Rhenium-based molecular rectangles as frameworks for ligand-centered mixed valency and optical electron transfer

Article abstract of DOI:10.1021/ja0473182

A series of six neutral, tetrametallic, molecular rectangles has been synthesized that have the form ([Re(CO)₃]₂BiBzlm) ₂-μ,μ′(LL)₂, where BiBzlm is 2,2′-bisbenzimidazolate and LL is a reducible, dipyridyl or diazine ligand. X-ray crystallographic studies of the six show that the rectangle frameworks, as defined by the metal atoms, range in size from 5.7 A × 7.2 A to 5.7 A × 19.8 A. The singly reduced rectangles are members of an unusual category of mixed-valence compounds in which the ligands themselves are the redox centers and interligand electronic communication is controlled by direct ligand orbital overlap rather than by superexchange through the metal ions. Despite nominally identical coordination-defined ligand positioning, the spectrally determined electronic strengths, H_ab², vary by roughly 100-fold. As shown by X-ray crystallography and computational modeling, the observed differences largely reflect detailed geometric configurational differences that can either facilitate or frustrate productive direct orbital overlap.

Full text of DOI:10.1021/ja0473182

Published on Web 09/17/2004
Rhenium-Based Molecular Rectangles as Frameworks for
Ligand-Centered Mixed Valency and Optical Electron Transfer
Peter H. Dinolfo, Mary Elizabeth Williams, Charlotte L. Stern, and Joseph T. Hupp*
Contribution from the Department of Chemistry, 2145 Sheridan Road, Northwestern UniVersity,
EVanston, Illinois 60208-3113
Received May 7, 2004; E-mail: jthupp@chem.northwestern.edu
Abstract: A series of six neutral, tetrametallic, molecular rectangles has been synthesized that have the
form ([Re(CO)3]2BiBzIm)2-µ,µ′-(LL)2, where BiBzIm is 2,2′-bisbenzimidazolate and LL is a reducible, dipyridyl
or diazine ligand. X-ray crystallographic studies of the six show that the rectangle frameworks, as defined
by the metal atoms, range in size from 5.7 Å × 7.2 Å to 5.7 Å × 19.8 Å. The singly reduced rectangles are
members of an unusual category of mixed-valence compounds in which the ligands themselves are the
redox centers and interligand electronic communication is controlled by direct ligand orbital overlap rather
than by superexchange through the metal ions. Despite nominally identical coordination-defined ligand
positioning, the spectrally determined electronic strengths, Hab2, vary by roughly 100-fold. As shown by
X-ray crystallography and computational modeling, the observed differences largely reflect detailed geometric
configurational differences that can either facilitate or frustrate productive direct orbital overlap.
Introduction
While the cavities of the new rectangles are too narrow to
encapsulate guest molecules, they are of finite width, with LL/
Molecular rectangles comprise one subset of a now quite large
collection of discrete cavity-containing supramolecular as-
semblies (squares, triangles, prisms, etc.) obtainable via metal-
ion coordination chemistry.1-5 Recently we reported on the
synthesis of rectangles featuring rhenium(I) sites organized in
pairs by doubly chelating planar ligands such as 2,2′-bisbenz-
imidazolate and then linked by rigid or semirigid difunctional
ligands (LL) such as 4,4′-bipyridine to close the cycle.6 Notably,
the 4,4′-bipyridine (bpy) containing rectangles exhibit well-
defined LL electrochemistry; depending on solvent, as many
as four electrons can be added in stepwise fashion with retention
of the rectangular configuration. Other examples of transition-
metal based rectangles include a tetrarhenium compound
featuring alternating pyrazine and bpy edges,7,8 tetrarhenium
compounds having bridging thiolates9 or alkoxides10,11 along
the short edges, and tetraplatinum compounds joined on the short
edge with anthracene.12-15
LL separation distances exceeding van der Waals contact
distances by ca. 1.5 to 2 Å, at least at the coordination sites.
We reasoned that the close proximity would facilitate through-
space electronic communication between ligands, potentially to
a much greater degree than coordination-bond-mediated interac-
tions (superexchange interactions). At the same time the metal
frameworks should define reasonably fixed cofacial ligand
geometries. To evaluate LL/LL electronic communication, we
have examined the rectangles spectroelectrochemically in the
1- charge state, i.e., a mixed-valence (MV) state with respect
to the ligands, LL. We find for most, but not all, of the
assemblies that electronic interactions are observable and their
strength is measurable based on the appearance of broad
absorption bands in the near-infrared region assignable as
intervalence transitions (IT), eq 1. IT band energies and shapes
can provide information about electronic coupling energies
(Hab), reorganization energies (λ), and activation free energies
(∆G*) for the corresponding thermal electron exchange reaction.
It is worth noting that most mixed-valence coordination
compounds feature metal ions rather than ligands as electron
donors and acceptors.16-22 Especially for second and third row
transition metals, the multiplicity of d orbitals, mixed and split
(1) Leininger, S.; Olenyuk, B.; Stang, P. J. Chem. ReV. 2000, 100, 853-907.
(2) Stang, P. J.; Olenyuk, B. Acc. Chem. Res. 1997, 30, 502-518.
(3) Fujita, M. Chem. Soc. ReV. 1998, 27, 417-425.
(4) Holliday, B. J.; Mirkin, C. A. Angew. Chem., Int. Ed. 2001, 40, 2022-
2043.
(5) Cotton, F. A.; Lin, C.; Murillo, C. A. Acc. Chem. Res. 2001, 34, 759-
771.
(12) Das, N.; Mukherjee, P. S.; Arif, A. M.; Stang, P. J. J. Am. Chem. Soc.
2003, 125, 13950-13951.
(6) Benkstein, K. D.; Hupp, J. T.; Stern, C. L. Angew. Chem., Int. Ed. 2000,
39, 2891-2893.
(13) Kaim, W.; Schwederski, B.; Dogan, A.; Fiedler, J.; Kuehl, C. J.; Stang, P.
J. Inorg. Chem. 2002, 41, 4025-4028.
(7) Rajendran, T.; Manimaran, B.; Liao, R.-T.; Lin, R.-J.; Thanasekaran, P.;
Lee, G.-H.; Peng, S.-M.; Liu, Y.-H.; Chang, I. J.; Rajagopal, S.; Lu, K.-L.
Inorg. Chem. 2003, 42, 6388-6394.
(14) Kuehl, C. J.; Huang, S. D.; Stang, P. J. J. Am. Chem. Soc. 2001, 123,
9634-9641.
(8) Rajendran, T.; Manimaran, B.; Lee, F.-Y.; Lee, G.-H.; Peng, S.-M.; Wang,
C. M.; Lu, K.-L. Inorg. Chem. 2000, 39, 2016-2017.
(9) Benkstein, K. D.; Hupp, J. T.; Stern, C. L. Inorg. Chem. 1998, 37, 5404-
5405.
(15) Kuehl, C. J.; Mayne, C. L.; Arif, A. M.; Stang, P. J. Org. Lett. 2000, 2,
3727-3729.
(16) Creutz, C. Prog. Inorg. Chem. 1983, 30, 1-73.
(17) Robin, M. B.; Day, P. AdV. Inorg. Chem. Radiochem. 1967, 10, 247-422.
(18) Crutchley, R. J. AdV. Inorg. Chem. 1994, 41, 273-325.
(19) Kaim, W.; Klein, A.; Gloeckle, M. Acc. Chem. Res. 2000, 33, 755-763.
(20) Demadis, K. D.; Hartshorn, C. M.; Meyer, T. J. Chem. ReV. 2001, 101,
2655-2685.
(10) Manimaran, B.; Rajendran, T.; Lu, Y.-L.; Lee, G.-H.; Peng, S.-M.; Lu,
K.-L. J. Chem. Soc., Dalton Trans. 2001, 515-517.
(11) Woessner, S. M.; Helms, J. B.; Shen, Y.; Sullivan, B. P. Inorg. Chem.
1998, 37, 5406-5407.
9
10.1021/ja0473182 CCC: $27.50 © 2004 American Chemical Society
J. AM. CHEM. SOC. 2004, 126, 12989-13001
12989

A R T I C L E S
Dinolfo et al.
Scheme 1
in energy by spin-orbit coupling, typically leads to multiple,
overlapping IT bands. The multiplicity and overlap can com-
plicate the estimation of Hab, λ, and ∆G* from band properties.
These complications should be absent for the rectangle-based
MV assemblies, as well as for most other ligand-centered mixed
valence (LCMV) compounds. Among the few examples of
LCMV compounds known to display measurable intervalence
absorptions are singly and doubly reduced tris-2,2′-bipyridyl
complexes of Ru and Ir.23,24 These are characterized by relatively
weak IT bands (ꢀ < 400 M-1 cm-1), a consequence of geometric
orthogonality (i.e., 90° orientation) of the planar donor ligand
and each of two acceptor ligands. Also displaying observable
but weak IT bands is [OsIII(bpy0)2(bpy-)]2+*, the lowest energy
excited-state form of this compound.20 Kitagawa and co-workers
Class II or borderline Class III (valence delocalized) on the
Robin and Day17 scale. Reported below are the syntheses, X-ray
crystal structures (neutral form), electrochemistry, and electronic
and vibrational spectroscopy of these compounds.
Results and Discussion
Synthesis and Initial Characterization. The general syn-
thesis of rectangles 1-6 is shown in Scheme 1 (abbreviations
correspond to ligand names, and numbers represent rec-
tangles).6,30,31 Equivalent quantities of [Re(CO)4]2BiBzIm (7)
(BiBzIm ) 2,2′-bisbenzimidazolate) and redox-active ligand,
LL, are refluxed in THF for 48 h. The resulting supramolecular
complexes are typically isolated in 43-96% yield.
have described a series of LCMV compounds, [CrIII(X4SQ)3-n
-
(X4Cat)n]-n (SQ ) semiquinonate, Cat ) catecholate, X ) Br
and Cl, n ) 0, 1, and 2), which have strong IT bands between
4500 and 6000 cm-1, mediated by the CrIII ions linking the
ligands.25-27 Recently Pratt and Stack have observed intense
(ꢀ ≈ 2500 M-1 cm-1) and moderately asymmetric IT bands
for two Cu salen compounds in their singly oxidized form where
the phenolic moieties on each side of the complex act as the
redox centers.28
The reactivity of 7 is an example of the well-known
translabilization effect of carbonyl ligands of Group VII
complexes.32-35 Reactions of this kind generally proceed through
a dissociative route. The substitution inertness of the doubly
chelating BiBzIm2- ligand, however, ensures that only one
coordination site per metal is opened, in this case cis to the
chelating ligand.36-38 The relatively high yields of closed cycles
(rectangles) point to the involvement of solvento complexes as
intermediates and equilibration to reach thermodynamic prod-
We find that the electronic interactions within the rectangle-
based MV assemblies are indeed dominated by direct through-
space LL/LL interactions rather than bond-mediated interactions.
Thus the rectangles comprise relatively rare examples of MV
systems that are describable by single-transition, two-state
treatments (as opposed to the three- and four-state treatments
appropriate for superexchange based electronic communication
and the multitransition treatments appropriate for metal-based
systems29). From Hush’s classical two-state theory, the available
mixed-valent rectangles span the range from essentially Class
I (noncommunicating, fully valence-localized redox centers) to
1
ucts: stable rectangular structures. H NMR, IR, and MS data
are consistent with the formation of discrete, highly symmetrical,
cycles and not open oligomers. Presumably, any partially reacted
or polymerized compounds remain in the reaction solution as
solvento species.
Figure 1 shows the carbonyl stretching region of the IR
spectra for 7, 2, and 4. Four bands are present for 7, consistent
with D2h symmetry. The spectrum for 2 shows only three CO
stretching bands, consistent with D2h symmetry. For rectangles
larger than 2, the local symmetry of the [Re(CO)3]2BiBzIm
(21) Ito, T.; Hamaguchi, T.; Nagino, H.; Yamaguchi, T.; Kido, H.; Zavarine, I.
S.; Richmond, T.; Washington, J.; Kubiak, C. P. J. Am. Chem. Soc. 1999,
121, 4625-4632.
(22) Elliott, C. M.; Derr, D. L.; Matyushov, D. V.; Newton, M. D. J. Am. Chem.
Soc. 1998, 120, 11714-11726.
(30) Benkstein, K. D.; Hupp, J. T.; Stern, C. L. J. Am. Chem. Soc. 1998, 120,
12982-12983.
(31) Benkstein, K. D.; Stern, C. L.; Splan, K. E.; Johnson, R. C.; Walters, K.
A.; Vanhelmont, F. W. M.; Hupp, J. T. Eur. J. Inorg. Chem. 2002, 2002,
2818-2822.
(32) Abel, E. W.; Wilkinson, G. J. Chem. Soc. 1959, 1501-1505.
(33) Wojcicki, A.; Basolo, F. J. Am. Chem. Soc. 1961, 83, 525-528.
(34) Angelici, R. J.; Basolo, F. J. Am. Chem. Soc. 1962, 84, 2495-2499.
(35) Angelici, R. J.; Basolo, F. Inorg. Chem. 1963, 2, 728-731.
(36) Atwood, J. D.; Brown, T. L. J. Am. Chem. Soc. 1976, 98, 3160-3166.
(37) Dobson, G. R. Acc. Chem. Res. 1976, 9, 300-306.
(38) Lichtenberger, D. L.; Brown, T. L. J. Am. Chem. Soc. 1978, 100, 366-
373.
(23) Heath, G. A.; Yellowlees, L. J.; Braterman, P. S. Chem. Phys. Lett. 1982,
92, 646-648.
(24) Coombe, V. T.; Heath, G. A.; MacKenzie, A. J.; Yellowlees, L. J. Inorg.
Chem. 1984, 23, 3423-3425.
(25) Chang, H.-C.; Kitagawa, S. Angew. Chem., Int. Ed. 2002, 41, 130-133.
(26) Chang, H.-C.; Mochizuki, K.; Kitagawa, S. Inorg. Chem. 2002, 41, 4444-
4452.
(27) Chang, H.-C.; Miyasaka, H.; Kitagawa, S. Inorg. Chem. 2001, 40, 146-
156.
(28) Pratt, R. C.; Stack, T. D. P. J. Am. Chem. Soc. 2003, 125, 8716-8717.
(29) Brunschwig, B. S.; Creutz, C.; Sutin, N. Chem. Soc. ReV. 2002, 31, 168-
184.
9
12990 J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004

Rhenium-Based Molecular Rectangles
A R T I C L E S
Table 1. Crystallographic Data for 1-6
compound
1
2
3
4
5
6
empirical
formula
formula
Re4C75H62-
N12O15.75
2128.2
Re2C27.5H14Cl-
N6O6
932.3
Re4C78H47Cl4-
N12O14
2262.9
Re4C64H32-
N20O12
2017.9
Re4C135.5H36-
N16O20
2952.6
Re4C94H52-
N16O24S2
2598.44
weight
crystal
yellow, block
red, block
yellow, needle
red, block
yellow, block
orange, needle
color, habit
crystal
0.102 × 0.056
× 0.044
0.180 × 0.166
× 0.080
0.412 × 0.050
× 0.026
0.284 × 0.222
× 0.158
0.212 × 0.148
× 0.118
0.456 × 0.032
× 0.030
dimensions
(mm3)
crystal
system
space
monoclinic
monoclinic
monoclinic
tetragonal
monoclinic
monoclinic
P21/c
P21/n
P21/c
I41/a
P21/n
P21/n
group
a (Å)
b (Å)
c (Å)
â (deg)
V (Å3)
Z
12.3741(10)
48.753(4)
24.198(2)
93.262(2)
14575(2)
8
13.830(3)
13.299(3)
19.931(5)
101.335(4)
3594.4(15)
4
12.401(4)
11.578(3)
26.649(8)
97.184(5)
3796.2(19)
2
30.8925(11)
30.8925(11)
26.6992(13)
91.420(3)
25480.3(18)
8
10.9266(19)
22.014(4)
29.322(5)
94.151(5)
7051(2)
2
11.000(3)
22.681(7)
19.552(6)
4865(3)
2
T [K]
153
1.94
6.699
0.968
153
1.723
6.846
1.047
153
1.980
6.571
1.065
153
1.052
3.828
1.027
153
1.391
3.487
1.071
153
F (calcd, g/cm3)
1.774
5.084
0.875
µ (cm-1
)
goodness-of-fit
on F2
Ra
Rw
0.0783
0.1490
0.0527
0.1654
0.0496
0.1451
0.0665
0.2873
0.0508
0.1604
0.0640
0.1795
b
a R(F) ) (∑Fo - Fc)/∑Fo. b Rw(Fo ) ) [∑[w(Fo - Fc )2]/∑wFo ]1/2
.
2
2
2
4
ethynl porphyrin ligands where van der Waals contact is
achieved over nearly the entire porphyrinic area of the LL/LL
pair.31
The crystal structure of 1 contains two distinct but very similar
rectangles in the asymmetric unit, both of which have dimen-
sions of 11.5 Å × 5.7 Å as defined by the Re atoms.39 These
are slightly larger than those found previously for the Mn
analogue (11.2 Å × 5.5 Å).6 Comparison of the structures of 1
and the Mn analogue6 shows that the dimensional differences
reflect both metal-ligand bond length and ligand geometry
differences. In 1, the Re-N bond lengths are approximately
0.1 Å longer than that of the Mn-N bond lengths which
accounts for the 0.2 Å difference along the short edge and part
of the 0.3 Å difference for the long edge. For the Mn version,
a near zero dihedral angle exists for the pair of rings comprising
each bipyridine ligand. The bpy ligands, however, bow inward,
thereby reducing the Mn-Mn separation distances along the
long edges of the rectangle.
The differences between the two crystallographic isomers of
1 are due to slight variations in the orientation of the bpy ligands
within the rectangular cycle. Curiously, the bpy rings in the
Mn analogue are essentially planar (dihedral angle between the
pyridine rings is less than 10°), while, in 1, the dihedral angles
range from 30° to 40°, values similar to those of unconstrained
biphenyl and 4,4′-bipyridyl compounds. The collinearity of the
pyridines within each twisted bpy ligand serves to place at a
maximum possible separation distance the pairs of Re atoms
defining the long edges of the rectangle. In the rectangle
environment, the dihedral twist serves to bring the otherwise
cofacial pyridyl rings of the pair of bpy ligands into van der
Waals contact (∼3.8 Å closest C-C distance). The geometry
around the Re atoms is distorted octahedral; the angles between
Figure 1. Carbonyl stretching region of the infrared spectra for 7 (solid
black line, top), 2 (long dashed blue line, middle), and 4 (short dashed
green line, bottom) taken in THF.
edges, C2V, governs the spectral behavior and the higher energy
transition is split, such that four distinct CO stretching bands
are observed for 3-6. 1H NMR spectra of 2-6 are also
consistent with the formation of highly symmetrical rectangular
structures. Upon rectangle formation, the R and â protons of
coordinated BiBzIm shift upfield by approximately 0.15 ppm,
while the resonances of the LL species generally shift downfield.
Only one set of LL proton resonances and one set of BiBzIm
resonances are seen for each assembly, consistent with the
formation of highly symmetrical closed cycles.
X-ray Crystal Structures. ORTEP drawings of the single-
crystal X-ray structures of 1-6 are shown in Figure 2. The
structures were solved by direct methods and corroborate the
proposed molecular structures. Table 1 contains crystal informa-
tion and refinement data. Table 2 lists representative Re-N bond
lengths. Common to the structure of all but the smallest
assembly are distortions from rectangular box geometries such
that the LL pairs achieve van der Waals contact. We have
previously observed an extreme example, based on pyridyl
(39) Benkstein, K. D. The Design, Synthesis and Evaluation of Transition Metal-
Based Cyclophanes for Host: Guest Applications. Ph.D. Thesis, North-
western University, Evanston, IL, 2000.
9
J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004 12991

A R T I C L E S
Dinolfo et al.
Figure 2. ORTEP drawings for the crystal structures of 1-6. Carbon atoms are shown in black, nitrogen in blue, oxygen in red and rhenium in green; the
hydrogen atoms and solvent molecules were removed for clarity. Displacement ellipsoids are drawn at the 50% probability level for 1, 40% for 2, 3, 5, and
6, and 30% for 4. The figure for 1 displays only one of the two rectangles present in the asymmetric unit.
the imidazolyl and pyridyl nitrogens (N-Re-N) range from
80.1° to 89.1°. The unit cell also contains eight disordered THF
molecules.
The structure of 2, in contrast to 1, is characterized by a
uniform LL/LL spacing (∼5.3 Å), reflecting the rigidity of the
cofacial pyrazine ligands. The rectangle framework dimensions,
as defined by the Re atoms, are 7.2 Å × 5.7 Å. Similar to 1,
the geometry of the Re coordination sphere is distorted
octahedral. The angles between the BiBzIm nitrogens and the
bpy pyridyl nitrogens (N-Re-N) range from 83.4° to 84.3°.
Somewhat different from the rest of the rectangles is an outward
bending of the BiBzIm chelating ligands, which also effects
the distortion of the Re centers. The X-ray measurements
additionally show that the unit cell of 2 contains disordered
dichloroethane and THF molecules.
The structures of 3 and 4 comprise moderately distorted
rectangles. The greater length of the diPyTz and dap ligands,
and the ease which coplanarity of ligand rings can be achieved,
allow LL pairs to come within van der Waals contact. The
framework dimensions of 3 and 4, as determined by the Re
9
12992 J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004

Rhenium-Based Molecular Rectangles
A R T I C L E S
Table 2. Re-N Bond Distances for 1-6
rectangles, the unit cell of 4 does not contain any additional
solvent molecules.
bond lengths (Å)
Single-crystal X-ray studies of 5 and 6 reveal similar
structures, in which the central benzene tetracarboxidiimide and
naphthalene tetracarboxidiimide rings are rotated in relation to
the Re coordinating pyridine groups (dihedral angles of ca. 45°
for 5 and ca. 55° for 6). The rotation allows the central rings to
reach van der Waals contact while retaining linearity along the
N-N-N-N ligand axis and avoiding the highly compressed
structures seen for 3 and 4. The dihedral displacements also
facilitate horizontal offsets of the central rings, such that less
than 25% and 40% cofacial area overlaps are attained for the
central portions of the ligands for 5 and 6, respectively. Offsets
of this kind are characteristic of π-stacking interactions, placing
comparatively electron-deficient hydrogens in close proximity
to the central electron-rich sections of the neighboring aromatic
rings. In 5, there is disorder in one of the pyridine units, such
that the ring is orientated either parallel (66% occupancy) or
orthogonal (34%) to the Re-Re plane. There does exist a slight
inward bending of the ligands that effects the geometry of the
Re centers. The angles between of the imidazolyl and pyridyl
nitrogens in 5 (N-Re-N) range from 83.2° to 85.8°, while these
angles in 6 range from 81.5° to 84.4°. In addition, the unit cell
of 5 contains six disordered THF molecules, and 6 contains
three disordered MeOH molecules and one disordered DMSO
molecule. The framework dimensions of 5 and 6, as determined
by there Re atoms, is 19.5 Å × 5.7 Å and 19.8 Å × 5.7 Å,
respectively.
Notable, across the series of 1-6 are the similarities of the
Re-N bond distances for both the pyridyl and imidazolyl
nitrogens, each with averages of 2.2 Å. These values are close
to those of other pyridyl-Re(CO)3 compounds40-42 and a
previously reported porphyrinic based rectangle.31 The similar
Re-N bond lengths for pyridyl versus imidazolyl nitrogens
suggest that the BiBzIm remains largely dianionic and does not
contribute enormous electron density to the cationic Re atoms.
Electrochemistry. Cyclic voltammograms (CV) of 1-6
display chemically reversible reduction waves; representative
CVs of 2, 3, and 4 are shown in Figure 3. Similar behavior has
been noted previously for 1 and related compounds.6,9-11,13,31,43
The waves are assigned as one-electron reduction reactions of
the LL ligands. Stepwise reduction of identical ligands reflects,
in part, the electrostatic cost of creating a negatively charged
ligand (or making an already negatively charged ligand more
negative) in the presence of another negatively charged ligand.
No oxidative waves are observed for 1-6 within the solvent
windows.
LL pyridyl N
−
Re
imidazolyl N−Re
1
Re1-N1
2.208(10)
2.259(12)
2.212(10)
2.215(11)
2.199(11)
2.233(12)
2.233(11)
2.200(10)
2.196(7)
2.205(7)
2.229(6)
2.231(7)
2.216(9)
2.229(9)
2.210(5)
2.222(5)
2.222(10)
2.218(10)
Re1-N2
2.207(10)
2.235(11)
2.199(11)
2.216(11)
2.195(10)
2.199(10)
2.191(11)
2.215(10)
2.197(11)
2.167(12)
2.218(10)
2.194(10)
2.202(10)
2.199(11)
2.227(10)
2.258(11)
2.210(8)
2.200(7)
2.220(7)
2.211(8)
2.213(7)
2.218(6)
2.218(6)
2.199(7)
2.220(10)
2.259(10)
2.242(10)
2.226(11)
2.217(6)
2.221(6)
2.210(6)
2.213(6)
2.246(11)
2.200(13)
2.192(10)
2.156(12)
Re1-N3
Re2-N4
Re2-N5
Re3-N8
Re3-N9
Re4-N10
Re4-N11
Re5-N14
Re5-N15
Re6-N16
Re6-N17
Re7-N20
Re7-N21
Re8-N22
Re8-N23
Re1-N1
Re1-N2
Re2-N3
Re2-N4
Re1-N1
Re1-N3
Re2-N2
Re2-N4
Re1-N3
Re1-N4
Re2-N1
Re2-N2
Re1-N1
Re1-N2
Re2-N4
Re2-N3
Re1-N1
Re1-N2
Re2-N3
Re2-N4
Re2-N6
Re3-N7
Re4-N12
Re5-N13
Re6-N18
Re7-N19
Re8-N24
Re1-N5
Re2-N7
Re1-N6
Re2-N5
Re1-N10
Re2-N9
Re1-N5
Re2-N8
Re1-N5
Re2-N8
2
3
4
5
6
atoms, are 11.2 Å × 5.7 Å and 15.3 Å × 5.7 Å, respectively.
While more extreme in the case of 4, the inward bowing of LL
ligands distorts the Re-BiBzIm short edges of both rectangles,
drawing the Re atoms toward the center of the molecule.
As a result of the bend in the central dap ligand, the
geometries around the Re centers in 3 are distorted octahedral.
The angles between nitrogens of the BiBzIm and the diazapyrene
ligands (N-Re-N) range from 80.0° to 84.5°, while the angles
between the BiBzIm nitrogens and the axial carbonyl ligands
(N-Re-C) are somewhat enlarged (94.8° to 97.5°), consistent
with an inward displacement of the Re atoms. Along with the
displacement of the Re centers, the BiBzIm ligand is curved
toward the center of the complex, with a mean deviation of 0.1
Å as defined by a plane through the BiBzIm bridge. The unit
cell of 3 contains one disordered 1,2-dichloroethane and one
disordered THF molecule.
In addition to the inward bending of the diPyTz ligands of
4, the central tetrazine rings are twisted slightly in relation to
the Re coordinating pyridine units (ca. 10-15°). This twist
facilitates the attainment of LL/LL van der Waals contact
while minimizing the need for further bending of the diPyTz
ligand and distortion of the rectangular shape. The residual
distortion of the octahedral Re centers is similar to that of 3;
the angles between nitrogens of the BiBzIm and the diazapyrene
ligands (N-Re-N) range from 81.0° to 83.8°, while the angles
between the BiBzIm nitrogens and the axial carbonyl ligands
(N-Re-C) range from 95.1° to 98.0°. Unlike the rest of the
Table 3 lists reduction potentials for 1-6 and the correspond-
ing free LL species versus ferrocenium/ferrocene, a reference
couple whose potential should be largely independent of solvent
composition. The ligand reductions for the rectangular as-
semblies are shifted to substantially less negative potentials in
comparison with free ligands.44-46 The differences are consistent
(40) Be´langer, S.; Hupp, J. T.; Stern, C. L. Acta Crystallogr., Sect. C: Cryst.
Struct. Commun. 1998, C54, 1596-1600.
(41) Be´langer, S.; Hupp, J. T.; Stern, C. L.; Slone, R. V.; Watson, D. F.; Carrell,
T. G. J. Am. Chem. Soc. 1999, 121, 557-563.
(42) Slone, R. V.; Hupp, J. T.; Stern, C. L.; Albrecht-Schmitt, T. E. Inorg. Chem.
1996, 35, 4096-4097.
(43) Hartmann, H.; Berger, S.; Winter, R.; Fiedler, J.; Kaim, W. Inorg. Chem.
2000, 39, 4977-4980.
(44) Braterman, P. S.; Song, J. I. J. Org. Chem. 1991, 56, 4678-4682.
(45) O’Reilly, J. E.; Elving, P. J. J. Am. Chem. Soc. 1972, 94, 7941-7949.
9
J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004 12993

A R T I C L E S
Dinolfo et al.
Figure 4. UV-visible absorption data for the neutral (solid black line),
1- (long dashed green line), and 2- (short dashed blue line) states for 1
and 2 taken in THF, with 0.1 M TBAPF6 as the supporting electrolyte.
orthogonal pyridine spacers.47,48 Unlike rectangles 1-6, cyclic
voltammograms of 7 feature chemically irreversible reduction
peaks, which are likely associated with reduction of the Re
centers.
Spectroelectrochemistry: Visible and Infrared Regions.
As shown in Figure 4 for 1 and 2, the UV-visible absorption
spectra of the neutral rectangles are dominated by high energy
(300-400 nm) bands associated with intraligand (BiBzIm)
transitions. In addition, spectra for 2 and 4 contain absorption
bands from 400 to 500 nm, which are associated with the pyrz
and diPyTz ligands. Ligand absorptions for 5 and 6 are higher
in energy and are obscured by the BiBzIm transitions. Table
S1 includes quantitative UV-visible absorption data for 1-6
in the 0, 1-, and 2- states. None of the molecular rectangles
are detectably emissive in solution at ambient temperatures.49
Support for the assignments above of electrochemical reduc-
tion waves as LL based is provided by spectroelectrochemical
measurements. As shown in Figure 4, when an appropriately
negative potential is applied to reduce 1 to the 1- state,
absorption bands at 396, 426, 598, and 644 nm evolve,
indicating the presence of the radical anion form of the 4,4′-
bpy ligand. The peaks are lower in energy than those reported
for the anionic form of the free ligand50 but closely approximate
to those of the radical cation of methyl viologen.51 The bands
for 1 increase in intensity, as expected, upon reduction to the
2- state and formation of a second anionic bipyridine. Hartmann
and Kaim have reported similar behavior for a cationic analogue
of 1.43 Molecular rectangle 3, in the 1- state, shows ligand
absorptions at 468, 460, and 610 nm, similar to values reported
for the N,N′-dimethyl-2,7-diazapyrene monocation.52 In the
singly reduced form of 4, absorptions at 450, 476, and 718 nm
match those of the N,N′-dimethyl-di-3,6-(4-pyridyl)-1,2,4,5-
Figure 3. Cyclic voltammograms of 2 (top), 3 (middle), and 4 (bottom) in
THF with 0.1 M TBAPF6 as the supporting electrolyte.
Table 3. Reduction Potentials (V vs Fc/Fc+)
/
1
-
-/
E1
2
-
-/
E2
3
-
-/
E3
4-
1
-
3-
rectangle E0
Kc
Kc
solventa
1/
2
1/
2
1/
2
1/2
1b
2
3
4
5
-1.47 -1.70 -2.28 -2.43 7.7 × 103 3.0 × 102 THF
-1.23 -1.48 -2.76 -2.89 3.7 × 104 2.0 × 102 THF
-1.56 -1.92
1.2 × 106
THF
-0.95 -1.30 -1.82 -2.08 8.2 × 105 2.5 × 104 THF
-0.86 -1.11
-0.70 -0.79
1.7 × 104
THF
DMSO
6
-1.33c
30
/
1-
-/2-
1/2
LL ligand
E0
E1
solventa
1/
2
bpyd
-2.39
-2.88
DMF
CH3CN
THF
THF
DMF
DMF
pyrze
-2.47g
-2.39
-1.81
-1.02g
-0.79g
dap
diPyTz
diPyBIf
diPyNIf
-2.50
-1.68g
-1.30g
a 0.1 M TBAPF6 used as supporting electrolyte. b Values taken from
ref 6. c 2-/3- and 3-/4- waves strongly overlap. d Values taken from
ref 44. e Value taken from ref 45. f Values from di-(2,5-di-tert-butylphenyl)-
substituted tetracarboximide dianhydrides used in place of dipyridyl
complexes, taken from ref 46. g Converted to Fc0/+ reference by subtracting
0.31 from reported value referenced to SCE.
(47) Hayes, R. T.; Wasielewski, M. R.; Gosztola, D. J. Am. Chem. Soc. 2000,
122, 5563-5567.
with the expectation of electrostatic stabilization of the nega-
tively charged, reduced forms of the ligands by the coordinated
rhenium cations. The positive shifts are smaller for 5 and 6,
most likely due to isolation of charge on the central tetra-
carboxydimide dianhydrides and away from the nominally
(48) Miller, S. E.; Lukas, A. S.; Marsh, E.; Bushard, P.; Wasielewski, M. R. J.
Am. Chem. Soc. 2000, 122, 7802-7810.
(49) The luminecense reported for 1 in ref 6 is incorrect; the observed emission
at ca. 600 nm was due to a contaminant.
(50) Shida, T. Electronic Absorption Spectra of Radical Ions; Elsevier: Am-
sterdam, 1988.
(51) Kosower, E. M.; Cotter, J. L. J. Am. Chem. Soc. 1964, 86, 5524-5527.
(52) Waldhoer, E.; Zulu, M. M.; Zalis, S.; Kaim, W. J. Chem. Soc., Perkin
Trans. 2 1996, 1197-1204.
(46) Gosztola, D.; Niemczyk, M. P.; Svec, W.; Lukas, A. S.; Wasielewski, M.
R. J. Phys. Chem. A 2000, 104, 6545-6551.
9
12994 J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004

Rhenium-Based Molecular Rectangles
A R T I C L E S
Figure 5. IR-SEC scans in the carbonyl region for 2 (left) and 4 (right) taking in THF with 0.1 M TBAPF6 as the supporting electrolyte. The top graphs
show the reduction from the 0 to 1- state, and the bottom graphs show the reduction from the 1- to 2- state for each compound.
tetrazine monocation.52 Finally, the ligand absorptions at 424,
494, 652, and 718 nm, for 51-, and 458, 564, 612, and 780 nm,
for 61-, parallel those of radical anions of closely related di-
tert-butylphenyl substituted benzenediimide and naphthalene-
diimide species.46 Absorption spectra for 3-6, in the neutral,
1- and 2- states, are shown in Figure S1.
Somewhat different behavior was encountered for 2 (Figure
4). A weak band at 422 nm, assigned as a rhenium-to-pyrazine
charge-transfer band, decreases in intensity upon reduction to
the 1- state. An even weaker band, now at somewhat lower
energy, is observed for the 2- form. Its origin is uncertain, but
assignment as a ligand-centered (pyrazine anion) transition is
one possibility.
Figure 6. Plot of average CO frequency shifts between neutral and 2-
states versus the area of the bridging ligand for 1-4 (blue circles) and 6
and 7 (red triangles). The solid line represents the linear regression for 1-4.
The plot illustrates the effect of charge localization on the bridging ligands
and Re(CO)3 centers as the size of the ligand increases. For 6 and 7, the
charge is localized on the central tetracarboxydimide rings. For these two,
the nominally orthogonal pyridyl linkers restrict electronic communication
with the Re(CO)3 centers (see semiempirical modeling below).
SEC measurements for 1-6 were also made in the CO
vibrational region of the infrared spectrum. Stretching frequen-
cies for carbonyls are well-known to report on relative electron
densities at coordinated metal centers, generally moving lower
as electron density increases. Figure 5 shows the IR-SEC data
for 2 and 4. The reduction of 2 to the 1- state breaks the D2h
symmetry, and two new bands appear at 2022 and 2009 cm-1
.
The bands around 1925 cm-1 are shifted to lower energy but
are not as well resolved. As the rectangle is further reduced to
the 2- state, the symmetry is restored and only three bands are
present, now at 2007, 1892, and 1879 cm-1. In contrast, the
addition of an electron to 4 does not affect the local symmetry
of the Re(CO)3 units sufficiently to generate resolvable new
bands. All four bands of the neutral complex shift to progres-
sively lower energy upon reduction to the 1- and then 2- states.
IR-SEC data for 1 and 3 are similar to those of 4, but with
larger shifts. The magnitudes of the shifts in stretching
frequency, ∆ν(CO), are highly dependent on the size of the
ligand reduced. As shown in Figure 6, where average ∆ν(CO)
values for the neutral versus 2- state of each rectangle are
plotted versus the area of LL, the smallest ligands engender
the largest shifts. The effect can be understood, in part, in terms
of dilution of the added charge on the metal-bound nitrogen
atoms by delocalization over the entire ligand. Additionally, for
5 and 6 where shifts are small (<4 cm-1), the reductions are
likely largely confined to the central tetracarboxydimide dian-
hydride fragments, making the charge changes at the pyridyl
nitrogens small.
Figure 7. NIR absorption scans for a 5.3 × 10-4 M solution of 1 in THF
with 0.1 M TBAPF6. The neutral state is shown as a solid black line, and
the 1- state, as a dashed green line.
Spectroelectrochemistry: Near-Infrared Regions. SEC
measurements were also made for 1-6 in the near-infrared
(NIR) region of the spectrum. Figures 7 and 8 contain spectra
obtained for 1 and 3 in THF and CH2Cl2, respectively. The
spectra reveal broad absorption bands centered at 5115 cm-1
for 11-, and 4255 cm-1 for 31-. The bands disappear upon
further reduction to the 2- state or upon restoration of the
neutral form. (Absorption spectra for the 2- state are not shown
due to difficulties in achieving full conversion to the doubly
9
J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004 12995

A R T I C L E S
Dinolfo et al.
solvent
Table 4. IT Band Absorption Data, Charge-Transfer Distances, and Calculated Electronic Coupling
νmax
ꢀ
∆ν1/
∆ν1/
Rgeom,calcd
(Å)
µ12
(eÅ)
R12
(Å)
Rab
(Å)
Hab
(cm-
2,measd
1
2,calcd
1
1
1
1
rectangle
(cm-
)
(M-1 cm-
)
(cm-
)
(cm-
)
)
R
11-
21-
31-
41-
5115
5350
4255
6595
3000
150
1300
1100
2230
3440
5.0
5.6
4.3
4.9
0.75
0.23
0.52
0.49
3.6
3.2
3.6
3.7
3.9
3.3
3.8
3.9
1000
400
600
800
0.20
0.07
0.14
0.12
THF
THF
CH2Cl2
THF
4500
2060
3385
3520
3140
3900
parallel the extent of ligand van der Waals interactions implied
by the X-ray crystal structures. Kc values for the rectangles are
included in Table 3 for the formation of the 1- MV state.
Compounds 3 and 4 have the highest values for Kc, 1.2 × 106
and 8.2 × 105, respectively, implying a significant amount of
electrostatic communication between reduced ligands. The
crystal structures of these two rectangles (in neutral form) show
the central bridging ligands in close proximity, near van der
Waals contact (Figure 2), allowing for strong interaction between
ligands.
The values for 1, 2, and 5 are somewhat lower: 7.7 × 103,
3.7 × 104, and 1.7 × 104, respectively. The X-ray crystal
structures of 1 and 2 reveal more rigid structures than those of
3 and 4, maintaining larger LL/LL separation distances. In the
case of 1,6 the dihedral displacements of pyridyl rings in relation
to each other enable the assembly to sustain linear N-N (LL)
axes, while the short length of the pyrazine ligand prevents the
inward bowing seen in 3 and 4. In the structure of 5, the spacial
proximity of the ligands presumably facilitates electrostatic
communication, but the horizontal offset of central rings keeps
Kc from reaching the values of 3 and 4. Like 5, the central
naphthalene rings of molecular rectangle 6 are in contact, but
it still is characterized by the smallest Kc value (30) of all the
rectangles examined. One factor may be the large horizontal
offset between the central rings. Another may be the larger LL
area, compared to 2 for example, over which the charge may
disperse, thereby reducing charge density and electrostatic
repulsions.
Figure 8. NIR-SEC absorption scans for an 8.8 × 10-5 M solution of 3 in
CH2Cl2 with 0.1 M TBAPF6. The neutral state is shown as a solid black
line, and the 1- state, as a dashed green line. The spikes in the absorption
spectra result from imperfect subtraction of absorption due to solvent and
electrolyte vibrational overtones.
reduced forms; typically only about 80-90% conversion was
obtained.) These bands clearly are electronic and are assigned
as ligand-based intervalence transitions (IT). Bands were also
found for assemblies 21- and 41- but were not detected for 51-
or 61- (nor were they observed for any of the triply reduced
rectangles). Table 4 contains IT band absorption data for 11-
1-, 31-, and 41-
Mixed Valence Analysis: Electrochemistry. The difference
,
2
.
in formal potential, ∆Ef, for identical redox sites reflects not
only electrostatic effects, as noted above, but also the degree
of electronic communication between the sites. By defining the
diabatic initial state of the 1- assembly as one with the added
electron fully localized on one of the two ligands, LL, and the
final state as one with the electron fully localized on the other,
the initial-state/final-state coupling energy, Hab, contributes the
quantity Hab2/λ to the difference in formal potential.53,54 On this
basis, the electronic coupling energy for 61- clearly must be
moderate or small.
For three of the rectangles, the table also lists values for
formation of a second mixed-valence form, a triply reduced
assembly; Kc′ eq 4.
-1
Kc′
2[-1,-2] y z [-1,-1] + [-2,-2]
(4)
(Suitable electrochemical data were not obtained for the other
three.) In each case, the 3- MV form is less stable than the 1-
form. From Coulomb’s law, the electrostatic contribution to the
comproportionation free energy should scale as the difference
in charge products for the redox sites divided by the distance
separating them. For the singly reduced form, the charge scaling
is nominally |(1-)(0) - 0.5[(0)(0) + (1-)(1-)]| ) 0.5. For
the triply reduced form, the charge scaling is nominally the
same: |(2-)(1-) - 0.5[(1-)(1-) - (2-)(2-)]| ) 0.5. The
separation distance, however, clearly should be greater for the
3- form because of the LL/LL Coulombic repulsion present
there but absent for the 1- form. Additionally, double reduction
of LL is likely to increase the contribution of quinoidal
resonance forms to the geometric and electronic structure of
the ligand. Quinoidal structures should be more rigid than
nonquinoidal ones. This effect should rigidify the rectangular
structure, increase the interligand separation distance and thereby
decrease the electrostatic contribution to Kc. Finally, increases
in separation distance will decrease Hab and further diminish
Regardless of its origin, Kc is a direct measure of the
thermodynamic stability of the mixed valence species with
respect to disproportionation into unreduced and doubly reduced
forms:
-1
Kc
2[0,-1] y z [0,0] + [-1,-1]
(2)
Expressed instead in terms of comproportionation, the relation
is
[0,-1]2
[0,0] × [-1,-1]
Kc )
) e∆EF/RT
(3)
where F is the Faraday constant. Table 3 lists Kc values for the
six singly reduced mixed-valence rectangles. These values vary
significantly with the identity of the bridging ligand and broadly
(53) Sutton, J. E.; Sutton, P. M.; Taube, H. Inorg. Chem. 1979, 18, 1017-
1021.
(54) Sutton, J. E.; Taube, H. Inorg. Chem. 1981, 20, 3126-3134.
9
12996 J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004

Rhenium-Based Molecular Rectangles
A R T I C L E S
Figure 9. PM3(tm)/UHF semiempirical geometry optimized structures of 11- (top) and 61- (bottom). Carbon atoms are shown in gray, nitrogen, in blue,
oxygen, in red, and rhenium, in green; the hydrogen atoms were removed for clarity. In both structures, the reduced ligand is on top. The left most images
are space filled models, viewed along the long axis of the rectangles, with the nearest [Re(CO)3]2BiBzIm (and pyridyl units for 61-) removed for clarity. The
middle and right images are stick models displaying the HOMO (centered on top ligand) and LUMO (centered on bottom ligand) orbitals.
Kc. Notably, none of the 3- forms displayed detectable
intervalence absorption intensity.
ionic assemblies exceed those obtained crystallographically for
the neutral assemblies.
Mixed Valence Analysis: Spectroelectrochemistry. Inter-
valence band energies (νmax) and intensities (ꢀmax) can be used
to estimate Hab. For class II compounds, assuming a two-state
description in the classical limit, Hush has derived the following:
In 1, the reduced form of 4,4′-bpy is predicted to adopt a
planar quinoidal structure as has been observed with the radical
cation of methyl viologen.63-65 The geometry-optimized struc-
ture (Figure 9) clearly shows a decrease in dihedral angle
between pyridine rings of the reduced side of the rectangle (21°,
versus 39° for the neutral side, bottom). The planarity of the
reduced 4,4′-bpy ligand engenders more rigidity in the rectangle
structure, increasing the minimum LL/LL separation distance
from 3.9 to 5.0 Å. Increases in LL/LL separation are also seen
in the calculated monoanionic structures of 3 and 4. In both
X-ray crystallographic structures, the nominally planar bridging
ligands bow inward sufficiently to achieve van der Waals
contact, but in the semiempirical models of the corresponding
1- compounds, the bridging ligands become more rigid, enlarg-
ing the LL/LL separation distance. Unlike 1, 3, and 4, the
changes for 2 are minimal. The pyrazine-pyrazine distance is
approximately 5.3 Å in the X-ray crystal structure and 5.6 Å in
the calculated monoanionic structure.
0.0206
Rab
Hab
)
υ
x
maxꢀmax∆ν1/2
(5)
where Rab is the diabatic electron-transfer distance and ∆ν1/2 is
the observed bandwidth at half-height.55,56 Additionally, the band
energy can be equated with the diabatic reorganization energy,
λ. In the class III limit (valence-delocalized limit; 2Hab > λ),
the relation is simpler:
Hab ) υmax/2
(6)
Application of eq 5 requires estimates for Rab. Uses of
geometric LL/LL separation distances, Rgeo, determined crys-
tallographically for the neutral compounds, are most likely
incorrect. Crystal packing forces may alter the LL/LL separation,
and the addition of electrons to the rectangles will most likely
engender other structural changes. At the same time, direct
electronic Stark effect measurements of the diabatic charge-
transfer distance often show that Rgeo oVerestimates Rab.57-62
In light of these facts, we employed PM3(tm)/UHF semiem-
pirical modeling to gain insight into the structures of the anionic
forms of these rectangles. Figure 9 shows the PM3(tm)/UHF
geometry-optimized structures for the radical anions of 1 and
6. Except for 2, the LL/LL separation distances of the monoan-
The calculated anionic structures of 5 and 6 help elucidate
the lack of observable intervalence absorption in NIR-SEC
experiments. In both mixed-valence structures, the central
tetracarboxydimide rings shift laterally in relation to each other,
in addition to spreading apart. These structural changes result
in less than 25% and 10% cofacial area overlap for 5 and 6,
respectively, likely accounting for the lack of optically signifi-
cant electronic coupling. Additionally, the calculated HOMO
and LUMO orbitals for 5 and 6 (Figure 9) reveal nodes at the
diimide nitrogen positions of the central rings, effectively
eliminating electronic communication between the central rings
and the Re-coordinating pyridyl groups. This is consistent with
the relatively small differences in reduction potentials between
the rectangle-coordinated and free versions of the ligand and
the minute CO frequency shifts seen in IR-SEC measurements.
(55) Hush, N. S. Prog. Inorg. Chem. 1967, 8, 391-444.
(56) Hush, N. S. Coord. Chem. ReV. 1985, 64, 135-157.
(57) Vance, F. W.; Williams, R. D.; Hupp, J. T. Int. ReV. Phys. Chem. 1998,
17, 307-329.
(58) Shin, Y. G. K.; Brunschwig, B. S.; Creutz, C.; Sutin, N. J. Am. Chem.
Soc. 1995, 117, 8668-8669.
(59) Shin, Y. K.; Brunschwig, B. S.; Creutz, C.; Sutin, N. J. Phys. Chem. 1996,
100, 8157-8169.
(60) Oh, D. H.; Boxer, S. G. J. Am. Chem. Soc. 1990, 112, 8161-8162.
(61) Oh, D. H.; Sano, M.; Boxer, S. G. J. Am. Chem. Soc. 1991, 113, 6880-
6890.
(63) Hester, R. E.; Suzuki, S. J. Phys. Chem. 1982, 86, 4626-4630.
(64) Poizat, O.; Sourisseau, C.; Corset, J. J. Mol. Struct. 1986, 143, 203-206.
(65) Wolkers, H.; Stegmann, R.; Frenking, G.; Dehnicke, K.; Fenske, D.; Baum,
G. Z. Naturforsch., B: Chem. Sci. 1993, 48, 1341-1347.
(62) Reimers, J. R.; Hush, N. S. J. Phys. Chem. 1991, 95, 9773-9781.
9
J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004 12997

A R T I C L E S
Dinolfo et al.
Table 4 lists the minimum LL/LL separation distances,
geom,calc, determined semiempirically for the singly reduced
localized assemblies. We tentatively assign 11-, 21-, 31-, and
41- as class II species and 51- and 61- as borderline class I
species.72,73 Also assigned as borderline class I, based on the
absence of observable IT bands, are the triply reduced versions
on the rectangles.
Hush also has derived an expression, applicable to class II
species, that relates intervalence bandwidth to band energy in
the classical limit:55
R
forms of the rectangles. These distances are not necessarily
equivalent to the required diabatic charge-transfer distances, Rab.
To estimate Rab we first made used of the PM3(tm)/UHF
semiempirical calculations to determine the ground-state dipole
moment, µg, for each rectangle. We, along with Nelsen and
Newton, have shown elsewhere under certain conditions,66 the
adiabatic charge-transfer distance, R12, can be directly equated
with 2µg (in other words, the change in dipole moment for the
transfer of an electron).67-69 The PM3(tm)/UHF semiempirically
calculated ground-state dipole moments for 11-, 21-, 31-, and
∆ν1/2
)
16 ln(2)k Tν
(10)
x
B
max
Absorption bands for 11- and 31-, in particular, are narrower
than expected (ca. 3400 and 3100 cm-1 calculated), while the
IT band for 21- is wider than expected (ca. 3600 cm-1
calculated). Additionally, the line shapes for 11-, 31-, and 41-
deviate from the Gaussian ideal, being narrower on the low-
energy side than on the high-energy side. The asymmetry could
signal the coupling of significant vibrational reorganization to
the intervalence excitation. Alternatively, the asymmetry might
be a manifestation of the “2Hab cutoff” effect described by
Lambert and No¨ll for classical systems, although the estimated
electronic coupling energies appear to be too small for the cutoff
effect to be important.29,74,75 Experimentally, narrower than
expected line shapes are often encountered when class II systems
approach the class III limit, conditions where the Born-
Oppenheimer condition may fail and intervalence excitation may
entail significant vibronic coupling. Notably, vibronic coupling
treatments typically predict narrower intervalence line shapes
than does Hush theory.76 In the pyrazine-based mixed-valence
assembly, 21-, the wider than expected bandwidth might reflect
problems in background spectral subtraction for this weakly
absorbing assembly, rather than a real discrepancy. If so, then
the estimate for Hab should be revised downward by ∼25%.
To gain additional insight into the significance of the
estimated electronic coupling energies, specifically in the context
of ground-state intramolecular electron transfer, adiabatic and
diabatic potential-energy surfaces were compared. Figure 10
shows the results for rectangles 11- and 31-. The surfaces were
generated using Hush theory29 using data contained in Table 4.
(Specifically, eqs 1a, 1b, 3a, and 3b of ref. 29 were employed.)
While a double-welled ground potential-energy surface is
retained in both cases, considerable barrier rounding exists for
electron exchange within 11-. For 21- the perturbation is much
smaller but still large enough to ensure adiabatic electron
exchange.
4
1- are 1.79, 1.62, 1.82, and 1.87 eÅ respectively. These were
then converted to diabatic distances (Rab) using experimental
intervalence energy and oscillator strength parameters, by the
generalized Muliken-Hush analysis as prescribed by Cave and
Newton.70,71 Briefly, the change in diabatic charge-transfer
distance can be determined from eq 7
Rab
)
(R )2 + 4(µ )2
(7)
x
12
12
where µ12 is the transition dipole moment in eÅ, as estimated
via eq 8 from the intervalence absorption parameters.
ꢀmax ∆ν1/2
µ12
)
(8)
νmax
x
Table 4 summarizes the distances obtained in this way.
Notably, both the diabatic and adiabatic charge-transfer distances
are shorter than the geometric donor-acceptor separation
distances, Rgeom,calcd. Table 4 also lists the electronic coupling
energies obtained from eq 5 using the calculated adiabatic
charge-transfer distances and measured intervalence absorption
spectra. Also listed are estimates of R, Hush’s delocalization
parameter, the amount of charge already transferred from donor
site to acceptor site in the ground electronic state. These were
obtained (again using the simple Hush theory) from
R ) Hab/υmax
(9)
All of the singly reduced rectangles appear to be valence-
(66) We assume, as is usually done, that R12 for thermal electron-transfer equals
R12 for optical electron transfer (a condition likely satisfied for two-state
systems). Under these circumstances, µg is equal to -µe (the dipole moment
for the intervalence excited state) provided that (1) a mirror plane exists,
(2) the plane lies between the donor site and the acceptor site, and (3) the
coordinate origin for dipole determination is located in the plane. The vector
difference between µg and µe (µg and -µg) then equals R12 after conversion
from units of debye to units of eÅ. The mixed-valence rectangles are very
well approximated by structures featuring mirror planes.
The coupling-energy estimates for the six mixed-valent
rectangles are broadly consistent with expectations from crystal-
lographic studies and semiempirical radical anion geometry
optimizations. The most strongly coupled systems, 1, 3, and 4,
are characterized in the neutral state by van der Waals (or near
(67) Electronic Stark effect methods can provide direct experimental measures
of charge-transfer induced changes in dipole moment and, therefore, R12
.
The measurements require rigid matices so that field-induced reorientation
of chromophores is prevented. Attempts to apply the technique to singly
reduced rectangles were largely unsuccessful. In most cases the compounds
reversibly disproportionated into neutral and doubly reduced compounds
when low-temperature glass formation was attempted. A possible cause
may be diminished solubility and precipitation of one of the forms at low
temperature. In two of six attempts with 1, disproportionation at 77 K°
was avoided and intervalence Stark spectra were recorded. The experiments
and subsequent analysis yielded R12 ) 1.5 eÅ and Rab ) 2.1 eÅ. The values
are somewhat smaller than those obtained from semiempirical calculations
and used in eq 5, suggesting that the assembly, while still a class II species,
is more strongly coupled than that indicated in the table. Using the Stark
data, admittedly from experiments that have been difficult to reproduce,
alternative estimates for Hab and R are 1830 cm-1 and 0.36.
(72) We interpret the absence of detectable intervalence absorption by 51- and
61- as evidence that Hab is considerably less than 400 cm-1 and probably
less than 100 cm-1. 100 cm-1 would be more than sufficient, however, to
ensure adiabatic thermal electron exchange kinetics. Indeed, dependent on
the nature of the exchange reaction dynamics, as little as 60 cm-1 would
be sufficient. 51- and 61- may well be class I like only in an optical sense.
(73) With these estimates, the issue of electronic-coupling contributions to
electrochemical potential differences, ∆Ef, can be revisited. Estimating these
contributions as Hab2/νmax, the values are small, ranging from ∼3 mV for
2 to ∼24 mV for 1.
(68) Johnson, R. C.; Hupp, J. T. J. Am. Chem. Soc. 2001, 123, 2053-2057.
(69) Nelsen, S. F.; Newton, M. D. J. Phys. Chem. A 2000, 104, 10023-10031.
(70) Cave, R. J.; Newton, M. D. Chem. Phys. Lett. 1996, 249, 15-19.
(71) Cave, R. J.; Newton, M. D. J. Chem. Phys. 1997, 106, 9213-9226.
(74) Lambert, C.; Noll, G. J. Am. Chem. Soc. 1999, 121, 8434-8442.
(75) Nelsen, S. F. Chem.sEur. J. 2000, 6, 581-588.
(76) Coropceanu, V.; Malagoli, M.; Andre, J. M.; Bredas, J. L. J. Am. Chem.
Soc. 2002, 124, 10519-10530.
9
12998 J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004

Rhenium-Based Molecular Rectangles
A R T I C L E S
associated double-chelating ligands largely limited to a structural
role. For these six systems, variations in coupling strength lead
to mixed-valence behavior ranging from nearly Class I (non-
communicating, fully valence-localized redox centers) to strongly
Class II on the Robin and Day scale, or roughly a 100-fold
2
variation in Hab despite nominally identical coordination-
defined geometries and the absence of any bridging unit. The
observed differences largely reflect detailed geometric configu-
rational differences that can either facilitate of frustrate produc-
tive direct orbital overlap. Because coupling is attained primarily
by direct donor-orbital/acceptor-orbital overlap (rather than
superexchange) and because mixed valency is based on ligands
rather than transition metal ions, the rectangles comprise systems
that can be logically described by Hush’s idealized two-state
model.
Figure 10. Calculated potential energy surfaces for 11- (left) and 21- (right)
based on the two-state Hush model. Adiabatic surfaces are shown as solid
lines, and diabatic surfaces, as dashed lines. The values for λ ) νmax and
Hab, for 11- and 21-, were taken from Table 4.
Experimental Section
van der Waals) contact between LL pairs, at least near the
centers of the ligands. While the LL/LL separation distances
(calculated) are somewhat greater for the mixed-valence rect-
angles, the ligands remain relatively planar and in good spatial
overlap to facilitate direct electronic interaction. The weakly
coupled systems, 5 and 6, also achieve van der Waals contact
in the neutral state but are offset in a fashion that reduces LL/
LL π*-orbital overlap. Rectangle 2, which is weakly but
detectably coupled, is constrained by ligand size and rigidity
from achieving LL/LL contact and good donor/acceptor orbital
overlap.
Finally, the comparatively small value of Hab for the pyrazine
based rectangle, 2, in comparison to its value for 1, 2, and 5,
strongly suggests that coupling for the latter three is obtained
primarily by direct LL/LL orbital overlap rather than through
rhenium-mediated superexchange. The Hab value for 3 can be
taken as an upper limit estimate of coupling obtainable by
superexchange. The rigid pyrazine ligand, comprising a single
ring doubly ligated to rhenium ions, represents an optimal
moiety for participation in superexchange, to the extent that it
can occur in a molecular rectangle configuration.
General Methods. Infrared spectra were recorded on a Bio-Rad
FTS-40 FTIR spectrometer. All NMR spectra were recorded on either
a Varian Mercury-400 MHz or Varian INOVA-500 MHz spectrometer,
and the chemical shifts were referenced to the standard solvent shift.
FAB LRMS were obtained in the Mass Spectrometry Laboratory,
School of Chemical Sciences, University of Illinois. Elemental analyses
were performed by Atlantic Microlabs, Inc. Norcross, GA. UV-vis-
NIR absorption spectra were measured using a Varian Cary 5000
spectrometer. MALDI LR-MS were performed using a Perseptive
Biosystems, Voyager DE-Pro MALDI-TOF. EI LRMS and HRMS were
performed using a VG70-250SE high-resolution mass spectrometer.
Materials. All commercial reagents were of ACS grade and used
without further purification. Tetrahydrofuran (THF), methylene chloride,
and dimethyl sulfoxide (DMSO) were purified using a two-column
solid-state purification system (Glasscontour System, Joerg Meyer,
Irvine, CA). 4-Aminopyridine, 1,4,5,8-naphthalenetetracarboxylic, 1,2,4,5-
benzenetetracarboxylic, pyrazine, and 4-cyanopyridine were purchased
from Aldrich. Hydrazine monohydrate was purchased from Fisher
Scientific. [Re(CO)4]2BiBzIm (2),6 2,7-diazapyrene,77,78 and 16 were
synthesized by literature procedures.
Di-3,6-(4-pyridyl)-1,2,4,5-tetrazine (8). 4-Cyanopyridine (5.5 g,
52.8 mmol), hydrazine monohydrate (25 mL), concentrated HCl (5 mL),
and deionized water (5 mL) were refluxed together for 2 h. After the
mixture cooled, 4.9 g of the orange, dihydro intermediate was isolated
by filtration. The precipitate was immediately added to 200 mL of
glacial acetic acid with stirring. A 35 mL aliquot of 30% HNO3 was
added dropwise to the brown solution. The reaction mixture turned
pink, and the oxidized product was isolated as the HNO3 salt by
filtration. Recrystallization from pyridine gave 3.44 g of the purple
product (52% overall yield). 1H NMR (400 MHz, CDCl3) 8.96 (d, J )
5 Hz, 4H), 8.52 (d, J ) 5 Hz, 4H). 13C NMR (400 MHz, CDCl3) 163.5,
151.1, 138.5, 121.3. EI-LRMS (m/z): [M]+ calcd for C12H8N6: 236.2,
found 236.2. EI-HRMS (m/z): [M]+ calcd for C12H8N6: 236.0810,
found 236.0809. Anal. Calcd for C12H8N6: C, 61.01; H, 3.41; N, 35.58.
Found: C, 60.64; H, 3.36; N, 35.60.
Conclusions
The molecular rectangles described here are examples of a
new type of ligand-centered mixed-valence compound that
employs a metal-organic framework to arrange the redox active
ligands in a manner that controls, at least in part, spatial overlap
and through-space interactions and, thus, electronic communica-
tion. X-ray crystallographic studies of the rectangles, in neutral
form, reveal a range of LL/LL geometrical configurations with
minimum interligand distances varying from van der Waals
contact (ca. 3.3 Å) to 5.3 Å. Cyclic voltammetry experiments
reveal chemically reversible single electron reductions that allow
for the addition of two to four electrons per rectangle. UV-
visible and IR-SEC measurements establish that the reductions
are LL based.
N,N′-Di-(4-pyridyl)-1,2,4,5-benzenetetracarboxydiimide (9). 1,2,4,5-
Benzenetetracarboxylic dianhydride (1.0 g, 4.6 mmol) and 4-aminopy-
ridine (1.08 g, 11.5 mmol) were added to a reaction flask. Anhydrous
dimethylformamide (DMF, 50 mL) (Aldrich) was added under N2, and
the solution was refluxed overnight under N2. After the reaction mixture
cooled in an ice bath, the white slurry was filtered and the product
was washed with CH2Cl2 and acetone. The product was then dried under
The singly and triply reduced assemblies are mixed-valence
species. For the triply reduced species, and two of the six singly
reduced, electronic coupling is weak as evidenced by the absence
of detectable intervalence absorption bands. Four of the singly
reduced LCMV species, however, do display weak (ꢀ ) 150
1
vacuum to yield 716 mg of a white powder (42% yield). H NMR
M-1 cm-1) to moderately intense (ꢀ ) 3000 M-1 cm-1
)
intervalence absorption bands in the extended near-infrared
region. We find that electronic coupling is chiefly governed by
through-space LL/LL interactions, with the Re atoms and
(77) Stang, P. J.; Cao, D. H.; Saito, S.; Arif, A. M. J. Am. Chem. Soc. 1995,
117, 6273-6283.
(78) Huenig, S.; Gross, J.; Lier, E. F.; Quast, H. Justus Liebigs Ann. Chem.
1973, 339-358.
9
J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004 12999

A R T I C L E S
Dinolfo et al.
(400 MHz, CF3COOD) δ 8.90 (d, J ) 5 Hz, 4H), 8.80 (d, J ) 5 Hz,
4H), 8.73 (s, 2H). 13C NMR (400 MHz, CF3COOD) δ 165.7, 151.6,
144.7, 139.4, 123.7, 122.8. MS-MALDI-TOF (m/z): [MH]+ calcd for
C20H10N4O4: 371.3; found 371.1. Anal. Calcd for C20H10N4O4: C,
64.87; H, 2.72; N, 15.13. Found: C, 64.63; H, 2.70; N, 14.94.
N,N′-Di-(4-pyridyl)-1,4,5,8-naphthalenetetracarboxydiimide (10).
Using the same procedure as that for 9, combining 1,4,5,8-naphthale-
netetracarboxylic dianhydride (1.0 g, 3.7 mmol) and 4-aminopyridine
(770 mg, 8.2 mmol) produced 1.29 g of 10 (82% yield). 1H NMR (500
MHz, CF3COOD) δ 9.12 (d, J ) 5 Hz, 4H), 9.03 (s, 4H), 8.37 (d, J )
5 Hz, 4H). 13C NMR (500 MHz, CF3COOD) δ 164.9, 154.8, 145.2,
135.0, 131.4, 129.7, 128.8. MS-MALDI-TOF (m/z): [MH]+ calcd for
C24H12N4O4: 421.1; found 421.4. Anal. Calcd for (C24H12N4O4)-
(H2O): C, 65.75; H, 3.22; N, 12.78. Found: C, 66.35; H, 3.03; N,
12.90.
General Rectangle Synthesis. Compounds 2-6 were synthesized
according to the method described by Benkstein, et al..6 Briefly,
stoichometric amounts of 2 and the bridging ligand, LL, were combined
with anhydrous THF (approximately 1-2 mL per mg of 2) under N2,
and the reaction mixture was refluxed for 2 days under N2. Removal
of ca. 75% of the THF and addition of either diethyl ether or hexanes
followed by filtration yielded the pure rectangle structures.
rectangles. Crystals of 1 were grown by diffusion of hexane into
saturated THF. Crystals of 2 were grown by diffusion of cyclohexane
into saturated 1,2-dichloroethane and THF. Crystals of 3 were grown
by diffusion of hexanes into saturated CH2Cl2. Crystals of 4 were grown
by diffusion of cyclohexane into a saturated 2-methyl-THF solution.
Crystals of 5 were grown from the slow diffusion of cyclohexane into
saturated THF. Crystals of 6 were grown by diffusion of methanol into
saturated dimethyl sulfoxide. Crystals were mounted on glass fibers
using oil and immediately placed in a cold nitrogen stream (153 K).
Data collection was done using a Bruker Smart 1000 CCD diffracto-
meter with Mo KR radiation (λ ) 0.710 69 Å) and processed using
the SMART and SAINT programs.79 An analytical integration absorp-
tion correction was applied to all data. The structures were solved by
direct methods. All structure solutions and refinement were done using
the teXsan crystallographic software package. For all structures, non-
hydrogen atoms were refined anisotropically and hydrogens were added
in idealized positions. The refinements of all structures were straight-
forward. The crystallographic details are shown in Table 1. Additional
information is provided in the Supporting Information.
Electrochemistry. All cyclic voltammetric experiments were per-
formed and analyzed using a CHI900 (CH Instruments, Austin, TX)
potentiostat. Electrolyte solutions (0.1 M tetrabutylammonium hexaflu-
orophosphate (TBAPF6) (>99%, Fluka)) were prepared with anhydrous
solvents with and nitrogen degassed prior to use. A Pt wire was used
as the counter electrode and a 2 mm diameter Pt or Au macro disk
electrode was used as the working electrode. A silver wire was used
as a pseudo-reference electrode, with ferrocene (Aldrich, purified by
sublimation) added as an internal reference at the end of each
experiment. All experiments were run under an nitrogen atmosphere.
Spectroelectrochemistry. Ultraviolet-visible-near-infrared spec-
troelectrochemical (UV-vis-NIR SEC) experiments were performed
using a home-built optically transparent electrode. A 1 mm quartz
cuvette was attached to a 14/20 ground joint with a stopcock and sample
holding bulb. A ground joint containing three tungsten leads was used
for the electrode connections. A Teflon coated wire was connected to
a 5 mm × 15 mm Pt mesh working electrode, and a Pt wire was used
as the counter electrode. The Ag wire pseudo-reference electrode was
coated with Teflon shrink tubing so that only the bottom (15 mm)
portion was exposed. The electrodes were arranged in the cell such
that the Pt mesh was in the optical path of the quartz cuvette and the
exposed portion of the Ag wire parallel to it. The Pt wire counter
electrode was placed above the quartz cuvette area (ca. 2 cm away
from the working electrode). Dry molecular rectangle and TBAPF6
powders were added to the cell and thoroughly deoxygenated.
Anhydrous solvent was then added, and the sample-electrolyte solution
was freeze-pump-thaw degassed 4 times. NIR and UV-vis absorption
scans were performed using a Varian CARY 5000 spectrometer while
the cell potential was maintained with a Princeton Applied Research
model 273 potentiostat.
([Re(CO)3]2BiBzIm)2-µ,µ′-(pyrazine)2 (2). Compound 7 (114 mg,
0.14 mmol) and pyrazine (11 mg, 0.14 mmol) yielded 51 mg of 3 (43%
yield) as a red powder. IR (THF, cm-1) 2027 (CtO), 1928 (CtO),
1
1915 (CtO). H NMR (400 MHz, CD2Cl2) δ 8.12 (s, 8H), 7.73 (q, J
) 3 Hz, 4H), 7.36 (q, J ) 3 Hz, 4H). FAB LRMS, calcd m/z 1705.6,
found m/z 1705.9. Anal. Calcd for Re4C48H24N12O12: C, 33.80; H, 1.42;
N, 9.85. Found: C, 34.73; H, 1.73; N, 9.44.
([Re(CO)3]2BiBzIm)2-µ,µ′-(2,7-diazapyrene)2 (3). Compound 7
(148 mg, 0.18 mmol) and 2,7-diazapyrene (36.5 mg, 0.18 mmol) yielded
138 mg of 2 as a yellow powder (79% yield). IR (CH2Cl2, cm-1) 1907
(CtO), 1922 (CtO), 2024 (CtO), 2027 (CtO). 1H NMR (500 MHz,
CD2Cl2) δ 8.57 (d, J ) 5 Hz, 8H), 8.02 (q, J ) 3 Hz, 8H), 7.56 (q, J
) 3 Hz, 8H), 7.19 (d, J ) 5 Hz, 8H). FAB LRMS, calcd m/z 1953.9,
found m/z 1954.0. Anal. Calcd for Re4C68H32N12O12: C, 41.80; H, 1.65;
N, 8.60. Found: C, 41.67; H, 1.71; N, 8.34.
([Re(CO)3]2BiBzIm)2-µ,µ′-(di-3,6-(4-pyridyl)-1,2,4,5-tetrazine)2 (4).
Compounds 7 (120 mg, 0.15 mmol) and 8 (34.2 mg, 0.15 mmol) yielded
141 mg of 5 as an orange powder (96% yield). IR (THF, cm-1) 2027
(CtO), 2021 (CtO), 1921 (CtO), 1906 (CtO). 1H NMR (400 MHz,
CD2Cl2) δ 7.87 (d, J ) 5 Hz, 8H), 7.60 (d, J ) 5 Hz, 8H), 7.45 (q, J
) 3 Hz, 8H), 7.15 (q, J ) 3 Hz, 8H). FAB LRMS, calcd m/z 2017.9,
found m/z 2018.7. Anal. Calcd for Re4C64H32N20O12: C, 38.09; H, 1.60;
N, 13.88. Found: C, 60.64; H, 3.36; N, 35.60.
([Re(CO)3]2BiBzIm)2-µ,µ′-(N,N′-di-(4-pyridyl)-1,2,4,5-benzenetet-
racarboxydiimide)2 (5). Compounds 7 (80 mg, 0.09 mmol) and 9 (35.8
mg, 0.09 mmol) yielded 94 mg of 6 as an orange powder (85% yield).
IR (THF, cm-1) 2027 (CtO), 2021 (CtO), 1919 (CtO), 1905 (Ct
O). 1H NMR (400 MHz, DMSO-d6) δ 8.16 (d, J ) 5 Hz, 8H), 8.01 (s,
4H), 7.82 (q, J ) 3 Hz, 8H), 7.53 (q, J ) 3 Hz, 8H), 7.32 (d, J ) 5
Hz, 8H). FAB LRMS, calcd m/z 2286.1, found m/z 2287.0. Anal. Calcd
for (Re4C80H36N16O20)(H2O): C, 41.70; H, 1.66; N, 9.73. Found: C,
41.97; H, 1.86; N, 9.32.
IR SEC experiments were performed using a reflectance spectro-
electrochemical cell based on a previously reported design.80 Spectra
were recorded on a Bio-Rad FTS-40 FTIR spectrometer and the
electrochemistry was controlled using a Princeton Applied Research
model 273 potentiostat. Samples were prepared using anhydrous
solvents with TBAPF6 (>99%, Fluka) as the supporting electrolyte and
freeze-pump-thawed three times prior to use. The solutions were
transferred to the SEC cell under N2 via airtight syringes. Blank spectra
were recorded using the electrolyte solutions.
([Re(CO)3]2BiBzIm)2-µ,µ′-(N,N′-dipyridyl-1,4,5,8-naphthalenetet-
racarboxydiimide)2 (6). Compounds 7 (82.8 mg, 0.1 mmol) and 10
(42 mg, 0.1 mmol) yielded 91 mg of 7 as an orange powder (76%
yield). IR (DMSO, cm-1) 2024 (CtO), 2018 (CtO), 1911 (CtO),
Molecular Modeling and Electronic Structure Calculations. PM3-
(tm)/UHF semiempirical calculations were performed using PC Spartan
Pro for Windows, Wave function Inc..81 Coordinates from the X-ray
1
1903 (CtO). H NMR (400 MHz, DMSO-d6) δ 8.28 (d, J ) 5 Hz,
8H), 8.04 (s, 8H), 7.84 (q, J ) 3 Hz, 8H), 7.52 (q, J ) 3 Hz, 8H), 7.31
(d, J ) 5 Hz, 8H). FAB LRMS, calcd m/z 2386.2, found m/z 2386.7.
Anal. Calcd for (Re4C88H40N16O20)(H2O): C, 43.96; H, 1.76; N, 9.32.
Found: C, 44.21; H, 1.97; N, 9.08.
Crystal Growth and Structure Determinations. X-ray quality
crystals of 1-6 were grown over 5 to 10 day periods by slow diffusion
of precipitating solvents into saturated solutions of the molecular
(79) SMART Version 5.054 Data Collection and SAINT-Plus Version 6.02A Data
Processing Software for the SMART System; Bruker Analytical X-ray
Instruments: Madison, WI, 2000.
(80) Zavarine, I. S.; Kubiak, C. P. J. Electroanal. Chem. 2001, 495, 106-109.
(81) PC Spartan Pro, Version 1.0.5 for Windows; Wave function Inc.: Irvine,
CA 62612, 2000.
9
13000 J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004

Rhenium-Based Molecular Rectangles
A R T I C L E S
crystal structures were used as starting points for the geometry
optimizations for the neutral and radical anion molecules. The center
of mass was used as the coordinate of origin for the ground-state dipole
moment calculations.
acknowledge the Office of Science, U.S. Department of Energy
(Grant No. DE-FG02-87ER13808) for financial support of our
research.
Supporting Information Available: UV-visible absorption
spectra for 3-6 in the 0, 1-, and 2- states and quantitative
UV-visible absorption data for 1-6 in the 0, 1-, and 2- states.
X-ray crystallographic files for 1-6 (CIF). This material is
available free of charge via the Internet at http://pubs.acs.org.
Acknowledgment. We acknowledge Kurt Benkstein for
providing the crystal structure of 1.39 We thank Prof. Clifford
P. Kubiak, J. Catherine Salsman, and Casey H. Londergan for
sharing their IR-SEC cell design and facilitating its fabrication
at the University of California at San Diego. We thank Dr. Fred
Arnold for guidance in semiempirical modeling. We gratefully
JA0473182
9
J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004 13001