Imp D1cp01738a

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

PCCP

View Article Online


PAPER View Journal | View Issue
Published on 28 June 2021. Downloaded by S.V. National Institute of Technology on 10/18/2022 10:59:05 AM.

The complex between molecular oxygen and an


organic molecule: modeling optical transitions to
Cite this: Phys. Chem. Chem. Phys.,
2021, 23, 15038 the intermolecular charge-transfer state†
Frederik Thorning, Kris Strunge, Frank Jensen * and Peter R. Ogilby *

The collision complex between the ground electronic state of an organic molecule, M, and ground state
oxygen, O2(X3Sg), can absorb light to produce an intermolecular charge transfer (CT) state, often
represented simply as the M radical cation, M+ , paired with the superoxide radical anion, O2 . Aspects
of this transition have been the subject of numerous studies for B70 years, many of which address
fundamental concepts in chemistry and physics. We now examine the extent to which the combination
of Molecular Dynamics simulations and electronic structure response methods can model transitions to
the toluene–O2 CT state. To account for the experimental spectra, we consider (a) the distribution of
toluene–O2 geometries that contribute to the transitions, (b) a quantitative description of intermolecular
CT, and (c) oxygen-induced local transitions in toluene that complement the CT transitions, specifically
transitions that populate toluene triplet states. We find that the latter oxygen-induced local transitions
Received 21st April 2021, play a prominent role on the long wavelength side of the spectrum commonly attributed to the
Accepted 28th June 2021 intermolecular CT transition. Our calculations provide a new perspective on the seminal discussion
DOI: 10.1039/d1cp01738a between R. S. Mulliken and D. F. Evans on the nature of O2-dependent transitions in organic molecules,
and bode well for modeling transitions to excited states with CT character in noncovalent weakly-
rsc.li/pccp bonded molecular complexes.

Introduction the O2(a1Dg) excitation energy, continue to provide a unique


exercise in the study of intermolecular electronic-to-vibrational
Interactions between an organic molecule, M, and molecular energy transfer.13,14
oxygen have played a key role in the study of noncovalent Studies of the interaction between 1M0 and O2(X3Sg) are
complexes for B70 years.1–4 This work has complemented challenging because this ground state complex is very weakly
related studies on noncovalent interactions between benzene, bound,15–17 which is consistent with Mulliken’s moniker of a
in particular, and molecular iodine.1,3,5 ‘‘collision complex’’.18 Nevertheless, such studies are facilitated
Much of the work on M and O2 has focused on cases where by a distinctive optical transition, first observed as a red shift in
one of the interacting partners is in an excited electronic the onset of light absorption in liquid solutions of M upon
state.2,6,7 For example, the quenching of a triplet state organic the introduction of O2.18–20 After much discussion, a key
molecule, 3M1, by ground state oxygen, O2(X3Sg), with energy component of this absorption was assigned to a transition
transfer to produce singlet molecular oxygen, O2(a1Dg), has from a 1M0–O2(X3Sg) collision complex to an intermolecular
provided insight into a variety of fundamental chemical and charge transfer, CT, state.18 Although the latter is often represented
physical principles (e.g., factors within the M–O2 complex that as a pair of radical ions, M+ O2 , implying complete electron
influence changes in electron spin angular momentum).6,8–12 transfer (Fig. 1), it is understood that a description involving
Likewise, ascertaining the roles played by the ground state of an partial charge transfer may be more appropriate.1,18 In some
organic molecule, 1M0, to induce the forbidden O2(a1Dg) - cases, the absorption spectrum due to 1M0 alone can be
O2(X3Sg) transition and then, independently, act as a sink for subtracted to indicate that the M–O2 CT transition is a broad
structureless band,15,17,21 which is consistent with a weakly
Chemistry Department, Aarhus University, DK-8000, Aarhus, Denmark. bound ground state complex to which many different 1M0 and
E-mail: frj@chem.au.dk, progilby@chem.au.dk
O2(X3Sg) orientations contribute. The effective concentration
† Electronic supplementary information (ESI) available: Details and results from
TD-DFT calculations; parameters for the AMOEBA polarizable force field; details
of this complex, which influences the intensity of the CT
of the MD simulation; MD-based spectra calculated using different sampling transition, is reflected in the magnitude of the equilibrium
methods. See DOI: 10.1039/d1cp01738a constant for M–O2 formation from M and O2.

15038 | Phys. Chem. Chem. Phys., 2021, 23, 15038–15048 This journal is © the Owner Societies 2021
View Article Online

Paper PCCP

the M–O2 complex.20,26–28 The production of O2(a1Dg) upon


irradiation into the M–O2 CT band has been of particular
interest in this regard (Fig. 1).20,26,29 This latter approach to
produce O2(a1Dg) complements the 3M1-sensitized production
of O2(a1Dg), and provides mechanistic insight for studies of
transitions between different states of the M–O2 complex.26,27
Published on 28 June 2021. Downloaded by S.V. National Institute of Technology on 10/18/2022 10:59:05 AM.

The available data are all consistent with the long-held under-
standing that mixing of the M–O2 CT state with other states of
the M–O2 complex plays a significant role in the photophysics of
oxygenated systems.2,6,7,9,10,26,30 For example, state mixing between
the CT state and the comparatively low-energy 1M0–O2(a1Dg) and
1
M0–O2(X3Sg) states facilitates the formally-forbidden transition
between O2(a1Dg) and O2(X3Sg).6,14,31,32
We have shown that the M–O2 CT state can also be populated
via two-photon absorption using visible light, which likewise
results in the production of an appreciable amount of O2(a1Dg)
(Fig. 1).29 Although this observation has ramifications for a
range of experimental studies, it also provides useful fodder
Fig. 1 Approximate relative energies of states in the complex between
for theoretical and computational studies of optical transitions
oxygen and an organic molecule xMy (the superscript x denotes electron
spin and the subscript y denotes either the ground state, 0, or the first in noncovalent complexes.
excited state, 1). Both spin-allowed one- and two-photon excitation to the Computational modeling of linear and nonlinear optical transi-
non-equilibrated Franck–Condon, FC, triplet CT state are shown (blue and tions has received much attention over the years.33 Appreciable
red arrows, respectively). The energy of the CT state depends on the advantages are realized with the use of response methods, which
ionization potential of M (i.e., the stability of the M radical cation). In liquids,
avoid sum-over-states approaches and errors that can arise if
the energy of the solvent-equilibrated, SE, CT state also depends on
the dielectric properties of M (gray bar). The green arrows illustrate one mixing with a particular excited state is not considered.34–37
non-radiative process that results in the formation of O2(a1Dg). Response methods implicitly consider all excited states, with
the principal requirement being an accurate ground state wave
function.34,38 Response methods based on Time Dependent
Other transitions in the M–O2 complex can play an appreci- Density Functional Theory (TD-DFT) are popular due to their
able role in the same spectral region as the M–O2 CT transition. computational efficiency,39 while corresponding methods based
These transitions reflect the way one of the interacting partners on coupled cluster reference wave functions provide a hierarchy
can perturb formally forbidden local transitions in the other of systematically more accurate results.2,37,40–42
partner. Thus, perturbation by M can increase the probability of When dealing with transitions to states with significant CT
the forbidden UV Herzberg transitions that populate highly character, TD-DFT often significantly underestimates excitation
excited states of oxygen (i.e., A3Su+, c1Su, and A 0 3Du).4,22 energies due to the charge being over-delocalized by typical
However, this phenomenon is perhaps most visibly exemplified exchange–correlation functionals,34,43 and calculated absorption
in the spectrum of liquid benzene where O2-induced 1M0 - 3M1 cross sections can be erroneous due to an incorrect assessment of
vibronic bands are clearly superimposed onto the broad CT the excited state dipole moment.40 One approach to address this
band.18,20,23,24 Distinguishing between these different transitions problem involves using so-called range-separated functionals
was a key issue in the seminal discussion between R. S. Mulliken where the DFT exchange functional is gradually modified to
and D. F. Evans on the nature of O2-dependent transitions in Hartree–Fock (HF) exchange as the interelectronic distance
organic molecules,18,25 with Mulliken focusing on the role of the increases.44 The interpolation between these two extremes is
CT state and Evans on the magnetic and spin properties of usually done by the error function with a range-separation
O2(X3Sg). parameter o that determines the distance where the transition
For common organic molecules, the ground state M–O2 from DFT exchange to HF exchange occurs.44,45 The o para-
complex has a net triplet spin reflecting the combination of a meter is often optimized to reproduce reference data for intra-
singlet organic component, 1M0, and triplet oxygen, O2(X3Sg). molecular CT states.45,46
In contrast, the combination of two triplet states in the Computational modeling of excitation energies in inter-
3
M1–O2(X3Sg) complex yields singlet, triplet, and quintet spin molecular complexes with CT character is a problem that is
states, which are nearly degenerate in the weakly interacting currently being addressed on a number of fronts.43,47–51 In the
limit. In this way, the oxygen-induced 1M0 - 3M1 transition is present study, we employ TD-DFT and Equation-of-Motion
readily seen as a spin-allowed process between M–O2 triplet Coupled-Cluster (EOM-CCSD)41 methods to model transitions
states. The Franck–Condon CT state will likewise have triplet to the CT state in the M–O2 system (Fig. 1), using toluene as the
spin. Subsequent intersystem crossing to a singlet CT state molecule M. This problem is not only influenced by the range-
coupled with nuclear relaxation (i.e., a solvent-equilibrated separation parameter o in the TD-DFT approach, but one must
state in liquids) can precede the formation of other states of also account for the shallow 1M0–O2(X3Sg) potential surface.

This journal is © the Owner Societies 2021 Phys. Chem. Chem. Phys., 2021, 23, 15038–15048 | 15039
View Article Online

PCCP Paper

The latter implies that transitions from a plethora of 1M0 and


O2(X3Sg) geometries must be considered and suitably averaged.
Success with these calculations, as judged by comparison to
experimental data, allows us to comment on the magnitude of
the equilibrium constant for the formation of the 1M0–O2(X3Sg)
‘‘collision’’ complex, a problem that has likewise challenged the
community for many years.15–18
Published on 28 June 2021. Downloaded by S.V. National Institute of Technology on 10/18/2022 10:59:05 AM.

Results and discussion


M–O2 CT absorption spectrum
Experimental data. To our knowledge, an M–O2 CT spectrum
has only been recorded for the benzene–O2 complex in the gas
phase, achieved by subtracting the spectrum of benzene itself
from the data recorded under oxygenated conditions.15,17,21
The resultant broad and structureless band has a maximum
Fig. 2 Four geometry-optimized structures of the ground state toluene–
at B215 nm.15,17,21 Analogous data obtained from frozen O2 complex used for preliminary calculations of the M–O2 CT spectrum.
matrices at 10 K yield a band maximum at B235 nm for the
benzene–O2 CT band and B245 nm for the toluene–O2 CT
band.52 Although the observed red shift in going from benzene the B215 nm expected from experimental data. If we assume
to toluene is arguably consistent with the expected change in that the toluene–O2 structures in Fig. 2 are representative of
the energy of the CT state (i.e., the toluene radical cation is ground state collision complexes, then this deviation from the
more stable), the band maxima of these low temperature data experimental data suggests limitations of the DFT model for
likely reflect discrete M and O2 orientations and intermolecular intermolecular charge-transfer excitations (vide supra).
distances that are only part of the distribution of M and O2 For electronic transitions in a non-centrosymmetric system,
geometries that characterize gas and liquid phase data. such as the M–O2 complex, selection rules indicate that the final
In liquid solutions of M, absorption attributed to the M–O2 states populated in a one-photon process should also be populated
CT transition appears as a bathochromic shift to the absorption in a two-photon process.57 Assuming that the abovementioned
onset of M itself (i.e., one observes a long-wavelength ‘‘tail’’ on limitation of DFT will apply equally to one- and two-photon
the absorption spectrum of M).18,20,26 Measuring the full CT calculations, we indeed observe that the respective spectra are
spectrum in this case is challenging since local transitions of the same (Fig. S2, ESI†). Limiting ourselves to the electric dipole
the highly concentrated M dominate and generally saturate the term of the multipole expansion, we calculate a one-photon molar
spectrometer, prohibiting subtraction to yield the CT band. To absorption coefficient of B2000–4000 M1 cm1 and a two-
our knowledge, the M–O2 CT spectrum in liquid M has never photon cross section of B5–8 GM, both at the maxima of the
been recorded in the same manner as the gas phase spectrum respective toluene–O2 absorption bands.
discussed above. Available liquid phase data for the toluene–O2
system only reveal the tail of the absorption spectrum, which Tools to accurately and efficiently model the toluene–O2 CT
extends out to wavelengths as long as B370 nm.20 transition
On the basis of these data, one criterion of computational Tuning the range-separation parameter x. The preliminary
success is to model a broad toluene–O2 CT absorption band calculations described above suggest that modification of the
with a maximum at B215–220 nm and a tail that extends out to range-separation parameter o may be required for modeling
B370 nm, with the added goal of providing molecular insight intermolecular CT transitions.
into the origin of this spectrum. Kozma, et al.58 have investigated the performance of differ-
Preliminary calculations: structures and energies. As a starting ent coupled cluster methods for predicting excitation energies
point, and inspired by previous work on related systems,12,30,53–55 of intermolecular CT transitions in complexes of two organic
we chose a few discrete geometries for the toluene–O2 complex. molecules. Taking the CCSDT-3/cc-pVDZ results of Kozma,
Energy optimization of these geometries at the oB97X-D/6-31G(d) et al. as our benchmark, we calculated CT excitation energies
level yielded the four structures shown in Fig. 2. Using TD-DFT for 10 of their intermolecular transitions using TD-DFT with the
with the oB97X-D range-separated functional optimized for oB97X-D functional and the aug-pcseg-259 basis set. For these
intramolecular CT excitations (o = 0.2 bohr1),46,56 we ascertained 10 transitions, using the standard value of o optimized for
that the lowest energy toluene–O2 CT transition at B310 nm was intramolecular CT (0.2 bohr1) yields a mean absolute deviation
associated with structure B in Fig. 2 (see ESI† for details). Although of B2 eV between the coupled cluster and DFT results, a value
the CT transition wavelengths for the other 3 structures in that far exceeds what is acceptable for computational assess-
Fig. 2 (B250–260 nm) are blue-shifted relative to this value of ment of a CT transition.46 Fig. 3 shows the mean absolute
310 nm, they nevertheless still occur at wavelengths longer than deviation for these 10 CT transitions as a function of different

15040 | Phys. Chem. Chem. Phys., 2021, 23, 15038–15048 This journal is © the Owner Societies 2021
View Article Online

Paper PCCP

expected rotational correlation times and radial diffusion


distances,71,72 coupled with the anticipated short lifetime of
the collision complex,18 all indicate that structures separated by
100 ps should not be correlated, leading to 100 independent
snapshots extracted from the 10 ns simulation.
We used three approaches to analyze the simulation:
Published on 28 June 2021. Downloaded by S.V. National Institute of Technology on 10/18/2022 10:59:05 AM.

(a) extracting a snapshot every 100 ps, we identified the one


toluene molecule that is closest to oxygen (i.e., the shortest
distance between an atom in toluene and one of the oxygen
atoms), leaving us with 100 different 1 : 1 complexes arguably
representatively placed on the toluene–O2 potential surface.
(b) Using the same 100 snapshots, we identified the two toluene
Fig. 3 Plot of the mean absolute deviation between intermolecular CT molecules that were closest to oxygen, leaving us with 100
state energies for 10 different CT states between two organic molecules different complexes with 2 : 1 stoichiometry. (c) For snapshots
computed using a high-level coupled cluster approach (CCSDT-3/cc-pVDZ)58 separated by 1 ps, we calculated the distance between the O2
and a TD-DFT approach (oB97X-D/aug-pcseg-2) as a function of the range-
center-of-mass and the toluene center-of-mass for every toluene
separation parameter o. The data show that a range separation parameter of
0.5 bohr1 is most appropriate for use when dealing with intermolecular CT
molecule (com-to-com distance), and selected the 1 : 1 complex
states. The error bars represent the standard deviation on the mean. with the shortest com-to-com distance. For a 10 ns simulation
window, this yields a distribution of com-to-com distances for
9900 structures that covers the range B3–6 Å. For each 100 ps
values of o, covering the range from 0.1 to 1.0 bohr1. Increasing window, we then selected one snapshot such that the total
o to 0.5 bohr1 reduces the deviation to 0.4 eV, which is distribution of com-to-com distances for 9900 structures was
comparable to the accuracy for valence excitations,60–63 and this reproduced as well as possible using 99 structures. With any of
value may be suitable for modelling intermolecular CT states. these sampling methods, if oxygen is situated near the edge of the
Kozma, et al.58 considered complexes of organic molecules simulated box, the nearest toluene molecule might be located in
with ground and CT states that have singlet spin, whereas the M–O2 the adjacent box. We account for this by redefining the boundaries
complex has ground and Franck–Condon CT states with net triplet of the box such that oxygen is placed in the center.
spin (Fig. 1). Because DFT often has problems calculating accurate Details of these sampling approaches are given in the ESI.†
energy differences of different spin states, it does not automatically
follow that the calibration in Fig. 3 carries over to the M–O2 system. Modeling M–O2 CT excitations
Thus, as outlined further below, we have chosen to compare TD-DFT computations. For each of the three sampling methods
TD-DFT results with those obtained using EOM-CCSD methods. used to assess the MD simulation, we calculated excitation
MD simulation for the M–O2 ground state. Rather than energies and the associated oscillator strengths for the toluene–
working with a given structure for the toluene–O2 ground state O2 structures using TD-DFT with the oB97X-D functional and the
complex, albeit with precedence and energy-optimization (vide range-separation factor of 0.5 bohr1 using the aug-pcseg-1 basis
supra), the shallow ground state potential surface implies that a set.59 Each of these calculations resulted in a spectrum showing
large number of toluene–O2 geometries will contribute to both discrete transitions (i.e., a ‘‘stick’’ spectrum, see ESI†). A similar
CT transitions and to O2-perturbed local transitions in toluene computation was performed for toluene in the absence of oxygen,
at ambient temperatures. Investigating these geometry effects using the molecular geometry in the complex. To emulate
requires a statistical sampling of possible structures and we experimental results, we generated structureless band spectra
have chosen to rely on Molecular Dynamics (MD) simulations by broadening each discrete transition with a Gaussian function
to capture a more statistically accurate picture of the ground (s width of 1000 cm1) and adding the results for all 100
state toluene–O2 structures. snapshots. We then subtracted the toluene spectrum from the
Although force field parameters for toluene are available,64 spectrum of oxygenated toluene to show the behavior unique to
parameters for molecular oxygen, to our knowledge, are not. With the toluene–O2 complex. Irrespective of the sampling method
the intent of using the AMOEBA Polarizable Force Field,64 we used to assess the MD simulation, the resultant spectra show an
derived parameters from information on the O–O bond length absorption band maximum of B215 nm (Fig. 4) which appears
and vibrational frequency,65 van der Waals interaction,66 the static to meet our criterion of computational success based on the
molecular polarizability,67 and the molecular quadrupole moment68 expectation from experimental data (vide supra). The observation
(details are provided in the ESI†). We generated a cubic simulation that complexes of 1 : 1 and 2 : 1 stoichiometry yield essentially the
box with 100 toluene molecules and one oxygen molecule using same result indicates that the molecule closest to oxygen is most
Packmol69 and used the Tinker 8 software package70 to perform a important in defining the M–O2 spectrum. This conclusion should
10 ns MD simulation under NVT conditions (see ESI† for details). facilitate attempts to correlate gas and solution phase data.
To establish a statistically relevant number of independent Weak point in the TD-DFT computations. Irrespective of the
toluene–O2 structures, we argue that a ‘‘snapshot’’ of the toluene– MD-derived toluene–O2 geometries used, our TD-DFT computations
O2 ensemble taken every 100 ps should be sufficient. Specifically, always revealed a separate group of appreciably red-shifted

This journal is © the Owner Societies 2021 Phys. Chem. Chem. Phys., 2021, 23, 15038–15048 | 15041
View Article Online

PCCP Paper
Published on 28 June 2021. Downloaded by S.V. National Institute of Technology on 10/18/2022 10:59:05 AM.

Fig. 4 TD-DFT computed absorption spectrum for the toluene–O2 complex


obtained by subtracting the computed spectrum for toluene from the
computed spectrum for toluene containing oxygen (see text). Computations
were performed using a range-separation parameter o of 0.5 bohr1 and the
data shown here were obtained for a 1 : 1 stoichiometry of toluene to oxygen.
The band maximum occurs at B215 nm. The inset shows an expanded view
of the transitions over the range 350–700 nm, with a band maximum at
B500 nm (note the change of intensity scale). Corresponding spectra
obtained when using the other two methods to assess the MD simulation
are similar and are shown in the ESI.†

transitions (B500 nm) with low oscillator strengths (o0.0001,


see inset of Fig. 4 and Fig. S4, ESI†). This part of the calculated
Fig. 5 (A) Coupled cluster computed spectrum for the toluene–O2
spectrum is readily assigned to local transitions in toluene that complex obtained using a 1 : 1 stoichiometry of toluene to oxygen with the
are perturbed by oxygen, in particular 1M0 - 3M1 transitions. same structures used for the calculation in Fig. 4. The distinct grouping of the
In our nonrelativistic approximation, these singlet–triplet red-shifted transitions, amplified in the insets, reflects oxygen-induced
transitions in isolated toluene are forbidden. However, as transitions to the local triplet states of toluene: 3M1 and 3M2 (see text).
(B) For discrete EOM-CCSD computed transitions in the toluene–O2
mentioned earlier, a transition from the triplet ground state
complex (see text), the plot indicates the extent of charge displacement.
complex, 3(1M0–O2(X3Sg)), to the triplet component of the Many transitions over the range B200–230 nm have appreciable charge
( M1–O2(X3Sg)) excited state is allowed. From experimental
1,3,5 3
displacement which is consistent with intermolecular CT, whereas the
data, the 1M0 (v = 0) - 3M1 (v = 0) transition in toluene occurs transitions at longer wavelengths are consistent with localized oxygen-
at 346 nm,73 which is appreciably blue-shifted from the induced transitions in toluene in which little charge displacement occurs.
B500 nm band shown in the inset of Fig. 4 obtained using
the TD-DFT calculations.
To help assess this observation, we performed coupled cluster more intense than the band assigned to the 1M0 - 3M1
calculations on the MD-derived 1 : 1 toluene–O2 structures (EOM- transition. In part, this could reflect the reasonable expectation
CCSD/aug-cc-pVDZ). In contrast to the TD-DFT calculations, the that the higher energy 1,3,5(3M2–O2(X3Sg)) states will more
EOM-CCSD results do not place these transitions at such low readily mix with both the M–O2 CT state and singlet excited
energies (Fig. 5A). Rather, they better represent experimental data states of toluene to which transitions are more allowed.
where a ‘‘tail’’ extends out to B370 nm on the long-wavelength The band maximum obtained from the EOM-CCSD spectrum
side of the toluene–O2 absorption band.20 Moreover, the EOM- occurs at B205 nm, which is slightly blue shifted relative to the
CCSD results show two distinct groupings of long-wavelength band maximum obtained in the TD-DFT calculation. Asserting
transitions (insets in Fig. 5A). On the basis of our preceding that our coupled cluster results are accurate, this deviation of
discussion, we assign the band that occurs over the range B310– B0.3 eV is well within the expected error of a DFT calculation.60–63
350 nm to transitions that populate the lowest energy triplet state However, given our approach to obtain these broadened spectra
in toluene, 3M1, which is in good agreement with the experimen- (vide supra), the inherent uncertainty associated with subtracting
tally observed band at 346 nm. By extension, we assign the band our modeled toluene spectrum from the spectrum of the toluene–
that occurs over the range B240–275 nm to transitions that O2 complex could also easily account for this difference in the
populate higher energy triplet states (e.g., 3M2). Although the band maxima.
accuracy of calculating oscillator strengths of such weak transitions The results shown in Fig. 4 and 5A clearly point to a limiting
is limited, our results model experimental data well. Moreover, it is feature of the TD-DFT approach. With an o value of 0.2 bohr1
clear that the band assigned to the 1M0 - 3M2 transition is appropriate to model intramolecular transitions in toluene,

15042 | Phys. Chem. Chem. Phys., 2021, 23, 15038–15048 This journal is © the Owner Societies 2021
View Article Online

Paper PCCP
Published on 28 June 2021. Downloaded by S.V. National Institute of Technology on 10/18/2022 10:59:05 AM.

Fig. 6 Illustration of how the TD-DFT calculated wavelength of two Fig. 7 Spectra of the toluene–O2 complex calculated at the EOM-CCSD/
selected transitions in a 1 : 1 toluene–O2 complex varies with the range- aug-cc-pVDZ level of theory, with the intermolecular distance ranging from
separation parameter o used in the calculation. When using o = 2.75 to 4.00 Å. The inset shows an expanded view of the band assigned to the
0.5 bohr1, the TD-DFT calculation for a transition to the intermolecular oxygen perturbed 1M0 - 3M1 transition. Oxygen does not appreciably perturb
CT state (solid blue line) comes close to the transition wavelength toluene at intermolecular distances greater than B4 Å (i.e., over this wave-
calculated using EOM-CCSD for the same toluene–O2 geometry (dashed length range, the spectra are superimposable; data not shown).
blue line). However, relative to this EOM-CCSD result, the TD-DFT
calculation shows a large redshift as o is decreased to a value commonly
used for intramolecular transitions, 0.2 bohr1. When using o = 0.2 bohr1, from the spectrum of pure toluene, clearly showing the B band at
the TD-DFT calculation for an oxygen-induced transition to the triplet state around 250 nm and the tail of the K band just above 200 nm.74
of toluene (solid red line), exemplifying a local intramolecular transition, These observations are consistent with expectation, and can be
yields a transition wavelength close to the EOM-CCSD result (dashed red ascribed to a decrease in the overlap between electron donor and
line). However, the TD-DFT result likewise shows a large redshift relative to
acceptor wave functions in Mulliken’s theory on CT in molecular
the EOM-CCSD result upon increasing o to the value of 0.5 bohr1 suitable
for an intermolecular transition.
complexes.1 The corollary to this is that the exact position of the
calculated CT band maximum may depend on details in the force
field parametrization used in the MD simulation to obtain the
TD-DFT does not properly model intermolecular toluene-to-oxygen M–O2 geometries. Complementary intermolecular distance plots,
CT transitions. With an o value of 0.5 bohr1 appropriate to model obtained using TD-DFT for both one- and two-photon transitions,
intermolecular toluene-to-oxygen CT transitions, TD-DFT does show the same general trend (see ESI†). However, as perhaps
not properly model intramolecular transitions in toluene with expected, the one- and two-photon transition probabilities scale
intensities perturbed by oxygen. We illustrate this phenomenon differently with the M and O2 intermolecular distance.
in Fig. 6. For the EOM-CCSD results shown in Fig. 7, the intensity
Thus, TD-DFT cannot accurately model the wide range of of the band at B330 nm assigned to the oxygen perturbed
optical transitions that characterize the toluene-oxygen collision 1
M0 - 3M1 transition decreases rapidly with increasing inter-
complex. Although changing the o value presents a valuable tool molecular distance, while the transition wavelength remains
for TD-DFT studies of pure CT transitions, accurate predictions almost unchanged. Again, this is consistent with expectation:
of a wide range of transitions in a given molecular system the perturbation of forbidden local transitions in an organic
requires more sophisticated computational methods. Indeed, molecule by oxygen, and vice versa, primarily influences transition
our results show that coupled cluster methods provide a more probabilities, not excitation energies.13
balanced description of both intermolecular CT transitions and As a corollary to these distance-dependent changes in the
O2-perturbed local transitions in toluene. spectra, we also find that, for a given intermolecular distance in
structure B of Fig. 2, the spectrum changes as oxygen is rotated
Exploring features of the toluene–O2 absorption band relative to the toluene axis defined by the pendant methyl group
Effect of intermolecular distance. Using the EOM-CCSD (see ESI†). Similar rotationally-mediated transition probabil-
method, and limiting ourselves to the complex with geometry B ities have been posited for other intermolecular complexes.50,55
in Fig. 2, we examined the effect of changing the intermolecular Assessing charge displacement. In our discussion, we imply
distance between toluene and oxygen, as measured from the that three types of transitions contribute to what has traditionally
center of mass in oxygen to the center of the aromatic system in been dubbed the M–O2 CT band. The intense peak at shorter
toluene. With an increase in the intermolecular distance, we find wavelengths consists of transitions with appreciable CT character
that (a) the energy of the CT transition increases, as reflected in a along with toluene-induced transitions to highly excited states of
systematic blue shift of the absorption band, and (b) the transition oxygen in which little charge displacement occurs. The absorption
oscillator strength decreases (Fig. 7). At an intermolecular tail at longer wavelengths reflects oxygen-assisted local transitions
separation distance of 4 Å, the spectrum is nearly indistinguishable in toluene in which little charge displacement likewise occurs.

This journal is © the Owner Societies 2021 Phys. Chem. Chem. Phys., 2021, 23, 15038–15048 | 15043
View Article Online

PCCP Paper

While no unique method exists for quantifying the CT


character of a transition, we used the index proposed by Bahers,
et al.75 to express the extent of charge displacement in the final
state populated by an optical excitation. To this end, and using
the EOM-CCSD results, we first removed discrete transitions
from the toluene–O2 spectrum that also appeared in the toluene
Published on 28 June 2021. Downloaded by S.V. National Institute of Technology on 10/18/2022 10:59:05 AM.

spectrum. It is assumed that transitions in the toluene–O2


spectrum closest in wavelength to those in the toluene spectrum
are the same (i.e., spin-allowed singlet–singlet local transitions
in toluene). Although 1M0 - 3Mn transitions in the oxygen-free
toluene spectrum have exactly zero intensity, they become
spin-allowed and gain intensity upon perturbation by oxygen
(vide supra). As a result, we retain transitions unique to the
toluene–O2 complex.
In Fig. 5B, we plot the distance between the barycenters of
the density increment and depletion due to the excitation (i.e., Fig. 8 Illustration of key parameters that influence the M–O2 CT absorp-
the CT index of Bahers, et al.75) against the excitation wavelength tion band, where ecom refers to the molar absorption coefficient of the
transition to the CT state, and Kcom refers to the equilibrium constant for
of the toluene–O2 transitions. As shown, we find that transitions
ground state complex formation.
at wavelengths longer than 240 nm show effectively no displace-
ment of charge, whereas many of the transitions at wavelengths
shorter than 240 nm show appreciable charge displacement. As equilibrium constant for complex formation, Kcom, as shown in
such, the former are consistent with a localized excited state of eqn (1) and Fig. 8 for a generic M–O2 complex.
toluene.  
Revisiting the Mulliken–Evans debate. Our calculations on M  O2 X3 Sg
Kcom ¼   (1)
the toluene–O2 complex indicate that oxygen-induced local ½M O2 X3 Sg
transitions in toluene play an appreciable role on the long
wavelength side of the absorption spectrum commonly attributed For one-photon transitions, it is common to rely on the Benesi-
to the intermolecular CT transition. This is consistent with Hildebrand77 or Drago-Rose78 treatments to separate Kcom from the
the phenomenon most visibly exemplified in the experimental molar absorption coefficient of the complex, ecom. Both techniques
spectrum of oxygenated liquid benzene where O2-induced exploit changes in sample absorbance as a function of a change in
1
M0 - 3M1 vibronic bands are clearly superimposed onto the the concentration of one of the interacting components. The
broad CT band.20,23,24 Based on this observation for benzene accuracy of these treatments, however, decreases appreciably
and our calculations on the toluene–O2 complex, it is reason- for weak complexes,79–81 and certainly for ‘‘collision’’ com-
able to assign similar, albeit very weak, ‘‘bumps’’ on the plexes, of which the M–O2 complex is an excellent example.15
long wavelength tail of the experimental toluene spectrum to When considering eqn (1) and the common approach of
O2-induced 1M0 - 3M1 vibronic bands.20 As such, our results using oxygen-saturated solutions of neat M, the concentration
support points made by both Mulliken and Evans in their of M (typically B10 M, or greater) far exceeds the concentration of
discussion on the nature of O2-dependent transitions in oxygen (typically B10 mM, or less82). This inequality mirrors what
organic molecules.18,25 is generally applied in the Benesi–Hildebrand and Drago-Rose
This wavelength-dependent superposition of different M–O2 treatments.15,77,78 In their 1960 paper on liquid solvents,
transitions has also been discussed in the context of experi- Tsubomura and Mulliken estimated ecom assuming that all
ments wherein O2(a1Dg) is produced upon irradiation into the dissolved oxygen formed the ground state M–O2 complex.18 In
M–O2 absorption band (Fig. 1).20,26,29 Although there is strong a recent paper on the two-photon excitation of the M–O2 CT
evidence to indicate that O2(a1Dg) can indeed be produced upon transition, the magnitude of the absorption cross section was
population of the M–O2 CT state (Fig. 1),20,26,76 it is likely that, estimated assuming that only 1% of the dissolved oxygen
for some molecules M under some experimental conditions, a formed the ground state M–O2 complex.29 In both cases, errors
fraction of the O2(a1Dg) produced is a consequence of an oxygen- on the absorption coefficients are large, reflecting the large uncer-
induced 1M0 - 3M1 transition followed by the 3M1 sensitized tainty in the concentration of the ground state M–O2 complex.
production of O2(a1Dg).29 With these limitations in mind, a way to independently
assess either ecom or Kcom would be a useful contribution.
The equilibrium constant for ground state complex formation
A key limitation in experimental studies of optical transitions of Using computational results to quantify Kcom
weakly bound intermolecular complexes is that, in the Lambert– With our computational model, one does not need to know the
Beer expression, one can only directly determine the product of concentration of the ground state M–O2 complex to quantify
the absorption coefficient with the concentration of the ground transition probabilities. Our present results thus facilitate an
state complex. The latter is often represented in terms of the independent assessment of Kcom for the toluene and oxygen

15044 | Phys. Chem. Chem. Phys., 2021, 23, 15038–15048 This journal is © the Owner Societies 2021
View Article Online

Paper PCCP

as exemplified by our present results on the toluene–O2 complex.


Our calculations show that, while the peak of what is commonly
viewed as an intermolecular M–O2 CT band indeed has sub-
stantial CT character, the long wavelength tail of the band is
better ascribed to oxygen-assisted local transitions to triplet states
of M. In this way, we achieve a compromise in the classic debate
Published on 28 June 2021. Downloaded by S.V. National Institute of Technology on 10/18/2022 10:59:05 AM.

between R. S. Mulliken and D. F. Evans on the nature of oxygen-


dependent transitions in organic molecules in which Mulliken
championed the importance of the CT state and Evans the spin
properties of O2(X3Sg). Our results thus indicate that, in some
cases, claims of O2(a1Dg) production from the M–O2 CT state may, in
fact, reflect O2(a1Dg) production via an oxygen-induced 1M0 - 3M1
transition followed by the 3M1 sensitized production of O2(a1Dg).
Through our M–O2 study, we highlight a weak point of using
time-dependent density functional theory, TD-DFT, to model
Fig. 9 For the toluene–O2 spectrum calculated using 100 1 : 1 complexes
systems in which both local transitions in M and intermolecular
obtained from an MD simulation, the plot shows the fraction of the
spectrum that is recovered as a function of the number of geometries CT transitions occur. To account for the latter, a range-separation
incorporated. The dashed line shows that B99% of the spectrum is parameter o suitable for intermolecular CT must be used, as
recovered when 54 of the 100 geometries are used. opposed to o values commonly used for intramolecular transi-
tions. The optimum o value for the two cases is sufficiently
different that TD-DFT with a single common o parameter falls
system. Given the inherent errors, we assume that this exercise short of giving an accurate representation of the full spectrum.
is equally applicable to both liquid and gas phase systems. In general, for noncovalent complexes in which intermolecular
To use our MD results in this regard, we exploit the general CT plays a role, it could be possible to combine information
ergodic theorem from statistical mechanics paraphrased to say for such different types of transitions with differently para-
that considering one molecule for a long time is equivalent to meterized TD-DFT methods. However, we have shown that
considering many molecules at one time.83 For each method coupled cluster methods allows one to handle these different
of assessing our MD simulation (vide supra), we obtain transitions with similar accuracy.
100 different toluene–O2 structures within the 10 ns simulation Using the results of our computed toluene–O2 CT spectrum,
window. We can then determine what fraction of these structures we estimate that the equilibrium constant for the formation of
must be considered to account for a given percentage of the the ground state toluene–O2 complex is B0.1 M1.
toluene–O2 spectrum, as quantified by the sum of the respective
oscillator strengths. To this end, we employ the same EOM-CCSD/
aug-cc-pVDZ spectrum of discrete transitions used to obtain the Computational methods
data shown in Fig. 5. Given the relative intensities of the toluene–O2
transitions (Fig. 5A), this exercise principally reflects a combination MD simulations were performed in Tinker 870 using the AMOEBA
of the transition to the CT state as illustrated in Fig. 8 along with force field64 and our own parametrization of molecular oxygen (see
toluene-induced transitions to higher excited states of oxygen. We ESI†). Initial structures were constructed using Packmol.69 Details
find that 43 of the structures account for B95% of the spectrum, of the MD simulations are given in the ESI.† All quantum
and 54 of the structures account for B99% of the spectrum (Fig. 9). chemistry computations were performed in Gaussian 16,84 except
This indicates that about half of the oxygen molecules present in a the two-photon computations, which were done in Dalton.85 One-
statistically viable ensemble of toluene and oxygen contribute to the photon TD-DFT computations were performed at the oB97X-D46/
light-absorbing collision complex at any given time. 6-31G(d)86 or oB97X-D/aug-pcseg-159 levels of theory with the latter
For an experiment in which [M] c [O2], we can correlate this basis set taken from the Basis-Set Exchange Library.87 Two-photon
fraction to the ratio [M–O2]/[O2] in eqn (1) and, with [M] = 9.45 M computations were performed at the CAM-B3LYP/aug-cc-pVDZ
for toluene, this yields an equilibrium constant Kcom of level of theory. Coupled cluster computations were performed at
B0.1 M1. This value of Kcom is appreciably smaller than the value the EOM-CCSD42/aug-cc-pVDZ88 level of theory.
of B3–6 M1 estimated from a Drago-Rose treatment of benzene
data,15 reinforcing the fact that the latter experimental approach is Conflicts of interest
indeed error prone for weakly bound collision complexes.79–81
The authors declare no competing financial interests.

Conclusions
Acknowledgements
The combination of molecular dynamics simulations and electronic
structure response methods is an approach to accurately model This work was supported by the Independent Research Fund of
optical transitions from weakly bound M–O2 collision complexes, Denmark. The numerical results presented were obtained at

This journal is © the Owner Societies 2021 Phys. Chem. Chem. Phys., 2021, 23, 15038–15048 | 15045
View Article Online

PCCP Paper

the Centre for Scientific Computing, Aarhus. The authors thank 17 J. J. Casero and J. A. Joens, Thermochemistry of the Gas
Mikkel Bregnhøj for helpful comments and Pier Paolo Poier for Phase Molecular Complex of Benzene with Oxygen, J. Phys.
assistance with the Tinker program. Chem. A, 1997, 101, 2607–2609.
18 H. Tsubomura and R. S. Mulliken, Molecular Complexes
and their Spectra. XII. Ultraviolet Absorption Spectra Caused
References
by the Interaction of Oxygen with Organic Molecules, J. Am.
Published on 28 June 2021. Downloaded by S.V. National Institute of Technology on 10/18/2022 10:59:05 AM.

1 R. S. Mulliken and W. B. Person, Molecular Complexes, Chem. Soc., 1960, 82, 5966–5974.
Wiley-Interscience, New York, 1969. 19 D. F. Evans, Molecular Association of Oxygen and Aromatic
2 M. J. Paterson, O. Christiansen, F. Jensen and P. R. Ogilby, Substances, J. Chem. Soc., 1953, 345–347.
Overview of Theoretical and Computational Methods Applied 20 R. D. Scurlock and P. R. Ogilby, Singlet Molecular Oxygen
to the Oxygen-Organic Molecule Photosystem, Photochem. (a1Dg) Formation upon Irradiation of an Oxygen (X3Sg)-
Photobiol., 2006, 82, 1136–1160. Organic Molecule Charge-Transfer Absorption Band, J. Phys.
3 M. Tamres and R. L. Strong, Contact Charge-Transfer Spectra Chem., 1989, 93, 5493–5500.
in Molecular Association, ed. R. Foster, Academic Press, 21 J. B. Birks, E. Pantos and T. D. S. Hamilton, The Benzene-
London, 1979, pp. 331–456. Oxygen Complex in the Vapour Phase, Chem. Phys. Lett.,
4 A. V. Baklanov and D. H. Parker, Weakly Bound Environment 1973, 20, 544–546.
of Molecular Oxygen as a Catalyst of Photooxidation, Kinet. 22 D. H. Parker, Laser Photochemistry of Molecular Oxygen,
Catal., 2020, 61, 174–197. Acc. Chem. Res., 2000, 33, 563–571.
5 J. T. Su and A. H. Zewail, Solvation Ultrafast Dynamics of 23 D. F. Evans, Magnetic Perturbation of the Lowest Triplet
Reactions. 14. Molecular Dynamics and ab Initio Studies of States of Aromatic Molecules by Dissolved Oxygen, Nature,
Charge-Transfer Reactions of Iodine in Benzene Clusters, 1956, 178, 534–535.
J. Phys. Chem. A, 1998, 102, 4082–4099. 24 D. F. Evans, Perturbation of Singlet-Triplet Transitions of
6 C. Schweitzer and R. Schmidt, Physical Mechanisms of Aromatic Molecules by Oxygen under Pressure, J. Chem.
Generation and Deactivation of Singlet Oxygen, Chem. Rev., Soc., 1957, 1351–1357.
2003, 103, 1685–1757. 25 D. F. Evans, Magnetic Perturbation of Singlet-Triplet Transi-
7 B. F. Minaev, Electronic Mechanisms of Molecular Oxygen tions. Part V. Mechanism, J. Chem. Soc., 1961, 1987–1993.
Activation, Russ. Chem. Rev., 2007, 76, 988–1010. 26 P.-G. Jensen, J. Arnbjerg, L. P. Tolbod, R. Toftegaard and
8 R. Schmidt, Photosensitized Generation of Singlet Oxygen, P. R. Ogilby, Influence of an Intermolecular Charge-Transfer
Photochem. Photobiol., 2006, 82, 1161–1177. State on Excited-State Relaxation Dynamics: Solvent Effect on
9 K. Kawaoka, A. U. Khan and D. R. Kearns, Role of Singlet the Methylnaphthalene-Oxygen System and its Significance for
Excited States of Molecular Oxygen in the Quenching of Singlet Oxygen Production, J. Phys. Chem. A, 2009, 113, 9965–9973.
Organic Triplet States, J. Chem. Phys., 1967, 46, 1842–1853. 27 D. J. McGarvey, F. Wilkinson, D. R. Worrall, J. Hobley and
10 A. Garner and F. Wilkinson, Quenching of Triplet States W. Shaikh, Picosecond Absorption Studies on the Role of
by Molecular Oxygen and the Role of Charge Transfer Charge Transfer Interactions in the Mechanism of Quenching
Interactions, Chem. Phys. Lett., 1977, 45, 432–435. of Triplet states by Molecular Oxygen, Chem. Phys. Lett., 1993,
11 F. Wilkinson, D. J. McGarvey and A. F. Olea, Excited Triplet 202, 528–534.
State Interactions with Molecular Oxygen: Influence of Charge 28 S. L. Logunov and M. A. J. Rodgers, Laser Flash Photolysis
Transfer on the Bimolecular Quenching Rate Constants and Studies of the Contact Complex between Molecular Oxygen
Yields of Singlet Oxygen for Substituted Naphthalenes in and 1-Methylnaphthalene, J. Phys. Chem., 1993, 97, 5643–5648.
Various Solvents, J. Phys. Chem., 1994, 98, 3762–3769. 29 M. Bregnhøj and P. R. Ogilby, Two-Photon Excitation of
12 B. F. Minaev, K. V. Mikkelsen and H. Ågren, Collision-Induced Neat Aerated Solvents with Visible Light Produces Singlet
Electronic Transitions in Complexes between Benzene and Oxygen, J. Phys. Chem. A, 2019, 123, 7567–7575.
Molecular Oxygen, Chem. Phys., 1997, 220, 79–94. 30 A. U. Khan and D. R. Kearns, Energetics of the Interaction of
13 M. Bregnhøj, M. Westberg, B. F. Minaev and P. R. Ogilby, Molecular Oxygen with Organic Molecules, J. Chem. Phys.,
Singlet Oxygen Photophysics in Liquid Solvents: Converging 1968, 48, 3272–3275.
on a Unified Picture, Acc. Chem. Res., 2017, 50, 1920–1927. 31 B. F. Minaev, Spin-Orbit Coupling of Charge-Transfer States
14 F. Thorning, F. Jensen and P. R. Ogilby, Modeling the Effect and the Mechanism for Quenching Singlet Oxygen by
of Solvents on Nonradiative Singlet Oxygen Deactivation: Amines, Theor. Exp. Chem., 1984, 20, 199–201.
Going beyond Weak Coupling in Intermolecular Electronic- 32 M. Bregnhøj, M. Westberg, F. Jensen and P. R. Ogilby, Solvent-
to-Vibrational Energy Transfer, J. Phys. Chem. B, 2020, 124, Dependent Singlet Oxygen Lifetimes: Temperature Effects
2245–2254. Implicate Tunneling and Charge-Transfer Interactions, Phys.
15 E. A. Gooding, K. R. Serak and P. R. Ogilby, The Ground State Chem. Chem. Phys., 2016, 18, 22946–22961.
Benzene-Oxygen Complex, J. Phys. Chem., 1991, 95, 7868–7871. 33 T. Helgaker, S. Coriani, P. Jørgensen, K. Kristensen, J. Olsen
16 J. R. Grover, G. Hagenow and E. A. Walters, Complexes of and K. Ruud, Recent Advances in Wave Function-Based
Oxygen with Benzene and Hexafluorobenzene, J. Chem. Methods of Molecular-Property Calculations, Chem. Rev.,
Phys., 1992, 97, 628–642. 2012, 112, 543–631.

15046 | Phys. Chem. Chem. Phys., 2021, 23, 15038–15048 This journal is © the Owner Societies 2021
View Article Online

Paper PCCP

34 F. Jensen, Introduction to Computational Chemistry, John 50 M.-S. Liao, Y. Lu, V. D. Parker and S. Scheiner, DFT
Wiley and Sons, Chichester, 2016. Calculations and Spectral Measurements of Charge-Transfer
35 T. D. Poulsen, P. K. Frederiksen, M. Jørgensen, K. V. Mikkelsen Complexes Formed by Aromatic Amines and Nitrogen Hetero-
and P. R. Ogilby, Two-Photon Singlet Oxygen Sensitizers: Quanti- cycles with Tetracyanoethylene and Chloranil, J. Phys. Chem. A,
fying, Modeling, and Optimizing the Two-Photon Absorption 2003, 107, 8939–8948.
Cross Section, J. Phys. Chem. A, 2001, 105, 11488–11495. 51 A. Schulz and C. R. Jacob, Description of Intermolecular
Published on 28 June 2021. Downloaded by S.V. National Institute of Technology on 10/18/2022 10:59:05 AM.

36 P. Salek, O. Vahtras, J. D. Guo, Y. Luo, T. Helgaker and H. Ågren, Charge Transfer with Subsystem Density-Functional Theory,
Calculations of Two-Photon Absorption Cross Sections by J. Chem. Phys., 2019, 151, 131103.
means of Density Functional Theory, Chem. Phys. Lett., 2003, 52 A. J. Rest, K. Salisbury and J. R. Sodeau, Contact Charge
374, 446–452. Transfer Interactions between Aromtic Molecules and Oxygen
37 M. J. Paterson, O. Christiansen, F. Pawlowski, P. Jørgensen, in Gas Matrices at 10 K, J. Chem. Soc., Faraday Trans. 2, 1977,
C. Hättig, T. Helgaker and P. Salek, Benchmarking Two- 73, 265–273.
Photon Absorption with CC3 Quadratic Response Theory, 53 G. Granucci and M. Persico, Benzene-O2 Interaction Potential
and Comparison with Density Functional Response Theory, from Ab Initio Calculations, Chem. Phys. Lett., 1993, 205,
J. Chem. Phys., 2006, 124, 054322. 331–336.
38 J. Olsen and P. Jørgensen, Linear and Nonlinear Response 54 G. E. Khalil and M. Kasha, Oxygen-Interaction Luminescence
Functions for an Exact State and for an MCSCF State, Spectroscopy, Photochem. Photobiol., 1978, 28, 435–443.
J. Chem. Phys., 1985, 82, 3235–3264. 55 T. D. Poulsen, P. R. Ogilby and K. V. Mikkelsen, The a1Dg 
39 M. Petersilka, U. J. Gossmann and E. K. U. Gross, Excitation X3Sg Transition in Molecular Oxygen: Interpretation of
Energies from Time-Dependent Density-Functional Theory, Solvent Effects on Spectral Shifts, J. Phys. Chem. A, 1999,
Phys. Rev. Lett., 1996, 76, 1212–1215. 103, 3418–3422.
40 M. T. P. Beerepoot, D. H. Friese, N. H. List, J. Kongsted and 56 J. Shee and M. Head-Gordon, Predicting Excitation Energies
K. Ruud, Benchmarking Two-Photon Absorption Cross of Twisted Intramolecular Charge-Transfer States with the
Sections: Performance of CC2 and CAM-B3LYP, Phys. Chem. Time-Dependent Density Functional Theory: Comparison
Chem. Phys., 2015, 17, 19306–19314. with Experimental Measurements in the Gas Phase and
41 S. Coriani, F. Pawlowski, J. Olsen and P. Jørgensen, Molecular Solvents Ranging from Hexanes to Acetonitrile, J. Chem.
Response Properties in Equation of Motion Coupled Cluster Theory Comput., 2020, 16, 6244–6255.
Theory: A Time-Dependent Perspective, J. Chem. Phys., 2016, 57 W. M. McClain, Two-Photon Molecular Spectroscopy, Acc.
144, 024102. Chem. Res., 1974, 7, 129–135.
42 H. Koch and P. Jørgensen, Coupled Cluster Response Functions, 58 B. Kozma, A. Tajti, B. Demoulin, R. Izsak, M. Nooijen and
J. Chem. Phys., 1990, 93, 3333–3344. P. G. Szalay, A New Benchmark Set for Excitation Energy of
43 S. N. Steinmann, C. Piemontesi, A. Delachat and C. Corminboeuf, Charge Transfer States: Systematic Investigation of Coupled
Why are the Interaction Energies of Charge-Transfer Cluster Type Methods, J. Chem. Theory Comput., 2020, 16,
Complexes Challenging for DFT?, J. Chem. Theory Comput., 4213–4225.
2012, 8, 1629–1640. 59 F. Jensen, Unifying General and Segmented Contracted
44 Y. Tawada, T. Tsuneda, S. Yanagisawa, T. Yanai and Basis Sets. Segmented Polarization Consistent Basis Sets,
K. Hirao, A Long-Range-Corrected Time-Dependent Density J. Chem. Theory Comput., 2014, 10, 1074–1085.
Functional Theory, J. Chem. Phys., 2004, 120, 8425–8433. 60 D. Jacquemin, V. Wathelet, E. A. Perpete and C. Adamo, Extensive
45 T. Tsuneda and K. Hirao, Long-Range Correction for TD-DFT Benchmark: Singlet-Excited States of Organic Molecules,
Density Functional Theory, Wiley Interdiscip. Rev.: Comput. J. Chem. Theory Comput., 2009, 5, 2420–2435.
Mol. Sci., 2014, 4, 375–390. 61 A. D. Laurent and D. Jacquemin, TD-DFT Benchmarks: A
46 J.-D. Chai and M. Head-Gordon, Long-Range Corrected Hybrid Review, Int. J. Quantum Chem., 2013, 113, 2019–2039.
Density Functionals with Damped Atom-Atom Dispersion Cor- 62 D. Jacquemin, E. A. Perpete, I. Ciofini and C. Adamo,
rections, Phys. Chem. Chem. Phys., 2008, 10, 6615–6620. Assessment of the oB97 Family for Excited-State Calculations,
47 Y. Mao, Q. Ge, P. R. Horn and M. Head-Gordon, On the Theor. Chim. Acta, 2011, 128, 127–136.
Computational Characterization of Charge-Transfer Effects 63 M. Isegawa, R. Peverati and D. G. Truhlar, Performance of
in Noncovalently Bound Molecular Complexes, J. Chem. Recent and High-Performance Approximate Density Func-
Theory Comput., 2018, 14, 2401–2417. tionals for Time-Dependent Density Functional Theory Cal-
48 M. M. Alam, M. Chattopadhyaya, S. Chakrabarti and K. Ruud, culations of Valence and Rydberg Electronic Transition
High-Polarity Solvents Decreasing the Two-Photon Transition Energies, J. Chem. Phys., 2012, 137, 244104.
Probability of Through-Space Charge-Transfer Systems: A Sur- 64 J. W. Ponder, C. Wu, P. Ren, V. S. Pande, J. D. Chodera,
prising In Silico Observation, J. Phys. Chem. Lett., 2012, 3, M. J. Schnieders, I. Haque, D. L. Mobley, D. S. Lambrecht,
961–966. R. A. DiStasio, M. Head-Gordon, G. N. I. Clark, M. E. Johnson
49 A. Otero-de-la-Roza and E. R. Johnson, Analysis of Density- and T. Head-Gordon, Current Status of the AMOEBA
Functional Errors for Noncovalent Interactions between Polarizable Force Field, J. Phys. Chem. B, 2010, 114,
Charged Molecules, J. Phys. Chem. A, 2020, 124, 353–361. 2549–2564.

This journal is © the Owner Societies 2021 Phys. Chem. Chem. Phys., 2021, 23, 15038–15048 | 15047
View Article Online

PCCP Paper

65 G. Herzberg, Molecular Spectra and Molecular Structure. 83 A. Nitzan, Chemical Dynamics in Condensed Phases: Relaxa-
I. Spectra of Diatomic Molecules, Van Nostrand Reinhold, tion, Transfer, and Reactions in Condensed Molecular Systems,
New York, 1950. Oxford University Press, Oxford, 2006.
66 R. Qi, Q. Wang and P. Ren, General van der Waals Potential 84 M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria,
for Common Organnic Molecules, Bioorg. Med. Chem., 2016, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, G. A.
24, 4911–4919. Petersson, H. Nakatsuji, X. Li, M. Caricato, A. V. Marenich,
Published on 28 June 2021. Downloaded by S.V. National Institute of Technology on 10/18/2022 10:59:05 AM.

67 D. Spelsberg and W. Meyer, Static Dipole Polarizabilities of J. Bloino, B. G. Janesko, R. Gomperts, B. Mennucci, H. P.
N2, O2, F2, and H2O, J. Chem. Phys., 1994, 101, 1282–1288. Hratchian, J. V. Ortiz, A. F. Izmaylov, J. L. Sonnenberg,
68 V. W. Couling and S. S. Ntombela, The Electric Quadrupole D. Williams-Young, F. Ding, F. Lipparini, F. Egidi, J. Goings,
Moment of O2, Chem. Phys. Lett., 2014, 614, 41–44. B. Peng, A. Petrone, T. Henderson, D. Ranasinghe, V. G.
69 L. Martinez, R. Andrade, E. G. Birgin and J. M. Martinez, Zakrzewski, J. Gao, N. Rega, G. Zheng, W. Liang, M. Hada,
Packmol: A Package for Building Initial Configurations for M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida,
Molecular Dynamics Simulations, J. Comput. Chem., 2009, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven,
30, 2157–2164. K. Throssell, J. A. Montgomery, Jr., J. E. Peralta, F. Ogliaro,
70 J. A. Rackers, Z. Wang, C. Lu, M. L. Laury, L. Lagardere, M. J. Bearpark, J. J. Heyd, E. N. Brothers, K. N. Kudin,
M. J. Schnieders, J.-P. Piquemal, P. Ren and J. W. Ponder, V. N. Staroverov, T. A. Keith, R. Kobayashi, J. Normand,
Tinker 8: Software Tools for Molecular Design, J. Chem. K. Raghavachari, A. P. Rendell, J. C. Burant, S. S. Iyengar,
Theory Comput., 2018, 14, 5273–5289. J. Tomasi, M. Cossi, J. M. Millam, M. Klene, C. Adamo,
71 G. R. Fleming, Chemical Applications of Ultrafast Spectro- R. Cammi, J. W. Ochterski, R. L. Martin, K. Morokuma,
scopy, Oxford University Press, New York, 1986. O. Farkas, J. B. Foresman and D. J. Fox, Gaussian16 Revision
72 P. R. Ogilby, Singlet Oxygen: There is Indeed Something B.01, Gaussian, Inc, Wallingford, CT, 2016.
New Under the Sun, Chem. Soc. Rev., 2010, 39, 3181–3209. 85 K. Aidas, C. Angeli, K. L. Bak, V. Bakken, R. Bast, L. Boman,
73 Y. Kanda and R. Shimada, The Triplet-Singlet Emission Spectra O. Christiansen, R. Cimiraglia, S. Coriani, P. Dahle, E. K.
of Toluene, o-. m- and p-Xylenes and Mesitylene in Cyclohexane Dalskov, U. Ekström, T. Enevoldsen, J. J. Eriksen, P. Ettenhuber,
Matrices at 90 K, Spectrochim. Acta, 1961, 17, 279–285. B. Fernández, L. Ferrighi, H. Fliegl, L. Frediani, K. Hald,
74 D. H. Williams and I. Fleming, Spectroscopic Methods in A. Halkier, C. Hättig, H. Heiberg, T. Helgaker, A. C. Hennum,
Organic Chemistry, McGraw-Hill, London, 1995. H. Hettema, E. Hjertenæs, S. Høst, I. M. Høyvik, M. F. Iozzi,
75 T. L. Bahers, C. Adamo and I. Ciofini, A Qualitative Index B. Jansı́k, H. J. A. Jensen, D. Jonsson, P. Jørgensen, J. Kauczor,
of Spatial Extent in Charge-Transfer Excitations, J. Chem. S. Kirpekar, T. Kjærgaard, W. Klopper, S. Knecht, R. Kobayashi,
Theory Comput., 2011, 7, 2498–2506. H. Koch, J. Kongsted, A. Krapp, K. Kristensen, A. Ligabue,
76 P. R. Ogilby, M. Kristiansen and R. L. Clough, Singlet Oxygen O. B. Lutnæs, J. I. Melo, K. V. Mikkelsen, R. H. Myhre,
Formation in a Solid Organic Polymer Upon Irradiation of the C. Neiss, C. B. Nielsen, P. Norman, J. Olsen, J. M. H. Olsen,
Oxygen-Polymer Charge-Transfer Band, Macromolecules, 1990, A. Osted, M. J. Packer, F. Pawlowski, T. B. Pedersen, P. F. Provasi,
23, 2698–2704. S. Reine, Z. Rinkevicius, T. A. Ruden, K. Ruud, V. V. Rybkin,
77 H. A. Benesi and J. H. Hildebrand, A Spectrophotometric P. Sałek, C. C. M. Samson, A. S. d. Merás, T. Saue, S. P. A. Sauer,
Investigation of the Interaction of Iodine with Aromatic B. Schimmelpfennig, K. Sneskov, A. H. Steindal, K. O. Sylvester-
Hydrocarbons, J. Am. Chem. Soc., 1949, 71, 2703–2707. Hvid, P. R. Taylor, A. M. Teale, E. I. Tellgren, D. P. Tew,
78 N. J. Rose and R. S. Drago, Molecular Addition Compounds A. J. Thorvaldsen, L. Thøgersen, O. Vahtras, M. A. Watson,
of Iodine. I. An Absolute Method for the Spectroscopic D. J. D. Wilson, M. Ziolkowski and H. Ågren, The Dalton
Determination of Equilibrium Constants, J. Am. Chem. Quantum Chemistry Program System, Wiley Interdiscip. Rev.:
Soc., 1959, 81, 6138–6141. Comput. Mol. Sci., 2014, 4, 269–284.
79 T. J. Beugelsdijk and R. S. Drago, Thermodynamic Data for the 86 W. J. Hehre, R. Ditchfield and J. A. Pople, Self-Consistent
Binding of Molecular Oxygen to Cobalt(II) Protoporphyrin IX Molecular Orbital Methods. XII. Further Extensions of
Dimethyl Ester, J. Am. Chem. Soc., 1975, 97, 6466–6472. Gaussian-Type Basis Sets for Use in Molecular Orbital
80 D. A. Deranleau, Theory of the Measurement of Weak Molecular Studies of Organic Molecules, J. Chem. Phys., 1972, 56,
Complexes. I. General Considerations, J. Am. Chem. Soc., 1969, 2257–2261.
91, 4044–4049. 87 B. P. Pritchard, D. Altarawy, B. Didier, T. D. Gibson and
81 W. B. Person, A Criterion for Reliability of Formation T. L. Windus, A New Basis Set Exchange: An Open Up-to-
Constants of Weak Complexes, J. Am. Chem. Soc., 1965, 87, Date Resource for the Molecular Sciemnces Community,
167–170. J. Chem. Inf. Model., 2019, 59, 4814–4820.
82 R. Battino, T. R. Rettich and T. Tominaga, The Solubility of 88 T. H. Dunning, Gaussian Basis Sets for use in Correlated
Oxygen and Ozone in Liquids, J. Phys. Chem. Ref. Data, 1983, Molecular Calculations. I. The Atoms Boron through Neon
12, 163–178. and Hydrogen, J. Chem. Phys., 1989, 90, 1007–1023.

15048 | Phys. Chem. Chem. Phys., 2021, 23, 15038–15048 This journal is © the Owner Societies 2021

You might also like