Solid-state changes in ligand-to-metal charge-transfer spectra of (NH 3)5RuIII(2,4-dihydroxybenzoate) and (NH 3)5RuIII(xanthine) chromophores

Article abstract of DOI:10.1021/ic800511g

Five distorted-octahedral complexes containing (NH₃) ₅Ru^IIIL ions, where L = 2,4-dihydroxybenzoate or a xanthine, have been studied using a combination of X-ray crystallography, solution and polarized single-crystal electronic absorption spectroscopy, and first principles electronic structure computational techniques. Both yellow (2) and red (3) forms of the complex (NH₃)₅Ru^IIIL, where L = 2,4-dihydroxybenzoate, as well as three xanthine complexes in which L = hypoxanthine-κN⁷(4), 7-methylhypoxanthine-κN ⁹(5), and 1,3,9-trimethylxanthine-κN⁷(6) were examined. In the solid state, some of these complexes exhibit split low-energy ligand-to-metal charge-transfer (LMCT) bands. Traditional solid-state effects, such as ligand π-π overlap or hydrogen bonding that might lead to splitting of electronic absorption bands, were probed in an attempt to identify the origins of these unusual observations. For comparison, companion studies were carried out for spectroscopically normal reference complexes of the same ligands. Time-dependent density-functional theory (TD-DFT) calculations, employing modified B3LYP-type functionals with increased contributions of exact exchange, attribute the color change in the crystalline complexes 2 and 3 to π(ligand) → Ru[d(π)] LMCT bands which, in the red form (3), arise from ligand donor π-orbitals split by strongly overlapping phenyl moieties in centrosymmetric (NH₃)₅Ru^III(2,4- dihydroxybenzoate) dimers. Complex 5 does not show split visible absorptions, whereas both the polarizations and energies of the split visible absorptions shown by 4 and 6 also suggest assignment as LMCT. No support is found for relating the split absorptions of 4 and 6 to the details of π-π xanthine overlap in the solid state; indeed, complex 4 enjoys considerably less π-stacking overlap than does 5. We feel compelled to attribute the split absorptions in crystalline 4 and 6 to the emergence of a LMCT transition originating in the carbonyl lone pair, potentially deriving intensity from the significant intramolecular N-H...O hydrogen bonding present in both 4 and 6 (but not in 5). The electronic structure calculations suggest an O(n) → Ru[d(σ*)] LMCT transition; however, this novel assignment must be considered tentative.

Full text of DOI:10.1021/ic800511g

Inorg. Chem. 2008, 47, 9813-9827
Solid-State Changes in Ligand-to-Metal Charge-Transfer Spectra of
(NH3)5RuIII(2,4-dihydroxybenzoate) and (NH3)5RuIII(xanthine) Chromophores
Karsten Krogh-Jespersen,*,† Robert T. Stibrany,† Elizabeth John,† John D. Westbrook,†
Thomas J. Emge,† Michael J. Clarke,‡ Joseph A. Potenza,*,† and Harvey J. Schugar*,†
Department of Chemistry and Chemical Biology, Rutgers, The State UniVersity of New Jersey,
New Brunswick, New Jersey 08903, and Department of Chemistry, Boston College, Chestnut Hill,
Massachusetts 02167
Received March 20, 2008
Five distorted-octahedral complexes containing (NH3)5RuIIIL ions, where L ) 2,4-dihydroxybenzoate or a xanthine,
have been studied using a combination of X-ray crystallography, solution and polarized single-crystal electronic
absorption spectroscopy, and first principles electronic structure computational techniques. Both yellow (2) and red
(3) forms of the complex (NH3)5RuIIIL, where L ) 2,4-dihydroxybenzoate, as well as three xanthine complexes in
which L ) hypoxanthine-κN7(4), 7-methylhypoxanthine-κN9(5), and 1,3,9-trimethylxanthine-κN7(6) were examined.
In the solid state, some of these complexes exhibit split low-energy ligand-to-metal charge-transfer (LMCT) bands.
Traditional solid-state effects, such as ligand π-π overlap or hydrogen bonding that might lead to splitting of
electronic absorption bands, were probed in an attempt to identify the origins of these unusual observations. For
comparison, companion studies were carried out for spectroscopically normal reference complexes of the same
ligands. Time-dependent density-functional theory (TD-DFT) calculations, employing modified B3LYP-type functionals
with increased contributions of exact exchange, attribute the color change in the crystalline complexes 2 and 3 to
π(ligand) f Ru[d(π)] LMCT bands which, in the red form (3), arise from ligand donor π-orbitals split by strongly
overlapping phenyl moieties in centrosymmetric (NH3)5RuIII(2,4-dihydroxybenzoate) dimers. Complex 5 does not
show split visible absorptions, whereas both the polarizations and energies of the split visible absorptions shown
by 4 and 6 also suggest assignment as LMCT. No support is found for relating the split absorptions of 4 and 6 to
the details of π-π xanthine overlap in the solid state; indeed, complex 4 enjoys considerably less π-stacking
overlap than does 5. We feel compelled to attribute the split absorptions in crystalline 4 and 6 to the emergence
of a LMCT transition originating in the carbonyl lone pair, potentially deriving intensity from the significant intramolecular
N-H · · · O hydrogen bonding present in both 4 and 6 (but not in 5). The electronic structure calculations suggest
an O(n) f Ru[d(σ*)] LMCT transition; however, this novel assignment must be considered tentative.
Introduction
relevant ligands.1-3 Although Ru(III) does not play a
prominent role in natural metallo-biochemistry, there are
promising ruthenium metallopharmaceuticals in clinical
trials,4 and the LMCT spectra of Ru(III) complexes can serve
as instructive analogues to those of related Cu(II) chro-
We report here novel differences between the solution and
single-crystal ligand-to-metal charge-transfer (LMCT) ab-
sorptions shown by pentammineRu(III) ions complexed to
2,4-dihydroxybenzoate or one of several xanthine ligands.
This study extends our long-time interest in the LMCT
spectra of Cu(II) and Ru(III) complexes of biologically
(1) Stibrany, R. T.; Fikar, R.; Brader, M.; Potenza, M. N.; Potenza, J. A.;
Schugar, H. J. Inorg. Chem. 2002, 41, 5203–5015.
(2) (a) Krogh-Jespersen, K.; Zhang, X.; Westbrook, J. D.; Fikar, R.; Nayak,
K.; Kwik, W. L.; Potenza, J. A.; Schugar, H. J. J. Am. Chem. Soc.
1989, 111, 4082–4091. (b) Krogh-Jespersen, K.; Westbrook, J. D.;
Potenza, J. A.; Schugar, H. J. J. Am. Chem. Soc. 1987, 109, 7025–
7031; and references cited therein.
(3) Kastner, M. E.; Coffey, K. F.; Clarke, M. J.; Edmonds, S. E.; Eriks,
K. J. Am. Chem. Soc. 1981, 103, 5747–5752.
(4) Clarke, M. J. Coord. Chem. ReV. 2003, 236, 207–231.
* To whom correspondence should be addressed. E-mail: krogh@
rutchem.rutgers.edu (K.K.-J.), potenza@rutchem.rutgers.edu (J.A.P.),
schugar@rutchem.rutgers.edu (H.J.S.).
†
Rutgers University.
Boston College.
‡
10.1021/ic800511g CCC: $40.75  2008 American Chemical Society
Inorganic Chemistry, Vol. 47, No. 21, 2008 9813
Published on Web 10/07/2008

Krogh-Jespersen et al.
subject of the structural, solution, and polarized single-crystal
spectroscopic and computational studies reported here.
Experimental Section
Preparation of the Complexes. [(NH3)5RuIII(2,4-dihydroxy-
benzoate)] · SO4 · 0.5(CH3)2NCHO (2) and [(NH3)5RuIII(2,4-
dihydroxybenzoate)] · SO4 · 2H2O (3) were prepared by following,
with minor variations, a published procedure for related com-
plexes.13
[(NH3)5RuIII(2,4-dihydroxybenzoate)]·SO4 ·0.5(CH3)2NCHO
(2). A mixture of 0.44 g (1.5 mmol) [(NH3)5RuCl]C12 in 4 mL of
H2O was stirred with 0.66 g (3.0 mmol) of CF3CO2Ag for 15 min
and filtered to remove the precipitated ionic chloride. The yellow
filtrate was added to a stirred mixture of 0.92 g (6 mmol) of 2,4-
dihydroxybenzoic acid, 0.11 g (2.8 mmol) of NaOH, 4 g of H2O,
2 g of dimethylformamide, and a small amount of freshly prepared
Zn(Hg) amalgam. A red-orange color developed rapidly, and the
mixture was filtered after 5 min. An additional gram of dimethyl-
formamide was added to the filtrate, which then was transferred to
a small beaker. Saturated aqueous (NH4)2SO4 was layered under
the red-orange solution. During overnight refrigeration, a yellow-
orange solid formed at the interface. A pipet was used to remove
the bottom layer; the solid was collected by filtration, twice
suspended in several milliliters of absolute ethanol, filtered, and
dried in air. The unoptimized yield of crude 2 was 0.44 g (62%);
additional product could have been isolated from the red-orange
filtrate.
Yellow, plate-like crystals of 2 were deposited after a filtered
solution of 80 mg of the crude material in 6 g of H2O, 0.4 g of
dimethylformamide, and 0.1 g of acetonitrile was allowed to
evaporate partially in air at room temperature. The crystals exhibited
light-yellow to dark-yellow dichroism when viewed under a
polarizing microscope. X-ray analysis (see below) showed these
crystals to have the composition 2. The combination of microscopic
examination and density measurement (1.786 g/mL calcd, 1.75(1)
g/mL measured by flotation in a mixture of heptane and 1,2-
dibromoethane) suggested that the recrystallized product was a pure
single phase.
Red six-sided plates of 3 deposited after filtered aqueous solutions
of recrystallized 2 were evaporated to dryness in air at room
temperature. Depending upon the thickness of the crystals, a red-
orange or pale-red/almost-colorless dichroism was observed. The
composition of this single phase was determined by X-ray crystal-
lography and density measurement (1.848 g/mL calcd, 1.86(1) g/mL
observed by flotation in a mixture of heptane and 1,2-dibromoet-
hane).
[(NH3)5RuIII(Hypoxanthine-KN7)]Cl3 ·3H2O(4)and[(NH3)5RuIII(7-
Methylhypoxanthine-KN9)]Cl3 (5). [(NH3)5RuIII(Hypoxanthine-
κN7)]Cl3 ·3H2O and [(NH3)5RuIII(7-Methylhypoxanthine-κN9)]Cl3
were prepared by following a published procedure3 and character-
ized for single-crystal spectroscopic studies as noted below.
[(NH3)5RuIII(1,3,9-trimethylxanthine-KN7)]·[ZnCl4]·Cl (6).
[(NH3)5RuIII(1,3,9-trimethylxanthine-κN7)]·[ZnCl4]·Cl was pre-
pared by variation of a published procedure that afforded the
trichloride salt.14
Figure 1. Schematic structures of the cations in 1-6.
mophores.2 Moreover, the excited states of Ru(II) and Cu(I)
complexes involved in metal-to-ligand charge transfer (MLCT)
also show useful similarities, which are being explored for
solar cell applications.5,6 Ru(III) and Cu(II) ions share the
features of being relatively strong oxidizing species, each
with a single d-vacancy that allows relatively low-energy
LMCT absorptions to occur from upper occupied ligand
orbitals. Ruthenium complexes are well suited for spectro-
scopic studies owing to their kinetic non-lability and pho-
tophysical properties, which have been used to probe long-
range electron transfer in biological systems7-10 and for
sensitizing dyes in solar cells.11,12
Some time ago, Stritar and Taube studied the electron-
transfer and spectroscopic properties of a number of pen-
tammineRuIII-carboxylate complexes.13 The [(NH3)5RuIII(4-
hydroxybenzoate)] complex (1) showed a low-energy
absorption band at approximately 400 nm, which was absent
in the spectra of the parent benzoate or aliphatic carboxylate
complexes. This absorption band was intensified and red-
shifted to 500 nm for the deprotonated phenolate analogue,
thus supporting a LMCT assignment. To explore this type
of LMCT in more detail, we have studied related complexes
containing (NH3)5RuIII units (Figure 1). Both yellow (2) and
red (3) forms of crystals containing the [(pentammineRuIII)2,4-
dihydroxybenzoate] unit have been characterized. These
crystals afford identical solution LMCT spectra but show
striking differences in their solid-state LMCT absorptions.
Equally striking solid-state effects on LMCT absorptions are
also shown by [(hypoxanthine-κN7)(NH3)5Ru]·3Cl·H2O (4)3
and [(1,3,9-trimethylxanthine-κN7)(NH3)5Ru] ·[ZnCl4]·Cl (6),
but not by [(7-methylhypoxanthine-κN9)(NH3)5Ru] ·3Cl (5).3
In contrast to complexes 4 and 6, strong intramolecular
hydrogen bonding between the (NH3)5RuIII unit and the
xanthine carbonyl group is absent in complex 5. These
curious phase-dependent features of LMCT spectra are the
(5) Armaroli, N. Chem. Soc. ReV. 2001, 30, 113–124.
(6) Scaltrito, D. V.; Thompson, D. W.; O’Callaghan, J. A.; Meyer, G. J.
Coord. Chem. ReV. 2000, 208, 243–266.
(7) Wishart, J. F.; Zhang, X.; Isied, S. S.; Potenza, J. A.; Schugar, H. J.
Inorg. Chem. 1992, 31, 3179–3181.
(8) Distefano, A. J.; Wishart, J. F.; Isied, S. S. Coord. Chem. ReV. 2005,
249, 507–516; and references cited therein.
A small amount of freshly prepared zinc amalgam, 75 mg of
[Ru(NH3)5Cl]Cl2 (0.26 mmol), 57 mg of CF3CO2Ag (0.26 mmol),
and 50 mg of 1,3,9-trimethylxanthine (0.26 mmol) were added
sequentially to 20 mL of distilled, deionized water, which had
previously been acidified to pH 4 with HCl and degassed under
(9) Gray, H. B.; Winkler, J. R. Q. ReV. Biophys. 2003, 36, 341–372.
(10) Chang, I.-J.; Lee, J. C.; Winkler, J. R.; Gray, H. B. Proc. Natl. Acad.
Sci. U.S.A. 2003, 100, 3838–3840.
(11) Polo, A. S.; Itokazu, M. K.; Murakami I, N. Y. Coord. Chem. ReV.
2004, 248, 1343–1361.
(12) Klein, C.; Nazeeruddin, M. K.; Di Censo, D.; Liska, P.; Gra¨tzel, M.
Inorg. Chem. 2004, 43, 4216–4226.
(13) Stritar, J. A.; Taube, H. Inorg. Chem. 1969, 8, 2281–2292.
(14) Clarke, M. J.; Taube, H. J. Am. Chem. Soc. 1975, 97, 1397–1403.
9814 Inorganic Chemistry, Vol. 47, No. 21, 2008

(NH3)5RuIII(2,4-dihydroxybenzoate) and (NH3)5RuIII(xanthine)
Table 1. Crystallographic Data for Structures 2, 3, and 6
argon. The mixture was stirred under continued argon purge for
30 min. Following filtration, air was bubbled through the green
filtrate for 30 min, during which time the color changed to cherry
red. This solution was concentrated to 10 mL, and the pH was
adjusted to 2.5 with HCl. The solution was transferred to a beaker
and placed in a closed jar containing absolute ethanol. After several
days at room temperature, vapor diffusion of the ethanol resulted
in the deposition of dichroic red plates as the major phase, along
with some elongated purple prisms and small, yellow rhomboids.
These three air-stable phases could be separated manually. The
dichroic red plates are the subject of the structural and spectroscopic
studies described below. Their bulk density, 2.01(1) g/mL, was
determined by flotation in a mixture of carbon tetrachloride and
1,2,3-tribromopropane.
Spectroscopic Measurements. Electronic-spectral measurements
were made using a computer-interfaced spectrophotometer, built
by Aviv Associates, that utilizes a Cary Model 14 monochromator
and cell compartment. Low-temperature (80 K) spectra were
measured using an Air Products optical Dewar. Polarization
measurements of 3, 4, and 5 were made using separate polarizers
in the sample and reference beams. Matched Glan-Thompson prisms
having a usable angular field of approximately 15° were rotated in
unison by a chain drive mechanism. Projections of electronic vectors
for the single-crystal spectra onto the molecular electronic axes of
the ruthenium chromophores were calculated from P(u||h) ) u -
(h·u)h/(h·h), where P is the projection, u is the desired crystal-
lographic vector, and h is the vector normal to the desired plane.15
Electronic spectra were processed and deconvoluted using the
Grams/AI software package.
A plate-like crystal of the “red” complex (3), 0.013 mm thick,
was placed on a quartz flat and masked with black electrical tape.
Microscopic examination was used to ensure the absence of light
leaks around the masked crystal. Orientation of the unit-cell axes
was established by X-ray diffractometry. The bc crystal face, the
only one that was prominently developed, was oriented perpen-
dicular to the light beam of the spectrophotometer, which was
directed along the a* reciprocal cell axis. Maximum (red) and
minimum (pale-yellow) absorbances of the visible bands were
observed with the electric vector of light oriented along the c and
b axes, respectively. These axial vectors also are optical extinction
directions imposed by the symmetry of the monoclinic cell.
Polarization studies along these directions are free from wavelength-
dependent dispersion effects.
Similarly, crystals of 4 (0.026 mm thick plate), 5 (0.011 mm
thick plate), and 6 (0.050 mm thick plate) were examined using a
diffractometer to verify their unit cells and orientations. The
prominent ab face of 4 showed orange/yellow dichroism, whereas
that of 5 showed a yellow/colorless dichroism; 6 exhibited a red/
yellow dichroism. Maximum and minimum absorbances of the
absorption(s) shown by both 4 and 5 were observed with the light
beam of the spectrophotometer directed along the c axis, and with
the electric vector of the light oriented, respectively, parallel to a
and parallel to b. The a and b axes also are optical extinction
directions for orthorhombic cells. For 6, the light was directed along
the c axis, but the maximum absorbance (red color) occurred with
the electric vector parallel to b, and the minimum absorbance
(yellow color) occurred with the electric vector parallel to a. Single-
crystal spectra of 2 were not measured. The space group for crystals
of 2, P1, does not impose any symmetry constraints on dispersion
effects with the result that extinction directions are wavelength
dependent, making the interpretation of polarized spectra difficult.
2
3
6
formula
C17H47N11
O17Ru2S2
943.92
C7H24N5
O10RuS
471.44
C8H25Cl5
N9O2RuZn
623.06
7.4212(3)
8.8688(3)
31.3964(12)
90.
Fw
a, Å
b, Å
c, Å
7.2458(5)
9.6065(6)
12.8376(8)
15.4850(8)
7.6969(4)
14.9386(8)
R, deg
ꢀ, deg
γ, deg
V, Å3
space group
Z
80.3290(10) 90.
88.7540(10) 107.845(1)
85.5150(10) 90.
90.
90.
877.47(10)
P1
1694.82(15)
P21/c
2066.42(13)
P212121
4
1
4
F
F
calc, g/cm3
obs, g/cm3
1.786
1.848
2.003
1.75(1)
1.06
1.86(1)
1.10
2.01(1)
13.53
0.38-1.00
µ, mm-l
trans factor
T, K
0.83-1.00
100(2)
9269/3/448
0.88-1.00
200(2)
5144/6/274
100(2)
data/restraints/
parameters used in
refinement
3194/317/245
largest peak, hole, e/Å3 3.43, -1.18 0.949, -0.482 2.392, -2.938
a
b
2
RF RwF
0.040, 0.104 0.031, 0.072
0.079, 0.185
a RF ) ∑||Fo| -|Fc||/∑|Fo|; selection criterion I > 2σ(I). RwF
)
b
2
2
2
2
2
{[∑[w(|Fo - Fc |)/∑[w(Fo )]}1/2; selection criterion all Fo .
X-ray Diffraction Studies. Diffraction measurements for 2, 3,
and 6 were made with a Bruker SMART CCD area detector system
diffractometer using φ and ω scans. For 2 and 3, monochromated
Mo KR radiation, λ ) 0.71073 Å, was used; for 6, data were
collected using monochromated Cu KR radiation, λ ) 1.54178 Å.
In all cases, data were collected over all of reciprocal space, and
to a resolution of 0.70 Å for 2 and 3, and of 0.84 Å for 6. Cell
parameters were determined using the SMART software.16a The
SAINT package16a was used for integration of data, Lorentz,
polarization, and decay corrections, and for merging data. Absorp-
tion corrections were applied using SADABS.16b
Structures were solved and refined on F2 using the SHELX
system and all data.16c,d Partial structures were obtained by direct
methods; the remaining non-hydrogen atoms in each structure were
located using difference Fourier techniques. H atoms were located
on difference Fourier maps or placed at calculated positions. For
H atoms whose thermal parameters were not refined, isotropic
temperature factors were set equal to 1.2-1.5 UN, where N is the
atom bonded to H. Views of the structures were prepared using
ORTEP32 for Windows.17a,b Crystal data and additional details of
the data collection and refinement for the three crystals studied here
are given in Table 1, and as Supporting Information.
Each of the three structures proved to be non-routine in some
way. Attempts to solve the structure of triclinic 2 in the space group
j
P1 with a disordered DMF solvate molecule were unproductive,
and the structure was solved and refined smoothly in space group
P1 with two cations, two anions, and the DMF solvate molecule
as the asymmetric unit/unit cell. Inspection of the correlation matrix
revealed a number of relatively large correlation coefficients arising
from the near centrosymmetric relationship of the two unique
(16) (a) SHELXTL (Version 6.10), Saint-Plus (Version 6.02), and SMART
(WNT2000 Version 5.622); Bruker AXS Inc.: Madison, WI, 2000. (b)
Blessing, R. H. Acta Crystallogr. 1995, A51, 33–38. (c) Sheldrick,
G. M. SHELXS-97. Acta Crystallogr. 1990, A46, 467–473. (d)
Sheldrick, G. M. SHELXL-97. A Computer Program for the Refinement
of Crystal Structures; University of Go¨ttingen: Go¨ttingen, Germany.
(17) (a) Farrugia, L. J. J. Appl. Crystallogr. 1997, 30, 565. (b) Burnett,
M. N.; Johnson, C. K. ORTEPIII, Report ORNL-6895; Oak Ridge
National Laboratory, Tennessee, U.S.A., 1996. (c) Crystal Structure
Refinement, IUCr Texts on Crystallography 8; Mu¨ller, P., Ed.; Oxford
University Press: Oxford, England, 2006; p 126.
(15) Sands, D. E. In Vectors and Tensors in Crystallography; Dover
Publications, Inc.: Mineola, NY, 1995; p 45.
Inorganic Chemistry, Vol. 47, No. 21, 2008 9815

Krogh-Jespersen et al.
cations. In the structure of 3, the sulfate anion exhibited a positional
disorder about one of its triad axes; as a consequence, three of the
oxygen atoms were split and were placed on partially occupied
sites. Crystals of 3, obtained from different batches over a period
of years, all exhibited this disorder. The salt 6 crystallized in the
chiral space group P212121 and could only be refined successfully
as a racemic twin. Such refinement necessitated the introduction
of a large number of distance and thermal constraints to facilitate
convergence, a procedure that is not uncommon for crystals with
this type of twinning.17c
Computational Details. All the computational data presented
here are based on calculations using density-functional theory
(DFT)18 methods as implemented in the GAUSSIAN03 series of
computer programs.19 A relativistic, 28e effective core potential
(ECP) and corresponding valence basis set (8s7p6d/[6s5p3d]) were
used for the Ru atom (SDD model);20 the second-row elements C,
N, and O carried all-electron, full double-ꢁ plus polarization function
basis sets (D95(d));21 and hydrogen atoms were assigned a double-ꢁ
21G basis set.22 Calculations of electronic ground-state properties
made use of the standard three-parameter hybrid exchange func-
tional of Becke23 (B3) and the correlation functional of Lee, Yang,
and Parr (LYP).24 Geometries of monomeric (NH3)5RuL+ (L )
deprotonated p-hydroxybenzoate; deprotonated 2,4-dihydroxyben-
zoate), (NH3)5RuL2+ (L ) acetate; benzoate; p-hydroxybenzoate;
2,4-dihydroxybenzoate) and (NH3)5RuL3+ (L ) hypoxanthine-κN7;
7-methylhypoxanthine-κN9; 1,3,9-trimethylxanthine-κN7) species,
a dimeric (NH3)10Ru2L24+ (L ) 2,4-dihydroxybenzoate) complex,
and the free ligands were fully optimized assuming vacuum
conditions.
Electronic transition energies and intensities (oscillator strengths,
f) were calculated using the time-dependent DFT (TD-DFT)
formalism25 and the ECP/basis sets just described. Assignment of
a particular electronic transition (π f π*, π f d LMCT, etc.) was
based on consideration of the magnitude of the oscillator strength,
the largest transition amplitude(s) for the excitation, and by
visualization of the contributing MOs. The experimental spectra
were measured in a condensed phase (aqueous solution or a mull)
and to facilitate comparisons of computed and experimental
transition energies electrostatic effects of the polar medium should
be incorporated into the TD-DFT wave functions. This was
accomplished via the self-consistent polarizable conductor model
(CPCM),26 always choosing water as the “solvent”.
Furthermore, we found the conventional B3LYP combination
of functionals to be suitable for excited-state calculations only on
the neutral organic ligands and on metal complexes carrying a net
+3 charge. For the calculations on metal complexes featuring
smaller net charge, it was necessary to make computational
adjustments to trace analogous electronic transitions through the
various complexes. With the B3LYP functionals, the transition
energies become too low (sometimes by more than 0.5 eV) in the
complexes which contain negatively charged ligands. Applying non-
hybrid functionals worsened the situation, in some cases making
the transition energies appear negative. Rather than introducing a
number of correction factors, we decided to increase the fractional
contribution of Hartree-Fock (exact) exchange in the hybrid
exchange functional above the 20% value prescribed in the B3
functional.23 The tendency for DFT to underestimate the band gap
for weakly interacting systems, and hence for TD-DFT to under-
estimate the electronic excitation energies, when local, time-
independent functionals are employed is well documented.27
Increasing the contribution of Hartree-Fock exchange to the overall
exchange functional increases the orbital energy separation between
the occupied and the unoccupied levels and thus increases the
computed transition energies, including the energies of the charge-
transfer states of interest here. The molecular orbitals and hence
configurations that contribute to a particular transition are not
significantly altered by this procedure; neither are the computed
intensities. A crude survey using (NH3)5RuL2+ (L ) acetate,
benzoate) as test molecules indicated that a combination of 40%
Hartree-Fock exchange (and hence 60% Slater local exchange)
was more suitable for calculations on excited states of metal
complexes with a net +2 charge, that is, the (NH3)5RuIII-L
complexes which contain a ligand carrying a single negative charge
(e.g., L ) p-hydroxybenzoate; 2,4-dihydroxybenzoate). The 40/60
fractional mix of Hartree-Fock/Slater exchange appeared to be
appropriate for excited states of the dimeric complex (overall +4
net charge) as well. For (NH3)5RuIII-L complexes containing a
ligand with a net charge of -2 (e.g., L ) deprotonated p-
hydroxybenzoate or deprotonated 2,4-dihydroxybenzoate), we found
that a 50/50 mix of Hartree-Fock/Slater exchange worked well.
Results and Discussion
Crystal Structures of 2, 3, and 6. The structures of 2
and 3 each contain [(NH3)5RuIIIDHB-]2+ cations (DHB- )
2,4-dihydroxybenzoate), sulfate anions, and either 0.5 dim-
ethylformamide (DMF) molecules (2) or two water molecules
(3), respectively, per asymmetric unit. Structure 6 contains
[(NH3)5RuIIITMX]3+ cations (TMX ) 1,3,9-trimethylxan-
thine), and both tetrachlorozincate and chloride anions in
equal amounts. Views of the four unique cations in the three
structures are shown in Figure 2, while Figure 3 shows a
general Cartesian coordinate system useful for describing
these and related structures, and for interpreting the electronic
spectra of their crystals. In each structure, five N(NH3) atoms
and an O(carboxylate) atom from DHB- or an N(imine) atom
from TMX complete a distorted octahedron about Ru. The
largest deviations from octahedral symmetry in structures 2
and 3 (Table 2) are associated with the DHB- ligands, as
evidencedbytheN(ax)-Ru-X(ligand)andN(eq)-Ru-X(ligand)
angles, both of which, on average, deviate substantially (ca.
5°) from their ideal values of 180° and 90°, respectively (the
magnitude of the deviations can be seen from the standard
deviations in the averages reported in Table 2). Using the
same angular measures, structure 6 exhibits more nearly a
regular octahedral coordination geometry than does 2 or 3.
(18) Parr, R. G.; Yang, W. In Density-Functional Theory of Atoms and
Molecules; University Press: Oxford, 1989.
(19) Gaussian 03, ReVision B.03: Frisch, M. J.; et al. Gaussian, Inc.:
Pittsburgh, PA, 2003; see Reference 1 in the Supporting Information
for the complete reference to Gaussian 03.
(20) Becke, A. D. J. Chem. Phys. 1993, 98, 5648–5652.
(21) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785–789.
(22) Andrae, D.; Ha¨ussermann, U.; Dolg, M.; Stoll, H.; Preuss, H. Theor.
Chim. Acta 1990, 77, 123–141.
(23) Dunning, T. H.; Hay, P. J. In Modern Theoretical Chemistry; Schaefer,
H. F., Ed.; Plenum: New York, 1976; pp 1-28.
(24) Binkley, J. S.; Pople, J. A.; Hehre, W. J. J. Am. Chem. Soc. 1980,
102, 939–947.
(25) Casida, M. E.; Jamorski, C.; Casida, K. C.; Salahub, D. R. J. Chem.
Phys. 1998, 108, 4439–4449.
(26) Cossi, M.; Rega, N.; Scalmani, G.; Barone, V. J. Comput. Chem. 2003,
24, 669–681.
(27) (a) See, for example, Tozer, D. J. J. Chem. Phys. 2003, 119, 12697–
12699. (b) Dreuw, A.; Weisman, J. L.; Head-Gordon, M. J. Chem.
Phys. 2003, 119, 2943–2946. (c) Perdew, J. P. Int. J. Quant. Chem.,
Quant. Chem. Symp. 1986, 19, 497–523.
9816 Inorganic Chemistry, Vol. 47, No. 21, 2008

(NH3)5RuIII(2,4-dihydroxybenzoate) and (NH3)5RuIII(xanthine)
canted 5-10° from the bisector toward certain axial ammine
ligands, as indicated by the N(2)-Ru-X(2)-C(1) torsion
angles in Table 2. For this geometry, the π systems of the
DHB- and TMX ligands have an appropriate orientation and
symmetry to interact strongly with the singly occupied
Ru(dxz) orbital (Figure 3). An ammine N-H group in each
of the cations in 2, 3, and 6 forms an intraionic hydrogen
bond to an oxygen atom of the DHB- or TMX ligand (Table
2), the location of which (Figure 4) is consistent with the
direction of canting noted above. Additionally, the cations
in2and3eachformintraionicO(carboxylate) · · ·H-O(hydroxyl)
hydrogen bonds (Figure 2).
In structures 2 and 3, the several species in the lattices
are linked by an extensive hydrogen-bond network involving
each of the ammine groups, all of the O(sulfate) atoms, both
solvate water molecules in the case of 3, and the O(aldehyde)
atom from the DMF solvate molecule in 2. Structure 6
contains no anions capable of forming strong hydrogen
bonds. Consequently, in addition to the intraionic hydrogen
bonds shown in Table 2, the cation in 6 forms a single,
interionic N(13)-H · · · O(2) hydrogen bond, as well as
2-
several weaker hydrogen bonds, to the chloride and ZnCl4
anions. Details of these hydrogen-bonding networks for 2,
3, and 6 are given as Supporting Information.
Packing diagrams for 2, 3, and 6 are shown in Figure 5.
The structure of 2 contains layers of alternating (NH3)5-
RuIIIDHB- cations and sulfate anions along the c cell
1
direction. In the layers near the planes x ) 0 and x ) /2,
the two unique cations, one in each layer, are approximately
perpendicular to each other. The DMF solvate molecules are
located near the corners of the unit cell outline. Intercation
distances suggest little, if any, π-π bonding among the
cations. Crystals of 3 contain layers of alternating sulfate
ions and water molecules centered about the x ) 0 plane,
separated by layers containing pairs of centrosymmetrically
related (NH3)5RuIIIDHB- cations stacked head-to-tail and
1
Figure 2. Structures of the cations in 2 (a), 3 (b), and 6 (c) showing the
atom numbering schemes. Dashed lines indicate intramolecular hydrogen
bonds. Thermal ellipsoids are shown at the 30% probability level.
centered about the plane x ) /2. In this arrangement, the
ammines, water molecules, and sulfate anions form a
“hydrophilic” layer with extensive hydrogen bonding. The
overlapping, substituted-phenyl rings are coplanar with an
interplanar separation of 3.596(3) Å, and, within a given
dimer, 12 C(phenyl) · · · C(phenyl) interionic distances in the
range 3.760(4) Å to 3.786(3) Å, all slightly shorter than the
value of 3.8 Å typically associated with π-π bonding.28 We
conclude that there is extensive π-π overlap in 3, but
essentially none in 2. In 6, layers along the c cell direction
containing cations, chloride, and ZnCl42- ions centered about
the plane x ) /4 alternate with layers of cations, ZnCl4
and chloride ions centered about the plane x ) /4; there is
no indication of π-π stacking among the cations.
Many of the structural features of 2 and 3 reported here,
including the slight deviations from octahedral symmetry of
the (NH3)5RuIII units, the staggered orientation of the DHB-
and TMX ligands with respect to the equatorial ammines,
and the existence of intraionic N-H · · · X(ligand) hydrogen
Figure 3. General coordinate system used for species 1-7. The plane of
the organic ligand defines the xy plane; the N atoms of the four “equatorial”
ammine ligands define the yz plane. “Equatorial” ammine ligands are
numbered 1-4, while the “axial” ammine ligand is numbered 5.
1
2-
,
3
Mean axial and equatorial Ru-N bond lengths in 2 and 3
are equal to within ( 0.01 Å, and appear to be marginally
shorter than those in 6.
Except for a 2-4° phenyl/carboxylate twist in the cations
of 2 and 3, as indicated by the torsion angles in Table 2, the
DHB- and TMX ligands are planar and are oriented such
that the plane of the ligand is staggered with respect to the
ammine groups (Figure 4). The planes of the ligands do not
bisect the N(ax)-Ru-N(ax) angles exactly; rather, they are
(28) Janiak, C. J. J. Chem. Soc., Dalton Trans. 2000, 3885, 3896.
Inorganic Chemistry, Vol. 47, No. 21, 2008 9817

Krogh-Jespersen et al.
Table 2. Metric Parameters for Structures 2, 3, and 6 (Å, deg)
2
3
6
Distances
N ) 2b
N ) 8
N ) 2
Ru-N(ax)a
Ru-N(eq)
2.091(17)
2.106(9)
2.023(5)
2.0840(19)
2.104(5)
2.0039(16)
N ) 1
N ) 4
N ) 1
2.104(10)
2.124(8)
2.094(11)
N ) 1
N ) 1
N ) 1
Ru-X(ligand)c
Angles
cis-N(eq)-Ru-N(eq)
trans-N(eq)-Ru-N(eq)
N(ax)-Ru-N(eq)
N(eq)-Ru-X(ligand)
N(ax)-Ru-X(ligand)
90.0(13)
178.2(6)
90.0(23)
90.2(37)
174.7(22)
N ) 8
N ) 4
N ) 8
N ) 8
N ) 2
90.0(19)
179.0(10)
90.3(12)
89.8(46)
174.37(8)
N ) 4
N ) 2
N ) 4
N ) 4
N ) 1
90.1(30)
177.4(4)
89.7(17)
90.3(13)
179.4(5)
N ) 4
N ) 2
N ) 4
N ) 4
N ) 1
Torsion Angles
N(2)-Ru-O(1)-C(1)d
Ru-O(1)-C(1)-C(2)
O(1)-C(1)-C(2)-C(3)
50.7(4), 40.9(4)
2.5(6), 6.9(6)
-4.2(6), -4.1(6)
52.5(2)
-2.7(4)
2.0(4)
35.1(16)
intramolecular hydrogen bonds
crystal
H bond
D-H
H· · ·A
D· · ·A
D-H· · ·A
2
2
2
2
3
3
6
6
O(3)-H· · ·O(2)e
N(5)-H· · ·O(2)
O(8)-H· · ·O(7)
N(8)-H· · ·O(7)
O(3)-H· · ·O(2)
N(1)-H· · ·O(2)
N(13)-H· · ·O(6)
N(14)-H· · ·O(6)
0.84
0.91
0.84
0.91
0.80(4)
0.91
0.89
0.89
1.91
2.28
1.82
2.43
1.87(4)
2.26
2.27
2.05
2.625(5)
2.981(6)
2.563(4)
3.031(6)
2.584(3)
2.976(3)
3.050(17)
2.822(17)
142
134
146
124
148(4)
135
146
144
a Axial (ax) and equatorial (eq) designations correspond, respectively, to atoms N(5), and N(1) through N(4), as shown in Figure 3. b Values of N refer
to the number of entries averaged; for such entries the average value and the standard uncertainty (su) of the mean are reported. c X refers to the ligating
atom of the DHB- and TMX ligands; that is, O for structures 2 and 3, and N for structure 6. d Atom numbering for the torsion angles is as sketched in Figure
3. e Atom numbering for the intra-ionic hydrogen bonds is as in Figure 2.
present as DHB2-. Deconvolution of the free DHB- UV-
spectrum (Figure 6a) indicates the presence of four intense
transitions in the 25,000-50,000 cm-1 region; the observed
transition energies are in excellent agreement with those
calculated for four intensity-carrying π f π* DHB- transi-
tions (Table 3). Solutions of 2 (or 3) at pH 5 (Figure 6b)
display, in addition to several of the UV absorptions shown
by free DHB-, two additional low-energy absorption bands
centered at 23,600 cm-1 and 31,200 cm-1, whose origins
and striking solid-state polychromism are discussed below.
Neither spin-forbidden nor spin-allowed ligand-field ab-
sorptions having molar extinction coefficients much greater
than 50 are expected at energies below ∼30,000 cm-1 for
Ru(III) complexes.13,30 The first prominent (ε > 500)
absorptions of (NH3)6Ru3+ and (NH3)5Ru(H2O)3+ complexes
Figure 4. View of the cation in 3 showing the relative orientation of the
DHB- ligand with respect to the equatorial ammine ligands. Dashed lines
indicate hydrogen bonds. This relative orientation prevails for structures
1-6.
occur at ∼ 36,000 cm-1 and have been assigned to LM-
CT.13,30 Substitution of the aquo ligand of this latter complex
by carboxylate gives rise to one new absorption band with
maximum at 34,100 cm-1, assigned as carboxylate f Ru(III)
LMCT.13 In aqueous solution, the maximum of this absorp-
tion systematically red-shifts to 33,900 cm-1 for acetate and
to 33,800 cm-1 for benzoate; further to 33,300 cm-1 for
p-hydroxybenzoate (p-HB-) and, as noted in Table 3, to
31,200 cm-1 for DHB-. The p-HB- complex exhibited
additional low energy absorption around 25,000 cm-1, a
feature not shown by the acetate or benzoate complexes. In
basic solution (pH not reported),13 the absorption at 25,000
cm-1 was significantly red-shifted to 20,000 cm-1 and
became slightly more intense (εmax > 530). This change was
bonds, when the ligand contains an appropriate acceptor,
pertain also to published structures 4 and 5.3
Electronic-Spectral and Computational Studies. Pen-
tammineRuIII(benzoate) Complexes (1-3). Deconvoluted
electronic spectra of 2, 3, and the free anion DHB- are
presented in Figure 6 and summarized in Table 3. The pKa
values for the first and second proton ionizations of 2,4-
dihydroxybenzoic acid (DHB) are 3.1 and 8.6, respectively.29
In pH 6.2 phosphate buffer, more than 99% of the free ligand
is thus present as DHB- and the insignificant remainder is
(29) Martell, A. E.; Smith, R. M. Critical Stability Constants, Vol. 5: First
(30) (a) Navon, G.; Sutin, N. Inorg. Chem. 1974, 13, 2159–2164. (b) Meyer,
T. J.; Taube, H. Inorg. Chem. 1968, 7, 2369–2379.
Supplement; Plenum Press: New York, NY, 1982; p 345.
9818 Inorganic Chemistry, Vol. 47, No. 21, 2008

(NH3)5RuIII(2,4-dihydroxybenzoate) and (NH3)5RuIII(xanthine)
ion becomes dyz dxy dxz . For the (NH3)5RuIII(acetate) species,
we calculate a LMCT transition at 34,800 cm-1 (f ) 0.011),
in very good agreement with the reported absorption
(shoulder) at 33,900 cm-1.13 The transition principally
involves excitation of an electron from a π orbital of the
acetate unit to the singly occupied Ru(dxz) orbital. The active
acetate π orbital is the out-of-phase combination of O(2pz)
orbitals, polarized so that the contribution to the donor orbital
is largest on the O atom bonded to Ru. However, the
transition does acquire some (π, π*) character from a
localized acetate excitation at higher energy. In the
(NH3)5RuIII(benzoate) species, this low-energy LMCT transi-
tion is computed at 31,950 cm-1 (f ) 0.075) and is described
fairly well solely as a single excitation from the out-of-phase
combination of acetate O[2p(π)] orbitals, slightly perturbed
by a phenyl MO, into the Ru(dxz) orbital.
2
2
1
The electronic-structure calculations for 1 predict π(p-
HB-) f Ru(dxz) LMCT transitions at 26,400 cm-1 (f )
0.057); 32,700 cm-1 (f ) 0.005), and 34,400 cm-1 (f )
0.021) (see Table 3). The p-hydroxyl group interacts strongly
with one member of the (initially) degenerate π-HOMOs of
the phenyl group, shifting it to higher energy. Electron
transfer from this perturbed and destabilized phenyl orbital,
which also contains a contribution from the carboxylate
moiety (viz., benzoate discussed above), into the Ru(dxz)
orbital gives rise to the lowest-energy LMCT absorption
computed at 26,400 cm-1 (observed about 25,000 cm-1). The
transition computed at 32,700 cm-1 originates in the other
high-lying π-orbital of the phenyl group, but it carries little
intensity and does not result in an experimentally resolved
absorption feature, because the ligand donor orbital does not
contain a significant contribution from the carboxylate
moiety; thus donor-acceptor overlap is poor. Finally, the
transition computed at 34,400 cm-1 is the one that is most
analogous to the single transitions observed in the (NH3)5RuIII-
acetate and benzoate complexes. The two computed transi-
tions at higher energy (32,700 cm-1 and 34,400 cm-1)
contribute together to the band observed as a shoulder near
33,000 cm-1. Upon ionization of the p-hydroxyl group of 1,
the TD-DFT calculations predict the appearance of a low-
energy transition at 20,800 cm-1 (f ) 0.040), clearly
assignable as (p-HB2-) f Ru(dxz) LMCT, which energeti-
cally provides an extremely good match to the red-shifted
absorption exhibited by 1 in alkaline solution (∼20,000
cm-1).13 Thus, the experimental observations made previ-
ously by Stritar and Taube strongly implied that the low-
energy electronic absorptions exhibited by 1 were due to
LMCT, and they are unequivocally confirmed as such by
our present calculations.
Figure 5. Packing diagrams for 2, 3, and 6. (a) View, approximately
perpendicular to (100), of the structure of 2. (b) Projection, along the b
axis, of the structure of 3. (c) Projection, along the b axis, of the structure
of 6. For clarity, H atoms have been omitted.
attributed to ionization of the phenolic proton and was
reversible. Although not explicitly noted, such spectral
behavior implies assignment of the absorption as LMCT, red-
shifted relative to the p-HB- complex by the increased
electron density of a p-HB2- ligand. Finally, the solution
spectra of free p-HB- were said to be little changed by
complexation to the (NH3)5Ru3+ unit of 2.13
The observations noted above strongly imply that the low-
energy absorptions displayed by 1 (and analogues) are due
to LMCT, and they are confirmed as such by our present
TD-DFT calculations. The coordinate system we use for the
near-octahedral (NH3)5RuIII-L complexes of interest here
is defined in Figure 3: a planar (or nearly planar) organic
ligand L is situated in the xy plane, approximately bisecting
two N-Ru-N angles, and four ammine N atoms define the
yz plane. Carboxylates and benzoates are strong π-donors,
which will preferentially interact with the dxz component of
the t2g d(π)-set of orbitals on the central metal ion; hence,
the idealized electronic configuration of the ground state RuIII
Relative to p-HB-, the additional phenolic OH substituent
present in DHB- increases the electron density in the
aromatic ring and further destabilizes the higher lying
phenylic π-orbitals. This causes the two observed low-lying
absorptions of the homologous DHB- complexes to red-shift
by a few thousand wavenumbers (Table 3, spectra of 2 vs
1). The calculations predict a somewhat larger (∼ 5,000
cm-1) red-shift to the π(DHB-) f Ru(dxz) LMCT transitions
in 2. The computed transitions are at 21,600 cm-1 and 28,400
Inorganic Chemistry, Vol. 47, No. 21, 2008 9819

Krogh-Jespersen et al.
Figure 6. Deconvoluted electronic spectra of (a) 6.72 × 10-4 M DHB- at room temperature in pH 6.2 phosphate buffer having a path length of 0.1 cm;
(b) 2.33 × 10-4 M 2 at room temperature in pH 5 aqueous solution having a path length of 1.0 cm; (c) 6.57 × 10-4 M 2 at room temperature in pH 10
aqueous solution having a path length of 1.0 cm; (d) a polarized spectrum of 3 at room temperature (3.783 M, pure single crystal, 0.013 mm thick) with the
electric vector parallel to c; and (e) same as (d), but with the electric vector parallel to b.
cm-1, whereas the deconvoluted spectrum shows peaks at
23,600 cm-1 and 31,200 cm-1. Contour plots of the principal
molecular orbitals involved are shown in Figure 7. The
transition of low energy originates in a DHB- (π)-donor
orbital, 7(a), and terminates in a Ru[d(π)] acceptor orbital,
7(c). The higher-energy excitation originates in a second
DHB- (π)-donor orbital of lower energy, 7(b), but terminates
in the same Ru[d(π)] acceptor orbital, 7(c). Deprotonation
of one of the phenolic groups to afford a (NH3)5RuIII(DHB2-)
species additionally destabilizes the ligand π-orbitals with
concomitant observed (Figure 6c) and calculated (for depro-
tonation at the 2-hydroxy site) red-shifts of the LMCT bands
by approximately 3,000 cm-1 (Table 3). To our knowledge,
it has not been established which hydroxy group in DHB-
has the lowest pKa, and the results from the TD-DFT
calculations for species deprotonated at the 2- or the
4-hydroxy site are quite similar.
ligand in the xy plane (Figure 3), the d-electron configuration
of the Ru(III) ion in a one-electron picture is formally
dyz dxy dxz1 (t2g5). The non-cubic ligand field, combined with
spin-orbit coupling effects, causes further modest splitting
and mixing of the Ru(III) orbitals.31,32 Consequently, the
formally half-occupied dxz orbital acquires some dxy and dyz
character, affording a mixed orbital still having predomi-
nantly dxz character but rotated somewhat toward dxy and dyz.
The percentage hole character of the Ru(III) dxz orbital could
be considerable (perhaps even greater than 50%) for strong
π-donor ligands such as DHB- and purines, that complete
the coordination geometries of the strongly π-accepting
(NH3)5RuIII units present in 2-6.32 The TD-DFT calculations
on 2-6 always predict two very low-lying d f d excitations
(1,000-3,000 cm-1 above the ground state), corresponding
to the alternative occupancies of the t2g set of orbitals. For
2
2
(31) Kaplan, D.; Navon, G. J. Phys. Chem. 1974, 78, 700–703.
(32) LaChance-Galang, K. J.; Doan, P. E.; Clarke, M. J.; Rao, U.; Yamano,
A.; Hoffman, B. M. J. Am. Chem. Soc. 1995, 117, 3529–3538.
In the near-octahedral ligand fields associated with the
observed site symmetries of 2-6 and a strongly π-donating
9820 Inorganic Chemistry, Vol. 47, No. 21, 2008

(NH3)5RuIII(2,4-dihydroxybenzoate) and (NH3)5RuIII(xanthine)
Table 3. Summary of Observed (Deconvoluted) and Calculated Electronic Spectra for 2,4-Dihydroxybenzoate (DHB-) and Complexes 1, 2, and 3a,b
solution or single crystal
mull
calc. (TD-DFT)
system
energy
εmax
∆ν1/2
energy
energy
f
assignment
c
DHB-
34,600
40,500
44,900
48,700
25,000
4,600
10,200
4,600
28,600
530
3,000
3,500
3,600
4,500
32,900
40,600
45,100
48,200
26,400
32,700
34,400
20,800
21,600
28,400
36,000
37,200
41,900
18,400
30,300
33,400
33,700
43,800
19,400
24,000
19,400
24,000
34,500
35,000
0.17
0.12
0.38
0.39
π f π*
π f π*
π f π*
π f π*
1d
0.057
0.005
0.021
0.040
0.044
0.016
0.121
0.226
0.188
0.058
0.031
0.104
0.178
0.055
0.074
0.008
0.074
0.008
0.042
0.122
π f dxz LMCT
π f dxz LMCT
π f dxz LMCT
π f dxz LMCT
π f dxz LMCT
π f dxz LMCT
π f π*
π f π*
π f π*
π f dxz LMCT
π f dxz LMCT
π f π*
33,000(sh)
2700
1d e
20,000
23,600
31,200
34,200
40,000
∼46,000
19,900
29,700
35,200
41,000
∼46,000
21,000
25,300
20,200
24,400
30,600
>530
500
,
2f
5,800
4,700
3,300
4,700
3,600
7,200
3,800
5,300
5,400
5,800
4,600
5,300
3,900
5,000
5,600
∼22,000(br)
1,200
5,900
9,200
12,500
900
1,900
8,200
12,000
11,000
310
2g
π f π*
π f π*
3h
3i
∼19,000(br)
Π+ f dxz LMCT
Π-f dxz LMCT
Π+f dxz LMCT
Π-f dxz LMCT
π f dxz LMCT
π f dxz LMCT
330
19
65
220
a Generally, only transitions with a computed oscillator strength (f) larger than 10-2 are shown; see Supporting Information for a more complete tabulation.
b Energies in cm-1; extinction coefficients (ε) in L mol-1 cm-1 c Measured in pH 6.2 phosphate buffer. d Reference.13 Calculations performed on
(NH3)5RuIII(HB1- e Alkaline solution. Calculations performed on (NH3)5RuIII(HB2-) with deprotonation at the 4-hydroxy site f Measured in pH 5 aqueous
.
)
solution. Calculations performed on (NH3)5RuIII(HB1- g Measured in pH 10 aqueous solution. Calculations performed on (NH3)5RuIII(DHB2-) with
)
deprotonation occurring at the 2-hydroxy site h Single crystal, electric vector parallel to c. i Single crystal, electric vector parallel to b.
Figure 7. Contour plots of principal orbitals involved in the DHB-(π) f Ru[d(π)] LMCT transitions in 2. The same isocontour value applies to all three
plots. (a) HOMO ꢀ spin-orbital, a DHB- (π)-donor orbital; (b) HOMO-1 ꢀ spin-orbital, a second DHB- (π)-donor orbital; and (c) LUMO ꢀ spin-orbital,
a Ru[d(π)] acceptor orbital. The transition amplitudes for the elementary excitations HOMO ꢀ f LUMO ꢀ and HOMO-1 ꢀ f LUMO ꢀ are very large and
dominate the overall intensities of the two low-lying ligand-to-metal charge-transfer transitions in 2 (Table 3).
simplicity in band assignments, we will continue to refer to
the potentially mixed RuIII d-vacancy as dxz.
or purine ligation also share the staggered orientation relative
to two of the cis-ammine groups located in the “yz” plane.
Solid-state spectra of 2 and 3 are expected to show modest
red-shifts in LMCT bands relative to those recorded in the
solution spectra. Stark-effect studies have revealed substantial
changes in dipole moments for the LMCT excitation of
Ru(III) complexes.33 For example, the pyridine f Ru(III)
LMCT absorption at 20,300 cm-1 shown by (NH3)5RuIII(4-
aminopyridine) carries a dipole-moment change of 18 ( 3
D.32 Because the dipole moment in the ground state of the
The electronic transitions in the aqueous hydroxybenzoate
complexes discussed above all appear to be well described
as ligand[(π)] f Ru[dxz] LMCT with no particular involve-
ment of Ru(dyz, dxy) orbitals in the final state. Moreover, these
transitions should be strongly polarized along the RuIII-L
bond direction,2b an expectation which mirrors the results
of polarized single-crystal spectroscopic studies reported
below for complexes 3-6, whose essentially planar DHB-
Inorganic Chemistry, Vol. 47, No. 21, 2008 9821

Krogh-Jespersen et al.
(NH3)5RuIII(DHB-) complex must be substantially greater
than that of the excited LMCT state, formally
Table 4. Squares of the Projection of the (NH3)5Ru-L Bond Vector on
the Crystallographic Axes
a
compound
3 (L ) 2,4-DHB-)
axis
projection
(NH3)5RuII(DHB) species, high-dielectric aqueous media
should preferentially stabilize the ground state. Lesser
preferential stabilization of the ground state in the lower
dielectric solid-state lattice ought to cause a red-shift in the
mull LMCT absorptions relative to those shown by the
corresponding solution complex. Indeed, the electronic
spectra of 2, measured in mineral-oil mulls (not shown), show
the lowest-energy LMCT absorption around ∼22,000 cm-1,
a red-shift of ∼1,600 cm-1 relative to the aqueous solution
spectra. Also, the mull spectra (not shown) of 3 are
considerably broader than those of 2 and presumably embody
a comparable red-shift. The deconvoluted low-energy ab-
sorptions of 3 are split by more than the calculated difference
in the absorptions for the two crystallographically unique
molecules of 2, consistent with the relative broadness of the
two spectra. Details are given in the following few paragraphs.
Examination of 3 by polarized microscopy revealed a
pronounced red/pale-yellow dichroism when the electric
vector of light was oriented, respectively, parallel to the c
and b crystal axes. The visible absorptions of the single
crystals are significantly broadened relative to those of
aqueous 2 or 3 (both afford the same solution complex) and
the polycrystalline mull spectra of 2, whose space group is
suboptimal for polarization studies (see Experimental Sec-
tion). Gaussian deconvolution of the broad visible absorption
band in Figure 6d suggests that it is comprised of two equally
broad, overlapping bands at ∼21,000 cm-1 and ∼25,000
cm-1, approximately centered about the lowest-energy solu-
tion LMCT absorption at 23,600 cm-1 (Table 3). Owing to
the substantially greater intensity of LMCT absorption at
∼31,000 cm-1, an estimate of the intensity of this band in
the solid-state spectrum could only be observed in the “pale-
yellow” orientation; the optical density for this absorption
was too high to measure in the “red” orientation.
a
b
c
a
b
c
a
b
c
a
b
c
0.9254
0.0083
0.0662
0.7196
0
0.2804
0.3365
0
0.6635
0.0978
0.8844
0.0178
4 (L ) N(7)-hypoxanthine)
5 (L ) N(9)-7-methylhypoxanthine)
6 (L ) N(7)-1,3,9-trimethylxanthine)
at approximately 21,000 cm-1 (Table 3). The large intensity
reductions, but not disappearance, of the absorptions when
the electric vector was rotated from the c to the b (Figure
6e) axis indicates that all three absorptions are strongly
polarized along the RuIII-O bond, yielding further support
to their assignment as LMCT transitions.
Could relatively minor structural differences among the
monomers be responsible for the spectral variations? As
reported above, at the DFT/B3LYP-optimized ground-state
geometry, the LMCT band for monomeric (NH3)5RuIII-
(DHB-) is computed at 21,600 cm-1 (f ) 0.044). In a
monomeric (NH3)5RuIII(DHB-) complex constructed using
the crystallographic coordinates for 3, the lowest-energy
LMCT absorption is computed at 22,300 cm-1 (f ) 0.034).
Similarly, considering the two non-equivalent (NH3)5-
RuIII(DHB-) units in 2 individually as molecular species,
we compute the lowest energy LMCT transition at 23,700
cm-1 in one unit (Ru-yellow-a, f ) 0.038) and at 21,500
cm-1 in the other (Ru-yellow-b, f ) 0.031). Thus, the spectral
changes predicted from slightly different coordination ge-
ometries present in various (NH3)5RuIII(DHB-) units are
modest. Barring extensive interactions among the units,
crystals of 2 would be predicted to show a broad absorption
band (centered near (23,700 cm-1 + 21,500 cm-1)/2 ∼
22,500 cm-1) because of the presence of two molecular units
in equal proportions absorbing with similar strengths.
Interestingly, the spectrum of 2 in a mull does show broad
absorption peaking ∼22,000 cm-1 (Table 3). In this analysis,
the colors of 2 and 3 should be very similar (yellow). This
is obviously not the case; however, minor geometric differ-
ences among the molecular units cannot be responsible for
the red-shifted LMCT band observed in crystalline 3.
Differences in dielectric constants between solution and
solid-state environments produce band shifts; the apparent
splitting of the lowest energy LMCT band shown by 3 and
the resulting polychromism must consequently originate from
differences in crystal packing between 2 and 3. As noted
above in the structural results section, 3, in contrast to 2,
crystallizes as dimers, which exhibit considerable spatial
overlap of the DHB- units. In an attempt to account for the
split LMCT absorption bands in 3 by computational means,
a (NH3)5RuIII(DHB-) dimer was constructed with maximum
spatial overlap of the two DHB- units and inversion
symmetry (molecular point group Ci). The geometry of the
dimer was optimized assuming that the two unpaired
For further analysis, the solid-state spectra must be
transformed to those obtainable if polarized light could
hypothetically be directed along molecular axes. For electric-
dipole transitions, band intensities are proportional to the
square of the transition moments, which depend upon squares
of the projections of the molecular axes along the crystal-
lographic axes (Table 4). For structure 3, the calculated
squares of the Ru-O(DHB-) bond vector projections along
the b and c axes are 0.0083 and 0.0662, respectively; the
major component of the Ru-O projection is 0.9254 along
the a axis. The Ru-O(DHB-) bond vector is largely parallel
to a and nearly orthogonal to b, as is apparent from the
crystal packing viewed down the c* axis (Figure 8a).
Unfortunately, the thin, plate-like morphology of 3 did not
permit measurement of the polarized spectra along a. From
the projections, the calculated reduction in oscillator strength
is 0.0662/0.0083, or 8.0, in fair agreement with an observed
decrease in molar extinction coefficients of 330/65 ∼ 5.1
for the absorption at approximately 25,000 cm-1, but not in
accord with the 16-fold (310/19) decrease of the absorption
(33) Shin, Y.-g. K.; Brunschwig, B. S.; Creutz, C.; Sutin, N. J. Am. Chem.
Soc. 1995, 117, 8668–8669; and references cited therein.
9822 Inorganic Chemistry, Vol. 47, No. 21, 2008

(NH3)5RuIII(2,4-dihydroxybenzoate) and (NH3)5RuIII(xanthine)
electrons, one on each Ru(III) ion, were triplet coupled to
form a 3Au ground state. The optimized Ru-Ru distance in
the computational dimer is 12.1 Å, while the experimentally
determined Ru-Ru distance in 3 is 10.8 Å; the optimized
distance between phenyl ring planes is 3.52 Å whereas the
experimentally observed separation is 3.596(3) Å. Because
DFT/B3LYP calculations do not include any treatment of
dispersive interactions,34 the good agreement nevertheless
obtained between experimental and computed dimer geom-
etries may indicate that the dominant interactions in the π-π
stacked dimer are electrostatic in nature.34b
The MOs of the dimer may be considered as in- and out-
of-phase combinations of the individual (NH3)5RuIII(DHB-)
monomer orbitals. Thus, the partially filled Ru(dxz) orbitals
(
form new orbitals Dxz ) Ru1(dxz) ( Ru2(dxz), and the two
highest doubly occupied orbitals combine into Π(
)
-
-
(
π(DHB1 ) ( π(DHB2 ). The Dxz orbitals are hardly split,
as expected considering the large spatial separation between
the two Ru centers; computed orbital energies of the two
components, which contain one electron each in the dimer
ground state, are Dxz-(ag) ) -3.27 eV and Dxz+(au) ) -3.28
eV. However, the Π( orbitals are significantly split, since
they are localized largely in the strongly overlapping phenyl
moieties. The computed orbital energies of Π+(ag) and
Π-(au) are -6.78 and -7.23 eV, respectively. Two of the
four possible elementary Π f Dxz excitations are forbidden
by symmetry (Π+ f Dxz-, Π-f Dxz+). However, one
allowed LMCT transition, which is predominantly the Π+
f Dxz+ excitation, is computed at 19,400 cm-1 (f ) 0.074);
a second allowed LMCT transition, which is predominantly
Π- f Dxz-, is calculated at 24,000 cm-1 (f ) 0.008). Relative
to the LMCT transition calculated for the monomer in
aqueous solution (∼21,600 cm-1), we compute a red-shift
of 2,200 cm-1 for the first transition and a blue-shift of 2,400
cm-1 for the second transition. The transition energies
computed from our model dimer match the polarized single
crystalpeakpositionswell(∼20-21,000cm-1and∼24-25,000
cm-1, Table 3).
We conclude that the crystal-packing arrangement in 3,
with its extensive spatial π-π overlap between the DHB-
rings of adjacent (NH3)5RuIII(DHB-) units, induces signifi-
cant splitting among the π-type DHB- orbitals in the ground
state of the (NH3)5RuIII(DHB-) dimers forming the unit cell
of 3. A spectroscopic consequence of this interaction is split
LMCT excited states in 3 and a difference in color between
crystalline 2 (yellow) and 3 (red). The TD-DFT calculations
with modified B3LYP-type functionals predict a low-energy
electronic spectrum containing two allowed transitions for
the dimer. The good agreement obtained for the computed
and experimental transition energies (Table 3) as well as the
separation between the absorption peaks (∼4,600 cm-1
computed, ∼4,000 cm-1 observed) suggests that differential
(excited vs ground state) dispersive interactions do not play
a significant role in determining the positions of the LMCT
states.
Figure 8. Partial packing diagrams of 3, 4, 5, and 6 for use in the
interpretation of their spectra: (a) 3 viewed along c*; (b) 4 viewed along c;
(c) 4 viewed along b; (d) 5 viewed along c; (e) 5 viewed along b; and (f)
6 viewed along c.
(34) (a) Kristyan, S.; Pulay, P. Chem. Phys. Lett. 1994, 229, 175. (b)
Cybulski, S. M.; Seversen, C. E. J. Chem. Phys. 2005, 122, 014117–
2946.
Inorganic Chemistry, Vol. 47, No. 21, 2008 9823

Krogh-Jespersen et al.
Figure 9. Electronic spectra and deconvoluted spectra of (a) 9.47 × 10-4 M 4 and 6.13 × 10-3 M 5 at room temperature in 50/50 H2O/glycerol made 0.15
M in HCl. The path length was 0.15 cm in both cases; (b) 0.026 mm thick plate of 4 at room temperature with the absorption parallel to a, deconvoluted
after subtracting the absorption parallel to b to correct approximately for the UV tail (the flattened absorption is not an artifact arising from stray light); (c)
5 at room temperature having a path length of 0.011 mm thick using a dummy band to approximate the UV tail for the deconvolution of absorption parallel
to a; and (d) polarized spectra of 6 (path length 0.05 mm) at room temperature. Following subtraction of the red absorption UV tail (approximated by
blue-shifting the yellow-polarized spectrum by 1400 cm-1; see left arrow), the resulting “difference” visible spectrum was used for deconvolution.
Table 5. Summary of Observed (Deconvoluted) and Computed Electronic Spectra for Complexes 4-6a,b
solution
polarized single crystal
calc. (TD-DFT)
system
energy
22,100
εmax
∆ν1/2
energy
εmax
∆ν1/2
energy
f
assignment
4
285
5,500
20,200
23,500
68c
2,700
4,900
22,300
22,800
31,000
38,000
22,600
33,650
34,700
19,800
22,100
28,100
36,900
<10-3
0.018
0.008
0.032
0.008
0.024
0.048
0.026
<10-3
0.011
0.037
O(n) f d(σ*) LMCT
π f dxz LMCT
π f dxz LMCT
π f dxz LMCT
π f dxz LMCT
π f dxz LMCT
π f dxz LMCT
π f dxz LMCT
O(n) f d(σ*) LMCT
π f π*
98c
33,600
1,370
5,400
5
6
23,000
31,300
141
2,480
5,800
3,000
23,300
188c
3,700
19,600
32,400
310
19,300
22,000
125d
58d
3,400
4,500
1,330
π f dxz LMCT
.
c Single-crystal, room-
a See Supporting Information for a more complete tabulation. Energies in cm-1; extinction coefficients (ε) in L mol-1 cm-1
b
temperature, electric vector parallel to the a axis. d Single-crystal, room-temperature, electric vector parallel to the b axis.
PentammineRuIII(purine) complexes (4-6). Deconvo-
luted solution and single-crystal spectra of 4-6 are presented
in Figure 9 and summarized in Table 5. Each complex
consists of (NH3)5RuIII moieties bound to the imine nitrogen
atoms of a purine ligand at either the N-7 (4, 6) or N-9 (5)
ring positions (Figure 1). Importantly, attachment of
(NH3)5Ru(III) to imine nitrogen atom N-7 places a carbonyl
oxygen atom of the planar purine ligand in a position to
hydrogen-bond to two cis-oriented ammine groups.14 In
contrast, the sole carbonyl oxygen of the hypoxanthine ligand
of 5 is well removed from possible intramolecular hydrogen
bonding by attachment of (NH3)5Ru(III) to N-9; in this case,
attachment of (NH3)5Ru(III) to N-7 is prohibited by methy-
lation at that site.14
by Clarke and Taube.14,35 Because quantification of the band
widths observed for the solution absorptions was required
for comparison to those observed in the single-crystal spectra,
solution spectra of 4 and 5 were remeasured (Figure 9a),
and they agree closely with the previously reported values.35
The spectra measured in glassed solutions at 80 K (not
shown) are little changed from those shown at room
temperature, aside from modest band sharpening at the lower
temperature. A striking feature shown by 4 and 6, but not 5,
is that the lowest-energy absorption either is substantially
broadened or clearly split in the solid-state spectra. Band
widths of the apparently single visible absorptions shown,
respectively, by solutions of 4 and 5 are presented in Table
Solution spectra of 4-6 exhibit two relatively low-energy
absorptions, which were assigned as purine f Ru(III) LMCT
(35) Clarke, M. J. Inorg. Chem. 1977, 16, 738–744.
9824 Inorganic Chemistry, Vol. 47, No. 21, 2008

(NH3)5RuIII(2,4-dihydroxybenzoate) and (NH3)5RuIII(xanthine)
5 for comparison with those determined from deconvolution
of their single-crystal spectra.
crystalline phase (bands at 19,300 cm-1 and 22,000 cm-1)
and, yet again, from the calculations on monomeric 6 there
is no sign of a second, low-energy transition carrying
measurable intensity. In the case of monomeric 6, the
transition computed closest in energy to the observed one is
at higher energy and near-zero intensity (22,100 cm-1, f <
10-4). We shall return below to these two “silent” transitions
in 4 and 6.
Inspection of Figures 8b and 8d reveals that the electric
vector of light polarized along the b axes of both 4 and 5 is
perpendicular to the respective Ru-N(purine) bond vectors.
However, the electric vector of light polarized along either
their a or their c axes has a substantial component along the
Ru-N(purine) bond vector, and this component is somewhat
larger along a for 4 than 5 (Figures 8c and 8e); the difference
is magnified in the calculated squares of these bond projec-
tions (Table 4). Thus, the low-energy absorptions shown by
4, 5, and 6 are strongly polarized along the Ru-N(purine)
bond directions. The absorptions shown by 4 and 5 es-
sentially vanish when the electric vector of polarized light
is directed along their b axes, onto which the Ru-N(purine)
bond vectors have zero projections. As is apparent from
Figure 8f, the absorptions shown by 6 greatly diminish when
the electric vector is rotated from b to a, a result which is
consistent with the Ru-N(purine) bond projection decreasing
from 0.884 to 0.098 (Table 4).
We now briefly describe some of the causes we have
considered to explain why the low-energy bands in crystalline
4 and 6, but not in crystalline 5, appear broad or split. In the
published structures of 4 and 5, and in the crystal structure
reported here for 6, the (NH3)5Ru(III)(xanthine) cations
utilize crystallographic mirror planes, which contain the
xanthine, the Ru(III) ion, and the N atom of the ammine
trans to the xanthine ligand (molecular Cs point group
symmetry). The crystal packing of 4 and 5 differs in that
complex 4 enjoys considerably less purine π-stacking overlap
than complex 5.3 When viewed along the a cell direction, 5
contains columns of pentammineRu(III)-hypoxanthine cat-
ions arranged centrosymmetrically, head-to-tail. In each
column, the cations are equally spaced (3.41 Å interplanar
spacing) and exhibit substantial overlap of the hypoxanthine
units. In contrast to 3, there are no dimeric units in 5. This
observation suggests that π-π interactions are not respon-
sible for the split electronic energy bands in 4 or in 6. Indeed,
in 6, the xanthine-containing cations pack in a herringbone
pattern with little indication of π-π interactions. In 4, the
cations stack in an arrangement similar to that in 5, but with
substantially less overlap of their hypoxanthine rings, strongly
suggesting that π-π interactions are not responsible for the
split electronic absorption band in the former system.
Calculations in which weak purine-purine ring pi-overlap
and interaction was simulated with the aid of an ethylenic
species positioned above various purine CdC bonds did not
produce any splitting in the low-energy purine-Ru band.
The spectra of the (NH3)5Ru(III)(purine) complexes poten-
tially contain n f π* transitions of relatively low energy
inherent to the purines, but these transitions have low
intensity and are not observed as individual spectral features
in solution. Importantly, the polarization of n f π* transi-
tions would be perpendicular to the purine plane and not in
agreement with the observed polarization of the split bands:
the b axes of 4 and 5 are perpendicular to the purine planes
(Figure 8c,e), and the low-energy band(s) vanish for this
polarization. Finally, the TD-DFT calculations position
purine n f π* transitions well above the observed bands in
energy.
The more recent experimental results presented above
support the prior assignment of these low-energy absorptions
as purine f Ru(III) LMCT.2,3 Moreover, as would be
expected for LMCT, the low-energy absorptions shown by
6 are red-shifted with respect to those shown by 4 and 5
owing to the electron-donating property of the additional
methyl groups present in 6 (Figure 1).
The electronically excited states of 4-6 were probed
through TD-DFT calculations. For 5, one low-lying purine(π)
f Ru(dxz) LMCT transition is predicted at 22,600 cm-1 (f
) 0.008); to higher energy, two closely spaced intensity-
carrying transitions are computed at 33,600 cm-1 (f ) 0.024)
and 34,500 cm-1 (f ) 0.08). In solution, the bands observed
for 5 (Table 5) are at 23,000 cm-1 and 31,300 cm-1,
respectively. The computed results for 4 present a similar
pattern with a low energy LMCT absorption predicted at
22,800 cm-1 (f ) 0.018), and two higher energy absorptions
at 31,000 cm-1 (f ) 0.008) and 38,000 cm-1 (f ) 0.032).
For 4, the bands observed in solution (Table 5) peak at
22,100 cm-1 and 33,600 cm-1, respectively. As mentioned
above, the lowest-energy band of 4 is split in the crystalline
phase (bands at 20,200 cm-1 and 23,500 cm-1). Although
the calculations on monomeric 4 do predict the presence of
a second, low-energy excited state at 22,300 cm-1 (ap-
proximately 500 cm-1 below the intensity-carrying transi-
tion), the transition to this state carries insignificant oscillator
strength (f < 10-4). The calculations on 6 predict a purine(π)
f Ru(dxz) LMCT band at 19,800 cm-1 (f ) 0.026), in
excellent agreement with the solution spectral data (19,600
cm-1, Table 5). The calculated value of ∼3,000 cm-1 for
the methyl-group induced red-shift to the low-energy spec-
trum of 6 relative to 4 matches the experimentally observed
value of ∼2,500 cm-1 well. To higher energy, the calcula-
tions predict a weak π f π* transition at 28,100 cm-1 (f )
0.011), and a second LMCT transition at 36,900 cm-1 (f )
0.037). These two computed transitions flank an experimen-
tally observed band peaking at 32,400 cm-1. As with
crystalline 4, the lowest-energy band of 6 is split in the
Next, we focused our attention on possible spectroscopic
ramifications of the unique position occupied by one oxygen
atom in complexes 4 and 6 vis-a-vis complex 5. Hydrogen-
bond interactions involving two ammines and the purine
carbonyl groups of 4 and 6 position the carbonyl oxygen
lone pair O(n) well for interaction with the Ru(III) ion
(Ru-O ∼ 3.6 Å, Figure 2c). The ground-state configurations
calculated by DFT for species 4 and 6 have, as expected,
the d-electron vacancy in the Ru(dxz) orbital (2A′′ state), and
hence the potential O(n) f Ru(dxz) LMCT transition would
Inorganic Chemistry, Vol. 47, No. 21, 2008 9825

Krogh-Jespersen et al.
inactive in the visible-near UV region, because the octahedral
t2g-eg* energy separation is large.2,30 However, given the
excellent spatial alignment of the O(n) donor orbital with
the d(σ*) set of orbitals in 4 and in 6, such LMCT excitations
could potentially be of low energy. Careful examination of
the data presented in Table 5 and Figure 9 indicates that the
new band appearing in crystalline 4 grows in to the red of
the solution band maximum, whereas the new band in
crystalline 6 appears to be developing to the blue of the
solution band. This is in full accord with the position in
which the O(n) f d(σ*) transition is computed relative to
the regular π(purine) f dxz LMCT excitation (toward lower
energy for 4, higher energy for 6; see above). Although the
X-ray data obtained for 6 demands symmetry in the “heavy
atom” skeleton, the hydrogen atoms were not located and
their distribution need not be of the same symmetry as the
crystalline environment.36 To explore spectroscopic conse-
quences of large amplitude motion, we arbitrarily placed one
of the hydrogen atoms participating in O-(H)N bonding
equidistant (∼ 1.40 Å) to N and O and recalculated the
electronic transitions. This type of distortion increased the
intensity of the O(n) f d(σ*) transition somewhat (f ∼ 10-4)
without significantly changing the excitation energy. Small
symmetry-destroying movements of the donor O-atom
preferentially toward one of the hydrogen-bonding ammines
also increased the transition intensity (to f ∼ 3 × 10-4).
Hence, we suggest that symmetry-destroying disorder as-
sociated with the O-(H)N bonds may provide the intensity
mechanism for the proposed O(n) f d(σ*) transition.37
Solid-state spectroscopic oddities reported previously
include the effects of conformational polymorphism,38 al-
lotropes showing different metal-ligand distances,39 stacking
changes which enable intermolecular charge-transfer absorp-
tions,40 and varying metal-metal distances shown by stack-
ing of planar Pt(II) complexes.41 The present study extends
this list to include split LMCT bands arising from interionic
π-stacking interactions of an aromatic donor ligand, as
characterized in 3. Although we offer a tentative assignment
for the band broadening/splitting in crystalline 4 and 6, the
true origin of this observed solid-state effect on what appears
to be purine f Ru(III) LMCT remains puzzling. We propose
intra-ionic hydrogen bonding and local environmental asym-
metry as a potential intensity-creating mechanism for the
otherwise prohibitively weak LMCT transitions in crystalline
4 and 6. In support of this suggestion, we note that hydrogen
bonds are known to exhibit anomalous behavior. In the solid
state, for example, asymmetric hydrogen bonds in a cen-
trosymmetric environment have been observed,36 while, with
a second system, a distortion of local symmetry caused by
dynamic hydrogen bonding effects was thought to be the
Figure 10. Contour plots of principal orbitals involved in the O(n) f
Ru[d(σ*)] transition in 6. The same isocontour value applies to all four
2
plots. (a) HOMO-3 R spin-orbital, O(n); (b) LUMO R spin-orbital, Ru(dx );
(c) HOMO-2 ꢀ spin-orbital, O(n); and (d) LUMO+2 ꢀ spin-orbital,
2
2
Ru(dy -z ). The transition amplitudes for the elementary excitations HOMO-3
R f LUMO R and HOMO-2 ꢀ f LUMO+2 ꢀ are similar in magnitude
and dominate the overall transition.
be weak for reasons of symmetry and would also be polarized
perpendicular to the N(purine) f Ru axis. As discussed
above, the real d-vacancy could contain (dxy, dyz) character
because of spin-orbit mixing, but an electron occupancy
2
change in the d-manifold to form a A′ reference ground
state (i.e., the unpaired electron now resides in the (dxy, dyz)
orbitals) in 4 or 6 leads to increases in total energy in both
species, even after geometry reoptimization. Furthermore,
2
TD-DFT calculations based on a A′ ground state do not
predict any low-energy LMCT transitions with significant
intensity, and two negative “d f d transition energies” (∼
-1000 cm-1) emerge from these calculations. While in-
tramolecular purine carbonyl interaction with Ru(dxy, dyz)
orbitals could provide a first rationalization of the band
broadening/splitting observed in the crystal spectra of 4 and
6, the TD-DFT calculations also suggest that LMCT transi-
tions of the type O(n) f Ru(dxy, dyz) carry negligible
oscillator strengths and are well removed toward higher
energy (>30,000 cm-1). Thus, the DFT calculations strongly
suggest that, in a one-electron picture, the ground states of
2
4-6 are best described as A′′ states.
The DFT optimized structures of both 4 and 6 confirm
the presence of two symmetric O-(H)N bonds in 4 and 6
with short interaction distances of about 1.87 Å; presumably,
such O · · · H(N) interactions are disrupted for complexes 4
and 6 in aqueous solution. In both monomeric 4 and 6, the
“silent” transition mentioned above is of the LMCT type with
proper x,z-polarization, and the excitation originates from
the proximal oxygen lone pair, O(n); surprisingly, however,
the transition terminates in formally antibonding Ru(III)
d-orbitals of local σ-symmetry. The relevant donor/acceptor
orbitals and elementary excitations are illustrated in Figure
10. We would tend to consider d(σ*) orbitals in near-
octahedral (NH3)5Ru(III)-L type ligation as charge-transfer
(36) Majerz, I.; Olovsson, I. Acta Crystallogr. 2007, B63, 748–752.
(37) Calculations with added diffuse and polarization basis functions on
O and on the hydrogens engaged in H-bonding also increase the
intensity of the O(n) f d(σ*) transition slightly.
(38) Yu, L. J. Phys. Chem. A 2002, 106, 544–550.
(39) Bennett, M. J.; Donaldson, P. B. Inorg. Chem. 1977, 16, 655–660.
(40) Desiraju, G. R.; Paul, I. C.; Curtin, D. Y. J. Am. Chem. Soc. 1977,
99, 1594–1601.
(41) Connick, W. B.; Henling, L. M.; Marsh, R. E.; Gray, H. B. Inorg.
Chem. 1996, 35, 6261–6265; and references cited therein.
9826 Inorganic Chemistry, Vol. 47, No. 21, 2008

(NH3)5RuIII(2,4-dihydroxybenzoate) and (NH3)5RuIII(xanthine)
cause of peak splitting in 13CCPMAS spectra.42 In solid
tropolone, dynamic hydrogen bonds were shown to provide
an efficient NMR relaxation mechanism,43 while, with proton
sponges, structural and NMR studies44 suggest that asym-
metric hydrogen bonding with a shallow double-minimum
potential well is the norm for these systems.
units induces significant splitting of the π-type orbitals in
the ground state and accounts for the observed split LMCT
excited states. The polarizations and energies of the broad/
split visible absorptions shown by 4 and 6 also suggest
assignment as LMCT. Complex 5 did not exhibit either split
visible absorptions or the intramolecular N-H· · ·O(carbonyl)
hydrogen bonding observed for 4 and 6. In contrast to 2 and
3, none of the complexes 4-6 form dimers in the solid state,
and the present study does not afford computational support
relating the split absorptions in crystalline 4 and 6 to the
details of π- π purine overlap. TD-DFT calculations suggest
that the additional bands observed for crystalline 4 and 6
may be O(n) f d(σ*) LMCT in nature with absorption
intensity gained from the presence of intraionic hydrogen
bonding. However, the computed overall weakness of these
transitions is of significant concern, and the assignment
should thus be considered tentative.
Conclusions
The (NH3)5RuIII(L) cations in the structures of 2-6 all
exhibit slight deviations from octahedral symmetry, a stag-
gered orientation of the nearly planar or planar ligands L
with respect to the equatorial ammines, and the presence of
intraionic N-H · · · X(ligand) hydrogen bonds when L con-
tains an appropriate acceptor. Differences among the crystals
include varying amounts of hydrogen bonding and π-π
stacking. The yellow (2) and red (3) forms of
(NH3)5RuIII(DHB-)SO4 ·solvate yield identical solution spec-
tra; hence, their different colors are attributed to differences
in crystal packing. In particular, in contrast to 2, 3 forms
centrosymmetric cation dimers which exhibit extensive π-π
overlap. Calculated (TD-DFT) energies and oscillator strengths
of the π f π* and π f dxz LMCT absorptions in the
UV-vis spectral regions for 1 (DHB-) and 2 agree well with
the experimental values obtained from the deconvoluted
solution spectra. Good agreement is also obtained for the
single-crystal spectrum of 3, in which extensive π-π overlap
between the DHB- rings of adjacent (NH3)5RuIII(DHB-)
Acknowledgment. We thank the reviewers for thoughtful
comments, which helped us improve the manuscript. The
present work elaborates on some of the pioneering studies
of Ru(III) charge-transfer absorptions from the laboratory
of the late Professor Henry Taube and is dedicated to his
memory.
Supporting Information Available: Crystallographic details,
including hydrogen bond data, in CIF format for the three structures
described in the manuscript; crystallographic details in CIF format
for the room-temperature structures of 2 and 3; complete ref 19;
DFT optimized geometries and TD-DFT excited-state data for 2-6.
This material is available free of charge via the Internet at
http://pubs.acs.org.
(42) Khan, M.; Brunklaus, G.; Enkelmann, V.; Spiess, H.-W. J. Am. Chem.
Soc. 2008, 130, 1741–1748.
(43) Detken, A.; Zimmermann, H.; Haeberlen, U.; Luz, Z. J. Magn. Reson.
1997, 126, 95–102.
(44) Stibrany, R. T.; Schugar, H, J.; Potenza, J. A. Acta Crystallogr. 2002,
E58, o1142–o1144; and references cited therein.
IC800511G
Inorganic Chemistry, Vol. 47, No. 21, 2008 9827