Experimental and theoretical studies of fused-ring persistent [1,2,4]thiadiazinyl radicals

Article abstract of DOI:10.1021/jo0486746

Five persistent radicals 1a-1e were generated by the oxidation of 4H-[1,2,4]thiadiazines 2a-2e and studied with ESR and UV-vis spectroscopy. Three of the radicals, 1a, 1d, and 1e, were generated in high yields (>90%) using either SO₂Cl₂/amine in toluene or AgO/K₂CO ₃ in a toluene/ MeCN mixture. Halogenated radicals 1d and 1e were sufficiently stable for chromatographic isolation and vacuum sublimation. The solution stability of the fluorinated 1d was measured at t_1/2 ≈ 4 months in the absence of oxygen, and 1e at t_1/2 ≈ 40 min in the presence of air. The crystal and molecular structures of 1d were determined by X-ray crystallography showing columns of parallel almost evenly spaced planar heterocycles connected by infinite ...N...S... chains. Cyclic voltammetry of Id and 1e shows reversible reduction waves at about 0 V and irreversible oxidations at about 1.2 V. Spectral and electrochemical properties of 1 were well reproduced by DFT methods.

Full text of DOI:10.1021/jo0486746

Exp er im en ta l a n d Th eor etica l Stu d ies of F u sed -Rin g P er sisten t
[1,2,4]Th ia d ia zin yl Ra d ica ls†
J o´zef Zienkiewicz‡ and Piotr Kaszynski*
Organic Materials Research Group, Department of Chemistry, Vanderbilt University,
Nashville, Tennessee 37235
Victor G. Young J r.
X-ray Crystallographic Laboratory, Department of Chemistry, University of Minnesota,
Twin Cities, Minnesota 55455
piotr.kaszynski@vanderbilt.edu
Received J uly 30, 2004
Five persistent radicals 1a -1e were generated by the oxidation of 4H-[1,2,4]thiadiazines 2a -2e
and studied with ESR and UV-vis spectroscopy. Three of the radicals, 1a , 1d , and 1e, were
generated in high yields (>90%) using either SO2Cl2/amine in toluene or AgO/K2CO3 in a toluene/
MeCN mixture. Halogenated radicals 1d and 1e were sufficiently stable for chromatographic
isolation and vacuum sublimation. The solution stability of the fluorinated 1d was measured at
t1/2 ≈ 4 months in the absence of oxygen, and 1e at t1/2 ≈ 40 min in the presence of air. The crystal
and molecular structures of 1d were determined by X-ray crystallography showing columns of
parallel almost evenly spaced planar heterocycles connected by infinite ‚‚‚N‚‚‚S‚‚‚ chains. Cyclic
voltammetry of 1d and 1e shows reversible reduction waves at about 0 V and irreversible oxidations
at about 1.2 V. Spectral and electrochemical properties of 1 were well reproduced by DFT methods.
In tr od u ction
Scant literature data shows that four persistent par-
tially chlorinated benzo[1,2,4]thiadiazinyls were gener-
ated by oxidation of the corresponding 4H-thiadiazines8
or reduction of the appropriate sulfiminyl chlorides.8,9 The
radicals were reported to persist for 24 h in solution in
the presence of air, but their isolation was not attempted.
Despite poucity of data, the reported stability was
encouraging and prompted us to investigate this general
class of [1,2,4]thiadiazinyls.
There is continuous interest in stable free radicals as
components of organic magnetic materials1 and molecular
conductors.2 In this context, we have been investigating
π-delocalized heterocyclic radicals as possible structural
elements for liquid crystals3,4 to study electric and
magnetic properties of free radicals in organized media.
Recently, we demonstrated the significant chemical
stability of three- and four-ring heterocycles5 and thus
suitability for functionalization to form discotic mesogens.
Our earlier attempt to develop a thiatriazinyl radical
system suitable for construction of calamitic liquid
crystals was unsucessful.6 Therefore, we focused on
benzo[1,2,4]thiadiazinyl and its derivatives, which upon
proper substitution at the 3 and 7 positions can form
molecules with elongated shapes conducive to the forma-
tion of liquid crystalline phases.7
* To whom correspondence should be addressed. Ph: (615) 322-3458.
Fax (615) 343-1234.
† Dedicated to Prof. J osef Michl on the occasion of his 65th birthday.
‡ On leave from University of Wroclaw, Poland.
(1) Rawson, J . M.; Palacio, F. In π-Electron Magnetism. From
Molecules to Magnetic Materials; Veciana, J ., Ed.; Springer: New York,
2001; Vol. 100, pp 93-128 and references therein.
(2) Barclay, T. M.; Cordes, A. W.; Haddon, R. C.; Itkis, M. E.; Oakley,
R. T.; Reed, R. W.; Zhang, H. J . Am. Chem. Soc. 1999, 121, 969-976
and references therein.
(3) Kaszynski, P. In Magnetic Properties of Organic Materials; Lahti,
P. M., Ed.; Marcel Dekker: New York, 1999; pp 305-324.
(4) Patel, M. K.; Huang, J .; Kaszynski, P. Mol. Cryst. Liq. Cryst.
1995, 272, 87-97.
(5) Benin, V.; Kaszynski, P. J . Org. Chem. 2000, 65, 8086-8088.
(6) Farrar, J . M.; Patel, M. K.; Kaszynski, P.; Young, V. G., J r. J .
Org. Chem. 2000, 65, 931-940.
Initially, we have concentrated on the evaluation of
stability of the radicals as a function of the ring structure
and finding an efficient method for their generation and
isolation. Thus, we investigated parent 3-phenylbenzo-
(7) Demus, D. In Handbook of Liquid Crystals; Demus, D., Goodby,
J . W., Gray, G. W., Spiess, H.-W., Vill, V., Eds.; Wiley-VCH: New York,
1998; Vol. 1, pp 133-187.
10.1021/jo0486746 CCC: $27.50 © 2004 American Chemical Society
Published on Web 10/07/2004
J . Org. Chem. 2004, 69, 7525-7536
7525

Zienkiewicz et al.
TABLE 1. Meth od s, Ma xim u m Yield s, a n d Op tim u m Tim es for Gen er a tion of Ra d ica ls 1 by Oxid a tion of 2
max yielda (time)
method
a
b
c
d
e
A: PbO2/K2CO3; benzene
B: AgO/K2CO3; benzene
C: AgO/K2CO3; MeCN/toluene 1:1
D: SO2Cl2/pyridine; toluene
26% (25 min)
29% (30 min)
94% (1.5 min)b
98%
38% (30 min)
46% (35 min)
2% (5 min)
0%
2% (105 min)
5% (120 min)
<2%
10% (40 min)
13% (50 min)
96% (1.5 min)c
93%c,d
7% (60 min)
11% (80 min)
91% (1.5 min)c
86%c,d
0%
a
b
Maximum concentration established by ESR. Maximum yield estimated by UV spectroscopy. ESR analysis shows 74% of the radicals
after 10 min. c ESR analysis of isolated solid. In some experiments EtMe2N was used instead of pyridine.
d
SCHEME 1
[1,2,4]thiadiazinyl (1a ) and the effect of incorporation of
nitrogen atoms (1b and 1c) and also tetrafluorination
F IGURE 1. Formation of 1e by oxidation of 2e with AgO/
(1d ) and tetrachlorination (1e)9 of the fused benzene ring
K2CO3 (filled symbols) and PbO2/K2CO3 (open symbols) in
benzene (diamonds) and MeCN/toluene (circles) followed by
on the radical stability. Keeping in mind the synthetic
constraints posed by the precursors and challenges in the
isolation of the pure liquid crystalline radicals, we focused
solely on oxidation of the corresponding 4H-[1,2,4]thia-
diazines with inorganic oxidants to generate the radicals.
Here we describe the generation of five radicals 1a -
1e by oxidation of the corresponding [1,2,4]thiadiazines
and isolation and extensive characterization of 1d and
1e. Subsequently, we report their spectroscopic charac-
terization (UV-vis and ESR) and electrochemical behav-
ior aided by DFT calculations. In the context of our
interest in radicals 1 as structural elements of liquid
crystalline materials, we also report extensive studies of
the generation methods for the radicals and their chemi-
cal stability.
ESR (diamonds) or intensity of absorption at 680 nm (circles).
The inset shows the initial 5 min of 2e oxidation with AgO/
K2CO3 in MeCN/toluene 1:1 mixture.
Gen er a tion a n d Isola tion of Ra d ica ls. Generation
of radicals by oxidation of 2 was investigated under
several conditions (Scheme 2), and the results are shown
in Table 1. Initially, a 10- to 20-fold excess of PbO2
(Method A) or AgO (Method B)12 in the presence of K2-
CO3 was used in dry benzene. The oxidation reactions
were generally slow, and the maximum concentration of
the radical was reached after about 30-90 min (Table 1
and Figure 1). AgO was found to be a generally more
effective oxidant than PbO2. The highest yield of the
radical, about 46%, was observed by ESR for the pyrido
derivative 1b, and the lowest spin concentrations were
recorded for the pyrazino derivative 1c (5%). The halo-
genated radicals 1d and 1e were generated in a maxi-
mum yield of about 10% after about 1 h. Interestingly,
the log of the observed spin yields correlate with the
calculated enthalpy of N-H homolytic bond dissociation
in the 4H-thiadiazines 2.13 In each case a significant
amount of starting thiadiazine 2 remained unreacted.
Using larger amounts of oxidant and base (up to 50 equiv)
had an insignificant effect on the spin yield of 1e. No 1a
was detected when Et3N was used as a base.
Resu lts
Syn th esis. The preparation of thiadiazines 2a -2d is
reported elsewhere.10 The tetrachloro analogue 2e was
prepared in about 38% overall yield using adaptations
of literature methods (Scheme 1).9,11 Thus, reduction of
crude sulfiminyl chloride 3e with dry 1-propanethiol gave
thiadiazine 2e in about 84% yield. The use of 1-pro-
panethiol instead of 4-ClC6H4SH used for a related
system11 significantly simplified purification of the thia-
diazine. The chloride 3e was obtained according to a
literature procedure by insertion of SCl2 into N-chloro-
N′- phenylbenzamidine in the presence of pyridine fol-
lowed by exhaustive chlorination with Cl2.9
When benzene was replaced with MeCN or with a 1:1
mixture of MeCN/toluene in oxidations of 2 with AgO/
K2CO3 (Method C), the rate of radical generation was
markedly improved but the decomposition rate was also
higher (Figure 1). The rapid (under 2 min) and almost
quantitative generation of the radicals was observed only
for thiadiazine derivatives 2a , 2d , and 2e. The nitrogen-
(8) Markovskii, L. N.; Talanov, V. S.; Polumbrik, O. M.; Shermo-
lovich, Y. G. Russ. J . Org. Chem. 1981, 17, 2338-2339.
(9) Shermolovich, Y. G.; Simonov, Y. A.; Dvorkin, A. A.; Polumbrik,
O. M.; Borovikova, G. S.; Kaminskaya, E. I.; Levchenko, E. S.;
Markovskii, L. N. Russ. J . Org. Chem. 1989, 25, 550-553.
(10) Zienkiewicz, J .; Kaszynski, P.; Young, V. G., J r. J . Org. Chem.
2004, 69, 2551-2561.
(12) AgO was used in generation of iminoxyl radicals: J ezierski, A.
Magn. Reson. Chem. 1989, 27, 130-133.
(13) Kaszynski, P. Molecules 2004, 9, 000.
(11) Levchenko, E. S.; Borovikova, G. S.; Borovik, E. I.; Kalinin, V.
N. Russ. J . Org. Chem. 1984, 20, 176-181.
7526 J . Org. Chem., Vol. 69, No. 22, 2004

Fused-Ring Persistent [1,2,4]Thiadiazinyl Radicals
SCHEME 2
filtered through Florisil even at ambient temperature
under nitrogen and concentrated giving the solid radicals
in high yields. Both of the halogenated radicals were
obtained as black-green solids, which could be stored in
a freezer under an inert atmosphere for several weeks
without detected decomposition. The spin concentration
measured after dissolving the solid samples indicated
about 90% purity of the radicals.
Analysis of the isolated samples showed that passing
through Florisil topped with a layer of silica gel did not
remove the unreacted thiadiazines 2d and 2e but re-
moved the decomposition products and ionic reagents.
Pure radicals 1d and 1e were obtained in about 10% yield
as black crystals by dynamic vacuum sublimation. The
solutions of sublimed radical 1e were green, whereas the
sublimed fluorinated radical 1d formed purple solutions,
presumably as a result of the presence of small amounts
of impurities.
Cr ysta l a n d Molecu la r Str u ctu r es. Black-purple
monoclinic crystals of the fluorinated radical 1d and
black-green triclinic crystals of 1e were grown by slow
gradient sublimation (>100 °C) at 10-3 Torr, and their
solid structures were determined by X-ray diffraction.15
All inspected crystals of 1d and 1e were twinned.
Selected interatomic distances are listed in Table 2, and
the molecular structure of 1d is shown in Figure 2.
Molecules of 1d are nearly planar, and the largest
deviation from planarity of the heterocyclic skeleton of
1.1(6)° is observed for the C(8)-C(9)-C(10)-C(6) angle
(Figure 2). The phenyl ring is nearly ideally coplanar with
the heterocycle, which results in short H‚‚‚N nonbonding
distances of about 2.44 Å. The solid-state geometry of 1d
is generally well reproduced by the DFT and to lesser
degree by HF/6-31G(d) methods. The overall mean dif-
ference between 16 experimental and DFT-calculated
containing heterocyclic radicals 1b and 1c were gener-
ated in trace quantities under these conditions, and rapid
formation of a brown-red-colored species was observed.
Oxidation of 2b and 2c with AgO or PbO2 in the absence
of base gave low concentrations of the corresponding
radicals. For instance, 1b was generated in about 12%
max yield after about 10 min, and 1e was observed in
60% yield after about 5 min. The use of PbO2 instead of
AgO or much less MeCN (e.g., 1:9 MeCN/toluene) was
significantly less effective and the reaction was slow and
incomplete as tested for 1e.
Analysis of the time-dependent spin concentration
curves using a kinetic equation for two consecutive
reactions14 showed that the rate of radical formation kf
is about 1 × 10-3 s-1 using Methods A and B and over
an order of magnitude faster when Method C is used.
The rate of radical decomposition (kd) is slower by a factor
ranging between 3 for 1a to over 100 for 1d and 1e with
Methods A and B and about 30 with Method C. The curve
fitting also shows that the maximum yields for the
radicals in the first two methods are moderate at best,
reaching about 50% for 1b, while Method C gives close
to quantitative yields of 1a , 1d , and 1e.
In an attempt to develop an alternative method for
generation of radicals, we investigated the reaction of 2
with SO2Cl2. After optimization of the reaction conditions,
it was found that simultaneous dropwise addition of
solutions of 0.5 equiv of SO2Cl2 and 1 equiv of pyridine
or Me2EtN to solutions of 2 gives the corresponding
benzo[1,2,4]thiadiazinyl derivatives 1a , 1d , and 1e in
reproducible high spin yields of about 90% (Table 1,
Method D). In contrast, no radical formation was detected
for the pyridio and pyrazino derivatives 1b and 1c.
Other oxidants such as K3Fe(CN)6 or K2S2O8 used in
aqueous-benzene or solid phase-benzene phase transfer
catalysis systems (Bu4NBr) were ineffective in the case
of 1b, and only slow accumulation of decomposition
products was observed.
bond lengths is ∆x ) 0.009 Å with a STD of 0.013, and
∆x ) 0.006 Å (STD ) 0.011) when the two bonds to
sulfur are excluded.
In the solid state, molecules of 1d are arranged in
ideally parallel interpenetrating mirror-imaged stacks
rotated by 166° (Figure 3). Within each such column, the
S-N bonds overlap, forming an infinite ladder-type
structure with the ladder “rungs” slipped by 0.37 Å and
an alternating S‚‚‚N distance of 3.193 and 3.212 Å
(Figure 3a). This separation between the ladder “rungs”
corresponds to about 93% of the van der Waals separa-
tion16 for the S and N atoms. Each column is related to
the neighboring one by an inversion center, and the
molecules form weak dimeric pairs with an intermolecu-
lar S(1)‚‚‚N(2) distance of 3.269 Å (or 95% of the sum of
the van der Waals radii).16
The high efficiency of generation of benzo[1,2,4]-
thiadiazine radicals (1a , 1d , and 1e) and acceptable
yields for 1b prompted us to investigate the isolation of
the four radicals in the solid state. Thus, cold solutions
of radicals generated using Methods C or D (Table 1)
were passed through short Florisil columns at -78 °C
under a flow of nitrogen. A solution of radical 1a partially
decolorized upon filtration and completely bleached after
10-15 min at about 0 °C. The pyridio radical 1b,
generated using Method B, was more stable than 1a and
could be isolated as a solid containing about 20% of the
original spin (by solution ESR measurement).
The availability of the crystal structure for thiadiazine
2d 10 provides an opportunity to examine structural
changes upon formation of the radical 1d . Thus the
removal of the 4-H atom from 2d results in planarization
(15) Crystal data for 1d: C13H5F4N2S monoclinic, C2/c, a ) 28.776(11)
Å, b ) 12.737(5) Å, c ) 6.363(3) Å, â ) 101.095(7)°, V ) 2288.5(15) Å3,
Z ) 8, T ) 173 (2) K, λ ) 0.71073 Å, R(F2) ) 0.0430 and Rw(F2) )
0.1037 (for 1741 reflections with I > 2σ(I)). Crystal data for 1e: C13H5-
Cl4N2S triclinic, P 1h, a ) 6.795(2) Å, b ) 8.362(3) Å, c ) 13.368(4) Å, R
) 71.791(4)°, â ) 83.326(4)°, γ ) 68.631(4)°, V ) 671.9(4) Å3, Z ) 2, T
) 173 (2) K, λ ) 0.71073 Å, R(F2) ) 0.0847 and Rw(F2) ) 0.2067 (for
1779 reflections with I > 2σ(I)). For details see Supporting Information.
(16) Rowland, R. S.; Taylor, R. J . Phys. Chem. 1996, 100, 7384-
7391.
The halogenated radicals 1d and 1e showed signifi-
cantly higher stability. The reaction mixtures could be
(14) Masel, R. I. In Chemical Kinetics and Catalysis; Wiley&Sons:
New York, 2001; pp 97-100.
J . Org. Chem, Vol. 69, No. 22, 2004 7527

Zienkiewicz et al.
calcda
TABLE 2. Selected Exp er im en ta l a n d Ca lcu la ted Bon d Len gth s a n d An gles for 1d
bond lengths [Å]
exptl
calcda
angles [deg]
exptl
S(1)-N(2)
N(2)-C(3)
C(3)-N(4)
N(4)-C(5)
C(5)-C(6)
S(1)-C(6)
C(5)-C(7)
C(6)-C(10)
C(3)-CPh
1.622(2)
1.352(4)
1.320(4)
1.368(4)
1.413(4)
1.733(3)
1.396(4)
1.385(4)
1.492(4)
1.656
1.342
1.339
1.360
1.423
1.757
1.418
1.391
1.489
S(1)-N(2)-C(3)
122.0(2)
128.7(2)
120.4(2)
124.7(3)
118.9(2)
105.22(13)
-0.5(6)
122.2
128.3
121.3
125.1
118.4
104.8
0.0
N(2)-C(3)-N(4)
C(3)-N(4)-C(5)
N(4)-C(5)-C(6)
C(5)-C(6)-S(1)
C(6)-S(1)-N(2)
S(1)-N(2)-C(3)-N(4)
N(2)-C(3)-C(11)-C(16)
N(4)-C(5)-C(6)-C(10)
0.8(6)
0.0
180.0
-179.4(4)
a
Gas-phase calculations at the B3LYP/6-31G(d) level of theory. The numbering scheme is shown in Figure 2.
F IGURE 2. Thermal ellipsoids diagram for 1d . Ellipsoids are
drawn at 50% probability and hydrogen atoms have arbitrary
radii. Selected molecular dimensions are listed in Table 2.
of the heterocyclic ring due to the change in the number
of π electrons from 8 to 7. Also, the removal of the
hydrogen makes it possible for the phenyl ring at the C(3)
position to adopt a coplanar orientation relative to the
heterocycle, and the whole structure becomes planar. The
bond lengths in the heterocyclic ring undergo significant
changes to reduce bond length alternation. Thus single
bonds such as S(1)-N(2) and C(3)-N(4) undergo contrac-
tion by nearly 0.06 Å, while the double bond N(2)-C(3)
is expanded to about the same extent. Other bonds are
adjusted to a lesser extent; most notably the S(1)-C(6)
and N(4)-C(5) bonds are contracted, and the C(5)-C(7)
bond is expanded by about 0.03 Å. The observed geo-
metrical changes are paralleled by changes in bond
orders shown for the 2d -1d pair in Figure 4. This is
consistent with aromatization of the heterocycle and the
general expectations for the electronic structure of six-
membered heterocyclic radicals.17
F IGURE 3. Partial packing diagram for 1d viewed along the
tilted ab plane (a) and along the c axis (b).
F IGURE 4. Wiberg bond order indices for 2d and 1d obtained
with the B3LYP/6-31G(d) method.
Differences in the molecular geometry and electronic
structures between 2d and 1d are reflected in the
packing diagrams of the two compounds. Arrangement
of molecules in the stacks of the 4H-thiadiazine 2d
appears to be governed by quadrupolar π-π interactions
between the fluorinated and nonfluorinated benzene
rings, as evident from the ∼35% overlap between the
rings.10 In contrast, interactions between the thiadiazine
rings appear to be the dominant factor in molecular
packing of radical 1d . This is evident from the nearly
ideal antiparallel arrangements of the N-S bonds and
generally good overlap of the thiadiazine rings but not
the benzene rings (Figure 3b).
The crystals of the tetrachloro radical 1e were gener-
ally of less than desirable quality and were plagued with
twinning. Three crystals were analyzed using either
conventional X-ray or synchrotron radiation, but none of
them gave data that would permit the structure solution
and refinement to R smaller than 8.5%. The X-ray
diffraction results indicate that the molecular structure
of 1e is as expected from theoretical predictions and
(17) Oakley, R. T. Prog. Inorg. Chem. 1988, 36, 299-391.
7528 J . Org. Chem., Vol. 69, No. 22, 2004

Fused-Ring Persistent [1,2,4]Thiadiazinyl Radicals
F IGURE 5. Partial packing diagram for 1e viewed along the
tilted a axis.
similar to that of 1d (Figure 2). The relatively large error
on the molecular dimensions does not allow for a detailed
comparison of 1e on the molecular level. The data
provide, however, useful information about crystal pack-
ing in 1e, which can be compared with that of 1d .
Molecules of 1e are, like 1d , arranged in antiparallel
interpenetrating stacks along the a axis. The arrange-
ment maximizes the overlap between benzene rings at
the expense of the overlap between the thiadiazinyls
(Figure 5). Unlike in the crystal structure of 1d , mol-
ecules of 1e in each stack are related to molecules in the
other stack and also in the neighboring column by an
inversion center. The intermolecular separation within
a column is about 3.40 Å, which is about 0.25 Å greater
than in 1d . The S‚‚‚S nonbonding distance for molecules
in neighboring columns is about 3.37 Å, which corre-
sponds to 94% of the sum of the van der Waals radii.16
Computational results show that other radicals 1a -
1c are also planar with Cs molecular symmetry. The
interatomic distances and angles of the thiadiazine ring
remain practically the same throughout the series with
a typical variation in the bond length of about 0.03 Å.
Most noticeably, the N(4)-C(5) distance varies from
1.358 Å for 1c and 1e to 1.369 Å for 1a . This presumably
reflects the π interactions between the nitrogen atom and
the other fused ring. For the relatively electron-rich benzo
derivative 1a the distance is longest, whereas for the
electron-deficient tetrachloro and pyrazino derivatives 1c
and 1e the distance is shortest. The analogous S(1)-C
distance is less sensitive to electronic effects in the fused
ring.
ESR Sp ectr oscop y. Room-temperature solution spec-
tra of radicals 1c and 1e exhibit five unresolved signals
arising from coupling to the thiadiazinyl nitrogen atoms
(Figure 6). In the benzo and pyrido derivatives 1a and
1b, each line of the quintet is split by interactions with
H atoms. The spectrum of 1d is significantly more com-
plicated as a result of additional coupling with fluorine
atoms. All spectra are centered at about 3475 G.
F IGURE 6. ESR spectra for 1 recorded in benzene at ambient
temperature. Solutions of 1a -1c were generated using Method
B, and solutions of 1d and 1e were obtained from sublimed
samples. The circle indicates the symmetry anomaly.
phenyl substituent), the correlation factors r were >0.99
for all pairs of simulated and experimental spectra except
for 1d (r ) 0.973). The resulting hfcc values were
assigned to the molecular structures on the basis of
trends in the calculated values and are listed in Table 3.
Spectra of the pyrido radical 1b consistently lacked
symmetry in the central part (Figure 6), and none of them
could be simulated. Therefore, the spectrum of 1b was
reflected about the midpoint defined by the pairs of
spectral peaks and assumed to be lying in the middle of
a small central peak (cf. spectrum of 1a ). The trans-
formed spectrum of 1b was simulated (r ) 1.00), and the
hfcc values are listed in Table 3.
All spectra except for 1b were numerically simulated
using initial hyperfine coupling constant (hfcc) values
obtained from B3LYP/cc-pVDZ//B3LYP/6-31G(d) calcula-
tions and appropriately scaled.18 After a least-squares fit
for nine parameters (including all hydrogen atoms of the
All five radicals show two similar couplings of about 6
and 5 G to the thiadiazinyl nitrogen atoms N(2) and N(4),
which results in the basic quintet in the spectrum.
(18) Kaszynski, P. J . Phys. Chem. A, 2001, 105, 7615-7625.
J . Org. Chem, Vol. 69, No. 22, 2004 7529

Zienkiewicz et al.
TABLE 3. Exp er im en ta l Isotr op ic h fcc [G] for Ra d ica ls
1
radical
1a
1b
1c
1d
1e
g
aN2
aN4
aX5
aX6
aX7
aX8
a
a
a
2.0056
5.92
4.86
1.31
0.02
2.06
0.32
0.47
0.20
0.37
2.0055
5.96
4.94
1.85
0.06
2.60
0.53
0.43
0.23
0.36
2.0058
5.57
4.49
0.62
0.07
1.82
0.29
0.36
0.17
0.24
2.0056
5.92
5.38
2.02
1.04
4.68
1.65
0.31
0.20
0.3
2.0057
5.74
4.71
0.09
0.00
0.26
0.16
0.35
0.18
0.25
b
b
b
b
H(Ph) ortho
H(Ph) meta
H(Ph) para
F IGURE 8. Absorption spectra of 2e (c ≈ 10-4 M) and 1e (c
≈ 10-3 M) recorded in toluene. Vertical bars represent calcu-
lated absorption bands with indicated symmetry and scaled
by Eexp ) 0.97Ecalcd - 0.12 [eV]. The height represents relative
oscillator strength.
a
Obtained from simulation of the experimental spectra to the
b
correlation value r > 0.99 (r ) 0.973 for 1d ). X is H, N, F (1d ),
or Cl (1e).
F IGURE 7. Total spin distribution in 1a calculated with the
B3LYP/cc-DVZ//B3LYP/6-31G(d) method. The numerals show
the heterocycle numbering system.
Additional splitting in the spectrum results from coupling
to atoms in positions 5 and 7, and in the case of 1d also
to F in position 8 (aF8 ) 1.65 G). The remaining hfcc are
small, typically <0.4 G, and they contribute to line
broadening. This is consistent with the spin distribution
in the heterocyclic ring shown for 1a in Figure 7.
F IGURE 9. Low energy portion of the absorption spectra for
selected radicals recorded in toluene. Sublimed 1d and 1e and
crude solutions of 1a and 1b were used. The vertical scale is
arbitrary and does not reflect the relative absorption coef-
ficients.
The spectrum of the pyrazino derivative 1c recorded
under various conditions consistently shows no fine
structure of the five broad lines despite the expected
relatively high aN5 and aH7 values. This can be attiributed
to the low yield of the radical (<4%) and high concentra-
tion of the unreacted precursor 2c (> 90%) in the crude
solution, and possible interactions between 1c and 2c.
These interactions may also affect the hfcc values, which
significantly deviate from those theoretically predicted.
This is consistent with a similar observation for 1a ;
solutions of crude 1a generated using Method A and
containing >50% of unreacted 2a show only five unre-
solved lines, whereas a nicely resolved spectrum of 1a
(Figure 6) was obtained for crude solutions generated
using Method C and containing mostly radical 1a .
large spin concentrations on the sulfur atoms (0.28; see
Figure 7).
Electr on ic Absor p tion Sp ectr oscop y. Oxidation of
thiadiazines 2 to radicals 1 results in the appearance of
new low energy and low intensity absorption bands above
500 nm. This is demonstrated for 1e in Figure 8 and
compared for four radicals in Figure 9. The halogenated
radicals 1d and 1e show four maxima above 450 nm
(Table 4). In the benzo derivative 1a only three such
maxima could be distinguished, while the pyrido deriva-
tive 1b has one broad featureless absorption band with
the maximum at 563 nm. Their longest wavelength
absorption bands range from 682 nm for the tetrachloro
radical 1e to 611 nm for the benzo derivative 1a . This
order is well reproduced by TD-DFT calculations, which
also show that the lowest energy band of the pyrido
derivative 1b should be about that for 1e (682 nm). The
analogous excitation energy for the pyrazino radical 1c
is predicted to be lowest in the series (Table 4).
A comparison of 11 experimental excitation energies
clearly identified for 1a , 1d , and 1e with those calculated
in a vacuum at the B3LYP/6-31G(d) level of theory shows
an excellent correlation (Figure 10). All computed ener-
gies are systematically overestimated by 0.17 eV. The
appropriately scaled theoretical excitation energies for
1e are shown in Figure 8 with sizes indicating relative
oscillator strength. The energies, calculated using diffuse
functions on heavy atoms (B3LYP/6-31+G(d)//B3LYP/6-
The hfcc’s obtained from simulations of experimental
spectra generally correlate well with the calculated
values. The only significant exceptions are the aH values
for 1c and the overestimated aF value assigned to F(6).
The scaling factor resulting from the best fit line for aH
values is very similar to that established earlier,18 but
for aN the factor is slightly smaller (0.73 vs 0.767).18 This
is due to systematic overestimation of the hfcc values by
about 0.4 G for the N(4) atom in all radicals 1. Neverthe-
less, overall good agreement with the previous scaling
factor and high correlation values (R2 ) 0.99) demon-
strate the validity of our protocol for prediction of hfcc
in this class of radicals.18 The calculated hfcc for the
chlorines are overestimated by a factor 1.86 (R2 ) 0.97).
All g values measured for the radicals are about 2.006
(Table 3), which reflect very similar and relatively
7530 J . Org. Chem., Vol. 69, No. 22, 2004

Fused-Ring Persistent [1,2,4]Thiadiazinyl Radicals
TABLE 4. Exp er im en ta la a n d Ca lcu la ted b Low est En er gy Electr on ic Absor p tion Ba n d s for Ra d ica ls 1
a [nm]
calcd
b [nm]
calcd
c [nm]
calcd
d [nm]
calcd
e [nm]
calcd
exptl
exptlc
exptld
exptl
exptl
609.7
585.7
505.7
561.7 (11) A′
556.3 (1) A′′
477.7 (4) A′
417.2 (20) A′
631.9 (1) A′′
558.1 (12) A′
468.4 (5) A′
432.4 (7) A′
702.3 (0.3) A′′
587.0 (15) A′
473.3 (3) A′
467.2 (6) A′
651.7
603.0
515.7
484.2
608.3 (15) A′
544.5 (1) A′′
488.4 (2) A′
453.0 (9) A′
682.4
628.9
545.2
508.5
624.8 (19) A′
573.3 (1) A′′
508.6 (2) A′
469.8 (26) A′
a
b
Obtained in toluene solution. TD-B3LYP/6-31G(d); in parentheses oscillator strength ×103. c One broad absorption band with a
d
maximum at 563 nm. Not studied experimentally.
the pure HOMO-2 to SOMO transition for 1a or HOMO-3
to SOMO transition for 1d and 1e. All orbitals involved
in these transitions are of π symmetry except for the
HOMO-2 for 1a , which includes the lone electron pairs
of the thiadiazine ring.
The lowest energy excitations in the pyrido and pyr-
azino derivatives 1b and 1c result largely from the
HOMO-1-SOMO transitions. In these two radicals the
HOMO-1 includes the thiadiazine ring lone pairs. The
next two lowest energy excitations in 1c appear to be
analogous to the two longest wavelength excitations in
the benzo derivative 1a. The HOMO-SOMO and SOMO-
LUMO transitions are the main components of the second
F IGURE 10. Experimental (recorded in toluene) versus
theoretical (TD-B3LYP/6-31G(d)) low energy absorption bands
for 1a , 1d , and 1e. The best fit line: Ecalcd ) 1.00Eexp+ 0.17
(R2 ) 0.984).
and third lowest energy absorption bands for 1b, respec-
tively.
2
All five radicals have the electronic state A′′ and the
unpaired electron resides on the π MO (-5.44 eV for 1a ,
-5.65 eV for 1b, -5.94 eV for 1c, -6.04 eV for 1d , and
-5.94 eV for 1e, according to the B3LYP/6-31+G(d)//
B3LYP/6-31G(d) method).
Ch em ica l Sta bility of Ra d ica ls. Generation of and
attempts to isolate radicals demonstrated that their
stability increases in the order 1a < 1b < 1d ≈ 1e. All
radicals appear to be stable in cold toluene solutions (-80
°C). Upon warming and exposure to air the green color
of the radicals bleaches, forming colorless solutions. The
only exception is the fluorinated radical 1d , for which
one of the minor decomposition products is purple with
a single maximum of absorption at 558 nm. This un-
identified product is formed in trace quantities during
sublimation of 1d and gives a pink color to the solution
of the sublimed material. Its presumed presence, how-
ever, was not clearly confirmed by UV-vis spectroscopy
(see Figure 9) or during solution of crystal structure (vide
supra).
The main decomposition products of radicals 1 in air
appear to be the corresponding S-oxides 4, which were
isolated and characterized for 1a and 1e. The nonvolatile
4e was also isolated in about 70% yield from a residue
after sublimation of radical 1e.
F IGURE 11. HOMO, SOMO, and LUMO contours for 1a
(B3LYP/6-31G(d,p))
31G(d)), are overestimated by about 0.3 eV and with a
lower correlation factor (R2 ) 0.972).
Computational analysis shows that the several lowest
energy transitions in radicals 1 involve the HOMO,
SOMO, and LUMO orbitals shown in Figure 11 for 1a .
For instance, the lowest energy excitations for the benzo
and tetrahalobenzo derivatives 1a , 1d , and 1e involve
largely the HOMO-SOMO transition with a smaller,
about 30%, component of the SOMO-LUMO transition.
The next lowest energy absorption band originates from
Quantitative ESR studies of stability were conducted
in benzene at ambient temperature using sublimed
halogenated radicals 1d and 1e, and solutions of 1a -1c
generated with AgO/K2CO3 (Method B). Analysis of the
kinetic data (Figure 12) shows that the stability of the
radicals follows the order 1c < 1a < 1b < 1e < 1d , which
is largely consistent with our qualitative observations
J . Org. Chem, Vol. 69, No. 22, 2004 7531

Zienkiewicz et al.
F IGURE 12. Intensity of ESR signal for radicals 1 as a
function of time recorder at abmbient temerature. Solutions
of crude 1a -1c and sublimed 1d and 1e in benzene were
degassed and filled with Ar. Open circle datapoints were
obtained for a sample of 1e kept under vacuum. Rate of
decomposition: 1a , k ) 112 × 10-4 h-1; 1b, k ) 81 × 10-4 h-1
;
1c, k ) 161 × 10-4 h-1; 1d , k ) 2.3 × 10-4 h-1; 1e, k ) 4.4 ×
10-4 h-1. Correlation coefficients R2 g 0.994. The inset shows
the initial 2 h of the measurement.
F IGURE 13. CV scan of 1d (top) and 1e (bottom) in CH2Cl2,
[n-Bu4N][PF6] supporting electrolyte and 0.10 V/min scanning
rate.
(vide supra). The most stable fluorinated radical 1d
decomposes over 60 times more slowly than the least
stable pyrazino derivative 1c and has a half-life in
solution of over 4 months. The stabilities of the unisolated
1a -1c probably represent the lower limit, based on the
results for 1e (vide infra).
TABLE 5. Solu tion Ha lf-Wa vea a n d Cell P oten tia ls for
Ra d ica ls 1d a n d 1e
red
1/2
ox
1/2
b
d
E
[V]
E
c [V]
Ecell [V]
radical CH2Cl2 MeCN CH2Cl2 MeCN CH2Cl2 MeCN
1d
1e
-0.11e
-0.17
g
+1.20
+1.20
+1.03
g
+1.26
+1.20
+1.20
g
f
-0.04
The kinetic analysis was performed on data acquired
1 h after the radical preparation to allow the radicals to
consume any residual oxygen. During the initial hour the
plots are nonlinear, as shown in the inset in Figure 12.
Samples prepared in the same manner but kept under
vacuum rather than Ar slowly leaked air, and the
nonlinear behavior persisted for many hours, especially
for 1a -1c.
a
Referenced to ferrocene set at +0.38 V in MeCN and +0.48 V
in CH2Cl2. Quasireversible behavior. c Irreversible behavior; Epc
b
ox
pc1/2
red
pc1/2
d
value listed. Ecell estimated as
E
- E
.
e Pre-peak at
+0.04 V. f Pre-peak at +0.08 V. Insufficient solubility.
g
V. Typical electrochemical waves for 1d and 1e are shown
in Figure 13, and half-wave potentials are listed in Table
5. The tetrafluoro radical 1d was analyzed in MeCN and
CH2Cl2, whereas the tetrachloro 1e was studied only in
CH2Cl2 because of its poor solubility in MeCN.
Cyclic voltammetry of the radicals shows pre-peaks,
which are 0.15 V (for 1d ) and 0.12 V (for 1e) more positive
than the appropriate anodic peaks. The pre-peak current
decreases proportionally to the increase of the scan rate,
which suggests surface adsorption of the reduced species.
Such behavior was observed before for other heterocyclic
radicals.19 The observed lower reduction potential for 1d
in MeCN than in CH2Cl2 is consistent with other experi-
mental data.20
The experimental redox potentials for 1d and 1e are
consistent with values estimated from the calculated and
appropriately scaled ionization potentials Ip, electron
affinities EA, and disproportionation energies in Table
6. Thus, the previously established correlations21 predict
a reduction potential of -0.05 V for 1d and a +0.06 V
difference in E
In another set of experiments, the stability of degassed
benzene solution of sublimed 1e was investigated using
UV-vis spectroscopy by monitoring the intensity of
absorption at 680 nm. During the first 5 h the kinetics
of radical decay appeared to be pseudo-zero-order with a
rate constant k ) 1.2 × 10-4 s-1, which suggests photo-
chemical instability of the radical. A similar analysis for
the crude unsublimed solid 1e showed about 10 times
faster decomposition of the radical. In contrast, a de-
gassed and subsequently aerated (15 s) solution of 1e
showed rapid decomposition during the initial 2 h, which
can be described as pseudo-first-order with a rate con-
stant k ≈ 1.1 h-1
.
The stability of 1e toward other reagents was briefly
investigated. Addition of about 10 equiv of EtOH, Et3N,
or AcOH to a toluene solution of 1e had little effect on
the decay of the absorption band at 680 nm, and the rate
was approximately the same as that when the sample
was exposed to air. In the case of EtOH and AcOH, the
main product of decomposition was identified as 4e,
whereas in Et3N generation of large amounts of starting
thiadiazine 2e was observed on the basis of TLC analysis.
red
1/2
between 1e and 1d in MeCN, while
the experimental values are -0.17 V and +0.07 V,
respectively. Also the estimated and experimental oxida-
Cyclic Volta m m etr y. Electrochemical analysis of the
two halogenated radicals 1d and 1e showed strongly
irreversible oxidation behavior with the cathodic poten-
tial peak at 1.2 V and a reversible reduction at about 0
(19) Boere´, R. T.; Moock, K. H.; Parvez, M. Z. Anorg. Allg. Chem.
1994, 620, 1589-1598.
(20) Boere´, R. T.; Roemmele, T. L. Coord. Chem. Rev. 2000, 210,
369-445.
(21) Kaszynski, P. J . Phys. Chem. A 2001, 105, 7626-7633.
7532 J . Org. Chem., Vol. 69, No. 22, 2004

Fused-Ring Persistent [1,2,4]Thiadiazinyl Radicals
TABLE 6. Ca lcu la ted a P a r a m eter s for Ra d ica ls 1
pounds, but only modest yields have been observed.5,8,22
The addition of MeCN to the reaction mixture signifi-
cantly increases the yield of the radicals and works
particularly well with AgO, a more potent oxidant
(Method C). This effect of MeCN cosolvent on the rate of
electron transfer from the substrate to the metal oxide
is presumably due to (i) increased polarity of the medium,
(ii) more complete deprotonation of the substrate, (iii)
coordination of the solvent to silver ions, or (iv) a
combination of the above. Experiments show that the
presence of a base and high dielectric medium is impor-
tant for fast rate and complete generation of radicals.
Unfortunately, the use of a polar aprotic solvent also
makes the base more reactive, which appears to have a
detrimental effect on the pyrido and pyrazino derivatives
1b and 1c. It is possible that this method will find more
general use in preparation of other radicals.
The observed trend in stability of the five radicals
established in Figure 12, 1c < 1a < 1b , 1e < 1d ,
generally is consistent with expectations, except for
pyrazino derivative 1c. Thus, incorporation of an N atom
into the benzene ring of 1a to form the pyrido derivative
1b increases the radical’s stability by a factor of 1.4.
However, contrary to expectations, the pyrazino deriva-
tive with one more N atom is less stable than the benzo
analogue 1a by the same factor. It would be interesting
to compare the stability of 1b to that of its isomer, which
could be generated from the known 4H-pyrido[2,3-e]-
[1,2,4]thiadiazine.23
radical
1a
1b
1c
1d
1e
Ip (vert)b [eV]
6.80
1.98
7.03
2.16
7.30
2.48
7.26
2.58
7.17
2.66
EAc [eV]
red
1/2
d
-0.50 -0.37 -0.13 -0.05 +0.01
123.2
E
[V]
∆E1SCFe [kcal/mol]
123.4
-6.9
116.5
5.04
1.1
122.6
-7.4
115.3
5.01
1.1
123.2
-8.9
114.3
4.98
1.0
114.6
-7.7
107.0
4.63
1.0
∆E2SCFf [kcal/mol]
Σ∆ESCF [kcal/mol] 115.2
Σ∆G298 [eV]
-8.0
g
h
5.04
1.1
+0.6
i
Ecell [V]
ox
j
+0.7
+0.9
+1.0
+1.0
E
[V]
1/2
a
B3LYP/6-31G(d) level of theory for Ip and B3LYP/6-31+G(d)//
B3LYP/6-31G(d) for all others. Calculated values are scaled using
factors in ref 21. Difference in SCF energies. c Difference in ZPE-
b
d
corrected SCF values. Reduction potential in MeCN obtained
from computed EA; ref 21. e Electron-transfer energy between two
radicals to form an ion pair. f Reorganization energy to form a
relaxed ion pair. g Total SCF energy for radical disproportionation.
h
i
Total free energy of radical disproportionation. Electrochemical
cell potential calculated from Σ∆ESCF
MeCN calculated from E ) Ecell + E
j
.
Oxidation potentials in
.
ox
1/2
red
1/2
ox
1/2
tion potentials E for 1d in MeCN are essentially the
same and about +1.0 V. The electrochemical window for
1d in MeCN is predicted to be bigger by about 0.2 V than
found experimentally, but considering the uncertainty of
the prediction21 and the irreversibility of the oxidation
process, the comparison is rather good.
A comparison of the series of five radicals 1a -1e in
Table 6 shows trends expected on the basis of the
electronic properties of the ring fused to the thiadiazine.
The most electron-rich benzothiadiazinyl 1a has the
lowest estimated ionization and oxidation potentials (6.80
eV and +0.6 V respectively). At the same time, 1a also
has the lowest EA and reduction potential in the series.
On the other hand, the tetrachloro 1e, closely followed
by tetetrafluoro derivative 1d , has the highest estimated
reduction and oxidation potentials and EA in the series.
The pyrazino derivative 1c has the highest value of Ip )
7.30 eV, which is higher by 0.50 eV than that for the
benzo derivative 1a . The estimated Ecell for all five
The significant stability of the tetrachloro derivative
1e was expected on the basis of ample literature ex-
amples.24 In contrast, π radicals fluorinated at the high
spin density sites are rare,25 and to our knowledge no
comparative studies of stability of the chlorinated and
fluorinated radicals have been performed. Our studies
demonstrate that fluorine atoms may be even more
effective substituents than chlorine in stabilization of π
delocalized radicals. This finding is advantageous for the
design of liquid crystals,7 since the smaller fluorine atoms
are much preferred to the larger chlorines.
The chemical stability of the halogenated radicals is
sufficient for further studies of liquid crystalline deriva-
tives of 1d and also 1e. Both radicals can be chromato-
graphed, and they show low sensitivity to protic solvents
but suffer from sensitivity to oxygen. Thus future work
and characterization of radical derivatives will require
an anaerobic atmosphere. The reaction of 1 with oxygen
takes place at the sulfur center and gives rise to the
S-oxides 4. Fortunately, the resulting S-oxides can, in
principle, be reduced to 4H-thiadiazines 2 with phos-
phines.26
radicals is similar and just above 1 V, while the electro-
red
1/2
chemical window moves from E
) -0.5 V for most
electron-rich system 1a toward positive values reaching
the estimated E ) 0.0 V for 1e.
red
1/2
The calculations show that radicals 1 have relatively
small disproportionation energies of about 122 kcal/mol
(about 5 eV). The exception in the series is the tetrachloro
derivative 1e, which has the lowest energy change upon
the formation of an ion pair. The value of 114.6 kcal/mol
is comparable with that calculated for phenalenyl radi-
cal,21 but all ions derived from radicals 1 still have a
significant reorganization energy of about 7-9 kcal/mol.
The lowest reorganization energies are found for ions
derived from three radicals 1b, 1c, and 1e, which
maintain the Cs molecular symmetry. The anions derived
from 1a and 1d are significantly puckered and also have
the highest reorganization energies of -8.0 and -8.9
kcal/mol, respectively.
Although our main focus is on liquid crystalline deriva-
tives of 1, we have also briefly considered the suitability
of 1 for molecular conductors. Our computational results
(22) For example: Miura, Y.; Oyama, Y.; Teki, Y. J . Org. Chem.
2003, 68, 1225-1234.
(23) Gilchrist, T. L.; Rees, C. W.; Vaughan, D. J . Chem. Soc., Perkin
Trans. 1 1983, 55-59.
(24) Ballester, M. Adv. Phys. Org. Chem. 1989, 25, 267-446 and
references therein.
(25) Ho¨fs, H.-U.; Bats, J . W.; Gleiter, R.; Hartmann, G.; Mews, R.;
Eckert-Maksic, M.; Oberhammer, H.; Sheldrick, G. M. Chem. Ber.
1985, 118, 3781-3804.
(26) Finch, N.; Ricca, S., J r.; Werner, L. H.; Rodebaugh, R. J . Org.
Chem. 1980, 45, 3416-3421.
Discu ssion
Method A (PbO2) is the most common way to generate
thioaminyl radicals from the corresponding N-H com-
J . Org. Chem, Vol. 69, No. 22, 2004 7533

Zienkiewicz et al.
in Table 6, supported by voltammetric studies, indicate
that the redox properties of the radicals are generally
favorable in the context of molecular conductors.28 Radi-
cals 1 appear to have a relatively small, about 1.1 V,
electrochemical window, whose position can be tuned
within at least 0.5 V. The values of Ecell for 1 and also
the disproprtionation energies of about 122 kcal/mol
compare favorably to 4-aryl-[1,2,3,5]dithiadiazolyls18,29
but are still significantly larger than those for recently
reported radicals.30 Particularly interesting in this con-
text is the tetrachloro radical 1e, which has the lowest
disproportionation energy in the series and a relatively
small expected electron-phonon coupling since both ions
preserve the Cs symmetry of the radical. It is likely,
however, that the redox properties of 1 and hence on-
site Coulomb repulsion can be optimized further. For
instance, placing the phenyl group at the 3 position
significantly lowered both values in the parent benzo-
oxygen in the solid state, but their solutions decompose
quickly in air. This indicates that both of the halogenated
radicals are suitable as structural elements for liquid
crystalline compounds. Such materials should be suf-
ficiently stable for future investigation, if air is rigorously
excluded. In this context, 4H-thiadiazine 2d , a precursor
to the apparently most stable fluorinated radical 1d , was
appropriately substituted with liquid-crystallinity-pro-
moting groups, and the corresponding radicals are being
studied.31
Com p u ta tion a l Deta ils
Quantum-mechanical calculations were carried out at the
B3LYP/6-31G(d) level of theory32,33 using the Linda-Gaussian
98 package34 on a Beowulf cluster of 16 processors. Geometry
optimizations were undertaken using appropriate symmetry
constraints and tight convergence limits. Vibrational frequen-
cies were used to characterize the nature of the stationary
points and to obtain thermodynamic parameters. Zero-point
energy (ZPE) corrections were scaled by 0.9806.35 The Wiberg
bond order indices were obtained using the NBO algorithm36
supplied in the Gaussian package.
The Fermi constants were calculated at the B3LYP/cc-
pVDZ//B3LYP/6-31G(d) and converted to hfcc’s. The aCl values
for the natural isotopic abundance of 35/37Cl were obtained
using a conversion factor of 420.183.
Electronic excitation energies for the radicals were obtained
at the B3LYP/6-31G(d) level using the time-dependent DFT
calculations37 supplied in the Gaussian package. Following
general recommendations,38 energy changes and differences
were derived as the differences of SCF energies of individual
species computed using the diffuse function-augmented
6-31+G(d) basis set at the geometries obtained with the
6-31G(d) basis set (single point calculations). Thermodynamic
corrections were obtained using the 6-31G(d) basis set.
21
[1,2,4]thiadiazinyl.
Experimental and theoretical results indicate a planar
geometry of the radicals, which combined with their
packing properties may be of interest for the design of
semiconductive solids. The crystal structure of 1d has a
rare, ladder-type infinite chain of almost evenly spaced
S-N rungs (Figure 3). A similar columnar arrangement
of the heterocycles is often found in dithiadiazolyls,27 but
unlike in 1d , the stacks are composed of dimeric pairs
with large alternations of the S‚‚‚N (and S‚‚‚S) distances
(Peierls-type distortion). This packing of 1d in columns
appears to be driven by the SOMO-SOMO overlap and
overall results in diamagnetic solids. This is evident from
the less than 1% of spin concentration found by ESR in
sublimed 1d at ambient temperature. Materials with
such low spin concentrations are not expected to exhibit
high electron conduction.
Exp er im en ta l Section
The crystal packing of 1e is distinctly different from
that of 1d . The larger size of chlorine atoms does not
allow for efficient SOMO-SOMO interactions within a
column, and molecules form dimeric pairs through in-
plane weak S‚‚‚S interactions. The crystallographic re-
sults for 1d and 1e suggest rich opportunities for the
engineering of solid-state structures and hence prop-
erties of substituted benzo[1,2,4]thiadiazinyls. A variety
of such structures can, in principle, be obtained easily
through a recently described method.10
X-band ESR spectra were obtained using modulation am-
plitude 0.10 G and spectral width of 100 G. Solutions in
distilled benzene were degassed by three freeze/pump/thaw
cycles. Spin concentration and yields were calculated by double
integration of the ESR signal of the sample and of a measured
amount of 4-hydroxy-TEMPO radical purchased from Aldrich
and assumed to be 100% pure. Samples were referenced using
strong pitch with g ) 2.0028. PbO2 and K2CO3 were dried in
a vacuum at 100 °C for 24 h, in the presence of P2O5. AgO
(31) Fryszkowska, A.; Zienkiewicz, J .; Sienkowska, M.; Kaszynski,
P. Unpublished results.
Con clu sion s
(32) Becke, A. D. J . Chem. Phys. 1993, 98, 5648-5652.
(33) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785-789.
(34) Gaussian 98, Revision A.9; Frisch, M. J .; Trucks, G. W.;
Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J . R.;
Zakrzewski, V. G.; Montgomery, J . A. J r.; Stratmann, R. E.; Burant,
J . C.; Dapprich, S.; Millam, J . M.; Daniels, A. D.; Kudin, K. N.; Strain,
M. C.; Farkas, O.; Tomasi, J .; Barone, V.; Cossi, M.; Cammi, R.;
Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J .;
Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.;
Rabuck, A. D.; Raghavachari, K.; Foresman, J . B.; Cioslowski, J .; Ortiz,
J . V.; Baboul, A. G.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz,
P.; Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J .; Keith, T.;
Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill,
P. M. W.; J ohnson, B.; Chen, W.; Wong, M. W.; Andres, J . L.; Gonzalez,
C.; Head-Gordon, M.; Replogle, E. S.; Pople, J . A. Gaussian, Inc.:
Pittsburgh, PA, 1998.
(35) Scott, A. P.; Radom, L. J . Phys. Chem. 1996, 100, 16502-16513.
(36) Glendening, E. D.; Reed, A. E.; Carpenter, J . E.; Weinhold, F.
NBO, version 3.1.
(37) Casida, M. E.; J amorski, C.; Casida, K. C.; Salahub, D. R. J .
Chem. Phys. 1998, 108, 4439-4449.
(38) Clark, T.; Chandrasekhar, J .; Spitznagel, G. W.; Schleyer, P.
v. R. J . Comput. Chem. 1983, 4, 294-301.
Among the five investigated fused-ring [1,2,4]thia-
diazinyl radicals only the halogenated derivatives 1d and
1e can be generated conveniently in high yields, isolated
chromatographically, and purified by vacuum sublima-
tion. The two radicals are practically inert to atmospheric
(27) For example: Banister, A. J .; Hansford, M. I.; Huptman, Z. V.;
Wait, S. T.; Clegg, W. J . Chem. Soc., Dalton Trans. 1989, 1705-1713.
Bricklebank, N.; Hargreaves, S.; Spey, S. E. Polyhedron 2000, 19,
1163-1166. Bond, A. D.; Haynes, D. A.; Pask, C. M.; Rawson, J . M. J .
Chem. Soc., Dalton Trans. 2002, 2522-2531.
(28) Garito, A. F.; Heeger, A. J . Acc. Chem. Res. 1974, 7, 232-240.
Haddon, R. C. Aust. J . Chem. 1975, 28, 2343-2351. Torrance, J . B.
Acc. Chem. Res. 1979, 12, 79-86.
(29) Boere´, R. T.; Moock, K. H. J . Am. Chem. Soc. 1995, 117, 4755-
4760.
(30) Beer, L.; Brusso, J . L.; Cordes, A. W.; Godde, E.; Haddon, R.
C.; Itkis, M. E.; Oakley, R. T.; Reed, R. W. Chem. Commun. 2002,
2562-2563.
7534 J . Org. Chem., Vol. 69, No. 22, 2004

Fused-Ring Persistent [1,2,4]Thiadiazinyl Radicals
was prepared according to a literature procedure39 and dried
as above at 50 °C.
different φ settings and a detector position of -28° in 2θ. Final
cell constants were calculated from 1617 (1d ) or 2515 (1e)
strong reflections from the actual data collection using SAINT
V6.35A.40 The intensity data for 1d were corrected for absorp-
tion using TWINABS.40 The intensity data for 1e were
corrected for absorption using SADABS.41 Both structures were
solved and refined using SHELXTL V6.12.41 All non-hydrogen
atoms were refined with anisotropic displacement parameters.
All hydrogen atoms were placed in ideal positions and refined
as riding atoms with relative isotropic displacement param-
eters.
Simulation of the ESR spectra was done with the PEST
program (version 0.96 for Windows; available at http//
epr.niehs.nih.gov/pest.html) using the scaled18 DFT results for
the input. The resulting hfcc’s were perturbed until the global
minimum for the fit was achieved. Spectra used for simulation
were generated by reflection of the left half of each spectrum.
Major absorption bands in IR spectra were assigned on the
basis of general trends and comparison with the results of DFT
calculations.
The space group for 1d was determined to be C2/c on the
basis of systematic absence intensity statistics. The specimen
selected for data collection was a nonmerohedral twin with
the twin law [-1 0 -1.75/0 -1 0/0 0 1], which corresponds to
a 180° rotation about direct axis (001).40 Additional reflection
manipulation was accomplished using program STRIP-
REDUNDANT V1.1,42 which removed redundant reflections.
In addition, all systematic absence violations were removed
using program SYSABSFILTER V1.1.42 There were 4767
reflections associated solely with the major twin component,
4763 reflections associated solely with the minor twin com-
ponent, and 6137 overlapping reflections.
The space group for 1e was determined to be P 1h; however
its unit cell is coincidentally close to the reduced cell corre-
sponding to a putative C-centered monoclinic cell. All three
examined specimens of 1e were pseudo-merohedral twins with
the twin law (by rows) [1 0 0/0 1 0/0 1 -1]. All residuals for 1e
were in the range R1 ) 0.085-0.090.
Electr och em ica l An a lysis. Gen er a l P r oced u r e. Cyclic
voltametry was performed at ambient temperature using a
CHI 660 electrochemical workstation (Austin TX) and a single-
compartment microcell equipped with a 2-mm Pt disk working
electrode, Pt wire counter electrode, and quasi-reference AgCl/
Ag electrode. Dry and degassed CH2Cl2 or MeCN (2 mL) was
transferred to a mixture of solid radical 1d or 1e (2 µmol) and
0.2 mmol of dry electrolytic grade [n-Bu4N][ClO4] (MeCN) or
[n-Bu4N][PF6] (CH2Cl2) placed in an electrochemical microcell
under Ar. Ar was passed through the resulting solution for 2
min. The electrochemical potential was scanned from -1.0 V
to +1.4 V at a rate of 100 mV/s with respect to the quasi-
reference AgCl/Ag electrode. At the end of the analysis a small
amounts of ferrocene was added, and the potential was
scanned again for referencing. Following a general procedure,19
the Fc/Fc+ couple was set to +0.38 V relative to SCE in MeCN
and +0.48V relative to SCE in CH2Cl2.
Sta bility Stu d ies. Gen er a l P r oced u r e. Meth od A. Solid
radical 1e (0.01 mmol) was dissolved in dry benzene (10 mL),
and the solution was degassed by three freeze/pump/thaw
cycles and finally filled with Ar. The solution was transferred
to an argon-filled UV cell equipped with a Teflon stopcock. For
measurements in the presence of air, a part of the stock
solution was placed in an ordinary UV cell, and air was slowly
passed for 15 s. The samples were constantly irradiated with
monochromatic light at 680 nm and the absorption at this
wavelength was measured every 30 s at room temperature.
Meth od B. Solutions of radicals 1a -1c in benzene were
generated according to Method B using AgO (reaction times:
20 min, 1a ; 40 min, 1b; 1c, 75 min) and filtered through a
microfilter. Solutions of 1d and 1e were prepared by dissolving
sublimed radicals. The resulting solutions in ESR tubes (c )
∼0.1 mM) were degassed by three freeze/pump/thaw cycles.
One set of tubes was kept under vacuum, and the other was
filled with dry Ar. Intensity of the ESR signal for each solution
was monitored at room temperature for several days.
1,5,6,7,8-P e n t a c h lo r o -3-p h e n y lb e n zo [1,2,4]t h ia d i-
a zin e9 (3e). A solution of freshly distilled SCl2 (1.12 g, 11
mmol) in CH2Cl2 (10 mL) was added dropwise to a stirred
Electr on ic Absor p tion Sp ectr a . Spectra of 1d and 1e
were recorded for solution of solid radicals in dry toluene at
ambient temperature. About 1 mM solutions of 1a and 1b were
generated in dry toluene with AgO/K2CO3, filtered through a
microfilter, and transferred to a UV cell.
Gen er a tion of Ra d ica ls. Meth od s A a n d B. Precursor 2
(0.01 mmol) was dissolved in degassed dry benzene (10 mL, 1
mM solution) and stirred over well-ground PbO2 (24 mg, 0.10
mmol, Method A) or AgO (12.5 mg, 0.10 mmol, Method B) and
K2CO3 (14 mg, 0.10 mmol) for 15 min under an atmosphere of
dry N2. The solution acquired a dark green color. To monitor
progress of the reaction, several samples, the first 5 min after
the beginning of the experiment, were taken from the reaction
mixture, and the amount of the radical was quantified by ESR
spectroscopy. Typically, a 0.5-mL sample was transferred to
an ESR tube using a syringe and a microfilter and carefully
degassed before taking the measurement.
Meth od C. Similar to Method B except for using a 1:1
mixture of MeCN and toluene as the solvent. Samples of the
solution were taken initially every 15 s to record the absorb-
ance at 610 nm for 1a , and 680 nm for 1e.
On a 10 mmol preparative scale, 1d and 1e were obtained
by oxidation of 2d and 2e respectively, with AgO/ K2CO3 at
ambient temperature. After vigorously stirring for 90 s, the
resulting black-green reaction mixture was diluted with
toluene and filtered through a short jacketed column (5 cm of
Florisil topped with 0.5 cm of silica) below 0 °C. Solvents were
removed under reduced pressure (∼1 Torr) at temperatures
below 0 °C, leaving a black-green solid of 1. Radicals 1d and
1e were sublimed in a vacuum (10-4 Torr) at 100 °C and 150
°C, respectively.
Meth od D. A solution of SO2Cl2 (0.005 mmol) in dry toluene
(0.5 mL) and solution of pyridine (0.01 mmol) or Me2EtN (0.01
mmol) in dry toluene (0.5 mL) were added simultaneously
dropwise (one drop every 3 s) to a stirred solution of precursor
2 (0.01 mmol) in dry toluene (10 mL) at -15 °C. The resulting
dark green solution of 1 was passed through a jacketed Florisil
column (6 × 1 cm) at -78 °C and washed with toluene using
a low pressure of N2. For isolation of radicals 1d and 1e an
additional 1-cm layer of silica gel was placed on the bottom of
the column. Solvents were removed under reduced pressure
(∼1 Torr) at temperatures below 0 °C to give a black-green
solid of 1.
X-r a y Cr ysta llogr a p h y. Black-purple (1d ) or black-green
(1e) crystals (approximate dimensions 0.25 × 0.13 × 0.10 mm3
for 1d and 0.40 × 0.08 × 0.04 mm3 for 1e) were placed onto
the tip of a 0.1-mm diameter glass fiber and mounted on a
CCD diffractometer for data acquisition at 173(2) K. Prelimi-
nary sets of cell constants were calculated from reflections
harvested from three sets of 50 (1d ) or 20 (1e) frames. This
produced initial orientation matrixes determined from 215 (1d )
or 78 (1e) reflections. The data acquisitions were carried out
using Mo KR radiation (graphite monochromator, λ ) 0.71073
Å) with frame durations of 60 (1d ) or 120 (1e) seconds and a
detector distance of 4.9 cm. Randomly oriented regions of
reciprocal space were surveyed to the extent of one sphere and
to a resolution of 0.84 Å. Four major sections of frames were
collected with 0.30° (1d ) or 0.50° (1e) steps in ω at four
(40) Bruker Analytical X-ray System, Madison, WI 2000.
(41) Bruker Analytical X-ray System, Madison, WI 2003.
(42) Brennessel, W. W.; Young, V. G., J r. Unpublished results.
(39) Hammer, R. N.; Kleinberg, J .; Holtzclaw, H. F., J r.; J ohnson,
K. W. R. Inorg. Synth. 1953, 4, 12-14.
J . Org. Chem, Vol. 69, No. 22, 2004 7535

Zienkiewicz et al.
solution of crude N-chloro-N′-phenylbenzamidine (2.31 g),
obtained from N-phenylbenzamidine (1.96 g, 10 mmol) and
tert-butyl hypochlorite (1.19 g, 11 mmol) in CH2Cl2 (30 mL),
and pyridine (1.58 g, 20 mmol) in CH2Cl2 (40 mL) at 0 °C.11
Chlorine (about 50 mmol) was passed through the reaction
mixture for 30 min. The mixture was stirred at 0 °C for 1 h
and left overnight at ambient temperature. The resulting red
precipitate was filtered to give 1.8 g (45% yield) of crude 3e,
which was used without further purification.
5,6,7,8-Tet r a ch lor o-3-p h en ylb en zo[1,2,4]-4H -t h ia d i-
a zin e (2e). Crude sulfiminyl chloride 3e (1.8 g, 4.5 mmol) was
added in one portion to a stirred solution of dry 1-propanethiol
(0.70 g, 9.2 mmol) in CH2Cl2 (50 mL) at ambient temperature.
The reaction mixture was stirred for 20 min, diluted with
CH2Cl2, and passed through a silica gel plug (CH2Cl2). Solvent
was evaporated, and the organic products were passed through
a short silica gel column using first a hexanes-CH2Cl2 mixture
in 10:1 ratio to remove dipropyl disulfide and then in 1:1 ratio
to elute 2e, which after recrystallization (hexanes-toluene,
5:1 ratio) gave 1.38 g (84% yield) of a yellow solid: mp 193-
195 °C; 1H NMR (300 MHz, CDCl3) δ 7.0 (brs, 1H), 7.42-7.55
(m, 3H), 7.62-7.65 (m, 2H); IR (neat) 3415 (N-H), 1626
(amidine), 1463 and 1349 (C6Cl4 ring) cm-1; MS m/e (relative
intensity) 364 (M+, 2), 103 (100). Anal. Calcd for C13H6-
Cl4N2S: C, 42.89; H, 1.66; N, 7.69. Found: C, 42.76; H, 1.69;
N, 7.57.
7.45 (td, J 1 ) 7.4, J 2 ) 1.2 Hz, 1H), 7.60-7.70 (m, 6H), 7.78
(d, J ) 7.5 Hz, 1H), 8.05-8.08 (m, 2H); IR (neat) 3245 and
3192 (N-H), 1031 (SdO) cm-1; HRMS m/e calcd for C13H11N2OS
243.0592, found 243.0597.
5,6,7,8-Tet r a ch lor o-3-p h en ylb en zo[1,2,4]-4H -t h ia d i-
a zin e-S-oxid e (4e). The S-oxide was isolated in about 70%
yield as the most polar colorless fraction (Rf ∼0.35 in AcOEt)
from a solid residue obtained after sublimation of 1e: mp >247
1
°C dec; H NMR (300 MHz, DMSO-d6) δ 7.62-7.74 (m, 3H),
8.19-8.22 (m, 2H); IR (neat) 2814 (br N-H), 1055 (SdO) cm-1
;
HRMS m/e calcd for C13H6Cl4N2OS 377.8955, found 377.8940.
Anal. Calcd for C13H6Cl4N2OS: C, 41.08; H, 1.59; N, 7.37.
Found: C, 41.45; H, 1.63; N, 7.14.
Ack n ow led gm en t. Financial support for this work
was received from the National Science Foundation
(CHE-0096827). Collection of synchrotrone crystal data
for 1e was performed at ChemMatCARS Sector 15 of
Argonne National Laboratory, which is principally
supported by the National Science Foundation/Depart-
ment of Energy (CHE0087817) and by the Illinois Board
of Higher Education. The Advanced Photon Source is
supported by the U.S. Department of Energy, Basic
Energy Sciences, Office of Science, under Contract
W-31-109-Eng-38. We thank Dr. David E. Cliffel for help
with analysis of the electrochemical data.
3-P h en ylben zo[1,2,4]-4H-th ia d ia zin e-S-oxid e (4a ). A
dark green solution of radical 3-phenylbenzo[1,2,4]thiadiazinyl
(1a ), prepared from 2a (34 mg, 0.15 mmol) with Method C,
was stirred at room temperature opened to air until the
solution became colorless (approximately 3 h). Solvents were
removed under reduced pressure, and the white solid residue
was extracted on a filtration funnel with solvents of gradually
increasing polarity starting with pentane. The main product
(26 mg, 72% yield) was obtained from the fraction washed with
CH2Cl2: mp >178 °C dec; 1H NMR (300 MHz, DMSO-d6) δ
Su p p or tin g In for m a tion Ava ila ble: Kinetic analysis of
the formation and decomposition curves for 1, experimental,
simulated, and difference spectra for 1, spin densities, calcu-
lated hfcc, details of computational results, and full crystal-
lographic results for 1d and 1e in CIF format. This material
is available free of charge via the Internet at http://pubs.acs.org.
J O0486746
7536 J . Org. Chem., Vol. 69, No. 22, 2004