Dimer radical cations of indole and indole-3-carbinol: Localized and delocalized radical cations of diindolylmethane

Article abstract of DOI:10.1021/jp201415z

Extending our previous study on the title species (J. Phys. Chem. A2010, 114, 6787), we investigated the dimer cations that are formed on oxidation of the glucobrassin derivatives indole-3-carbinol (I3C) and diindolylmethane (DIM) and of parent indole (I). Radiolysis in ionic liquid and Ar matrices shows that, at sufficiently high concentrations and/or on annealing the solid glasses, intense intermolecular charge-resonance (CR) absorption bands in the NIR herald the formation of sandwich-type dimer cations. The molecular and electronic structure of these species is modeled by calculations with the double-hybrid B2-PLYP-D density functional method which yields predictions in good accord with experiment. The radical cation of DIM also shows a CR band, but unlike in the case of I and I3C, its occurrence is not dependent on the concentration but instead on the solvent: in ionic liquid the CR band is initially absent and arises only on annealing, whereas in Ar matrices it is present from the outset and undergoes blue shifting and sharpening on annealing. These puzzling findings are rationalized on the basis of B2-PLYP-D calculations which predict that neutral DIM exists in the form of two conformers, present in different relative amounts in the two experiments, which on vertical ionization form distinct radical cations, a nonsymmetric one where the odd electron is largely localized on one of the two indole moieties and one with C₂ symmetry where charge and spin are completely delocalized over both halves of the molecule, thus giving rise to an intramolecular CR transition. On annealing, the nonsymmetric cation relaxes to a similarly delocalized structure with C _s symmetry, thus explaining the observed increase and the shift of the CR band. We believe that DIM^?+ represents the first example of a radical cation which can exist under the same conditions as a localized and a delocalized complex cation.

Full text of DOI:10.1021/jp201415z

ARTICLE
pubs.acs.org/JPCA
Dimer Radical Cations of Indole and Indole-3-carbinol: Localized and
Delocalized Radical Cations of Diindolylmethane
Anna Bzoch-Mechkour and Thomas Bally*
Department of Chemistry, University of Fribourg, CH-1700, Fribourg, Switzerland
Andrzej Marcinek
Institute of Applied Radiation Chemistry, Technical University, 90-924 Lodz, Poland
S Supporting Information
b
ABSTRACT: Extending our previous study on the title species
(J. Phys. Chem. A 2010, 114, 6787), we investigated the dimer
cations that are formed on oxidation of the glucobrassin
derivatives indole-3-carbinol (I3C) and diindolylmethane
(DIM) and of parent indole (I). Radiolysis in ionic liquid and
Ar matrices shows that, at sufficiently high concentrations and/
or on annealing the solid glasses, intense intermolecular charge-resonance (CR) absorption bands in the NIR herald the formation of
sandwich-type dimer cations. The molecular and electronic structure of these species is modeled by calculations with the double-
hybrid B2-PLYP-D density functional method which yields predictions in good accord with experiment. The radical cation of DIM
also shows a CR band, but unlike in the case of I and I3C, its occurrence is not dependent on the concentration but instead on the
solvent: in ionic liquid the CR band is initially absent and arises only on annealing, whereas in Ar matrices it is present from the outset
and undergoes blue shifting and sharpening on annealing. These puzzling findings are rationalized on the basis of B2-PLYP-D
calculations which predict that neutral DIM exists in the form of two conformers, present in different relative amounts in the two
experiments, which on vertical ionization form distinct radical cations, a nonsymmetric one where the odd electron is largely
localized on one of the two indole moieties and one with C2 symmetry where charge and spin are completely delocalized over both
halves of the molecule, thus giving rise to an intramolecular CR transition. On annealing, the nonsymmetric cation relaxes to a similarly
delocalized structure with Cs symmetry, thus explaining the observed increase and the shift of the CR band. We believe that DIM•þ
represents the first example of a radicalcation which can exist underthe sameconditions asa localized and a delocalizedcomplexcation.
Scheme 1. MO Picture of the Electronic Structure of Dimer
Radical Cations (M)2
I. INTRODUCTION
•þ
It is a well-known property of planar π systems, such as
polyenes and (hetero)aromatic compounds, that upon ionization
•þ
under suitable conditions they can form π-complexes (M)2
,
composed of a radical cation M•þ and its parent, neutral M in
which spin and charge are fully delocalized over both moieties.1
Such dimer radical cations are bound by substantial covalent or
resonance contributions ΔEres, arising mainly from the interac-
tion between the HOMO of M and the SOMO of M•þ as
depicted in Scheme 1.
•þ
Spectroscopically (M)2 can be identified by having broad,
intense, near-infrared bands which are attributed to so-called
2
“charge-resonance” (CR) transitions of energy ECR
.
These
transitions can be understood on the basis of the generic MO
diagram in Scheme 1, which shows how the MOs of M and M•þ
split as the two species interact. The lowest energy electronic
excitation in the resulting dimer radical cations is that between
the bonding and the antibonding combination of the two
occupied frontier MOs. Charge-resonance (CR) transitions are
usually quite intense because the oscillator strengths for excita-
tions of that type are high and very broad because the distance
between the interacting π systems increases strongly on CR
Received: February 13, 2011
Revised:
April 26, 2011
Published: June 07, 2011
r
2011 American Chemical Society
7700
dx.doi.org/10.1021/jp201415z J. Phys. Chem. A 2011, 115, 7700–7708
|

The Journal of Physical Chemistry A
ARTICLE
Scheme 2
target bombarded at 40 kV/40 mA. The following net processes
are thought to occur
Ar þ hνðXÞ f Ar•þ þ eꢀ
CH2Cl2 þ eꢀ f CH2Cl• þ Clꢀ
Ar•þ þ S f S•þ þ Ar ðS ¼ I or DIMÞ
ð1Þ
ð2Þ
ð3Þ
excitation, as in such transitions an electron is promoted from
an MO that is bonding between the two moieties to one that is
antibonding. In addition to CR bands, dimer radical cations(M)2
Measurements in Ionic Liquid. Glassy solutions of indole
derivatives in a mixture of 1-butyl-3-methylimidazolium hexafluoro-
phosphate and methylene chloride (BMIMþPF6ꢀ:CH2Cl2, 1:1
v/v) were prepared by immersing room-temperature solutions into
liquid nitrogen in cuvettes with an optical path length of 0.5ꢀ
3 mm. These cuvettes were then placed in a liquid nitrogen-cooled
cryostat (Oxford Instruments), where any temperature between 77
and 300 K can be maintained by controlled heating. The optical
absorption spectra were measured on a Cary 5 (Varian) spec-
trometer. The samples mounted in the cryostat were irradiated with
4 μs electron pulses from an ELU-6 linear accelerator. The spectra
presented in the figures of this paper are invariably difference spectra
(after minus before irradiation).
•þ
show “local” transitions which are related to those of the monomer
radical ions M•þ but are usually shifted by some amount.
In the course of our work on the oxidation products of two
naturally occurring glucobrassicin metabolites, indole-3-carbinol
(I3C) and diindolylmethane (DIM),3 we noticed that in
samples of high concentration (in Ar matrices) and/or upon
annealing of matrices containing the radical cations a broad
NIR band appears, indicating that these compounds can
form dimer radical cations. In the case of DIM the question
arises whether that transition arises by intra- or intermolecular
complexation.
We have shown previously that ionic liquids, many of which form
stable transparent glasses at 77 K,6 can be used for generating radical
ions by radiolysis. Recently, we found that the glass quality does not
In this paper we present the absorption spectra of these dimer
radical cations, obtained by annealing frozen ionic liquid solutions
of parent indole (I, I3C, or DIM) irradiated with electron pulses or
X-irradiated Ar matrices (I, DIM). We assign these spectra in
terms of possible structures of the complex dimer radical cations
(I, I3C), and we discuss in particular the conditions under which
DIM forms a radical cation where the charge is localized in one
indole moiety or where it is delocalized over both of them. The
discussion is based on quantum chemical calculations by methods
which are able to account for the mix of covalent and nonbonding
interaction that prevail in dimer cations.
ꢀ
change upon mixing BMIMþPF6 with CH2Cl2, which has the
advantage that it improves the solubility of many precursors and leads
to a higher yield of radical cations on radiolysis due to its ability to
scavenge electrons by dissociative attachment.4
Quantum Chemical Calculations. The geometries of all
species were optimized first by the well-established B3LYP/
6-31G* method by which stationary points were also character-
ized by frequency calculations. Following that all geometries
were reoptimized by the double-hybrid dispersion-corrected B2-
PLYP-D density functional method,7,8 as implemented in the
Gaussian 09 suite of programs,9 using Dunning’s-type correla-
tion-consistent cc-pVDZ basis set.10,11 This method is designed
to account both for the covalent bonding as well as for mid- and
long-range correlation effects which contribute to the binding in
molecular complexes such as dimer cations. As analytic second
derivatives are not available yet for double-hybrid functional
methods, frequency calculations were not possible at the
B2-PLYP level. The π-dimer radical cations of I and I3C
were optimized using the BoysꢀBernardi counterpoise (CP)
method12,13 to correct for basis set superposition errors (BSSE).
Excited state energies were calculated by density functional-based
time-dependent linear response theory14 using the TD-B2-PLYP
method.
2. EXPERIMENTAL AND COMPUTATIONAL METHODS
Materials. Indole and its derivatives were obtained from Sigma/
Aldrich and purified by sublimation before use. The ionic liquid,
1-butyl-3-methylimidazolium hexafluorophosphate (BMIMþPF6ꢀ),4
was prepared according to literature procedures.
Matrix Isolation and Spectroscopy.5 Ground crystals of
indole (I) were placed in a U-tube attached to the inlet system
of an ATR closed cycle cryostat and carefully degassed. A 9:1
mixture of argon and nitrogen (where the latter is added to
improve the optical quality of matrices) was passed at ca. 0.5 Torr
through the U-tube held at 8 °C and slowly deposited on a CsI
window maintained at ca. 20 K. The much less volatile DIM was
sublimed from a double-walled glass tube fitted to the vacuum
shroud of the cryostat. The opening of the inner tube which faces
the cold CsI window is surrounded by four nozzles through
which Ar is directed to the same window to ensure efficient
mixing with the substance emanating from the heated tube.
To deposit DIM to a similar concentration to that of I in the
matrix, the inner glass tube was resistively heated to 122 °C.
Unfortunately, I3C could not be volatilized to a sufficiently high
concentration to form dimers (and hence dimer cations after
ionization) in the resulting matrices.
3. RESULTS AND DISCUSSION
3.1. Dimer Radical Cations of Indole, (I)2•þ, and Indole-3-
carbinol, (I3C)2•þ. Figure 1a shows the absorption spectrum of
ionized indole in a BMIMþPF6ꢀ/CH2Cl2 matrix. The intense
peak at 330 nm and the double-humped band at 560/605 nm
have been previously attributed to the radical cation of indole.3
Softening of the irradiated frozen solution by controlled increase
of the temperature to 120ꢀ140 K allows for slow diffusion to
take place. This leads to the decay of the absorption bands of I•þ
(Figure 1b and 1c), which is accompanied by the buildup of a
broad structureless band extending from 600 to 1400 nm and
To generate radical cations, the Ar/N2 mixture was doped with
about 1% of CH2Cl2, which acts as an electron scavenger. The
sample was exposed to 90 min of X-irradiation from a tungsten
7701
dx.doi.org/10.1021/jp201415z |J. Phys. Chem. A 2011, 115, 7700–7708

The Journal of Physical Chemistry A
ARTICLE
Figure 1. Spectrum obtained (a)ꢀon radiolysis of a 0.02 M solution of
I in a 1:1 mixture of BMIMþPF6 and CH2Cl2 at 77 K and (b) after
annealing at 120 K for 30 min, (c) 130 K for 30 min, (d) 135 K for
15 min, and (e) 135 K for 55 min; black bars represent results from TD-
DFT calculations for the most stable dimer radical cation of I (see text
for discussion).
Figure 2. Spectrum obtained (a) on radiolysis of a frozen 0.02 M
solution of I3C in BMIMþPF6ꢀ/CH2Cl2 at 77 K and (b) after annealing
at 160 K for 10 min, (c) 160 K for 30 min, (d) 160 K for 50 min, and (e)
165 K for 10 min; black and red bars represent results from TD-DFT
calculations for two the most stable dimer radical cation of I3C (see text
for discussion).
peaking at ca. 980 nm. After the absorptions of I•þ have decayed
completely (Figure 1e), a new band at 525 nm and a small peak at
365 nm can be discerned. Upon annealing to >140 K, all three
bands decay concomitantly. We attribute the spectrum in
Figure 1e to the dimer radical cation complex formed by
•þ
a neutral molecule of indole and its radical cation, (I)2
.
For I3C the sequence of events is similar: the absorption
spectrum of I3C in a BMIMþPF6ꢀ/CH2Cl2 matrix (Figure 2a)
shows also an intense peak at 330 nm and a wider, double-
humped band peaking at ca. 600 nm which we assigned to
I3C•þ.3 Upon annealing of the glass, first to 160 K and finally to
165 K, a decay of the absorption bands of I3C•þ is observed
(Figure 2bꢀe). At the same time a broad NIR band extending
from 600 to 1600 nm and peaking at ca. 1200 nm grows in. In
addition, we can discern a band peaking at 565 nm which may
already be contained in the spectrum obtained after radiolysis
or could have arisen on annealing (Figure 2e). We attribute the
spectrum in Figure 2e to the I3C dimer radical cation. The
concomitant decrease of these bands on annealing above 165 K
indicates that they belong to the same species.
Figure 3. (a) Spectrum of Iþ• obtained in Ar matrix (1:1000) at 10 K;
(b) the same at concentration of 1:200; (c) spectrum of I ionized in
BMIMþPF6ꢀ/CH2Cl2 after annealing at 135 K for 55 min.
To identify possible structures of the dimer cations, (I)2•þ and
(I3C)2•þ, we resorted to quantum chemical calculations. The
geometries of different dimer radical cation structures were
optimized by the B2-PLYP-D/cc-pVDZ method, accounting
for the BSSE correction (cf. Experimental and Computational
Methods section). The most stable of these structures (I)2•þ and
(I3C)2•þ are shown in Tables 1 and 2, respectively.
First, we note that these dimer cations are bound by up to
1 eV, which shows that forces much larger than those which
prevail in van der Waals complexes hold these species together.
For (I)2•þ we found that structure A, where the best overlap
between the two indole rings prevails, has the lowest energy.
In an effort to shed more light on these processes, we decided
to subject I to ionization in an Ar matrix at 10 K. If such a matrix
containing a high concentration of I is subjected to X-radiolysis, a
•þ
broad CR band due to (I)2 appears in the NIR next to the
absorptions of monomeric I•þ (see Figure 3b). Interestingly, this
CR band is broader and red shifted by about 0.23 eV compared to
that obtained on annealing an ionic liquid glass (cf. Figure 3c),
which may indicate that under these experimental conditions
•þ
other and perhaps more conformations of (I)2 are formed.
However, annealing of the Ar matrix to 35 K did not affect the
shape and position of this band.
7702
dx.doi.org/10.1021/jp201415z |J. Phys. Chem. A 2011, 115, 7700–7708

The Journal of Physical Chemistry A
ARTICLE
Table 1. Relative (Erel) and Binding (Ebind) Energies of
Different Structures of (I)2
Table 2. Relative (Erel) and Binding (Ebind) Energies of
Different Structures of (I3C)2
•þ
•þ
* Relative to the sum of energy of neutral I3C (ꢀ477.887461 Ha) and
I3C•þ(ꢀ477.625022 Ha).
* Relative to the sum of the energy of neutral I (ꢀ363.459928 Ha) and
I•þ (ꢀ363.187207 Ha).
great similarity to structure A of (I)2•þ, show good agreement
with the experimentally observed CR band. Nevertheless, it
cannot be excluded that upon annealing of ionic liquid glass we
•þ
also obtain some of structure B of (I3C)2
.
Interestingly, one of the most stable structures (D) is not of the
π-complex type but seems to be bound by H bonds. We also
did TD-DFT calculations for the NIR transitions of all five
•þ
The electronic structure of conformers A of (I)2 and
(I3C)2•þ is presented in Table 3 in terms of excitations involving
the molecular orbitals shown in Figure 4. As expected for both
(I)2 and (I3C)2 the NIR band corresponds to the first
excited state, which is a charge-resonance transition, attained
essentially by promotion of an electron from the HOMOꢀ1 to
the singly occupied MO (SOMO). These orbitals are composed
of the bonding and antibonding combinations of the singly
•þ
structures of (I)2 listed in Table 1, and we found that the
•þ
•þ
electronic transitions predicted for the most stable structure A
are in the best agreement with the experimental spectrum (see
black bars in Figure 1; other transitions are listed in the Sup-
porting Information).
The similarity of the spectra in Figures 1e and 2e indicates that
•þ
the structure of (I3C)2•þ may be akin to that of (I)2•þ, although
occupied MOs of monomeric radical cations of I for (I)2
and I3C for (I3C)2•þ, respectively. The band in the visible
region, which corresponds to the fourth excited state, is due a so-
called “local” transition related to those of the monomer radical
cations of I•þ and I3C•þ, respectively. Allowed transitions below
400 nm cannot be associated with leading excitations.
our calculations (Table 2) predict that a NꢀH O(H) bonded
3 3 3
structure (B) actually has the lowest energy. However, TD-DFT
calculations predict that this structure has only absorptions
below 600 nm (see red bars in Figure 2), whereas the predictions
for structure A of (I3C)2•þ (black bars in Figure 2), which shows
7703
dx.doi.org/10.1021/jp201415z |J. Phys. Chem. A 2011, 115, 7700–7708

The Journal of Physical Chemistry A
ARTICLE
Table 3. Excited States of Structures A of (I)2•þ (bold) and
(I3C)2•þ (italic) Calculated by TD-B2PLYPD/cc-pVDZ
energy
excited state eV
nm
oscillator strength
main excitation
first
1.10 1126
1.01 1229
1.27 978
1.39 891
1.70 730
1.80 688
2.31 536
2.30 538
3.17 391
3.13 396
3.19 388
3.14 395
3.47 357
3.43 362
3.71 334
3.74 331
0.1595
0.1542
0
61β f 62β 0.98948
77β f 78β 0.99046
60β f 62β 0.99408
76β f 78β 0.99288
59β f 62β 0.97374
75β f 78β 0.94981
58β f 62β 0.96791
74β f 78β 0.94052
highly mixed
second
third
0
0.001
0.0013
0.0703
0.0768
0.0258
0.0093
0.0011
0.019
0
fourth
fifth
highly mixed
sixth
highly mixed
highly mixed
seventh
eighth
57β f 62β 0.92114
73β f 78β 0.82567
highly mixed
0
0.004
0.0026
highly mixed
3.2. Radical Cations Derived from DIM. In orderꢀto dissolve
DIM to a concentration of 0.01 M in BMIMþPF6 the ionic
liquid has to be mixed again with an equal part of CH2Cl2. The
spectrum obtained after radiolysis of this solution at 77 K
(Figure 5a) shows a very similar double-humped band around
600 nm, as observed with I under the same conditions (cf.
Figure 1a). This band can readily and unequivocally be assigned
to a DIM radical cation (DIM•þ) where spin and charge are
localized in one of the indole moieties, while the other indole is
just a “spectator” substituent.
Upon annealing the sample, this spectrum changes in a way
which indicates formation of several secondary products, some of
which we previously identified,3 i.e., the red dye urorosein (the
band peaking at 490 nm in the spectrum in Figure 5e) and the
C-centered radical of DIM (assigned to a small peak at 420 nm).
A third species, which we did not discuss in the previous paper,
has a broad, structureless band in the NIR region characteristic
for charge-resonance (CR) transitions of dimer radical cations
(cf. section 3.1), and very similar to the band observed for the
dimer cation of I3C in BMIMþPF6ꢀ(cf. Figure 2).
This would seem to indicate that a similar species is formed,
i.e., that the indole moiety of an ionized DIM molecule which
carries the spin (and formally the charge) forms a complex with
one of the indole moieties of a nearby neutral DIM, with the
remaining two indole moieties serving as uninvolved substitu-
ents. However, we believe that this is not the case, because the
exact same NIR band appears on annealing independently of the
concentration of DIM (down to 5 mM, below which no bands
can be detected), whereas 20 mM solutions are needed to obtain
clearly discernible dimer cations on annealing of solutions of
ionized I and I3C. Also, one would expect the NIR band to be
present from the outset in concentrated solutions, if it is due to
intermolecular complex formation. Thus, we looked for alter-
native explanations.
•þ
Figure 4. KohnꢀSham molecular orbitals of (I)2 (numbers to the
•þ
left) and (I3C)2 (numbers to the right) involved in the electronic
transitions of these species (cf. Table 3).
and thus form an intramolecular dimer radical cation is not
unprecedented (a simple example is the radical cation of
norbornadiene15 where two olefins interact intramoleculary).
As the CH2 link is too short to allow for effective through-space
overlap of the π systems of the two indole moieties, this effect will
lead to a smaller splitting of their frontier MOs than in (I)2•þ. It is
therefore to be expected that the CR band in DIM•þ will be at
lower energy, which is indeed what is observed. However, what is
intriguing about the observations reported in Figure 5 is that a
localized radical cation (which has no CR transitions) seems to be
converted into a delocalized form (which has a strong CR
transition) on annealing, a rather exceptional phenomenon that
calls for an explanation.
As in the case of I we decided to ionize DIM also in an Ar
matrix at 10 K, which resulted in the spectrum shown in
Figure 6a. In contrast to the experiment in the ionic liquid glass,
this spectrum comes in two distinct parts, one of which is almost
identical to that of I•þ in Ar (cf. Figure 3a), whereas the other
consists of a very broad, CR-type NIR band. It is important to
note that, judging from the intensities of comparable IR bands,
the concentration of DIM in the Ar matrix was similar to that of
I before ionization, i.e., it was such that intermolecular dimer
cations are not preformed in the matrix (which is the case at much
higher concentrations, cf. Figure 3b). We therefore believe that the
NIR band of ionized DIM in Ar is not due to intermolecular
dimer cations, which are not formed under such conditions.
The occurrence of a CR band in a radical cation containing two
identical chromophores which may interact through space or via
orbitals of the bridging moiety (a CH2 group in the case of DIM)
However, this CR band peaks at significantly lower energy (λmax
1700 nm) than the one in ionic liquid glass (λmax ≈ 1300 nm)
≈
7704
dx.doi.org/10.1021/jp201415z |J. Phys. Chem. A 2011, 115, 7700–7708

The Journal of Physical Chemistry A
ARTICLE
Figure 6. (a) Spectrum of DIM ionized in an Ar matrix at 10 K and
(b) after annealing at 20 K for 15 min, (c) 23 K for 15 min, (d) 25 K for
15 min, and (e) 25 K for 15 min. The wiggles in the NIR are due to
interferences. (Inset) Spectrum of DIM (a0) and I (b0) ionized in Ar
matrix at 10 K.
Figure 5. Spectrum obtained (a) on radiolysis of a 0.01 M solution of
DIM in a 1:1 mixture of BMIMþPF6ꢀ and CH2Cl2 at 77 K and (b) after
annealing at 135 K for 10 min, (c) 135 K for 30 min, (d) 135 K for
40 min, and (e) 145 K for 10 min; black, red, and blue bars represent
results from TD-B2PLYP calculations on DIM radical cations (see text
for discussion).
correlating the movements of the electrons, HartreeꢀFock (or
CASSCF) theory invariably predicts localized radical cations to be
more stable than delocalized ones. Accounting for these correla-
tioneffects by means of many-body perturbation theory (MBPT),
such as in the popular MP2 (or CASPT2) method, fails because
the change on going from a fully localized to a fully delocalized
wave function is too large to be handled as a perturbation. On the
other hand, density functional theory artifactually overestimates
the stability of delocalized radical cations, because delocalization
minimizes the spurious self-interaction of the odd electron which
arises by the incomplete cancellation of the classical Coulomb and
the exchange energy for a single electron for all currently known
exchange functionals.16
In addition to these molecule-inherent effects, solvation will
favor localization of the partial charge in radical cations (which to
some degree accompanies spin localization) and thus create a bias in
favor of the localized form of ionized DIM. In view of all this we
cannot hope to make a reliable theoretical prediction of the
relative stability of the localized and delocalized forms of ionized
DIM, even in the gas phase, let alone in a condensed medium.
What we can, however, do is model the delocalized form of
ionized DIM by DFT. Thus, we explored the conformational
space of DIM and its radical cation by B2-PLYP-D/cc-pVDZ. In
Figure 7 the conformational minima of DIM and of its radical
cation are presented. The neutral has two equilibrium conforma-
tions of very similar energy, one with no symmetry, where the
planes of the two indole moieties are nearly perpendicular,
DIM(C1), and one of nearly C2 symmetry, DIM(C2).
and extends much further toward the IR range. Upon annealing
above 20 K, we first observe a concomitant decay of the sharp
band system at 600 nm and a low-energy tail (above 2000 nm)
while the CR transition sharpens and shifts to higher energies.
Subsequent annealing up to 25 K results in further narrowing and
blue shifting of the NIR band with no further decay of the 600 nm
band system.
As there can be no doubt that the 600 nm band system must be
assigned, as in the above case of the experiments in ionic liquid, to
a localized DIM radical cation and that the intense NIR band
belongs most probably to a delocalized form of the same cation
(because formation of intermolecular dimer cations can be
practically excluded at the concentrations which prevailed in
the Ar matrix), we are led to conclude that the two forms actually
coexist in Ar matrices. It thus seems that the energetic advantages
of localizing or delocalizing spin and charge, respectively, strike a
very delicate balance in the case of DIM•þ, a balance which can
tip one way or the other depending on subtle differences in
geometry or the environment.
In the ionic liquid the balance seems to be initially in favor of
the localized form, perhaps because this enjoys better solvation in
this polar solvent. However, the conformational freedom gained
by DIM•þ on annealing appears to allow this compound to attain
geometries where the delocalized form is more stable, as
indicated by the appearance of the NIR band, again at lower
energy than in Ar because in this solid conformational restric-
tions prevent complete relaxation even upon annealing.
On removing an electron from the HOMO, the vertically
formed radical cation of DIM(C1) relaxes to a structure which
has nearly Cs symmetry while that obtained by ionization of
DIM(C2) keeps its 2-fold axis of symmetry and relaxes to the
structure denoted C2-a in Figure 7. However, this structure is
protected by a barrier of only 2.85 kcal/mol from further
relaxation to structure C2-b, which lies ca. 1.28 kcal/mol lower
in energy. All the structures can be characterized by the two
CꢀCꢀCꢀC dihedral angles d1 and d2 around the bonds linking
the two indole moieties to the bridging CH2 group, which are
given in Figure 7.
4. THEORETICAL STUDIES ON LOCALIZED AND DELO-
CALIZED DIM•þ
Unfortunately, the balance between localized and delocalized
forms of radical cations such as DIM•þ cannot be assessed
unambiguously by current quantum chemical methods: as delo-
calization of the spin is “driven” by the energy gained upon
7705
dx.doi.org/10.1021/jp201415z |J. Phys. Chem. A 2011, 115, 7700–7708

The Journal of Physical Chemistry A
ARTICLE
Figure 8. MOs of DIM•þ involved in the electronic transitions of its
three conformers (cf. Table 4).
Figure 7. Equilibrium structures of DIM and DIM•þ (relative energies
in kcal/mol).
states of reactant conformers connect adiabatically to the excited
states of product conformers and vice versa.
This is illustrated in Figure 9, which shows that in both ground
states the SOMO is antibonding across the CH2 group but as one
twists the right indole moiety the SOMO becomes bonding, i.e., one
ends up in an excited state of the product radical cation. As
geometries along the reaction coordinate have no symmetry, this
“orbital following” can be avoided without breaking any symmetry.
However, in order to pass from a reactant to a product conformer
ground state surface, the coefficients on one of the indole moieties
must go to zero, so they can re-emerge with the opposite sign on the
product side. Physically, this corresponds to a complete localization
of the spin in one-half of the molecule at or near the transition state.
However, such a transition, from a species with delocalized
spin to one where the spin is localized in one-half of the molecule,
cannot be achieved by DFT or MBPT methods,17 so this
transition state cannot be located using currently available
methodology. On the basis of the numbers in Figure 9 (which
come from TD-DFT calculations)18 we predict that the energy of
TS (CS f C2-b) should be less than 10 kcal/mol above the
conformers of delocalized DIM•þ. Thus, it cannot be excluded
that the Cs and C2-b conformers interconvert to some extent
before the incipient radical cations are thermalized.
Even if we accept the above assignments, several questions
remain: (a) why does the broad CR band in the NIR only appear
after annealing in the ionic liquid glass, whereas it is present from
the outset in the Ar matrix spectrum; (b) what is the nature of the
low-energy tail in the NIR band, which disappears upon slightly
raising the temperature, in Ar, and (c) what is the origin of the ca.
0.14 eV shift of the CR transition of DIM•þ between annealed
ionic liquid and Ar (cf. Figure 10).
In order to answer these questions, we have to address the
issue of what happens on vertical ionization of the most stable
conformer of DIM, C1 in Figure 7. Surely, at this geometry the
spin will not be distributed evenly between the two indole
moieties and may in fact remain localized on one of the two
until the geometry has a chance to relax and at least approach that
of the species labeled Cs in Figure 7. If we assume that this
Table 4. Main Electronic Transitions Predicted by TD-
B2PLYP/cc-pVDZ for DIM•þ
energy
conformer
eV
nm
oscillator strength
main excitation
Cs
0.91 1357
0.39 3191
2.09 593
0.94 1313
0.1976
0.0964
0.0517
0.2485
64β f 65β 0.98111
64β f 65β 0.99676
61β f 65β 0.97324
64β f 65β 0.97681
C2-a
C2-b
In the C2 and the Cs conformations of DIM•þ charge and spin
are of course fully delocalized over both indole moieties. For both
the Cs and the C2-b conformations of the radical cation, TD-
B2PLYP-D calculations predict an intense CR-type transition
around 1300 nm (cf. Table 4), arising from HOMOꢀ1 to
SOMO (64 f 65) excitation (see MOs in Figure 8). These
predictions are in very good agreement with the observed NIR
band obtained upon annealing in ionic liquid glass (cf. Figure 5).
In contrast, for the initially formed C2-a conformer a much
weaker CR transition is predicted at ca. 3000 nm. The reduced
splitting of the two highest MOs is due to an inversion of the
“natural sequence” of orbitals (the bonding combination of
indole HOMOs lies above the antibonding one, see Figure 8)
as a consequence of strong through-bond interaction involving
pseudo-π MOs of the bridging CH2 group.
In order to localize transition states between all three con-
formers of DIM•þ we generated a 3D potential energy surface by
scanning the two CꢀCꢀCꢀC dihedral angles d1 and d2 (see
Figure S2 in the Supporting Information). This surface indicates
where the transition states are expected to lie, and we had indeed
no problem locating that for the interconversion of the two C2
conformers, shown in Figure 7 as TS(C2-a f C2-b). In the search
for the transition states linking the Cs to either of the two C2
conformers we ran, however, into a problem in that the ground
7706
dx.doi.org/10.1021/jp201415z |J. Phys. Chem. A 2011, 115, 7700–7708

The Journal of Physical Chemistry A
ARTICLE
Figure 9. Apparent state crossing between the C2-b and the Cs
conformers of DIM•þ; numbers are based on (TD)B2PLYPD calcula-
tions (the two ground states are within 0.25 kcal/mol).
Figure 11. Processes that occur in Ar matrix and ionic liquid glass upon
radiolysis and further annealing.
to get DIM into the gas phase for Ar matrix experiments,
the material must be heated to ca. 120° where ΔGcalcd,395K
=
1.01 kcal/mol, K = 0.21, i.e., 27% of the DIM is present as
conformer C2. As vertical ionization of this conformer leads
already to a symmetric (i.e., delocalized) radical cation, this may
explain the presence of a CR band in the Ar matrix experiments
right after ionization of DIM (if ΔH is larger than the calculated
value, the effect will be more pronounced).
The presence of the thermally labile low-energy tail of the NIR
transition in Ar may be rationalized by the initial presence after
ionization of some DIM•þ in the metastable conformation C2-a
which has a CR transition at much lower energy than the other
conformations (cf. Table 4). Slight annealing will permit relaxation of
this to the more stable conformation C2-b, which is formed directly
on annealing the ionic liquid glass to 160 K. Alternatively, one may
argue that the transition between the localized C1 and the delocalized
Cs conformation is not a stepwise one but that intermediary, partially
localized conformations may be attained and initially trapped in Ar
(but not in the ionic liquid glass). These conformations are expected
to have CR/CT transitions at lower energy than the symmetric, fully
relaxed ones. Our experimental data do not allow us to distinguish
between these two possible interpretations.
Finally, we propose that the shift of the CR band between the
ionic liquid and Ar may be due to one or both of two effects: (a)
the presence of substantial amounts of conformer C2-b of
DIM•þ, which may have a lower energy CR transition than
conformer Cs, in the Ar matrix experiments (see above) and/or
(b) enhanced stabilization of the CR excited state of DIM•þ by
the highly polarizable Ar environment.
Figure 11 sums up the processes that we believe to occur upon
ionization of DIM in an ionic liquid glass and in Ar, respectively,
based on the experimental spectra and DFT calculations.
Figure 10. (a) Spectrum obtained onꢀradiolysis of a 0.01 M solution of
DIM in a 1:1 mixture of BMIMþPF6 and CH2Cl2 after annealing at
145 K for 10 min; (b) spectrum of DIM ionized in Ar matrix after
annealing at 25 K for 15 min. Note that the peak at 19 000 cmꢀ1 is due to
the urorosein cation.3
relaxation, which involves a 70° rotation of one of the indole
moieties, is inhibited in solid matrices and can only take place
upon their gradual annealing, this may explain why the initial
spectrum obtained upon ionization of DIM in an ionic liquid
glass and a part of the initial spectrum obtained in Ar has the
characteristics of a radical cation where only one of the indole
moieties is ionized while the other is acting as a “spectator”
substituent.19 Upon annealing, that spectrum would be gradually
replaced by one of a species with delocalized spin and charge, i.e.,
the CR band that is characteristic for such species.
So why is this CR band present from the outset in Ar but arises
only after annealing the ionic liquid? An answer to this question
might be found in the thermochemistry of the equilibrium
between the two conformers of neutral DIM (ΔHcalcd = þ
0.26 kcal/mol, ΔScalcd = ꢀ1.91 cal Kꢀ1 molꢀ1 for C1 f C2).
In the ionic liquid experiment, the equilibrium will be rapidly
established, at least during theinitial phasesof the cooling process.
At 77 K, where ΔGcalc,77K = 0.41 and K = 0.07 (assuming that ΔH
and ΔS do not change between 298 and 77 K), 93% of the DIM is
present as conformer C1, i.e., the localized radical cation is
expected to predominate largely before its relaxation (and con-
comitant delocalization) to the Cs structure. In contrast, in order
7707
dx.doi.org/10.1021/jp201415z |J. Phys. Chem. A 2011, 115, 7700–7708

The Journal of Physical Chemistry A
ARTICLE
5. CONCLUSIONS
(9) Frisch, M. J.; Trucks, W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
M. A.; Cheeseman, J. R.; Montgomery, J. J. A.; Vreven, T.; Kudin, K. N.
Gaussian 09, Rev. A02; Gaussian, Inc.: Wallingford, CT, 2009.
(10) Dunning, T. H., Jr. J. Chem. Phys. 1989, 90, 1007.
(11) Kendall, R. A.; Dunning, T. H., Jr.; Harrison, R. J. J. Chem. Phys.
1992, 96, 6796.
(12) Boys, S. F.; Bernardi, F. Mol. Phys. 1970, 19, 553.
(13) Simon, S.; Duran, M.; Dannenberg, J. J. J. Chem. Phys. 1996,
105, 11024.
(14) Casida, M. E. In Recent Advances in Density Functional Methods,
Part I; Chong, D. P., Ed.; World Scientific Publ. Co.: Singapore, 1995.
(15) Nelsen, S. F.; Trieber, D. A.; Wolff, J. J.; Powell, D. R.; Rogers-
Crowley, S. J. Am. Chem. Soc. 1997, 119, 6873.
(16) Lundberg, M.; Siegbahn, P. E. J. Chem. Phys. 2005, 122, 224103.
(17) M€uller, B.; Bally, T.; Gerson, F.; de Meijere, A.; von Seebach, M.
J. Am. Chem. Soc. 2003, 125, 13776.
(18) Calculations indicate that the rotational barrier around the
indoleꢀCH2 bond in DIM•þ and hence the driving force for relaxation
of the C1 structure—once the bond lengths and angles have adopted the
optimal values—is quite small. Unfortunately, it is difficult to arrive at
reasonable estimates for this quantity by calculations because the
transition from a localized to a delocalized radical cation that accom-
panies this relaxation cannot be modelled reliably by any method that
can be applied to a molecule the size of DIM.
(19) In the parlance of mixed valence compounds, species such as
the localized DIM radical cation would be called a RobinꢀDay class I
and the delocalized one a class III system. In spite of the fact that
electronic excitation of such delocalized systems involves no net charge
transfer whatsoever, the CR transitions in such systems are called
“intervalence charge-transfer” (IVCT) absorptions which lends to some
confusion.
We investigated the radical cations obtained on pulse radi-
olysis of frozen ionic liquid solutions of parent indole (I), indole-
3-carbinol (I3C), and diindolylmethane (DIM) or on X-irradia-
tion of Ar matrices of I and DIM, respectively. On annealing or at
high concentration, I and I3C form dimer radical cations which
distinguish themselves by intense charge-resonance (CR) bands
in the NIR range which were assigned on the basis of quantum
chemical calculations. The behavior of DIM on ionization is
more complicated: initially it forms a radical cation where the
spin is localized in one of the two indole moieties that are
connected by a CH2 group. On annealing, this species is
gradually transformed into one where spin and charge are
symmetrically delocalized over both indole moieties and which
shows a CR band similar to that of the dimer cations of I and I3C.
Due to the increased presence of a higher lying symmetric
conformer of neutral DIM prior to ionization in the Ar matrix
experiments, a CR band is already seen right after ionization.
Thus, DIM is one of the rare cases where a localized and a
delocalized radical cation may coexist. The subtleties of the DIM
radical cation are discussed on the basis of quantum chemical
calculations of the potential energy surfaces and the excited
states.
’ ASSOCIATED CONTENT
S
Supporting Information. Pdf file containing (a) the
b
complete citation of the Gaussian program, (b) a table and a figure
showing the excited states of conformer B of (I3C)2•þ, (c) a table
with the equilibrium conformations of DIM and its radical cation,
(d) the B3LYP/6-31G* potential energy surface generated by
scanning the two dihedral angles d1 and d2 in DIM•þ; separate
text file containing the geometries and energies of all stationary
points mentioned in the paper and detailed results of excited
state calculations. This material is available free of charge via the
Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION
Corresponding Author
*E-mail: Thomas.Bally@unifr.ch.
’ ACKNOWLEDGMENT
This work is part of project no. 200020-121747 of the Swiss
National Science Foundation. A.M. acknowledges support from
the Polish Ministry of Science and Higher Education (no. N205
003 32/0258).
’ REFERENCES
(1) Badger, B.; Brocklehurst, B. Trans. Faraday Soc. 1969, 65, 2576.
(2) Badger, B.; Brocklehurst, B. Trans. Faraday Soc. 1970, 66, 2939.
(3) Bzoch-Mechkour, A.; Bally, T.; Sikora, A.; Michalski, R.; Marcinek,
A.; Gebicki, J. J. Phys. Chem. A 2010, 114, 6787.
(4) Czerwinska, M.; Sikora, A.; Szajerski, P.; Zielonka, J.; Adamus, J.;
Marcinek, A.; Bednarek, P.; Bally, T. J. Org. Chem. 2000, 71, 5312.
(5) Bally, T. In Reactive Intermediate Chemistry; Moss, R. A., Platz,
M. S., Jones, M., Eds.; Wiley & Sons: New York, 2004.
(6) Marcinek, A.; Zielonka, J.; Gebicki, J.; Gordon, C. M.; Dunkin,
I. R. J. Phys. Chem. A 2001, 105, 9305.
(7) Grimme, S. J. Chem. Phys. 2006, 124, 034108.
(8) Schwabe, T.; Grimme, S. Phys. Chem. Chem. Phys. 2007, 9, 3397.
7708
dx.doi.org/10.1021/jp201415z |J. Phys. Chem. A 2011, 115, 7700–7708