Photocatalytic Water Splitting with the Acridine Chromophore: A

Jul 28, 2015 - In addition to the well-known excited states of the acridine chromophore, excited states of charge-transfer character were identified, ...
0 downloads 0 Views 3MB Size
Download PDF
Article pubs.acs.org/JPCB

Photocatalytic Water Splitting with the Acridine Chromophore: A Computational Study Xiaojun Liu,†,‡ Tolga N. V. Karsili,† Andrzej L. Sobolewski,§ and Wolfgang Domcke*,† †

Department of Chemistry, Technische Universität München, D-85747 Garching, Germany Key Laboratory of Luminescence and Optical Information, Institute of Optoelectronic Technology, Beijing Jiaotong University, 100044 Beijing, China § Institute of Physics, Polish Academy of Sciences, PL-02668 Warsaw, Poland ‡

ABSTRACT: The hydrogen-bonded acridine−water complex is considered as a model system for the exploration of photochemical reactions which can lead to the splitting of water into H• and OH• radicals. The vertical excitation energies of the lowest singlet and triplet excited states of the complex were calculated with the CASSCF/CASPT2 and ADC(2) ab initio electronicstructure methods. In addition to the well-known excited states of the acridine chromophore, excited states of charge-transfer character were identified, in which an electron is transferred from the p orbital of the H2O molecule to the π* orbital of acridine. The low-energy barriers which separate these reactive charge-transfer states from the spectroscopic states of the acridine−water complex have been characterized by the calculation of two-dimensional relaxed potential-energy surfaces as functions of the H atom-transfer coordinate and the donor (O)−acceptor (N) distance. When populated, these charge-transfer states drive the transfer of a proton from the water molecule to acridine, which results in the acridinyl-hydroxyl biradical. The same computational methods were employed to explore the photochemistry of the (N-hydrogenated) acridinyl radical. The latter possesses low-lying (about 3.0 eV) ππ* excited states with appreciable oscillator strengths in addition to a low-lying dark ππ* excited state. The bound potential-energy functions of the ππ* excited states are predissociated by the potential-energy function of an excited state of πσ* character which is repulsive with respect to the NH stretching coordinate. The dissociation threshold of the πσ* state is about 2.7 eV and thus below the excitation energies of the bright ππ* states. The conical intersections of the πσ* state with the ππ* excited states and with the electronic ground state provide a mechanism for the direct and fast photodetachment of the H atom from the acridinyl radical. These computational results indicate that the H2O molecule in the acidine−H2O complex can be dissociated into H• and OH• radicals by the absorption of two visible/ultraviolet photons.

1. INTRODUCTION

The commonly accepted explanation of the pronounced solvent dependence of the fluorescence quantum yield of Ac is a solvent-dependent ordering of the lowest excited singlet states. While the bright 1ππ* state seems to be the lowest excited singlet state of the Ac chromophore in water and other hydrogen-bonding solvents, the essentially dark 1nπ* state is assumed to be below the 1ππ* state in the gas phase and in nonpolar organic solvents. Rapid radiationless relaxation of the 1 ππ* state to the dark 1nπ* state may effectively quench the fluorescence of the 1ππ* state.3−5,18 The mechanisms of the H atom-abstraction reactivity of photoexcited Ac with certain solvents, on the other hand, remain enigmatic until today. The attempts of identifying the reactive excited state of Ac in alcohols and some hydrocarbons were remarkably inconclusive. Any of the 1nπ*, 1ππ*, 3nπ*, or

The spectroscopy and photochemistry of acridine (Ac), the Nheterocyclic analogue of anthracene, in the gas phase, in the crystalline state, and in various solvents has been the subject of extensive investigations over about four decades, from the 1950s to the 1980s. Ac is nonfluorescent or very weakly fluorescent in the crystalline state and in organic solvents, such as hexane or benzene.1−7 In aqueous solution, Ac is a weak base in the electronic ground state, but becomes a significantly stronger base in excited electronic states.8,9 The fluorescence quantum yield of protonated acridine, AcH+, is much higher than that of Ac. In H-atom-donating solvents, such as alcohols, Ac is both fluorescent and photoreactive, abstracting H atoms from the solvent when irradiated with light near 355 nm, the maximum of the lowest absorption band.6,10−14 Acridan (AcH2), the so-called C-type and N-type acridinyl (AcH•) radicals, as well as AcOH• radicals were identified as reaction products.10,15−17 © 2015 American Chemical Society

Received: May 20, 2015 Revised: July 23, 2015 Published: July 28, 2015 10664

DOI: 10.1021/acs.jpcb.5b04833 J. Phys. Chem. B 2015, 119, 10664−10672

Article

The Journal of Physical Chemistry B ππ* states of Ac was proposed as the reactive state by some authors, but was excluded by others.6,10−14 Apparently, the indirect evidence obtained from the observation of final reaction products in different solvents under varying conditions was insufficient for a reliable identification of the elementary reaction mechanisms. In the 1970s, picosecond time-resolved laser spectroscopy became available and was applied to the investigation of the electronic relaxation dynamics of Ac in the gas phase and in several solvents.19−21 While the picosecond relaxation kinetics of excited states could be resolved in time, no additional insights into the mechanisms of the photoreactivity of Ac with solvents were obtained. In the 1980s, multiphoton ionization of Ac in aqueous solution was investigated with intense nanosecond pulsed lasers.15,22 Hydrated electrons, (e−)aq, were generated in an apparently biphotonic process. We are not aware of more recent femtosecond time-resolved investigations of the excited-state relaxation dynamics and the photoreactivity of the Ac chromophore. The nature of the photoreactive electronic states and the mechanisms of the excited-state H atom-abstraction reaction from solvents are still unknown. In the present work, we explored the photoreactivity of Ac in the Ac−H2O hydrogen-bonded complex with ab initio computational methods. We choose water rather than methanol or ethanol as the ligand, since the ultimate goal of the present work is the direct light-induced decomposition of water into H• and OH• radicals using simple aromatic chromophores as photocatalyzers. The present study is the continuation of previous work on the photochemistry of the pyridine−H2O complex and the photochemistry of the pyridinyl radical.23,24 Whereas pyridine and the pyridinyl radical absorb in the far UV and therefore are inconvenient substrates for laser-induced chemistry, Ac and AcH• absorb in the near UV and in the visible ranges of the spectrum, respectively. In the first step of the envisioned photocatalytic cycle, the Ac chromophore acts as a photobase, abstracting a hydrogen atom from the hydrogenbonded H2O molecule

active-space self-consistent-field (CASSCF) method,25 the CASPT2 (second-order perturbation theory with respect to the CASSCF reference) method,26 as well as with the ADC(2) method.27,28 ADC(2) is a many-body Green’s function method which is closely related to the approximate second-order singles-and-doubles coupled-cluster (CC2) method.29 Geometries of conical intersections and saddle points, excitation energies, minimum-energy reaction paths, and excited-state PE surfaces were determined with the ADC(2) method. Although ADC(2) is a single-reference method, it has been found to be reliable for the prediction of vertical excitation energies of singly excited states of closed-shell systems.30 In particular, ADC(2) PE surfaces are well behaved near conical intersections of exited electronic states, although the method may fail near conical intersections of excited states with the electronic ground state. Therefore, most of the calculations of the present work, in particular the excited-state geometry optimizations, were performed with the ADC(2) method. Being derived by diagrammatic perturbation theory for systems with a closedshell ground state,27 the ADC(2) method is not readily applicable to radicals. The calculations for the AcH• radical were therefore performed with the multiconfiguration CASSCF and CASPT2 methods. The reaction paths were constructed as so-called relaxed scans. For the calculation of the reaction path for the H atomtransfer process from water to Ac, the bond length of the OH bond of the water molecule was chosen as the driving coordinate, while all other internal nuclear coordinates of the complex were relaxed in the electronic state under consideration. The distance between the O atom of H2O and the N atom of Ac was taken as the second driving coordinate in the calculation of two-dimensional PE surfaces in the barrier region of the H-transfer reaction. The reaction path for the photodetachment of the hydrogen atom from the AcH• radical was constructed as a rigid scan of the NH bond length, since the relaxation of the other internal coordinates is of little importance in this case. The saddle points for the H atomtransfer reactions in the singlet and triplet states of the Ac− H2O complex were estimated from the two-dimensional relaxed PE surfaces. The minimum-energy geometries of conical intersections were determined using the CIOpt program of Martinez and co-workers.31 In the CASSCF calculations for the Ac−H2O complex, 14 electrons were distributed in 13 orbitals, including the five highest π and five lowest π* orbitals of Ac, the n orbital of the N atom, one p orbital of the O atom, and the lowest σ* orbital of water. The energies of the S0 state and the lowest 1ππ* and 1 nπ* excited states were averaged in the calculations of the singlet states. In the calculations of the triplet states, the energies of the S0 state and the lowest 3ππ* and 3nπ* states were averaged. The active space of the CASSCF calculations for the AcH• radical consisted of 9 electrons in 10 orbitals: the three highest π orbitals and the three lowest π* orbitals, one σ orbital, and one σ* orbital of the NH bond as well as the highest ring-centered σ orbital. The CASPT2 calculations were carried out as single-state calculations. A level shift of 0.3 au was employed in the CASPT2 calculations. The MP2 and ADC(2) calculations were carried out with the TURBOMOLE program package,32 making use of the resolution-of-the-identity (RI) approximation.33 The CASSCF and CASPT2 calculations were performed with the MOLPRO program package.34 Cs symmetry was enforced throughout the calculations, except for the optimization of the ground-state

3

Ac−H 2O + hv → Ac*−H 2O

(1)

Ac*−H 2O → AcH• + OH•

(2)

•

The resulting AcH radical is the N-hydrogenated acridinyl radical (usually referred to as the C-radical in the literature15). If the surplus H atom can efficiently be photodetached from the AcH• radical AcH• + hv → Ac + H•

(3)

the chromophore Ac is recovered and thus becomes a photocatalyzer. It should be noted that the AcH• radical acts as a photoacid in eq 3. The net effect of the reactions (eqs 1−3) is the decomposition of the H2O molecule in the Ac−H2O complex into H• and OH• radicals by the sequential absorption of two photons. The photoinduced H atom-abstraction reaction, eq 2, may occur in the singlet manifold of the Ac− H2O complex or, after intersystem crossing (ISC), in the triplet manifold. Therefore, the potential-energy (PE) surfaces of both manifolds were characterized in the present investigation.

2. COMPUTATIONAL METHODS The second-order Møller−Plesset (MP2) method was employed for the determination of the ground-state equilibrium geometries of Ac, the Ac−H2O complex, and the AcH• radical. Vertical excitation energies were calculated with the complete10665

DOI: 10.1021/acs.jpcb.5b04833 J. Phys. Chem. B 2015, 119, 10664−10672

Article

The Journal of Physical Chemistry B equilibrium geometry and of conical intersections. Dunning’s correlation-consistent split-valence double-ζ basis set with polarization functions on all atoms (cc-pVDZ)35 was employed in the calculations for the Ac−H2O complex. For the AcH• radical, the augmented aug-cc-pVDZ basis was employed, since diffuse basis functions are essential in this case due to the Rydberg character of the 2πσ* state.

structure is 0.098 eV higher than the energy of the lowestenergy structure. Because the calculation of excited states and their PE surfaces is significantly facilitated in Cs symmetry, we enforced Cs symmetry (coplanarity of H2O and Ac) in the present calculations, if not indicated otherwise. The excitation energies, electronic structures, and transition dipole moments of the low-lying singlet and triplet excited states of Ac were determined by Rubio-Pons et al.36 with the CASPT2 method. In the present context, the two lowest 1ππ* states, the lowest 1nπ* state, as well as the corresponding triplet states are of interest. The vertical excitation energies of the six lowest excited states of the Ac−H2O complex, calculated with the ADC(2) and CASPT2 methods, are given in Table 1. For comparison, the vertial excitation energies of the isolated Ac chromophore are also given. The present vertical excitation energies for Ac differ marginally from those of ref 36 due to a slightly smaller active space, a different basis set, and the use of single-state CASPT2 rather than multistate CASPT2. In the present work, the active space has been carefully optimized to describe the locally excited states of Ac and the charge-transfer (CT) excited states of the Ac−H2O complex in a balanced manner, rather than to yield the highest acuracy for the vertical excitation energies for isolated Ac. The two lowest excited singlet states of the Ac−H2O complex are the 11ππ*(Lb) and 21ππ*(La) states with vertical excitation energies of 3.36 and 3.63 eV at the CASPT2 level. According to the CASPT2 results, they are slightly red-shifted relative to the 1ππ* states of isolated Ac. The 11nπ* excited state is predicted to lie above the two lowest 1ππ* states both in Ac and in Ac−H2O. As expected, the 11nπ* state is significantly blue-shifted (by 0.33 eV) by hydrogen bonding of Ac with water. The ADC(2) excitation energies for ππ* excited states are higher than those predicted by CASPT2 by several tenths of an electronvolt, which is a generally observed trend.30 The CASPT2 vertical excitation energy of the 11ππ* state of Ac is in excellent agreement with experiment.7 The two 1ππ* states of Ac−H2O have moderate oscillator strengths (≈0.1) like the 1 ππ* states of isolated Ac, while the 1nπ* state is essentially dark. The lowest excited state of the Ac−H2O complex is the 13ππ* state with an excitation energy of 2.19 eV at the CASPT2 level. The energy of the 3nπ* state, 4.04 eV, is much higher.

3. RESULTS 3.1. Ground-State Equilibrium Structure and Vertical Excitation Energies of the Ac−Water Complex. The lowest-energy structure of the Ac−H2O complex optimized at the MP2 level is shown in Figure 1a. The H2O molecule is

Figure 1. (a) Equilibrium geometry of the electronic ground state of Ac−H2O hydrogen-bonded complex (lowest-energy structure). (b) Structure of Ac−H2O obtained with Cs symmetry constraint.

hydrogen-bonded as H atom donor to the N atom of Ac. The H2O molecule is oriented out-of-plane with respect to the Ac chromophore and is slightly tilted in the plane due to a weak hydrogen bond between the O atom of H2O and the H atom of the neighboring CH group of Ac, see Figure 1a. The calculated bond length RNH of the Ac−H2O hydrogen bond is 1.951 Å. There exists an additional local minimum on the ground-state PE surface in which the H2O molecule is coplanar with Ac (Cs symmetry, see Figure 1b). The hydrogen-bond length of the Cs structure is longer than the hydrogen-bond length of the lowest-energy structure by 0.040 Å. The energy of the Cs

Table 1. Vertical Excitation Energies (in eV) and Oscillator Strengths (in Parentheses) of the Excited Triplet and Singlet States of Acridine and the Acridine−Water Complex acridine ADC(2) CASPT2 CASSCF others

13ππ*

23ππ*

3

nπ*

13ππ*

23ππ*

3.65 3.68 4.25 3.59a

2.42 2.27 2.45 2.48a acridine

3.68 3.32 3.94 3.32a

3.87 4.04 4.87

2.42 2.19 2.51

3.67 3.22 3.87

acridine−water

nπ*

1 ππ*

2 ππ*

3.92(0.001) 3.98 4.52(0.005) 3.87(0.003)a

3.76(0.089) 3.56 4.08(0.033) 3.58(0.120)a 3.5b

3.90(0.046) 3.75 5.38(0.096) 3.77(0.101)a

1

ADC(2) CASPT2 CASSCF others

acridine−water

nπ*

3

1

1

nπ*

11ππ*

21ππ*

4.12(0.001) 4.31 5.15(0.006)

3.70(0.092) 3.36 4.03(0.036)

3.90(0.050) 3.63 4.85(0.090)

1

Reference 36. MS-CASPT2 with CAS (12e,13o). bReference 7. Absorption spectrum of acridine (1 × 10−4 M) in deaerated water at room temperature and pressure, at pH = 11. a

10666

DOI: 10.1021/acs.jpcb.5b04833 J. Phys. Chem. B 2015, 119, 10664−10672

Article

The Journal of Physical Chemistry B Like in the singlet manifold, the 3ππ* states of Ac−H2O are red-shifted, and the 3nπ* states are blue-shifted by hydrogen bonding of Ac with water. The molecular orbitals involved in the six relevant excited states of the Ac−H2O complex are displayed in Figure 2. It can

The relaxed scans for the H atom transfer reaction from H2O to Ac were constructed at the ADC(2) level (see section 2). We emphasize that it is essential to optimize the geometry with respect to the remaining internal coordinates of the complex. Otherwise, substantial barriers would be predicted for the H atom-transfer reaction.37 The reaction coordinate is the bond length ROH of the OH group of water involved in hydrogen bonding with the N atom of Ac. The resulting energy profiles of the singlet and triplet excited states are shown in Figure 3a,b,

Figure 2. Frontier molecular orbitals involved in the lowest 1nπ*, 1 ππ*, 3nπ*, and 3ππ* excited states of the planar Ac−H2O complex. Orbitals 6a″, 7a″, and 8a″ are occupied π orbitals, and orbital 44a′ is the occupied n orbital, while orbitals 9a″, 10a″, and 11a″ are unoccupied π* orbitals.

Figure 3. Energy profiles of minimum-energy reaction paths for hydrogen transfer from water to acridine. (a) Singlet states. (b) Triplet states.

respectively. Small values of the OH bond length (ROH ≈ 1.0 Å) correspond to the equilibrium geometry of the Ac−H2O complex. Large values of the OH bond length (ROH ≈ 2.0 Å) correspond to AcH•−OH• biradicals. The energy profiles to the left of the vertical dashed line at 1.2 Å in Figure 3 were calculated for a reaction path which is optimized in the electronic ground state (black dots in Figure 3). The energy profiles to the right of the vertical dashed line were calculated for a reaction path which was optimized in the lowest ππ* state of CT character (red squares in Figure 3). The energies of the remaining states were calculated at the geometries of these optimized reaction paths. The discontinuities of the electronic PE profiles at the vertical dashed line in Figure 3 reflect the different geometries of these reaction paths in other internal coordinates. The energy profiles to the left of the dashed line represent the ordering of the excited states in the Franck−Condon (FC) region of the Ac−H2O complex. In the singlet and triplet manifolds, the lowest two ππ* states are located below the lowest nπ* state. In the triplet manifold, the 13ππ* state is much lower in energy than all other excited states, as expected for extended π-conjugated systems. In the FC region, the PE functions of all excited states are parallel to the ground-state PE function. These states are thus nonreactive with respect to H atom transfer from the H2O molecule to the chromophore. The ππ* and nπ* PE functions for ROH > 1.2 Å in Figure 3 represent CT states in which an electron has been transferred from H2O to Ac. While these CT states are located above the locally excited states of the Ac chromophore in the FC region, they are strongly stabilized (by about 1.5 eV) by the transfer of a proton from water to Ac (the proton follows the electron). In the electronic ground state, on the other hand, the transfer of a proton from H2O to Ac is energetically highly unfavorable. The ground-state energy increases by more than 4 eV (at the ADC(2) level as well as at the CASPT2 level). The stabilization of the CT states and the destabilization of the S0 state lead to

be seen that the π orbitals (6a″, 7a″, 8a″) and the π* orbitals (9a″, 10a″, 11a″) are localized on the Ac chromophore. They do not mix with the p orbital of π symmetry on the H2O molecule. The wave functions of the 13ππ* and 11ππ* states correspond to the promotion of an electron from the highest occupied π orbital (8a″) to the lowest unoccupied π* orbital (9a″). The wave function of the 21ππ* state is a mixture of configurations involving the 7a″(π) → 9a″(π*) and 8a″(π) → 10a″(π*) excitations. The wave function of the 23ππ* state is a mixture of the 6a″(π) → 9a″(π*) and 8a″(π) → 11a″(π*) excitations. In contrast to the π and π* orbitals, the n orbital (44a′) involved in the 1nπ* and 3nπ* states is substantially delocalized over the Ac and H2O moieties, see Figure 2. As a result, the 1nπ* and 3nπ* states exhibit partial CT character (from H2O to Ac). It is the stabilization of the n orbital by its delocalization over Ac and H2O which leads to the significant blue-shift of the nπ* excitation energy in the singlet and triplet manifolds. 3.2. Minimum-Energy Reaction Paths and PE Profiles for Excited-State H Atom Transfer in the Ac−H2O Complex. Evidence for the possibility of H atom transfer from protic solvent molecules to electronically excited Nheteroaromatic chromophores was found earlier with firstprinciples computational methods for the pyridine−ammonia37 and the pyridine−water complexes.23,24,38 The reaction is mediated by electronic states of solvent-to-chromophore CT character. The energies of these CT states are stabilized in energy by the transfer of a proton from the solvent to the chromophore. The transfer of the proton neutralizes the electronic charge separation of the CT states. This so-called electron-driven proton-transfer reaction37 results in electronic states of biradical character (with one unpaired electron on the solvent molecule and one on the chromophore).37 10667

DOI: 10.1021/acs.jpcb.5b04833 J. Phys. Chem. B 2015, 119, 10664−10672

Article

The Journal of Physical Chemistry B

has C1 symmetry, with the OH• radical being oriented nearly perpendicular to the plane of Ac. 3.3. PE Surfaces in the Vicinity of the Barrier for Hydrogen-Atom Transfer in the Ac−H2O Complex. The one-dimensional energy profiles of the minimum-energy paths for H atom transfer in Figure 3 indicate that the PE surfaces of the locally excited states are adiabatically connected with the PE surfaces of the corresponding CT states via low barriers. The most relevant nuclear coordinates for the description of the H atom-transfer dynamics are the proton-transfer coordinate ROH and the distance of RON of the H atom donor (O) and acceptor (N) atoms. To characterize the barriers separating the reaction path in the S0 state from the reaction path in the CT states, relaxed two-dimensional PE surfaces were calculated; that is, for fixed ROH and RON the energies of the excited states were optimized with respect to the remaining internal coordinates of the complex. Four PE surfaces in the barrier region for H atom transfer were thus obtained for the lowest 1ππ*, 1nπ*, 3ππ*, and 3nπ* states. They are shown as contour plots in Figure 6a− d, respectively. The four contour plots of Figure 6 exhibit the trough corresponding to the energy minimum of the locally excited state in the FC region in the upper left corner. The corresponding AcH•−OH• biradical is represented by the deep valley in the lower right corner. The location of the saddle point for the H atom-transfer reaction is indicated by the red star. The 1ππ* PE surface has a minimum in the FC region which is about 3.5 eV above the S0 minimum. The FC minimum of the 1nπ* PE surface is 0.1 eV below the FC minimum of the 1ππ* state. The 3nπ* PE surface lies slightly (≈0.1 eV) below the 1nπ* PE surface in the FC region. As mentioned above, the 3ππ* PE surface is much lower in energy in the FC region and exhibits a minimum about 2.2 eV above the S0 minimum. We estimated the location and the energy of the saddle points on the 1ππ*, 1nπ*, 3ππ*, and 3nπ* PE surfaces of Ac− H2O from the relaxed two-dimensional PE surfaces. The estimated barrier heights of the 1ππ*, 1nπ*, 3ππ*, and 3nπ* surfaces with respect to the FC minima are 0.3, 0.2, 0.8 and 0.3 eV, respectively. The barrier heights of the 1ππ*, 1nπ*, and 3 nπ* states are nearly the same as the corresponding barrier heights for the Py−H2O complex.23,24 The energy of the locally excited 3ππ* state, on the other hand, is particularly low in Ac− H2O. As a consequence, the 3ππ* surface exhibits a significantly higher barrier for H atom transfer of about 0.8 eV (Figure 6c). 3.4. Ground-State Equilibrium Structure and Vertical Excitation Energies of the Acridinyl Radical. The equilibrium geometry of the AcH• radical in the D0(2A″) ground state exhibits C2v symmetry. The vertical excitation energies of the four lowest excited states of the AcH• radical, calculated with the CASSCF and CASPT2 methods, are given in Table 2. The vertical excitation energy of the lowest 2ππ* excited state is 2.65 eV at the CASPT2 level. This state does not carry oscillator strength. The vertical excitation energies of the following two 2ππ* excited states are 2.67 and 3.00 eV at the CASPT2 level. The oscillator strengths of the latter states are comparable to those of the 1ππ* states of acridine. The third excited state is the lowest 2πσ* state with a vertical excitation energy of 3.15 eV. Like the lowest 2ππ* excited state, this state is a dark state. The molecular orbitals involved in the excited electronic states of the AcH• radical obtained by an unrestricted Hartree− Fock calculation are displayed in Figure 7. For clarity, only the

crossings of the energies of the CT states with the energy of the S0 state. At the CASPT2 level, the intersection of the 1,3ππ* (1,3nπ*) CT state with the ground state occurs at ROH ≈ 1.7 Å (ROH ≈ 1.8 Å) at an energy of about 2.9 eV (3.1 eV) above the S0 minimum. For all ROH > 1.8 Å, the energies of the CT states are below the energy of the closed-shell S0 state, see Figure 3. After the transfer of the proton, the CT states represent neutral AcH•−OH• biradical states. The structure of the biradical at ROH = 2.1 Å is shown in Figure 4a. The molecular orbitals at

Figure 4. (a) Equilibrium geometry of the AcH•−OH• biradical. (b) Molecular structure of the five-state conical intersection.

Figure 5. Frontier molecular orbitals of the AcH•−OH• biradicals.

this geometry are shown in Figure 5. The π orbital is the pz orbital on the O atom of H2O, while the π* orbital is localized on Ac (it is essentially identical with the 9a″ orbital in Figure 2). The n orbital is primarily a px orbital on the O atom of H2O with minor admixtures of the n orbital of Ac (44a′ in Figure 2). At ROH = 2.1 Å, the energies of the 1,3ππ* (1,3nπ*) chargetransfer states are 2.9 eV (3.1 eV) above the S0 minimum at the CASPT2 level. CASPT2 consistently predicts somewhat higher energies of the biradical states than ADC(2). The biradicals with singlet/triplet coupled spins are degenerate since the unpaired electrons are spatially separated. When out-of-plane vibrational displacements are taken into account, the crossings of the energies of the 1,3ππ* (1,3nπ*) biradical states with the energy of the S0 state become conical intersections. We located two three-state conical intersections involving the 1nπ*/1ππ*/S0 states and the 3nπ*/3ππ*/S0 states, respectively, by geometry optimization at the CASSCF level without symmetry constraint (green stars in Figure 3a,b). The structural parameters of the Ac−H2O complex at the nπ*/ππ*/ S0 three-state intersections in the singlet and triplet manifolds are identical. Due to singlet−triplet degeneracy in this region of the PE surface, the green stars in Figure 3 actually mark a fivestate degeneracy: the 1nπ*, 3nπ*, 1ππ*, 3ππ*, and S0 states have identical energies at this geometry. The geometry of the fivestate conical intersection is shown in Figure 4b. The complex 10668

DOI: 10.1021/acs.jpcb.5b04833 J. Phys. Chem. B 2015, 119, 10664−10672

Article

The Journal of Physical Chemistry B

Figure 6. PE surfaces of the lowest 1ππ*(a), 1nπ*(b), 3ππ*(c), and 3nπ*(d) electronic states of the Ac−H2O complex in the vicinity of the barrier for hydrogen transfer from water to acridine.

Table 2. Vertical Excitation Energies (in eV) and Oscillator Strengths (in Parentheses) of the Excited States of the Acridinyl Radicala CASPT2 CASSCF

12ππ*

22ππ*

32ππ*

2.65 3.58(0.000)

2.67 3.52(0.017)

3.00 4.24(0.064)

πσ*

2

3.15 2.82(0.000)

a

The four lowest states of A″ symmetry and the three lowest states of A′ symmetry were state-averaged in the CASSCF calculation.

α-spin orbitals are shown in order of increasing energy. The 9a″ orbital is the singly occupied molecular orbital (SOMO) in the electronic ground state of AcH•. The lowest 12ππ* excited state results from the excitation of an electron from the 8a″(π) orbital to the SOMO. The 22ππ* excited state corresponds to the 8a″ → 10a″ electronic excitation. The 32ππ* excited state is mainly a 8a″ → 11a″ excitation. The 2πσ* excited state results from the promotion of an electron from the 8a″(π) orbital to the unoccupied 41a′(σ*) orbital. Figure 7 shows that this orbital, in contrast to the π and π* orbitals, is localized outside the molecular frame of AcH•. As a consequence, the 2πσ* state possesses a rather large dipole moment of 11.1 D. One can also see that the σ* orbital is antibonding with respect to the NH bond. 3.5. PE Functions for Excited-State Hydrogen Detachment from the Acrdinyl Radical. The PE profiles of the four lowest electronic states of the acridinyl radical as functions of the NH bond length are shown in Figure 8. For clarity, only two of the three closely spaced 2ππ* PE functions are shown in the figure. While the PE functions of the 2ππ* excited states are parallel to the PE function of the D0 state, the 2πσ* state exhibits a very different PE function. The 2πσ* PE function is repulsive apart from a low barrier at ≈ 1.3 Å and intersects the

Figure 7. Frontier molecular orbitals involved in the lowest excited states of the acridinyl radical. Orbitals 7a″ and 8a″ are doubly occupied π orbitals; orbital 9a″ is the singly occupied molecular orbital; orbitals 10a″, 11a″ are unoccupied π orbitals; and orbital 41a′ is the unoccupied σ* orbital. Spin orbitals for α spin are shown.

D0 energy at RNH ≈ 1.7 Å. Asymptotically (RNH > 3.0 Å), the CASPT2 energy is 2.7 eV above the minimum of the D0 state, while the bond dissociation energy of the D0 state is larger than 5 eV (see Figure 8). The large dissociation energy of the D0 state reflects the fact that the H atom is covalently bonded in the AcH• radical (the unpaired electron density is on carbon atoms, see the 9a″ orbital in Figure 7). The dissociation energy of the 2πσ* state, on the other hand, is very low because the 10669

DOI: 10.1021/acs.jpcb.5b04833 J. Phys. Chem. B 2015, 119, 10664−10672

Article

The Journal of Physical Chemistry B

Ac. The energy profiles of these states as functions of the proton-transfer coordinate are shown on the right-hand side of Figure 3a,b for the singlet and triplet manifolds, respectively. These CT states are difficult to find in the FC region of the Ac−H2O complex, because they are strongly mixed with spectroscopic ππ* and nπ* states and their energies are very sensitive to the geometry of the complex. However, they quickly become the lowest singlet and triplet excited states of the Ac−H2O complex when a proton moves from H2O to Ac, see Figure 3. The transfer of the proton neutralizes the electronic charge separation, resulting in AcH•−OH• biradical states. As Figure 3 shows, the biradical states are lower in energy than the closed-shell ground state for ROH > ≈1.6 Å. They thus represent chemically stable species in this region of the PE surface. Because the unpaired electrons are spatially separated, the singlet and triplet biradical states are essentially degenerate. In the vicinity of the dashed vertical line in Figure 3, the PE surfaces of the spectroscopic states are connected with the PE surfaces of the CT states via low-energy barriers. As illustrated in Figure 6a−d, there exists a saddle point (red star) on each of the four energy surfaces which separates the excited-state energy minimum in the FC region (upper left corner) from the valley leading to the AcH•−OH• biradical (lower right corner). The barrier heights with respect to the FC minima are approximately 0.4 eV (1ππ*), 0.3 eV (1nπ*), 0.8 eV (3ππ*), and 0.3 eV (3nπ*). While the barriers of the 1ππ*, 1nπ*, and 3nπ* states are similar, the 3ππ* surface is different. As a consequence of the low energy of the 3ππ* state in the FC region (Figure 3b), the barrier separating the FC region from the CT region is rather high (0.8 eV). Figures 3 and 6 together provide a qualitative mechanistic picture of the photoreaction of Ac with an hydrogen-bonded H2O molecule either in the singlet manifold, or, after ISC, in the triplet manifold. The peak of the H atom transfer barrier on the 1ππ* surface (3.9 eV) lies 0.1 eV below the vertical excitation energy of the 1La state (ADC(2) results). The peak of the barrier on the 1nπ* surface (3.6 eV) is lower than the barrier on the 1ππ* surface. Depending on the efficiency of nπ*−ππ* vibronic coupling, the H atom transfer barrier can thus be overcome by above-barrier dynamics or by H atom tunneling. For barriers of the order of 0.3 eV, tunneling times of the order of 10−100 ps are expected. According to these estimates, the H atom-abstraction reaction may directly occur from the photoexcited singlet states of Ac on picosecond time scales. Because the time scale of ISC is of the same order of magnitude in Ac,19−21 the H atom-transfer reaction may alternatively take place on the triplet PE surfaces after ISC. The peak of the barrrier on the 3nπ* surface is slightly lower (3.6 eV) than the singlet barriers. The 3nπ* surface may thus promote efficient H atom transfer. If ISC to the 3ππ* surface takes place and vibrational energy redistribution is fast, the system may become trapped in the low-lying and long-lived minimum of the 3ππ* state, which seems to be a process with substantial quantum yield for Ac in aqueous solution.3,4,6,16 The accuracy of our calculated excitation energies and H atom transfer barriers is not sufficient to make quantitative predictions of reaction rates and quantum yields. The data provide, however, a comprehensive qualitative picture of the photochemistry of Ac in aqueous solution. The results explain why the “reactive state” in the H atom-abstraction reaction from alcohols could not be agreed upon in the original literature.10−14 In fact, the reaction may take place in any of the

Figure 8. Energy profiles for the photodetachment of the hydrogen atom from the acridinyl radical, calculated at the CASPT2 level. The three lowest states of A″ symmetry and the three lowest states of A′ symmetry were state-averaged in the CASSCF calculation. For clarity, only two of the three closely spaced 2ππ* energy profiles are shown.

highly stable closed-shell Ac molecule is formed by the photodissociation reaction. As a result, the PE function of the 2 πσ* state cuts through the PE function of the ground state at a much lower energy than in related closed-shell photoacids, such as pyrrole, indole, or aniline. While the 2πσ* state of AcH• cannot be excited directly by light, it can be populated by vibronic coupling with the bright 22ππ* and 32ππ* states. When H atom tunneling through the low barrier of the 2πσ* state PE function is taken into account, a photon energy of about 3.0 eV should be sufficient for the direct photodetachment of the hydrogen atom from the AcH• radical (see Figure 8).

4. DISCUSSION The ordering of the vertical singlet excitation energies of the Ac chromophore is predicted as 11ππ*(Lb), 21ππ*(La), and 11nπ* by the ADC(2) and CASPT2 calculations, in agreement with earlier calculations of Pons et al.36 The two 1ππ* states are located below the lowest 1nπ* state at the ground-state equilibrium geometry. Upon geometry optimization, the 1nπ* state becomes the lowest excited singlet state of Ac.36 In the Ac−H2O complex, the 1ππ* states are slightly red-shifted (at the CASPT2 level), while the 1nπ* state is blue-shifted by about 0.3 eV. The 1Lb and 1La states are thus beyond doubt the lowest excited singlet states of Ac in aqueous solution. The origin of the blue-shift of the 1nπ* state upon complexation with water is the stabilization of the n orbital by delocalization of this orbital over the Ac and H2O moieties in the Ac−H2O complex (see Figure 2). Like in the singlet manifold, the two 3ππ* states are slightly red-shifted in the Ac−H2O complex, while the 3nπ* state is blue-shifted (CASPT2 results). Whereas the 23ππ* and 13nπ* states lie just a few tenth of an electronvolt below the corresponding singlet states, the 13ππ* state is more than 1 eV lower in energy than the 11ππ* state. It is by far the lowest excited electronic state in Ac and in the Ac−H2O complex (see Figure 3). The main finding of the present computational studies is the identification and characterization of hitherto unknown 1ππ*, 1 nπ*, 3ππ, and 3nπ* CT excited states which can drive the transfer of a proton from the O atom of water to the N atom of 10670

DOI: 10.1021/acs.jpcb.5b04833 J. Phys. Chem. B 2015, 119, 10664−10672

Article

The Journal of Physical Chemistry B nπ*, 1ππ*, 3nπ*, or 3ππ* PE surfaces and the microscopic processes of nπ*−ππ* vibronic coupling, ISC, above-barrier dynamics, and barrier-tunneling dynamics compete with each other. For ROH beyond the H atom transfer barriers located between 1.05 and 1.30 Å (see Figure 6), the singlet and triplet energy surfaces quickly become degenerate due to the biradicalic character of the excited electronic states. Since singlet−triplet degeneracy enhances ISC, singlet biradicals of nπ* or ππ* character may efficiently be converted into triplet biradicals during the H atom transfer reaction. While the H atom transfer reaction is expected to be partially aborted on the singlet biradical surfaces at the conical intersection with the S0 state, leading to internal conversion to the ground state of the Ac−H2O complex, wave packets on the triplet surfaces will cross the S0 surface at the conical intersection essentially without perturbation, because the SO coupling is weak. The conversion of singlet biradicals into triplet biradicals during the H atom transfer reaction is thus expected to enhance the efficiency of water photooxidation by the Ac chromophore. The existence of the long-lived 3ππ* state in Ac−H2O indicates the possibility of a two-photon excitation process, wherein the first photon excites one of the 1ππ* states. After ISC and internal conversion to the long-lived 3ππ* state, a triplet−triplet absorption20,39,40 may take the system to a higher triplet state, from which a barrierless H atom-abstraction reaction readily can take place. While more input of excitation energy is required in this scheme, the advantages of this scheme are the presumably optimal values of the quantum yields of all processes. We turn now to the second step of the water-splitting reaction, eq 3. The mechanism of the H atom photodetachment reaction from the AcH• radical is illustrated by Figure 8. While the 2ππ* excited states of the radical are bound with respect to the NH bond length, the PE function of the 2πσ* state is, apart from a low barrier, repulsive. It crosses the PE functions of the 2ππ* states as well as the PE function of the D0 state. These apparent crossings are conical intersections, since the degeneracy at the crossing point is lifted by out-of-plane vibrational modes. It is an important finding of the present work that both the vertical excitation energy as well as the dissociation energy of the 2πσ* state are unusually low in the AcH• radical. Although the 2πσ* state cannot be excited directly by light, it can be populated via the excitation of the two bright 2ππ* states with excitation energies near 3.0 eV. The vibronically induced predissociation of these 2ππ* states by the 2πσ* state opens a channel for direct and fast (that is, nonstatistical) H atom photodetachment via the conical intersection of the 2πσ* state with the D0 state. Thanks to the unusually low dissociation energy of the 2πσ* state, the AcH• radical can presumably be efficiently photodissociated with visible light. While H atom photodetachment via repulsive 1πσ* states is an experimentally41,42 as well as theoretically43−45 well-characterized phenomenon in closed-shell photoacids, this photoreaction seems to be experimentally as well as theoretically unexplored for radicals. A specific feature of the photodissociation via 2πσ* states in radicals is the generic existence of a low-lying dark 2 ππ* state, arising from the excitation of an electron from the highest doubly occupied π orbital to the singly occupied π orbital (the 12ππ* state in Table 2). The PE surface of this lowest 2ππ* state has to be crossed diabatically by the photoexcited wave packet for direct photodissociation to 1

occur, see Figure 8. Since experimental spectroscopic studies are very difficult for radicals, theoretical first-principles dynamics investigations of the photodissociation of AcH• and related radicals are needed.

5. CONCLUSIONS The mechanisms of the H atom-abstraction photochemistry of Ac in aqueous solution were explored with ab initio computational methods, adopting the hydrogen-bonded Ac− H2O complex as a model system. We constructed minimumenergy reaction paths, their energy profiles, and two-dimensional relaxed PE surfaces for H atom transfer from water to Ac in the lowest ππ* and nπ* singlet and triplet excited states. The results reveal the mechanisms by which photoexcited Ac can abstract an H atom from water via an electron-driven protontransfer process. The resulting AcH• and OH• radicals are chemically stable species. The AcH• radical can be photodissociated with another low-energy photon (≈3.0 eV). The photodissociation of AcH• regenerates the Ac chromophore, which thus becomes a photocatalyzer. In this way, a water molecule can be decomposed into free H• and OH• radicals by the sequential absorption of two visible/UV photons. The available experimental data indicate that Ac can photoabstract H atoms from hydrocarbons and alcohols, albeit not from water.3,6,12−14 However, the oxidation potential of Ac can be manipulated by substitution, e.g., by substitution with electron-withdrawing groups. Electron-withdrawing substituents lower the energy of the CT states, and thus the barriers for H atom transfer from water to the chromophore. While Ac absorbs at 355 nm, which is near the high-energy cutoff of sunlight at the surface of earth, it is the lead structure of a large variety of organic dyes which absorb in the visible, such as acridine orange, acridine yellow, or benzacridine. Our results thus indicate that Ac is a promising model system for photocatalysis which may be systematically developed toward a simple, cheap, and efficient photocatalyzer for solar water splitting.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by the National Natural Science Foundation of China (X.L., Grant No. 61177017), by the National Science Center of Poland (A.L.S., Grant No 2012/04/ A/ST2/00100), and by a grant of the Deutsche Forschungsgemeinschaft (W.D.). A.L.S. acknowledges support by the DFG cluster of excellence “Munich Centre for Advanced Photonics”.

■

REFERENCES

(1) Bowen, E. J.; Holder, N. J.; Woodger, G. B. Hydrogen Bonding of Excited States. J. Phys. Chem. 1962, 66, 2491−2492. (2) Gafni, A.; Brand, L. Excited State Proton Transfer of Acridine Studied by Nanosecond Fluorometry. Chem. Phys. Lett. 1978, 58, 346−350. (3) Kasama, K.; Kikuchi, K.; Yamamoto, S.; Koji, U.; Nishida, Y.; Kokubun, H. Relaxation Mechanism of Excited Acridine in Nonreactive Solvents. J. Phys. Chem. 1981, 85, 1291−1296.

10671

DOI: 10.1021/acs.jpcb.5b04833 J. Phys. Chem. B 2015, 119, 10664−10672

Article

The Journal of Physical Chemistry B (4) Kasama, K.; Kikuchi, K.; Nishida, Y.; Kokubun, H. Deactivation Mechanism of Excited Acridine and 9-Substituted Acridines in Water. J. Phys. Chem. 1981, 85, 4148−4153. (5) Diverdi, L. A.; Topp, M. R. Subnanosecond Time-Resolved Fluorescence of Acridine in Solution. J. Phys. Chem. 1984, 88, 3447− 3451. (6) Kikuchi, K.; Kasama, K.; Kanemoto, A.; Koji, U.; Kokubun, H. Reaction and Deactivation of Excited Acridine in Ethanol. J. Phys. Chem. 1985, 89, 868−871. (7) Ryan, E. T.; Xiang, T.; Johnston, K. P.; Fox, M. A. Absorption and Fluorescence Studies of Acridine in Subcritical and Supercritical Water. J. Phys. Chem. A 1997, 101, 1827−1835. (8) Weller, A. Fast Reactions of Excited State Molecules. Prog. React. Kinet. 1961, 1, 188−214. (9) Rehm, D.; Weller, A. Kinetics of Fluorescence Quenching by Electron and H-Atom Transfer. Isr. J. Chem. 1970, 8, 259−271. (10) Kellmann, A.; Dubois, J. T. Photoreactive State of Acridine in Solution. J. Chem. Phys. 1965, 42, 2518−2522. (11) Vander Donckt, E.; Porter, G. Role of the 3(n-π*) State in the Photoreduction of Acridine. J. Chem. Phys. 1967, 46, 1173−1175. (12) Wilkinson, F.; Dubois, J. T. Reactive State in the Photoreduction of Acridine in Ethanol. J. Chem. Phys. 1968, 48, 2651−2654. (13) Koizumi, M.; Ikeda, Y.; Iwaoka, T. Reactive State of Acridine in the Photoreduction in Alcohols. J. Chem. Phys. 1968, 48, 1869. (14) Whitten, D. G.; Lee, Y. J. Photochemistry of Aza Aromatics. Identification of the Reactive Intermediate in the Photoreduction of Acridine. J. Am. Chem. Soc. 1971, 93, 961−966. (15) Kellmann, A.; Tfibel, F. Radicals Produced from the LaserInduced Photoionization of Acridine in Solution. J. Photochem. 1982, 18, 81−88. (16) Kellmann, A. Intersystem Crossing and Internal Conversion Quantum Yields of Acridine in Polar and Nonpolar Solvents. J. Phys. Chem. 1977, 81, 1195−1198. (17) Ito, O.; Ito, E.; Yoshikawa, Y.; Watanabe, A.; Kokubun, H. Preferential Solvation studied by the Fluorescence Lifetime of Acridine in Water-Alcohol Mixtures. J. Chem. Soc., Faraday Trans. 1996, 92, 227−230. (18) Madej, S. L.; Okajima, S.; Lim, E. C. ″Proximity Effects″ in Radiationless Transitions of Aromatic Molecules with Nonbonding Electrons. J. Chem. Phys. 1976, 65, 1219−1220. (19) Sundstrom, V.; Rentzepis, P. M.; Lim, E. C. Electronic Relaxation of Acridine as Studied by Picosecond Spectroscopy. J. Chem. Phys. 1977, 66, 4287−4293. (20) Hirata, Y.; Tanaka, I. Build-Up of T-T Absorption of Acridine and Phenazine in the Gas Phase Following Singlet Excitation with a Picosecond Pulse. Chem. Phys. 1977, 25, 381−385. (21) Noe, L. J.; Degenkolb, E. O.; Rentzepis, P. M. Dependence of Radiationless Relaxation in Acridine on Solvent Using Picosecond Spectroscopy. J. Chem. Phys. 1978, 68, 4435−4438. (22) Kellmann, A.; Tfibel, F. Laser-Induced Two-Photon Ionization of Acridine in Aqueous Solution. Chem. Phys. Lett. 1980, 69, 61−65. (23) Liu, X.; Sobolewski, A. L.; Borelli, R.; Domcke, W. Computational Investigation of the Photoinduced Homolytic Dissociation of Water in the Pyridine-Water Complex. Phys. Chem. Chem. Phys. 2013, 15, 5957−5966. (24) Liu, X.; Sobolewski, A. L.; Domcke, W. Photoinduced Oxidation of Water in the Pyridine-Water Complex: Comparison of the Singlet and Triplet Photochemistries. J. Phys. Chem. A 2014, 118, 7788−7795. (25) Shepard, R. The Multiconfiguration Self-Consistent Field Method. Adv. Chem. Phys. 1987, 69, 63−200. (26) Andersson, K.; Roos, B. O. Multiconfigurational Second-Order Perturbation Theory. In Modern Electronic Structure Theory; Yarkony, D. R., Ed.; World Scientific: Singapore, 1995; Vol. 1; pp 55. (27) Schirmer, J. Beyond the Random-Phase Approximation: A New Approximation Scheme for the Polarization Propagator. Phys. Rev. A: At., Mol., Opt. Phys. 1982, 26, 2395−1416. (28) Trofimov, A. B.; Schirmer, J. An Efficient Polarization Propagator Approach to Valence Electron Excitation Spectra. J. Phys. B: At., Mol. Opt. Phys. 1995, 28, 2299−2324.

(29) Christiansen, O.; Koch, H.; Jørgensen, P. The Second-Order Approximate Coupled Cluster Singles and Doubles Model CC2. Chem. Phys. Lett. 1995, 243, 409−418. (30) Wormit, M.; Rehn, D. R.; Harbach, P. H. P.; Wenzel, J.; Krauter, C. M.; Epifanovski, E.; Dreuw, A. Investigating Excited Electronic States Using the Algebraic Diagrammatic Construction (ADC) Approach of the Polarization Propagator. Mol. Phys. 2014, 112, 774−784. (31) Levine, B. G.; Coe, J. D.; Martinez, T. J. Optimizing Conical Intersections Without Derivative Coupling Vectors: Application to Multistate Multireference Second-Order Perturbation Theory. J. Phys. Chem. B 2008, 112, 405−413. (32) Ahlrichs, R.; Bär, M.; Häser, M.; Horn, H.; Kölmel, C. Electronic Structure Calculations on Workstation Computers: The Program System TURBOMOLE. Chem. Phys. Lett. 1989, 162, 165− 169. (33) Hättig, C.; Weigend, F. CC2 Excitation Energy Calculations on Large Molecules Using the Resolution of the Identity Approximation. J. Chem. Phys. 2000, 113, 5154−5161. (34) Werner, H. J.; Knowles, P.; Knizia, G.; Manby, F. R.; Schütz, M.; Celani, P.; Korona, T.; Lindh, R.; Mitrushenkov, A.; Rauhut, G. MOLPRO 2006.1, a Package of Ab Initio Programs; University College Cardiff Cons. Lt.: Cardiff, Wales, 2006. (35) Dunning, T. H., Jr. Gaussian Basis Sets for Use in Correlated Molecular Calculations. I. The Atoms Boron Through Neon and Hydrogen. J. Chem. Phys. 1989, 90, 1007−1023. (36) Rubio-Pons, Ò .; Serrano-Andrés, L.; Merchán, M. Theoretical Insight into the Photophysics of Acridine. J. Phys. Chem. A 2001, 105, 9664−9673. (37) Sobolewski, A. L.; Domcke, W. Computational Studies of the Photophysics of Hydrogen-Bonded Molecular Systems. J. Phys. Chem. A 2007, 111, 11725−11745. (38) Reimers, J. R.; Cai, Z. Hydrogen Bonding and Reactivity of Water to Azines in Their S1(nπ*) Electronic States in the Gas Phase and Solution. Phys. Chem. Chem. Phys. 2012, 14, 8791−8802. (39) Nishida, Y.; Kikuchi, K.; Kokubin, H. Protolytic Reactions of Acridine in the Triplet State. J. Photochem. 1980, 13, 75−81. (40) Periasamy, N. The T3(π, π*) State of Acridine. Chem. Phys. Lett. 1983, 99, 322−325. (41) Ashfold, M. N. R.; Cronin, B.; Devine, A. L.; Dixon, R. N.; Nix, M. G. D. The Role of πσ* Excited States in the Photodissociation of Heteroaromatic Molecules. Science 2006, 312, 1637−1640. (42) Ashfold, M. N. R.; King, G. A.; Murdock, D.; Nix, M. G. D.; Oliver, T. A. A.; Sage, A. G. πσ* Excited States in Molecular Photochemistry. Phys. Chem. Chem. Phys. 2010, 12, 1218−1238. (43) Sobolewski, A. L.; Domcke, W. Conical Intersections Induced by Repulsive 1πσ* States in Planar Organic Molecules: Malonaldehyde, Pyrrole and Chlorobenzene as Photochemical Model Systems. Chem. Phys. 2000, 259, 181−191. (44) Lan, Z.; Domcke, W.; Vallet, V.; Sobolewski, A. L.; Mahapatra, S. Time-dependent Quantum Wave-Packet Description of the 1πσ* Photochemistry of Phenol. J. Chem. Phys. 2005, 122, 224315. (45) Vallet, V.; Lan, Z.; Mahapatra, S.; Sobolewski, A. L.; Domcke, W. Photochemistry of Pyrrole: Time-dependent Quantum WavePacket Description of the Dynamics at the 1πσ*-S0 Conical Intersections. J. Chem. Phys. 2005, 123, 144307.

10672

DOI: 10.1021/acs.jpcb.5b04833 J. Phys. Chem. B 2015, 119, 10664−10672