Structural control of dithiazolyl radicals: Case studies on 3′- and 4′-cyano-benzo-1,3,2-dithiazolyl, NCC6H3S2N

Article abstract of DOI:10.1016/j.jorganchem.2006.12.020

Dithiazolyl radicals with π-stacking motifs have attracted particular interest because of their ability to exhibit spin-switching between diamagnetic distorted π-stacks and paramagnetic regular π-stacked structures through a solid state phase transition.

Full text of DOI:10.1016/j.jorganchem.2006.12.020

Journal of Organometallic Chemistry 692 (2007) 2750–2760
www.elsevier.com/locate/jorganchem
Structural control of dithiazolyl radicals: Case studies on
30- and 40-cyano-benzo-1,3,2-dithiazolyl, NCC6H3S2N
*
Antonio Alberola, Jonathan Burley, Rebecca J. Collis, Robert J. Less, Jeremy M. Rawson
Department of Chemistry, The University of Cambridge, Lensfield Road, Cambridge CB2 1EW, UK
Received 14 October 2006; received in revised form 6 December 2006; accepted 6 December 2006
Available online 23 December 2006
Abstract
Dithiazolyl radicals with p-stacking motifs have attracted particular interest because of their ability to exhibit spin-switching between
diamagnetic distorted p-stacks and paramagnetic regular p-stacked structures through a solid state phase transition. Previous studies
indicate that inclusion of electronegative heteroatoms into the backbone favours lamellar structures. This methodology has been
extended to the synthesis and characterisation of the title compound, 40-cyanobenzo-1,3,2-dithiazolyl (4-NCBDTA). Its electronic struc-
ture is probed through DFT calculations, cyclic voltammetry and EPR spectroscopy and its crystal structure determined by X-ray pow-
der diffraction at room temperature. Variable temperature SQUID magnetometry reveals that 4-NCBDTA undergoes two phase
transitions, each exhibiting bistability; a high temperature phase transition occurs at room temperature (TCfl = 291 K, TC› = 304 K,
DT = 13 K); whilst the low temperature phase transition occurs below liquid nitrogen temperatures (TCfl = 37 K, TC› = 28 K;
DT = 9 K).
ꢀ 2007 Elsevier B.V. All rights reserved.
Keywords: 1,3,2-dithiazolyl radicals
1. Introduction
Whilst there has been some success in crystal structure pre-
diction, [6] it is arguable that a large number of structures
are rationalised in a ‘post facto’ manner. In this article we
describe how we have utilised basic concepts in crystal
engineering in order to construct new materials with well-
defined physical properties. We shall also see that there
are still a number of challenges which must be resolved if
we are to proceed beyond a superficial approach to struc-
ture design.
Our work in this area stemmed from observations of
bistability in the thiazyl radical TTTA (1) in which a
solid-solid phase transition occurs between a diamagnetic
‘low spin’ distorted p-stack phase and a paramagnetic ‘high
spin’ regular spaced p-stack [7]. This system is of particular
interest since the temperatures for the low spin ! high spin
transition and high spin ! low spin transitions are inequiv-
alent. The consequence is a wide region of bistability for 1
in which both ‘low spin’ diamagnetic and ‘high spin’ para-
magnetic phases are stable. For 1, TC(fl) = 230 K and
TC(›) = 320 K providing a 90 K window of bistability
There has been considerable interest in recent years in
the development of molecular materials in which the bulk
physical properties can be tailored at the molecular level
[1]. Of considerable interest have been the physical proper-
ties associated with non-centrosymmetric structures such as
piezo-, pyro- and ferro-electric materials as well as non-lin-
ear optic activity [2]. In other areas molecular materials
have provided examples of electrically conducting [3] and
magnetic materials [4]. In all these cases the physical prop-
erties are intimately associated with their solid state struc-
ture. As synthetic chemists, the manipulation of both the
connectivity and electronic structure at the molecular level
is usually within our capabilities, yet control of the weaker
intermolecular forces (van der Waals, dipolar and electro-
static interactions) still offers a considerable challenge [5].
*
Corresponding author. Tel.: +44 1223336319; fax: +44 1223336319.
E-mail address: jmr31@cus.cam.ac.uk (J.M. Rawson).
0022-328X/$ - see front matter ꢀ 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.jorganchem.2006.12.020

A. Alberola et al. / Journal of Organometallic Chemistry 692 (2007) 2750–2760
2751
which encompasses room temperature. Work by Oakley
has shown that other dithiazolyl radicals such as PDTA
(2) and QDTA (3) exhibit similar behaviour [8]. Work by
Awaga has shown that spin-switching in 1 can additionally
be induced by light irradiation or application of pressure
[9]. Structurally radicals 1–3 exhibit certain similarities,
specifically the formation of p-stacked structures. Crystal-
lographic studies indicate that the change in the magnetic
response is associated with a structural distortion of the
p-stack [7,8]. Our target has therefore been to develop
new bistable materials through the systematic design of
well-defined p-stacked structures. Particular objectives
were (i) to prepare layer-like motifs; (ii) to investigate
whether we are able to predict the nature of the intermolec-
ular interactions within layers; (iii) establish whether there
are further structural considerations which must be taken
into account in order to develop new bistable switching
materials.
therefore be incorporated in order to stabilise p-stacked
structures. These directional interactions embrace a large
number of the so-called supramolecular synthons such as
O–Hꢀ ꢀ ꢀO, N–Hꢀ ꢀ ꢀO as well as weaker C–Hꢀ ꢀ ꢀO and C–
Hꢀ ꢀ ꢀN hydrogen-bonding motifs [10]. Hydrogen bonds
are by no means exclusive and other structure-directing
motifs have been identified such as CNꢀ ꢀ ꢀX (X = I, Br,
Cl) [11]. In the case of thiazyl chemistry, Sꢀ ꢀ ꢀN contacts
are prevalent and Sꢀ ꢀ ꢀN contacts between heterocyclic
rings e.g. in thiadiazolyl rings [12] and CNꢀ ꢀ ꢀS interactions
in dithiadiazolyl rings have been successfully used as struc-
ture-directing motifs [13].
3. A comparison of aromatic and N-substituted dithiazolyl
radicals
The structures of benzodithiazolyl (BDTA) [14] and
naphthyldithiazolyl (NDTA) [8] are presented in Fig. 1
and exhibit herringbone motifs typical of simple aromatics.
Substitution of two C–H units by isoelectronic N atoms in
the aromatic core of these two radicals chemically gener-
ates the isoelectronic pyrazine and quinoxaline derivatives
(PDTA and QDTA). Both these radicals provide addi-
tional functionality for electrostatic interactions and both
PDTA and QDTA radicals reported by Oakley adopt
layer-like motifs in which the additional heterocyclic ring
N atoms can be involved in electrostatically favourable
interactions of the form C–Hꢀ ꢀ ꢀN or Sꢀ ꢀ ꢀN (Fig. 2) [8].
N
N
N
N
N
N
S
S
S
S
S
S
.
N
.
N
.
N
S
1
2
3
2. Structures of aromatic p-systems
A number of studies on planar p-aromatics have indi-
cated a strong tendency to adopt either of two basic struc-
tures; either a p-stack or a herringbone motif [10]. Detailed
studies on benzene and related systems indicate that for
aromatic hydrocarbons there are three types of intermolec-
ular interactions; p ꢀ ꢀ ꢀ p, C–Hꢀ ꢀ ꢀp and C–Hꢀ ꢀ ꢀH–C. Of
these the C–Hꢀ ꢀ ꢀp interaction is ca. 12 kJ/mol whereas p–
p interactions are ca. 8 kJ/mol [10]. As a consequence the
majority of neutral aromatics adopt the more thermody-
namically stable herringbone motif. This preference for
herringbone motifs means that additional interactions must
4. Development of new spin-transition dithiazolyl radicals
In order to prepare new dithiazolyl radicals capable of
undergoing spin transitions analogous to TTTA and
related p-stacked structures such as PDTA and QDTA
we targeted the two novel radicals 3-NCBDTA and 4-
NCBDTA in which the electronegative CN group was
available for in-plane interactions to either the aromatic
C–H or heterocyclic S atoms. A preliminary account of
the synthesis and characterisation of 3-NCBDTA has
recently been reported [15].
Fig. 1. Crystal structures of (a) BDTA and (b) NDTA.

2752
A. Alberola et al. / Journal of Organometallic Chemistry 692 (2007) 2750–2760
Fig. 2. Crystal structures of (a) PDTA and (b) QDTA viewed perpendicular to the stacking axis whilst (c) and (d) are views of PDTA and QDTA
respectively parallel to the stacking axis.
4.1. Synthesis of 4-NCBDTA
Subsequent dissolution of the sulfenyl chloride in
CH2Cl2 followed by condensation with Me3SiN3 generated
the 40-cyanobenzo-1,3,2-dithiazolylium chloride, [4-
NCBDTA]Cl in good yield. This was then reduced with
Ag powder in liquid SO2 to generate the desired 4-
NCBDTA which was subsequently purified by vacuum
sublimation (10ꢁ1 Torr, 70–25 ꢁC) (Scheme 1).
The synthesis of 4-NCBDTA utilised the same method-
ology previously applied to the isomeric 3-NCBDTA [19].
In both cases the key benzo-substituted sulfenyl chloride
was prepared in two steps from the parent dichlorobenzo-
nitrile and ‘‘Less’ reagent’’, [tBuS]Na, which we have found
provides a mild route into sulfenyl chloride chemistry;
[15,16] Treatment of dichlorobenzonitrile with two equiva-
lents of [tBuS]Na in DMI (1,3-dimethyl-2-imidazolidinone)
led to the dithiolate under mild conditions. The tert-butyl
derivative is considerably bulky and deprotection of the
thiolate occurs readily by chlorination with Cl2 in CCl4
at 0 ꢁC. In comparison, reductive deprotection of thiolates
to generate dithiols tends to require harsh conditions e.g.
Na/l Æ NH3 which may not always support chemically sen-
sitive functional groups [17].
4.2. Electrochemical studies on 4-NCBDTA
Solutions of both 3-NCBDTA and 4-NCBDTA were
studied by cyclic voltammetry and revealed reversible oxi-
dation waves to the corresponding cations. These are pre-
sented in Fig. 3. The presence of the electron-
withdrawing CN group leads to a stabilisation of the radi-
cal centre and a higher oxidation potential in relation to
BDTA [+0.15 V vs SCE]; [13] 3-NCBDTA and 4-
Cl
S
S
Bu
Bu
2 [tBuS]Na
S
S
Cl
Cl
Cl2
CCl4
DMI
C
Cl
C
C
N
N
N
Me3SiN3
CH2Cl2
+
S
S
S
Ag
Cl-
.
N
N
S
l. SO2
C
C
N
N
Scheme 1. Synthetic methodology for formation of 4-NCBDTA.

A. Alberola et al. / Journal of Organometallic Chemistry 692 (2007) 2750–2760
2753
1.50
1.00
0.50
0.50
0.40
0.30
0.20
0.10
0.00
0.00
-0.50
-1.00
-1.50
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
-0.10
-0.20
-0.30
-0.40
-0.50
-0.60
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
E/V(SCE)
E/V(SCE)
Fig. 3. Cyclic voltammograms of 3-NCBDTA (left) and 4-NCBDTA (right).
NCBDTA show reversible 1 eꢁ 0/+1 couples with E1/2
0.491 V and 0.447 V respectively vs SCE.
=
of cyano-substitution on the electronic structure and redox
behaviour of these derivatives. The calculated spin density
distributions are presented in Fig. 5.
4.3. EPR studies on 3-NCBDTA and 4-NCBDTA
The inclusion of the electron-withdrawing CN group
leads to a small reduction in the calculated spin density
at the SNS component of the dithiazolyl ring (91.1% of
the calculated spin density residing on the S2N fragment
in BDTA cf. 85.8% and 88.1% in 3-NCBDTA and 4-
NCBDTA respectively). Nevertheless the unpaired spin
density at the dithiazolyl ring N atom is little affected at
46.3 1.0% and the spin density at the cyano-N atom is
small in both derivatives. As a consequence these small
changes in spin density at the spin active 14N nuclei are
unlikely to be manifested in changes in EPR spectra. How-
ever the energy of the singly-occupied molecular orbital is
lowered significantly for both cyano-derivatives indicating
that they should be more readily generated by 1eꢁ reduc-
tion of the corresponding dithiazolylium cation. The
invariance in EPR spectra and trends in electrochemical
behaviour are in good agreement with these predictions.
The calculated molecular geometries for both cyano-
substituted radicals are in reasonable agreement with those
EPR studies on frozen solutions of both compounds in
THF were recorded at X-band and revealed a rhombic
character typical of dithiazolyl radicals [18]. Both spectra
revealed large hyperfine coupling to the dithiazolyl ring
N but no hyperfine coupling to the nitrile N atom could
be resolved indicating couplings similar to or less than
the inherent linewidth of the EPR spectra (2.5 G). A typical
spectrum is shown in Fig. 4. Both spectra could be simu-
lated using similar sets of parameters (Table 1) indicating
very small changes in spin density at the spin active 14N
nuclei.
4.4. Theoretical calculations on 3-NCBDTA and
4-NCBDTA
DFT calculations were performed on BDTA, 3-
NCBDTA and 4-NCBDTA in order to probe the effect
Fig. 4. Frozen solution EPR spectrum in dichloromethane at 220 K and simulation using parameters in Table 1 of 3-NCBDTA.

2754
A. Alberola et al. / Journal of Organometallic Chemistry 692 (2007) 2750–2760
Table 1
Anisotropic EPR parameters used to simulate frozen solution spectra of 3-NCBDTA and 4-NCBDTA
g-Tensor
3-NCBDTA
4-NCBDTA
aN
3-NCBDTA/G
4-NCBDTA/G
g1
g2
g3
Ægæ
2.0130
2.0051
2.0017
2.0066
2.0146
2.0067
2.0033
2.0082
a1N
2.5
2.5
29.4
11.5
2.5
2.5
28.3
11.1
a2N
a3N
ÆaNæ
a
b
c
+ 3 . 2
+ 5 . 7
+ 4. 1
+ 2 1 . 7
+ 2 1 . 2
+1 9. 2
S
S
S
+ 2 . 4
+ 2 .4
+ 2 . 0
+ 4. 6
+0 . 8
- 0 . 6
- 1 . 1
- 1 . 5
-1 . 5
- 0. 7
+ 4 7. 3
+ 4 5. 3
+ 46 . 6
N
N
N
+ 3. 1
- 0. 3
-0 . 6
+3 . 2
+ 4 . 0
+5 . 2
S
S
S
N
+ 2 2 . 1
+ 1 9 . 3
+ 22 . 3
+ 1. 9
- 0. 2
+ 1 . 3
N
HOMO = -4.102 eV
HOMO = -4.662 eV
HOMO = -4.698 eV
Fig. 5. Spin density distribution and HOMO energy calculated for: (a) BDTA; (b) 3-NCBDTA and (c) 4-NCBDTA.
˚
determined by X-ray diffraction (Table 2) although the
BP86/DN* basis set appears to slightly overestimate most
bond lengths by ca. 1–2%.
between molecules in different layers (dCNꢀ ꢀ ꢀH = 3.056 A,
Fig. 7a). These CNꢀ ꢀ ꢀH contacts are short and comparable
with those seen in co-crystals of sym-tricyanobenzene [20a].
Indeed recent work [15] has suggested that the cyano-nitro-
gen atom is better able to act as a hydrogen bond acceptor
than more conventional acceptors such as amines and thus
short contacts to weak H-bond donors such as aromatic C–
H groups are probably significant.
At 300 K the unit cell of 3-NCBDTA is halved. Whilst it
retains the monoclinic P21/c setting there is now a single
molecule in the asymmetric unit and a regular spacing
4.5. Crystal structure of 3-NCBDTA
The crystal structures of 3-NCBDTA have been
reported previously [15] but a brief description is included
here for comparison to 4-NCBDTA and within the context
of the discussion of the calculated molecular electrostatic
isopotential maps (Section 4.7). At 180(2) K, 3-NCBDTA
crystallises in the monoclinic space group P21/c with two
molecules in the asymmetric unit. These radicals form cofa-
cial p*–p* dimers with intradimer Sꢀ ꢀ ꢀS contacts of 3.263
˚
(3.665 A) of radicals parallel to the crystallographic a-axis.
Within the bc-plane the radicals retain both the Sꢀ ꢀ ꢀN and
CNꢀ ꢀ ꢀH–C contacts identified in the low temperature
phase. The Sꢀ ꢀ ꢀN contacts are now located about an inver-
sion centre but slightly longer than in the low temperature
˚
and 3.347 A.These adopt a distorted p-stacked motif along
the crystallographic a-axis. The inter-dimer Sꢀ ꢀ ꢀS contacts
˚
˚
along this p-stacking direction are 3.886 and 3.971 A.
structure with Sꢀ ꢀ ꢀN = 3.322 A. The corresponding in-
Within the bc-plane dimers are linked via alternating sets
plane CNꢀ ꢀ ꢀH–C contacts are also located about an inver-
˚
of Sꢀ ꢀ ꢀN contacts between dithiazolyl rings (dSꢀ ꢀ ꢀN
=
sion centre. The CNꢀ ꢀ ꢀH contacts (2.818 A) now lie mid-
˚
˚
3.149–3.287 A) and CNꢀ ꢀ ꢀH–C contacts (2.645 A) which
propagate a ribbon-like motif parallel to the crystallo-
graphic b-axis (Fig. 6). The Sꢀ ꢀ ꢀN contacts are similar to
or a little less than the sum of the van der Waals radii of
way between the intra-plane and inter-plane contacts
observed in the low temperature phase.
4.6. Crystal structure of 4-NCBDTA
˚
S and N (3.2 A for in-plane contacts) [19]. The molecular
planes of the two crystallographically independent mole-
cules are inclined at 18.97ꢁ and 18.69ꢁ with respect to bc-
plane. The twisting of the molecules out of the bc-plane
affords an additional set of longer CNꢀ ꢀ ꢀH–C interactions
Attempts to grow suitable crystals of 4-NCBDTA for
single-crystal X-ray diffraction at a range of temperatures
above and below the high temperature phase transition
region (identified from SQUID data to be between 291

A. Alberola et al. / Journal of Organometallic Chemistry 692 (2007) 2750–2760
2755
Table 2
Heterocyclic bond lengths and angles for 3-NCBDTA and 4-NCBDTA
with those determined by DFT theory
3-NCBDTA
4-NCBDTA
Observed
Observeda
Calculated
1.417
Calculated
1.419
˚
Bond length (A)
C–C
C–S
S–N
N–S
S–C
1.394(8)b
1.399(4)
1.397(4)
1.394(7)
1.394(9)
1.742(6)b
1.746(3)
1.749(3)
1.760
1.667
1.665
1.753
1.742(8)
1.742(6)
1.751
1.668
1.671
1.763
1.634(6)b
1.642(3)
1.650(2)
1.633(7)
1.632(7)
Fig. 6. Structure of 3-NCBDTA at 180(2) K viewed parallel to the a-axis.
1.673(6)b
1.655(3)
1.650(3)
1.673(6)
1.673(6)
the normal effects due to thermal expansion there is no
clear evidence from the diffraction data of any phase tran-
sition on warming from 250 to 350 K. The absence of
super-lattice reflections below 290 K (expected for a Peierls
distortion of the one-dimensional p-stack) may reflect dif-
fuse/weaker scattering from the super-lattice reflections.
Structure solution from powder diffraction data has
developed rapidly over the last 10 years [21] and has been
applied to a range of both organic and inorganic materials,
although it is still far from being a routine technique. Nev-
ertheless we have been successful in applying this method-
ology to a number of thiazyl radicals [22] and it proved
possible to solve the structure of 3 from X-ray powder dif-
fraction data. Full details of the structure solution of 3 are
presented in the Supporting Information.
1.737(5)b
1.743(3)
1.742(3)
1.738(9)
1.738(8)
Bond angle (ꢁ)
C–S–N
S–N–S
N–S–C
S–C–C
100.0(3)b
100.45(13)
99.82(13)
99.78
100.01(11)
100.02(12)
100.03
114.09
99.52
114.5(3)b
114.29(14)
114.48(14)
114.26
99.64
114.59(15)
114.62(14)
98.4(3)b
98.95(13)
99.16(13)
98.39(13)
98.35(14)
114.4(4)b
114.4(2)
114.1(2)
113.57
112.75
114.37(14)
114.40(15)
113.18
113.17
The powder pattern at 330 K was indexed on a mono-
clinic cell (P21) whose volume indicated two crystallo-
graphically independent molecules in the asymmetric
unit. Structure solution and refinement revealed a structure
with striking similarities to 3-NCBDTA. Specifically the
molecules adopt a layer-like structure within the ab-plane.
Close contacts comprise a set of CNꢀ ꢀ ꢀH–C interactions
C–C–S
112.6(5)b
111.8(2)
112.4(2)
112.6(3)
112.6(5)
a
Number marked.
Corresponds to the high temperature phase.
b
˚
with similar geometric parameters (dCNꢀ ꢀ ꢀH = 2.554 A) to
and 304 K) have so far proved fruitless. However sublimed
samples provided good quality X-ray powder diffraction
patterns over the temperature range 250–350 K. Besides
3-NCBDTA albeit with shorter C–Nꢀ ꢀ ꢀH contacts. How-
ever, whilst in-plane Sꢀ ꢀ ꢀN contacts are retained, the cen-
trosymmetric Sꢀ ꢀ ꢀN motif in 3–NCBDTA is replaced by
Fig. 7. Structure of 3-NCBDTA viewed (a) in the ac-plane at 180(2) K and (b) in the ab-plane at 300(2) K.

2756
A. Alberola et al. / Journal of Organometallic Chemistry 692 (2007) 2750–2760
a near linear N–Sꢀ ꢀ ꢀN–S contact which generates a chain
parallel to the crystallographic b-axis (dSꢀ ꢀ ꢀN = 2.880 and
˚
3.062 A, cf. sum of the van der Waals radii for S and N
˚
at 3.20 A) [19]. The result is a set of two crystallographi-
cally independent but co-parallel chains parallel to the
crystallographic b-axis with each chain comprising Sꢀ ꢀ ꢀN
contacts. Independent chains are then linked via
CNꢀ ꢀ ꢀH–C contacts (Fig. 8). Like 3-NCBDTA molecules
are twisted with respect to these layers. In 4-NCBDTA
one is inclined at 7.34ꢁ whereas the second independent
molecule is inclined at 21.81ꢁ.
Perpendicular to the stacking axis radicals form a regu-
˚
lar p-stack with Sꢀ ꢀ ꢀS separations of 3.619 A (equivalent to
Fig. 9. View of 4-NCBDTA illustrating the interlayer CNꢀ ꢀ ꢀH–C
the crystallographic c-axis) and comparable to that
contacts.
observed for the high temperature phase of 3-NCBDTA
˚
(3.665 A at 300(2) K). The twisting of the molecules out
of the molecular plane facilitates additional short inter-
˚
plane CNꢀ ꢀ ꢀH–C contacts (dCNꢀ ꢀ ꢀH–C = 2.588 A) of similar
geometry to the in-plane contacts. These contacts generate
a helical arrangement along the crystallographic c-axis
(Fig. 9).
4.7. Analysis of the structures of 3-NCBDTA and
4-NCBDTA
The MEPs of these two radicals are illustrated in
Fig. 10. In both cases the cyano group bears the largest
partial negative charge with smaller contributions on the
heterocyclic dithiazolyl ring N atom. In both cases the
molecular structures would appear to favour either of the
electrostatically favourable CNꢀ ꢀ ꢀS or CNꢀ ꢀ ꢀH–C interac-
tions and a CNꢀ ꢀ ꢀH–C interaction appears as a recurrent
theme. Notably in both PDTA and QDTA the correspond-
ing pyridyl-Nꢀ ꢀ ꢀH–C interactions appeared structure-
directing. Whilst the CNꢀ ꢀ ꢀH–C motif persists in both 3-
NCBDTA and 4-NCBDTA, the centrosymmetric Sꢀ ꢀ ꢀN
dimer motif (common to thiazyl systems) [12] is present
Fig. 10. Molecular electrostatic isopotential surfaces for (a) 3-NCBDTA
and (b) 4-NCBDTA.
in 3-NCBDTA but absent in 4-NCBDTA. This would sug-
gest that the CNꢀ ꢀ ꢀH–C contact is more favourable.
In 3-NCBDTA each molecule offers one CN and one C–
H group for this interaction and the shortest of the inter-
molecular CNꢀ ꢀ ꢀH interactions generate dimers close to
the bc-plane. In 4-NCBDTA each molecule again offers
one CN group and a C–H hydrogen bond donor. However
Fig. 8. View of (a) 4-NCBDTA in the ab-plane and (b) crystallographically independent molecules emphasize the two sets of independent Sꢀ ꢀ ꢀN chains
linked via the CNꢀ ꢀ ꢀH–C contacts.

A. Alberola et al. / Journal of Organometallic Chemistry 692 (2007) 2750–2760
2757
the molecules of 4-NCBDTA are twisted out of the ac-
0.0009
0.0008
0.0007
0.0006
0.0005
0.0004
0.0003
0.0002
0.0001
0.0000
a
plane so as to form layers linked via CNꢀ ꢀ ꢀH contacts
(Fig. 9). We believe that these different methods of propa-
gation of the CNꢀ ꢀ ꢀH contacts lie at the origin of the subtle
differences in their magnetic properties (section 5.10).
4.8. Magnetic studies on 3-NCBDTA and 4-NCBDTA
The magnetic properties of p-stacked DTA radicals are
typically reflected in a diamagnetic-paramagnetic phase
transition which is associated with a solid-state transforma-
tion from regular p-stack to distorted p-stack motif [23].
With the exception of 3-NCBDTA (which we reported
recently [19]) the phase transition is typically associated
with a region of bistability i.e. there is a temperature regime
in which both diamagnetic and paramagnetic phases can
co-exist and in which the magnetic response is strongly
dependent upon sample history [7,8]. In some instances
the temperature region of bistability can be large (>90 K).
Magnetic susceptibility measurements on 4-NCBDTA
were made on a Quantum Design SQUID magnetometer
in an applied field of 5000 G between 5 K and 350 K. Data
were corrected for both sample diamagnetism (Pascal’s
constants) and the sample holder.
0
10
20
30
40
50
Temperature (K)
0.0003
0.0002
0.0001
0.0000
-0.0001
b
On heating a sample of 4-NCBDTA from 270 to 350 K
an abrupt phase transition is observed at 304 K (31 ꢁC) to a
paramagnetic state. The value of vT then increases steadily
200
250
300
350
Temperature (K)
up to 350 K, reaching a value of 0.04 emu.K.Oeꢁ1molꢁ1
,
Fig. 11. Temperature dependence of v for both 3-NCBDTA upon heating
(d) and cooling (s) and 4-NCBDTA upon heating (j) and cooling (h) in
(a) the low temperature regime (<50 K) and (b) the high temperature
regime (200–350 K).
just 10% of that expected for an S = 1/2 paramagnet. This
is consistent with a very strongly correlated electronic
regime and comparable to the value of 0.16
emu.K.Oeꢁ1molꢁ1 at 350 K for 1 [7]. This phase transition
was found to be reversible (Fig. 11a) but, unlike 3-
4.9. Mechanism of phase-switching
NCBDTA, it does exhibit thermal hysteresis (TCfl
291 K, TC› = 304 K, DT = 13 K).
=
The further development of spin-switching materials
based on dithiazolyl radicals relies heavily on developing
an understanding of the mechanism of the phase transition
and the origins of the bistability. The origin of the thermal
hysteresis has been associated with a large activation bar-
rier to structural reorganisation from an enthalpically sta-
bilised distorted p-stack and an entropically stabilised
regular stack. Previous work by Oakley has focused on
the cleavage and reformation of inter-layer Sꢀ ꢀ ꢀN contacts
between molecules as the likely origin of this activation
energy in TTTA and PDTA [8]. The inter-layer contacts
arise through tilting of the molecule with respect to the
stacking axis in the regular stacked phases and formation
of in-plane contacts in the low temperature distorted p-
stack structures.
In the case of 3-NCBDTA there is no hysteresis
(DT < 1 K) associated with the phase transition indicative
of a minimal reorganisation energy. Indeed the electrostat-
ically favourable intermolecular Sꢀ ꢀ ꢀN and CNꢀ ꢀ ꢀH con-
tacts in 3-NCBDTA lie within the layers with the two
independent molecules twisted by just 3.7ꢁ from coplana-
rity in the low temperature phase and coplanar in the high
temperature phase. The corresponding Sꢀ ꢀ ꢀN and CNꢀ ꢀ ꢀH
Upon cooling the sample of 4-NCBDTA undergoes a
second magnetic phase transition from a diamagnetic
phase to a second paramagnetic phase. The discontinuity
in v associated with this transition mitigates against con-
tamination with an S = 1/2 Curie spin [24]. This transition
also exhibits thermal hysteresis (TCfl = 37 K, TC› = 28 K;
DT = 9 K) (Fig. 11b) and is comparable to the low temper-
ature phase transition observed in 3-NCBDTA
(TCfl = 40 K, TC› = 25 K; DT = 15 K). Previous studies
on spin-transition compounds [7,8,25] typically favour
the entropically stabilised paramagnetic regime at higher
temperature. The phase transition to a paramagnetic phase
upon cooling is therefore highly unusual. Moreover the
thermal hysteresis in this low temperature regime is also
anomalous; the phase transition temperature for both 3-
NCBDTA and 4-NCBDTA upon warming occurs below
the phase transition for the cooling process. This should
be contrasted with other bistable materials in which the
phase transition on warming typically occurs above the
phase transition for the cooling process, as is the case for
the high temperature phase transition in 4-NCBDTA (Sec-
tion 4.9).

2758
A. Alberola et al. / Journal of Organometallic Chemistry 692 (2007) 2750–2760
Fig. 12. Propagation of electrostatically favourable CNꢀ ꢀ ꢀH–C motifs generating (a) discrete centrosymmetric dimers in 3-NCBDTA and (b) chains in 4-
NCBDTA.
interstack contacts between layers are substantially longer.
Consequently the structural reorganisation via a distortion
along the a-axis can occur without cleavage of the strong
inter-stack contacts.
Both 3-NCBDTA and 4-NCBDTA exhibit phase transi-
tions between regular and distorted p-stacked motifs. In
3-NCBDTA this occurs without thermal hysteresis at
250 K whereas in 4-NCBDTA a small region of thermal
hysteresis is observed (TCfl = 291 K, TC› = 304 K, DT =
13 K). The absence of bistability in 3-NCBDTA and pres-
ence in 4-NCBDTA has been attributed to the activation
energy associated with the cleavage of interlayer Sꢀ ꢀ ꢀN
and particularly CNꢀ ꢀ ꢀH contacts during the solid state
transformation. Whilst layer-like structures favour spin-
switching, the evidence here suggests that rotation of the
molecular plane with respect to the normal to the stacking
direction facilitates interstack/interlayer contacts which
appears important for bistability.
In addition to these high-temperature phase transitions,
both 3-NCBDTA and 4-NCBDTA exhibit low-tempera-
ture transitions with thermal hysteresis. In both cases the
behaviour is unprecedented, with the spin transition driven
to the paramagnetic phase upon cooling. In addition the
phase transition exhibits an unusual ‘inverse’ behaviour
of the hysteresis, the origins of which are under further
investigation.
In the case of 4-NCBDTA only the structure of the high
temperature phase is unambiguously identified. Here there
are two crystallographically independent molecules. Whilst
one molecule lies close to the ab-plane (7.34ꢁ), the second is
twisted more substantially out of this plane (21.81ꢁ). This
gives rise to a set of Sꢀ ꢀ ꢀN contacts close to the molecular
˚
plane (dSꢀ ꢀ ꢀN = 3.062 A) and between different planes
˚
(dSꢀ ꢀ ꢀN = 3.355 A). In addition the twisting also facilitates
CNꢀ ꢀ ꢀH contacts in which the CNꢀ ꢀ ꢀH contacts between
molecules in neighbouring layers are near equivalent
˚
dCNꢀ ꢀ ꢀH = 2.554 and 2.588 A and also shorter than those
in 3-NCBDTA. A comparison of the CNꢀ ꢀ ꢀH contacts in
3-NCBDTA and 4-NCBDTA are presented in Fig. 12.
Whilst 3-NCBDTA can undergo a lattice distortion with-
out cleavage of intermolecular CNꢀ ꢀ ꢀH contacts, dimerisa-
tion in 4-NCBDTA is likely to proceed via cleavage and
reformation of the interstack CNꢀ ꢀ ꢀH contacts.
At the present time there is no structural information on
the low temperature phase transitions observed for both
materials. However the similarities in both the positions
of their phase transitions and their solid-state motifs would
indicate that the low temperature processes in both cases
are likely to be similar. Indeed the limited thermal energy
available would indicate that this structural rearrangement
must follow a low energy pathway which disrupts the p*–p*
dimerisation motif. A glide between layers is plausible but
further low temperature studies are required.
Further studies are underway to probe the low temper-
ature phase transitions in both 3-NCBDTA and
4-NCBDTA, as well as prepare further bistable materials
in which the molecular orientations are tailored to modify
the window of bistability.
6. Experimental
6.1. Synthesis of 3-NCBDTA
5. Conclusions
2,3-Dichlorobenzonitrile (2 g, 11.6 mmol) was dissolved
in 1,3-dimethyl-2-imidazolidinone (20 ml). Sodium butyl
thiolate (4.7 g, 43 mmol) was added and the reaction mix-
ture heated to 60 ꢁC for 24 h. After allowing the reaction
mixture to cool to room temperature, water (100 ml) was
added and the mixture extracted with diethyl ether
(3 · 75 ml). The organic extracts were combined and
washed with water and brine, and the solvent removed in
vacuo. The product was recrystallised from diethylether
to afford pure C6H3(CN)(StBu)2 (1.9 g, 57%). Analysis
found: C 67.43; H 7.58; N 5.04; C15H21S2N requires: C
An analysis of the structures of simple neutral aromatic
molecules proves to be instructive in the design of dithiaz-
olyl radicals, leading to a controlled predilection for either
p-stacked or herringbone motifs. This work has been
extended to the synthesis of two cyano-functionalised ben-
zodithiazolyl radicals which exhibit layer-like motifs. The
importance of the CNꢀ ꢀ ꢀH contact is identified in the calcu-
lated molecular electrostatic isopotential maps and
reflected in the observed p-stacked structures.

A. Alberola et al. / Journal of Organometallic Chemistry 692 (2007) 2750–2760
2759
64.47; H 7.57; N 5.01%. dH (CDCl3) 1.32 (s, 18H), 7.41 (t,
Acknowledgements
1H), 7.70 (dd, 1H), 7.95 (dd, 1H). m/z (EI) 279.1 (M+,
100%).
We would like to thank the EPSRC for financial support
(R.J.L. and R.J.C.); Homerton College and Jesus College
Cambridge for Fellowships (A.A. and J.B. respectively);
Prof. Coronado at the University of Valencia for use of
the SQUID magnetometer and; Prof. Sanders (University
of Cambridge) for access to the PC Spartan Pro software
for preliminary DFT calculations.
Chlorination of C6H3(CN)(StBu)2 (1.0 g, 3.5 mmol) in
CCl4 (20 ml) under ambient conditions yielded an orange
solution of C6H3(CN)(SCl)2. The solvent was removed in
vacuo and the oily residue redissolved in CH2Cl2 (20 ml).
This was treated with Me3SiN3 (0.4 ml, 5 mmol) to yield
the salt [3-NCBDTA][Cl]. Reduction with Ag powder in
l. SO2 (20 ml) yielded 3-NCBDTA (0.236 g, 35%) which
was purified by vacuum sublimation. Analysis found: C
49.46; H 1.73; N 15.24; C7H3S2N2 requires: C 49.9; H
1.68; N 15.63%.
References
[1] A.M. Reed, J.M. Tour, Sci. Am. 86 (2002) 282;
M.D. Ward, Chem. Soc. Rev. 24 (1995) 121.
[2] Ch. Bosshard, K. Sutter, Ph. Pretre, J. Hulliger, M. Florsheimer, P.
Kaatz, P. Gunter, Organic Nonlinear Optical Materials, Gordon and
Breach, Basel, 1995.
6.2. Synthesis of 4-NCBDTA
3,4-Dichlorobenzonitrile (2 g, 11.6 mmol) was dissolved
in 1,3-dimethyl-2-imidazolidinone (20 ml). Sodium butyl
thiolate (4.7 g, 43 mmol) was added and the mixture heated
to 60ꢁC for 24 h. After cooling room temperature, water
(100 ml) was added and the mixture extracted with diethyl
ether (3 · 75 ml). The organic extracts were combined and
washed with water and brine, and the solvent removed in
vacuo. The product was recrystallised from diethylether to
afford pure C6H3(CN)(StBu)2 (2 g, 61%). Analysis found:
C 64.81; H 7.63; N 4.98. C15H21S2N requires: C 64.47; H
7.57; N 5.01%. dH (CDCl3) 1.41 (s,18H), 7.47 (dd, 1H),
7.68 (d,1H), 7.85 (d, 1H). m/z (EI) 279.1 (M+, 100%).
Chlorination of C6H3(CN)(StBu)2 (1.0 g, 3.5 mmol) in
CCl4 (20 ml) under ambient conditions yielded an orange
solution of C6H3(CN)(SCl)2. The solvent was removed in
vacuo and the oily residue redissolved in CH2Cl2 (20 ml).
This was treated with Me3SiN3 (0.4 ml, 5 mmol) to yield
the salt [4-NCBDTA][Cl]. The salt was reduced with Ag
powder in l. SO2 (20 ml) to yield 4-NCBDTA (0.314 g,
46%) which was purified by vacuum sublimation. Analysis
found: C 46.91; H 1.70; N 15.62. C7H3S2N2 requires: C
49.91; H 1.68; N 15.63%.
[3] B.E. Bowler, A.L. Raphael, H.B. Gray, Prog. Inorg. Chem. 38 (1990)
259.
[4] J.S. Miller, A.J. Epstein, Angew. Chem. Int. Ed. Engl. (1994) 385.
[5] M.D. Hollingsworth, Science (2002) 2410.
[6] See for example: J.P.M. Lommerse, W.D.S. Motherwell, H.L.
Ammon, J.D. Dunitz, A. Gavezzotti, D.W.M. Hofmann, F.J.J.
Leusen, W.T.M. Mooij, S.L. Price, B. Schweizer, M.U. Schmidt,
B.P. van Eijck, P. Verwer, D.E. Williams, Acta Crystallogr. B 56
(2000) 697.
[7] W. Fujita, K. Awaga, Science (1999) 281;
G.D. McManus, J.M. Rawson, N. Feeder, J. van Duijn, E.J.L.
McInnes, J.J. Novoa, R. Burriel, F. Palacio, P. Oliete, J. Mater.
Chem. 11 (2001) 1992.
[8] J.L. Brusso, O.P. Clements, R.C. Haddon, M.E. Itkis, A.A. Leitch,
R.T. Oakley, R.W. Reed, J.F. Richardson, J. Am. Chem. Soc. 126
(2004) 14692;
J.L. Brusso, O.P. Clements, R.C. Haddon, M.E. Itkis, A.A. Leitch,
R.T. Oakley, R.W. Reed, J.F. Richardson, J. Am. Chem. Soc. 126
(2004) 8256;
T.M. Barclay, A.W. Cordes, N.A. George, R.C. Haddon, R.T.
Oakley, T.T.M. Palstra, G.W. Patenaude, R.W. Reed, J.F. Richard-
son, H. Zhang, J. Chem. Soc., Chem. Commun. (1997) 873.
[9] H. Matsuzaki, W. Fujita, K. Awaga, H. Okamoto, Phys. Rev. Lett.
91 (2003) 017403;
W. Fujita, K. Awaga, Chem. Phys. Lett. 393 (2004) 150.
[10] G.R. Desiraju, in: Crystal Engineering: The Design of Organic
SolidsMaterials Science Monographs, vol. 54, Elsevier Press, 1989.
[11] See for example A.D. Bod, J. Griffiths, J.M. Rawson, J. Hulliger,
Chem. Commun. (2001) 2488, and refs therein. See also [10] and
references therein.
[12] T. Chivers, K. McGregor, M. Parvez, I. Vargas-Baca, T. Ziegler,
Can. J. Chem. (1995) 1380.
[13] A.J. Banister, N. Bricklebank, I. Lavender, J.M. Rawson, C.I.
Gregory, B.K. Tanner, W. Clegg, M.R.J. Elsegood, F. Palacio,
Angew. Chem., Int. Ed. Engl. 35 (1996) 2533.
[14] E.G. Awere, N. Burford, C. Mailer, J. Passmore, M.J. Schriver, P.S.
White, A.J. Banister, M. Oberhammer, L.H. Sutcliffe, J. Chem. Soc.,
Chem. Commun. (1987) 66.
[15] A. Alberola, R.J. Collis, S.M. Humphrey, R.J. Less, J.M. Rawson,
Inorg. Chem. 45 (2006) 1903.
[16] A. Alberola, R.J.Collis, R.J. Less, J.M. Rawson, J. Organomet.
Chem, this issue, doi:10.1016/j.jorganchem.2006.11.048.
[17] D. Sellmann, K.P. Peters, R.M. Molina, F.W. Heinemann, Eur. J.
Inorg. Chem. 5 (2003) 903.
6.3. Theoretical calculations
6.3.1. Experimental
Preliminary UHF DFT calculations (LSDA/pBP86/
DN*) were carried out within Spartan Pro [26]. This
approach utilises a perturbative Becke-Perdew (pBP86)
procedure [27] within the local spin density approximation
LSDA. Instead of Gaussian basis sets, Spartan Pro utilises
atomic solutions supplemented with d-type functions for
heavy atoms, including numerical polarisation (DN*). It
is generally considered that the pBP86 method compares
most closely to the Gaussian B88-P86 method, whilst the
DN* basis set is close to 6-311+G*. Comparisons of
DFT methods can be found in articles cited in Ref. [28].
Initial trial geometries were determined using semi-empiri-
cal methods (PM3) and then optimised freely. All
structures were stationary points with no imaginary fre-
quencies consistent with energy minima.
[18] S.M. Mattar, Chem. Phys. Lett. 300 (1999) 542;
Y.-L. Chung, S.A. Fairhurst, D.G. Gillies, G. Kraft, A.M.L.
¨
Krebber, K.F. Preston, L.H. Sutcliffe, G. Wolmershauser, Magn.
Reson. Chem. 30 (1992) 774.

2760
A. Alberola et al. / Journal of Organometallic Chemistry 692 (2007) 2750–2760
[19] S.C. Nyburg, C.H. Faerman, Acta Crystallogr Sect. B 41 (1985) 274.
[20] R. Taylor, O. Kennard, J. Am. Chem. Soc. 104 (1982) 5063;
D.S. Reddy, B.S. Goud, K. Panneerselvan, J. Chem. Soc., Chem.
Commun. (1993) 663;
[23] A. Alberola, J.M. Rawson, A.L. Whalley, J.Mater.Chem 16 (2006) 2560.
[24] A small number of paramagnetic defect sites would be expected to
give rise to a Curie tail (a steady increase in v upon cooling). The
discontinuity in v on both heating and cooling is clear evidence of a
reversible phase transition.
N. Ziao, J. Graton, C. Laurence, J.Y. Le Questel, Acta Crystallogr.,
Sect. B 57 (2001) 850.
[25] J. van den Brink, New J. Phys. 6 (2004) 201.
[21] R.B. Hammond, M.J. Jones, S.A. Murphy, K.J. Roberts, E.D.L.
Smith, H. Klapper, H. Kutzke, R. Docherty, J. Cherryman, R.J.
Roberts, P.G. Fagan, Mol. Crys. Liq. Crys. 356 (2001) 389.
[22] See for example: A.M.T. Bell, J.N.B. Smith, J.P. Attfield, J.M.
Rawson, K. Shankland, W.I.F. David, New J. Chem. (1999) 565, We
have also successfully used XRPD to solve structures of 1 in high and
low temperature phases and b-p-NCC6F4 CNSSN (J.M. Rawson,
A.D. Bond, J. Burley et al., unpubl. work).
[26] PC Spartan Pro. Wavefunction Inc., 18401 Von Karman Avenue,
Suite 370, Irvine, CA 92612, USA. Available from: <http://wave-
function.com/>.
[27] A.D. Becke, Phys. Rev. A 38 (1988) 3089;
J.P. Perdew, Phys. Rev. B. 33 (1986) 8822.
[28] J.M. Nedelec, L.L. Hench, J. Non-cryst. Solids (2000) 106;
R.J. Meier, E. Koglin, Macromol., Theory Sim. 13 (2004) 133;
E. Lewars, Can. J. Chem. 78 (2000) 297.