7. The observed ring puckerings are represented by the envelope
form Ih and the twist forms Ii and Ij; the envelope Ih is apparently adopted
by the uncomplexed 1 and 2, whereas the twist forms Ii and Ij are
preferred by the thiazolidine rings 3a and 3b, respectively in the HgCl2
complex 3. (The HgBr2 complex 4 is nearly planar.)
Fig. 2. ORTEP diagram of the asymmetric unit cell of the N-acetyl-2-
phenyl-1,3-thiazolidine 2 at 50% ellipsoidal probability. (Crystallographic
numbering is shown.)
Table 6
Puckering of the 1,3-thiazolidine ring in 1–4 as measured by the indicated
torsion angles (in degree (ꢁ), cf. Scheme 7)a
Compound
S1–C2
N3–C2
N3–C4
C5–C4
S1–C5
1
À33.3(3)
À34.7(3)
15.3(2)
À21.0(3)
À21.9(3)
À21.1(3)
15.1(3)
16.0(2)
7.2(1)
44.0(1)
22.5(2)
À5.4(1)
À8.4(1)
À37.7(4)
À33.3(4)
À29.6(3)
36.7(4)
11.4(1)
7.1(1)
40.6(4)
39.7(4)
8.4(1)
33.7(4)
19.4(1)
16.0(2)
2
21.3(3)
À36.4(4)
2.2(1)
17.7(3)
18.5(3)
3a
3b
4a
4b
a
These correspond to the following dihedral angles in the given order of
atoms (sequentially in columns 2–6): C5–S1–C2–N3, C4–N3–C2–S1, C2–N3–
C4–C5, S1–C5–C4–N3, C2–S1–C5–C4.
Table 7
Puckering in the thiazolidine rings in 1–4 in terms of the total puckering
amplitude Q(T), phase angle u(2) and ka,b
´
˚
Compound
Q(T) (A)
u(2)
k
1
0.5016 (6)
0.4770 (2)
0.3930 (4)
03660 (4)
0.2220 (3)
0.2440 (2)
343.1 (7)ꢁ
351.7 (5)
29.3 (5)
96.8 (5)
17.2 (3)
8.8 (5)
10
10
1
3
0.5
0.25
2
3a
3b
4a
4b
a
Fig. 3. ORTEP diagram of the asymmetric unit cell of the mercuric
chloride complex 3 of N-(4-nitrobenzoyl)-1,3-thiazolidine at 50% ellipsoi-
dal probability. (Crystallographic numbering is shown.)
Based on the method of Cremer and Pople, Refs. [26] and [27].
k = [u(2)]/36 should be an integer for an envelope conformation.
b
Previous reports had indicated the existence of various
envelope and twist forms for the 1,3-thiazolidine ring,
although with the nitrogen atom often in an out-of-plane
position [6–8]. (This would bring the nitrogen lone pair
of electrons into antiperiplanarity with the S1–C2 bond.)
In 1–4 the nitrogen lone pair is part of an amide p system,
and the nitrogen centre is not tetrahedral but (essentially)
trigonal, so antiperiplanarity (with S1–C2) cannot be
defined in the conventional sense. However, the general
stereoelectronic requirement of overlap between the amide
p system and the S1–C2 antibonding (r*) orbital (cf. Sec-
tion 3.8), is fulfilled in various conformations including
the planar (as indicated by models), although to varying
extents.
Fig. 4. ORTEP diagram of the asymmetric unit cell of the mercuric
bromide complex 4 of N-benzoyl-1,3-thiazolidine at 50% ellipsoidal
probability. (Crystallographic numbering is shown.)
envelope form for 1, 2, 3a and 3b, but not for 4a and 4b.
(These computations were performed with the PLATON
program developed by Spek [27].)
The above p–r* overlap is most efficiently attained when
the N–C–S system is planar, asin Ih (adopted by 1 and 2, with

S. Chandrasekhar et al. / Journal of Molecular Structure 837 (2007) 118–131
127
olidines (95ꢁ) [16] and other thiazolidines (106.5ꢁ) [8], is
likely due to partial result of sp2 hybridisation at N3.
The bond angles at the mercury atoms are compatible
with a trigonal bipyramidal (TBP) geometry (VII) [28] in
3 and a tetrahedral geometry (VIII) [28] in 4 (Table 9,
Scheme 9). This is because 3 is dimeric and pentacoordinat-
ed at mercury, whereas 4 is monomeric and tetracoordinat-
ed at mercury. Thus, the dimeric 3 is comprised of two TBP
moieties linked via a nearly rectangular (HgCl)2 unit.
(Apparently, Hg(II) prefers octahedral coordination,
although other geometries are also known [28].)
The Cl–Hg-Cl angles of 83ꢁ and 173ꢁ in 3 indicate an
approximate TBP structure, whereas the smaller Br–Hg–
Br angle in 4 (ꢀ125ꢁ) indicates an approximate tetrahedral
one (Table 9). In 3 (VII) the apices are occupied by two
chlorine atoms (‘Clap’) and the equatorial positions by
two sulphur (‘Seq’) and one chlorine (‘Cleq’) atom. The
S–Hg–S angles in 3 (99ꢁ) and 4 (84ꢁ) are compatible with
these assignments.
O
Ar
Ik
3
X
Torsion
Cl ~ 90o
Br ~ 700
N
S
4
X2Hg
Ik
Scheme 8. The orientation of the mercury atom relative to the thiazolidine
ligand in the mercury(II) complexes 3 and 4 is indicated in Ik. The N3–C2–
S-Hg torsions of ꢀ90ꢁ (in 3) and ꢀ70ꢁ (in 4) suggest sp2 hybridization at
the sulphur. Coordination to the mercury atom occurs via a p-type orbital
on sulphur (not shown), the free lone pair in an sp2-like orbital more or
less in the plane of the thiazolidine ring (shown in black) enabling the
sulphur-to-nitrogen GAE in the planar ring.
C5 at the ‘flap’). In Ii, adopted by 3a, N3 occupies the ‘flap’, so
the S1–C2 r* orbital is at an angle to the C2–N3–C6 plane
(comprising the amide p system); this accords with the (mod-
est) pyramidalisation of N3 observed in 3a (cf. Section 3.8
above). In Ij, adopted by 3b, S1 occupies the ‘flap’, and again
the S1–C2 r* orbital is at an angle to the C2–N3–C6 plane,
although this meets the stereoelectronic requirement for
the sulphur-to-nitrogen GAE (vide infra).
The two thiazolidine moieties of the HgBr2 complex 4
are puckered rather similarly, and indicate a nearly planar
ring geometry, possibly because the sulphur-to-nitrogen
GAE dominates in them. The N3–C2–S–Hg moiety sub-
tends a dihedral angle of ꢀ70ꢁ in both 4a and 4b (cf. Ik,
Scheme 8). This indicates that the hybridization at the sul-
phur atom is somewhat close to sp2, with the coordinating
mercury atom bonding with a lone pair in the p orbital.
This would leave the other free lone pair in an sp2 orbital
and more or less in the plane of the ring, and thus antiperi-
planar to the C2–N3 bond: this would meet the stereoelec-
tronic requirements for the sulphur-to-nitrogen GAE.
In 3 too, the N3–C2–S–Hg torsion is similar at ꢀ90ꢁ,
thus again indicating sp2-like hybridization at the sulphur
atom (cf. Scheme 8). Thus, this seems to be the general
mode of operation of the sulphur-to-nitrogen GAE in the
complexes 3 and 4.
The tetrahedral (rather than square planar) geometry in
4 is indicated by the closely similar bond angles (108ꢁ and
113ꢁ) involving a given bromine atom and the two sulphur
atoms (or vice versa, cf. Table 9). However, the widely dif-
fering Br–Hg–Br and S–Hg–S angles (125ꢁ and 84ꢁ, respec-
tively) indicate a distorted tetrahedron at the mercury.
3.11. Orientation of the amide carbonyl group
The C6@O group of the amide moiety is almost always
oriented antiperiplanar to the C2–N3 bond (Il, Scheme 10).
Thus, the C2–N3–C6@O dihedral angle is ꢀ180ꢁ (Table 10),
indicating overlap between the antibonding r* orbital of
the C6–O r bond and the electron density of the C2–N3
bond (strictly the ‘HOMO’). (A similar effect, involving
an alkyl-oxygen lone pair, exists in esters and lactones
[1].) The lone exception is the case of 3b in which the sul-
phur-to-nitrogen GAE dominates: thus, the C2–N3 bond
would be partly broken so the above stereoelectronic over-
lap would be weak (the charge at N3 being delocalized by
amide resonance).
3.10. Bond angles and distances at S, N3 and Hg
The above trends in the crystalline lattice, however, are
at variance with the results of the solution state studies by
NMR, which indicate that rotation around the amide N3–
C6 bond is rapid at room temperatures (cf. Sections 2.4 and
3.7).
The annular bond angles at the sulphur atom are ꢀ90ꢁ
(cf. Table 8), similar to tetrahydrothiophenes (ꢀ95ꢁ) [15].
The marginally larger values in 4 are possibly due to the
ꢀ sp2 hybridization at the sulphur (vide supra.) The large
annular bond angle at N3 (ꢀ 115ꢁ) relative to N-acylpyrr-
3.12. Possible relevance to the mechanism of action of
penicillin
Table 8
Annular bond angles at the ring sulphur (S1) and nitrogen (N3) atoms in
the 1,3-thiazolidine derivatives 1–4
The key step in the mechanism of action of the penicil-
lins (and the b-lactam antibiotics, in general) involves the
cleavage of the b-lactam ring, via nucleophilic attack at
its carbonyl group by a serine hydroxyl group of a bacterial
transpeptidase enzyme [5,29] (cf. IX, Scheme 11). This also
effectively (and irreversibly) inactivates the enzyme which is
critically required for the construction of the bacterial cell
wall, thereby destroying the bacterium itself. Although this
Compound
Angle at S1 (ꢁ)
Angle at N3 (ꢁ)
1
91.3(1)
89.5(2)
92.8(1)
92.0(1)
98.2(3)
100.3(3)
114.5(2)
115.5(2)
111.0(3)
115.5(4)
115.2(3)
113.4(2)
2
3a
3b
4a
4b

128
S. Chandrasekhar et al. / Journal of Molecular Structure 837 (2007) 118–131
Table 9
Relevant bond angles at the mercury atom in the thiazolidine-HgX2 complexes 3 and 4 (typical values only, cf. Scheme 9)
Compound
‘X’
Angle X–Hg–X (ꢁ)
Angle S–Hg–S (ꢁ)
Angles X–Hg–S (ꢁ)
3
4
a
Cl
Br
172.7(4),a 82.9(3)b
124.4(9)
98.7(3)c
83.9(2)
91.9(3)b 93.1(3)b 97.2(1)c
107.9(2), 112.6(2)
Apical–apical.
Apical–Equatorial.
Equatorial–Equatorial.
b
c
R
S
O
E+
Br Br
Hg
N
N
O
Ar
S
Hg
Cl
R
CO2H
S
S
Ar
S
S
OH
Cl
S Hg
N
N
N
O
N
Cl
N
O
O
O
S
CO2H
O
CO
NH
N
Cl
Ar
Ph
Ph
Ar
HN
O
X
VIII
VII
TRANSPEPTIDASE
Scheme 9. The geometries at the mercury atom in the mercury(II)
complexes 3 and 4: the dimeric 3 adopts an approximate trigonal
bipyramidal geometry as shown in VII, the central (HgCl)2 unit – the site
of dimerisation – forming a rectangle; however, the monomeric 4 adopts
an approximate tetrahedral geometry as shown in VIII.
IX
Scheme 11. Nucleophilic attack at the b-lactam carbonyl group in a
penicillin by a serine hydroxyl group on a bacterial transpeptidase enzyme
(IX), leading to the cleavage of the lactam and the irreversible inactivation
of the enzyme (not shown). The possible activation of the b-lactam
carbonyl group to nucleophilic attack (as in IX), via electrophilic
coordination at the sulphur atom and an enhanced nitrogen-to-sulphur
GAE is depicted in X. (E+ represents the electrophile, which may be a
proton, Zn(II), etc., in the enzymic case).
Ar
O
Ar (R)
O
3
N
3
2
N
2
S
Cl
S
Hg
polarisable, and that electrophilic coordination at the sul-
phur atom can enhance the nitrogen-to-sulphur GAE (cf.
X, Scheme 11).
Cl
Im
Il
Scheme 10. The orientation of the N3-acyl group in 1–4 is nearly always as
shown in Il, (C@O antiperiplanar to C2–N3), except in 3b in which they are
nearly synperiplanar, as shown in Im.
It would be surprising if this possible mode of activation
of the b-lactam antibiotics were not employed at the relevant
enzyme active site. Although the toxicity of mercury(II) rules
out its possible involvement as an electrophilic activator, two
plausible alternatives may be mentioned: these are hydrogen
bonding and zinc(II), the latter being a more reactive – and
biocompatible – congener of Hg(II) in Group II B of the peri-
odic table. Although further work may throw light on these
possibilities, it is noteworthy that zinc(II) has indeed been
implicated in the action of a class of b-lactamases (analogs
of the transpeptidases in terms of their reactivity towards
b-lactams) [32].
Table 10
The orientation of the N3-acyl group in 1–4 as indicated by the C2–N3–
C6@O dihedral angle (cf. Scheme 10)
Compound
C2–N3–C6@O torsion (ꢁ)
1
169.1(9)
174.0(8)
À161.6(5)
6.8(2)
167.0(7)
167.9(8)
2
3a
3b
4a
4b
indicates that the activity of the b-lactam antibiotics is
related to their susceptibility to nucleophilic cleavage, it is
noteworthy that they are stable to oral ingestion and pass
through the digestive tract relatively unscathed. Thus, they
possess a kinetic stability that is apparently lost at the
transpeptidase active site, the basis of which has not been
clarified.
An interesting possibility – that motivated this study
and is apparently supported by it – is the diminution of
the b-lactam amide resonance via an enhanced nitrogen-
to-sulphur GAE at the active site [30,31]. The above results
indicate that the N–C–S moiety in 1,3-thiazolidine is highly
4. Conclusions
Crystal structures of the four N-acyl-1,3-thiazolidine
derivatives 1–4 indicate that both the nitrogen-to-sulphur
and the sulphur-to-nitrogen GAE’s operate in them,
although the former appears to dominate. Electrophilic
coordination by mercury(II) at the sulphur atom apparent-
ly enhances the nitrogen-to-sulphur GAE in general. These
effects are masked in some cases by other modes of charge
transfer. An enhanced nitrogen-to-sulphur GAE diminish-
es the amide resonance of the N-acyl group. It seems pos-

S. Chandrasekhar et al. / Journal of Molecular Structure 837 (2007) 118–131
129
Table A.2
sible that such an effect would form an important part of
the mode of action of the b-lactam antibiotics.
Crystal data
HgCl2 complex: 3
20H20Cl2Hg1N4O6S2
225047
748.0
293(2)
Mo Ka
0.7107
HgBr2 complex: 4
Formula
C
C20H22Br2Hg1N2O1S2
225048
746.9
293(2)
Mo Ka
CCDC No.
Formula weight
Temperature (K)
Radiation
Wavelength (A)
Crystal system
Acknowledgements
We thank CSIR (New Delhi) and DST (New Delhi) for
generous financial support of this work, including the CCD
facility (DST) in the Division of Chemical Sciences, IISc.
We thank Dr. Jyoti R. Kavali for preliminary investiga-
tions, and the referees for critical comments and helpful
suggestions.
˚
0.7107
Triclinic
P À 1
Triclinic
P À 1
Space group
˚
a (A)
6.540(3)
13.466(6)
14.210(7)
80.35(1)
83.19(1)
88.65(1)
1225.06(21)
2
5.907(9)
12.597(20)
15.426(26)
101.76(1)
100.64(3)
90.18(6)
1113.40(43)
2
˚
b (A)
˚
c (A)
a (ꢁ)
b (ꢁ)
c (ꢁ)
Appendix A
3
˚
Volume (A )
This section contains a brief note on the lattice packing
in 1–4, important crystallographic data obtained in this
study in tabulated form (Tables A.1 and A.2), and plots
of the relevant bond length data in Table 2 (Figs. 5–8, men-
tioned in Section 3.1.).
Z
Density (g/cc)
Abs. Coeff. (mmÀ1
F(000)
hmin,max
hmin,max, kmin,max
lmin,max
Number of
reflections
2.03
6.716
723.8
1.5,26.4
(À8,8), (À16,16),
(À17,17)
9503
2.25
10.805
707.8
1.6,24.7
(À6,6), (À14,13),
(À18,18)
5164
)
,
A note on the lattice packing in crystalline 1–4
Number unique
reflections
Number of
parameters
4787
396
2582
262
There appears to be no uniform pattern to the lattice
packing in the above thiazolidine derivatives [14].
Table A.1
Refinement method Full matrix least
Full matrix least
squares on F2
0.099
0.090
0.231
squares on F2
0.028
0.024
0.063
Crystal data
N-(4-Nitrobenzoyl): 1
10H10N2O3S1
225,046
290(2)
Mo Ka
0.7107
N-Benzoyl-2-phenyl: 2
R_all
R_obs
Formula
C
C11H13N1O1S1
227,436
290(2)
Mo Ka
0.7107
CCDC No.
Temperature (K)
Radiation
wR2_all
wR2_obs
Dqmin,max (e A
0.061
0.221
À3
˚
)
À0.626,0.857
À3.010,2.703
˚
Wavelength (A)
Goodness-of-fit
1.03
1.055
Crystal system
Monoclinic
P21/n
12.721(8)
5.937(4)
14.810(9)
90.00
110.42(1)
90.00
1048.32(43)
4
Orthorhombic
P212121
9.295(6)
9.482(6)
11.842(7)
90.00
90.00
90.00
1043.72(11)
4
Space group
˚
a (A)
˚
b (A)
˚
c (A)
a (ꢁ)
b (ꢁ)
c (ꢁ)
1.48
1.47
1.46
1.45
1.44
3
˚
Volume (A )
Z
Density (g/cc)
Abs. Coeff. (mmÀ1
F(000)
1.51
0.301
495.9
1.32
0.275
439.9
)
hmin,max
hmin,max, kmin,max
lmin,max
1.8, 25.3
(À15,13), (À7,7),
(À17,17)
7229
2.8, 26.4
(À11,11), (À11,11),
(À14,14)
8177
,
Number of
reflections
Number unique
reflections
Number of
parameters
1910
177
2101
179
1.43
1.78
Refinement method
Full matrix leastsquares Full matrix leastsquares
1.79
1.80
1.81
1.82
1.83
1.84
on F2
on F2
C-S bond distance
R_all
R_obs
0.139
0.107
0.042.0.
0.037
Fig. 5. A plot of the S1–C2 distance (x-axis) vs. the C2–N3 distance (y-axis)
in 1–4 (cf. Table 2). The straight line (coloured circles) excludes the point
indicated by the arrow (representing 4a, in which both modes of the GAE
are unusually enhanced). (The straight line obeys the equation:
y = 2.406 À 0.523x, with correlation coefficient ‘R’ = 0.772.)
wR2_all
wR2_obs
Dqmin,max (e A
Goodness-of-fit
0.263
0.247
0.081
0.079
À3
˚
)
À0.401, 0.489
À0.222, 0.144
1.236
1.12

130
S. Chandrasekhar et al. / Journal of Molecular Structure 837 (2007) 118–131
1.84
1.83
1.82
1.81
1.80
1.79
1.78
1.36
1.34
1.32
1.30
1.28
1.43
1.44
1.45
1.46
1.47
1.48
1.49
1.28
1.30
1.32
1.34
1.36
C-N bond distance
C-N bond distance
Fig. 6. A plot of the C2 À N3 distance (x-axis) vs. the N3–C6 distance (y-
axis) in 1–4 (cf. Table 2). The straight line (coloured circles) excludes the
point indicated by the arrow (representing 4b). (The straight line obeys the
equation: y = 3.922 À 1.825x, with correlation coefficient ‘R’ = 0.919.)
Fig. 8. A plot of the N3–C6 distance (x-axis) vs. the S1–C2 distance (y-axis)
in 1–4 (cf. Table 2). The straight line (coloured triangles) excludes the
point indicated by the arrow (representing 4b). (The straight line obeys the
equation: y = À0.493 + 1.704x, with correlation coefficient ‘R’ = 0.630.)
1.240
1.235
1.230
1.225
1.220
1.215
relatively lower amide resonance in the case of the com-
plexes 3 and 4 seems to induce p–p stacking.)
All the derivatives 1–4 also display numerous (weak)
hydrogen bonds involving C–H moieties in both the aro-
matic and thiazolidine rings as the usual donor, and an oxy-
gen atom of the amide carbonyl or nitro group (when this is
present) as the acceptor. In the case of 2 a hydrogen atom a
to the carbonyl atom functions as a donor to several accep-
tors, including the thiazolidine sulphur atom. In the case of
the complexes 3 and 4 hydrogen bonds involving the halo-
gen atoms as the acceptors are also evident. These relatively
weak (but numerous) interactions appear to contribute con-
siderably to the lattice packing in all the above cases.
1.26
1.28
1.30
1.32
1.34
1.36
1.38
C-N bond distance
References
Fig. 7. A plot of the N3–C6 distance (x-axis) vs. the C6@O distance (y-
axis) in 1–4 (cf. Table 2). The straight line (coloured circles) includes all
points and obeys the equation: y = 1.505 À 0.209x, with correlation
coefficient ‘R’ = 0.888.
[1] A.J. Kirby, The Anomeric and Related Stereoelectronic Effects at
Oxygen, Springer Verlag, Berlin, 1983.
[2] E. Juaristi, G. Cuevas, Tetrahedron 48 (1992) 5019.
[3] S. Chandrasekhar, Arkivoc xiii (2005) 37.
[4] J.V. Metzger, in: A.R. Katritzky, C.W. Rees, K.T. Potts (Eds.),
Comprehensive Heterocyclic Chemistry, vol. 6, Pergamon Press,
Oxford, 1984, p. 273, and references cited therein.
[5] G. Lowe, in: Sir Derek Barton, W.D. Ollis, E. Haslam (Eds.),
Comprehensive Organic Chemistry, vol. 5, Pergamon Press, Oxford,
Vol. 5, 1979, p. 289.
[6] R.C. Cambie, G.R. Clark, T.C. Jones, P.S. Rutledge, G.A. Strange,
P.D. Woodgate, Aust. J. Chem. 38 (1985) 745.
[7] H.E. Howard-Lock, C.J.L. Lock, P.S. Smalley, Can. J. Chem. 63
(1985) 2411.
[8] F. Baert, M. Muller, D. Barbry, D. Couturier, Acta Crystallogr. B43
(1987) 538.
[9] E.L. Eliel, S.H. Wilen, L.N. Mander, Stereochemistry of Organic
Compounds, Wiley, New York, 1994, p. 550.
[10] G.W. Stacy, P.L. Strong, J. Heterocycl. Chem. 5 (1968) 101.
[11] I.R. Schmolka, P.E. Spoerri, J. Am. Chem. Soc. 79 (1957) 4716.
[12] B.S. Furniss, A.J. Hannaford, V. Rogers, P.W.G. Smith, A.R.
Tatchell, Vogel’s Textbook of Practical Organic Chemistry, fourth
ed., Longmans Group, Harlow, 1978, p. 683.
Although stacking and hydrogen bonding interactions
determine the lattice packing, these are different in each
case.
Thus, the parent 4-nitrobenzoyl derivative 1 exhibits
stacking between the aromatic ring and the nitro group
(involving both the nitrogen and oxygen atoms). The par-
ent N-acetyl-2-phenyl derivative 2 shows hydrogen bond-
ing interactions involving the aromatic p system and an
aromatic C–H moiety. However, in the mercuric chloride
complex 3, p–p stacking involving the nitrophenyl rings
are apparent in the asymmetric unit. Similar p–p stacking
involving the phenyl rings of the N-benzoyl moiety are
apparent in the mercuric bromide complex 4, apart from
hydrogen bonding interactions between the aromatic p sys-
tem and an aromatic C–H moiety. (Thus, interestingly, the

S. Chandrasekhar et al. / Journal of Molecular Structure 837 (2007) 118–131
131
[13] N.A. Bell, T.N. Branston, W. Clegg, L. Parker, E.S. Raper, C.
Sammon, C.P. Constable, Inorg. Chim. Acta 319 (2001) 130.
[14] Full details of our study have been deposited with the Cambridge
Crystallographic Data Centre, 12 Union Road, Cambridge CB2 1EZ,
U.K., (e-mail: deposit@ccdc.cam.ac.uk), and can be obtained by
quoting the following depository numbers (sequentially for 1–4):
CCDC 225046, CCDC 227436, CCDC 225047, CCDC 225048.
[15] R.T. LaLonde, L. Codacovi, H. Cun-heng, X. Cang-fu, J. Clardy,
B.S. Krishnan, J. Org. Chem. 51 (1986) 4899.
[16] M.G.B. Drew, A.V. George, N.S. Isaacs, H.S. Rzepa, J. Chem. Soc.
Perkin Trans. 1 (1985) 1277.
[17] J. March, Advanced Organic Chemistry, fourth ed., Wiley, New
York, 1992, p. 24, and references cited therein.
[21] J. Buckingham (Ed.), Dictionary of Organic Compounds, fifth ed.,
vol. 5, Chapman and Hall, New York, 1982, p. 5326.
[22] C.I. Braenden, Arkiv. Kemi. 22 (1964) 495.
[23] T. Ohwada, T. Achiwa, I. Okamoto, K. Shudo, Tetrahedron Lett. 39
(1998) 865.
[24] M.I. Page, Adv. Phys. Org. Chem. 23 (1987) 165.
[25] E.L. Eliel, S.H. Wilen, L.N. Mander, Stereochemistry of Organic
Compounds, Wiley, New York, 1994, p. 758.
[26] D. Cremer, J.A. Pople, J. Am. Chem. Soc. 97 (1975) 1354–
1358.
[27] A.L. Spek, Acta Crystallogr. Sect A 46 (1990) C34.
[28] N.N. Greenwood, A. Earnshaw, Chemistry of the Elements, second
ed., Elsevier, Amsterdam, 1997, pp. 914, 1217.
[29] S. Coulton, I. Franc¸ois, Prog. Med. Chem. 31 (1994) 297.
[30] (a) S. Chandrasekhar, S. Srinivasa Gopalan, J. Chem. Res. (M) (2002)
1139;
[18] B.M. Pinto, R.Y.N. Leung, in: G.R.J. Thatcher (Ed.), The
Anomeric Effect and Associated Stereoelectronic Effects, ACS
Symposium Series, vol. 539, American Chemical Society, Washing-
ton, DC, 1993, p. 146.
(b) J. Chem. Res. (S) (2002) 532.
[19] J.M. Sprague, A.H. Land, in: R.C. Elderfield (Ed.), Heterocyclic
Compounds, vol. 5, Wiley, New York, 1957, pp. 700–702.
[20] J. March, Advanced Organic Chemistry, fourth ed., Wiley, New
York, 1992, p. 212 and references cited therein.
[31] K. Gopalaiah, Ph.D. Thesis, Indian Institute of Science, May
2005.
[32] M.S. Wilke, A.L. Lovering, N.C.J. Strynadka, Curr. Opin. Microbiol.
8 (2005) 525.