Anthraquinones derivatives have widely been used in the dye industry, biological research, pharmacy, and medicine in the last few decades for their photosensitization and anticancer ability.^[1–3] Amino-substituted anthraquinones are used to composite porphyrin-anthraquinone hybrids, which are efficient photocleavage agents of DNA.^[4–7] The key to efficient photocleavage is the charge transfer efficiency.^[8–13] The –NH₂ and –NO₂ groups play a role in a strong electronic donor and acceptor in intramolecular charge transfer, respectively. When an –NH₂ and/or –NO₂ group is directly bonded to a polycyclic aromatic system, the photophysics and photochemistry of the molecule are drastically changed by this substituent. Molecular systems with –NH₂ and/or –NO₂ group are of great interest and widely studied for their charge transfer character.^[14–18] The intramolecular charge transfer (ICT) of compounds containing the –NH₂ group, 4-dimethylaminobenzonitrile (DMABN) and related molecules, are widely studied by experiments as well as quantum calculations.^[19–24] The dual fluorescence of DMABN in polar solvents is observed and explained as an intramolecular charge transfer phenomenon.^[19] It consists of emissions from a local excited (LE) and ICT states.^[25] Several models of the reaction were proposed to reveal the influence of the molecular structure on the ICT state: ICT state with perpendicularly twisted amino and benzonitrile moieties (TICT),^[26] ICT state with pyramidalized amino group (WICT),^[27] ICT state with a rehybridized nonlinear C–CN group (RICT),^[28] and planar ICT state (PICT).^[29] However, the twisted intramolecular charge transfer (TICT) model is widely accepted.^{[19–22,30,31]} A great majority of theoretical studies indicate that the –NH₂ and/or –NO₂ group twists as the reaction coordinates along which the ICT reaction occurs.^[24,32] Köhn et al. pointed out that the twisting of the dimethylamino group is the dominant reaction coordinate by using correlated CC2 calculations.^[24]

The amino anthraquinone derivatives are used as catalytic DNA photonucleases for their special charge and structure character.^[5,33–37] 1-aminoanthraquinone (1-NH₂-AQ) is the simplest anthraquinone derivative and has two chromophoric groups, namely anthraquinone and the amino groups. The amino (–NH₂) group in 1-NH₂-AQ is a strong electron donor and induces an ICT property after an excitation. The ICT character with the excitation to the S₁ state has been proved by using Shpol’skii matrix fluorescence and fluorescence excitation spectra.^[38,39] From CCl₄ to more polar solvents, the fluorescence lifetimes of 1-NH₂-AQ were determined to be 400 ps ∼2000 ps, and fluorescence quantum yields were measured to be 0.006 ∼ 0.0715 in different solvents by steady absorption and time-resolved spectra.^[40,41] It is interesting that no dual fluorescence was observed in various solvents, but the absorption and fluorescence spectra, fluorescence quantum yields are strongly correlative to solvent polarity. The phenomenon of no dual fluorescence was also observed in other ICT molecular systems.^[17,18,42] Although the ICT character has been observed in several experiments in 1-aminoanthraquinone, its theoretical description and discussion are still absent.

Quantum chemical calculations are important tools for theoretical researches.^[43–45] In this paper, the electronically excited states of 1-NH₂-AQ are studied theoretically using ab initio methods with focusing on the mechanism for TICT and the following radiationless processes. Optimized geometries of the ground and excited states are calculated and discerned by several methods with different basis sets. The mechanism of TICT is proposed by dipole moments and the conformational relaxation with the twisted angle of the amino group. The energies and potential surface of the excited states involved in radiationless relaxation are described and discussed in detail. The deactivation of the S₁ state is deduced to be due to the contribution of both intersystem crossing (ISC) and internal conversion (IC) channels.

All calculations in this work were performed with the Gaussian 09 program.^[46] In the gas phase, geometry optimizations at ground state are performed using B3LYP, B3PW91, and MP2 in conjunction with 6-311G and Lanl2dz basis sets, respectively. All optimized geometries were confirmed to be stationary points by vibrational frequencies analysis. The molecular orbitals of 1-NH₂-AQ were calculated using the B3LYP with the 6-311G basis set, and the optimization of the first excited singlet state was also performed using the TDDFT/B3LYP method with the 6-311G basis set. In gas and solvents, the energies, oscillator strengths, and dipole moments of the ground and excited states were calculated as the twisted angle of the –NH₂ group changes. Energies and oscillator strengths of the ground and excited states were obtained under the optimized S₁ structure in order to explain the radiationless relaxation. Thus, the effect of the bulk solvent dielectric on the ground-state geometry and on the excited-state vertical energy were modeled by performing self-consistent reaction field (SCRF) calculations using the polarizable continuum model (PCM) with the integral equation formalism.

The energies and dipole moments of the ground state in the gas phase are calculated by several methods with different basis sets and listed in Table 1. However, all optimized geometries are determined to be planar structures with Cs symmetry and confirmed to be stationary points by vibrational frequencies analysis. The lowest ground state energy is given by DFT/B3LYP/6-311G.

Table 1.

Table 1.

Table 1. Values of energy (in unit a.u., the unit a.u. is short for atomic unit) and dipole moment (D) of the ground state in the gas phase, calculated by several methods with different basis sets. . Method/basis set Energy/a.u. Dipole moment (D) DFT/B3LYP/6-311G –744.1210 2.382 DFT/B3LYP/Lanl2dz –744.0368 2.380 DFT/B3PW91/6-311G –743.8241 2.402 DFT/B3PW91/Lanl2dz –743.7548 2.401 MP2/6-311G –741.1487 1.710 MP2/Lanl2dz –740.9021 1.734

Table 1. Values of energy (in unit a.u., the unit a.u. is short for atomic unit) and dipole moment (D) of the ground state in the gas phase, calculated by several methods with different basis sets. .

The vertical excitation energies, oscillator strengths, orbital transition coefficients, and the percentages of the transition orbits of the singlet excited states in a planar form in the gas phase are calculated by using several methods with different basis sets, and the results are listed in Table 2. The energy of the S₁ state 2.6756 eV at the B3LYP/6-311G level is the closest to the experimental value, which is determined to be 2.6325 eV using time-resolved fluorescence spectroscopy.^[47] Since the B3LYP/6-311G level gives a minimum energy of the ground state and is the most consistent energy value of the S₁ state, the other calculations in this work are performed at the same method level. All calculations of the planar structure reveal that the lowest excited S₁ state is a bright state and mainly originates from a 59←58 transition, whereas the second excited S₂ state mainly derives from a 59←57 transition and is a dark state for its zero transition strength.

Table 2.

Table 2.

Table 2. Values of vertical excitation energy (in unit eV), oscillator strength (f), orbital transition coefficient (k), and the percentage of the transition orbits (%) of 1-NH2-AQ for optimized geometry of the S0 state in the gas phase, calculated using the TDDFT method with different basis sets. . State Transition Value B3LYP/6-311G B3LYP/Lanl2dz B3PW91/6-311G B3PW91/Lanl2dz S1 58–59 eV 2.6757 2.6849 2.6825 2.6970 f 0.1239 0.1260 0.1249 0.1278 k 0.70155 0.70139 0.70167 0.70154 % 98.43 98.38 98.47 98.43 S2 57–59 eV 2.8895 2.8350 2.8732 2.8169 f 0 0 0 0 k 0.68609 0.68521 0.68727 0.68646 % 94.14 93.90 94.47 94.25

Table 2. Values of vertical excitation energy (in unit eV), oscillator strength (f), orbital transition coefficient (k), and the percentage of the transition orbits (%) of 1-NH2-AQ for optimized geometry of the S0 state in the gas phase, calculated using the TDDFT method with different basis sets. .

Optimized structures of the ground and first excited singlet states of 1-NH₂-AQ in the gas phase are shown in Figs. 1(a) and 1(b), respectively. The lowest excited singlet state geometry is only optimized in gas for simplicity and the energy gap between the S₀ and S₁ state is only 0.6049 eV for the optimized geometry. The optimized geometry of the S₁ state is no longer planar, but twisted. The amino group is almost perpendicular to the anthraquinone plane. The dihedral angle between the –NH₂ group and anthraquinone for the optimized structure of the S₁ state becomes about 90°. The vector of the dipole moment and the charge of the nitrogen atom are also displayed in Fig. 1. The dipole moment of the optimized S₁ state is more than three times to the dipole moment of the optimized S₀ state. Such a large change of the transition dipole moment manifests a charge transfer process in the molecule upon an excitation. The charge quantity of the nitrogen atom at the optimized S₀ and S₁ states are 0.906 e⁻ and 0.650 e⁻, respectively. This also supports the charge transfer process. The molecular orbits of the highest occupied molecular orbital (HOMO)-1, HOMO and the lowest unoccupied molecular orbital (LUMO) at optimized S₀ and S₁ structures were discussed in detail in the literature.^[48] The calculations show that the excitation to the S₁ state is strongly allowed with the main contribution of the transition from the HOMO to the LUMO. For the optimized geometry of the S₀ state, the HOMO is largely localized in the amino group, whereas the LUMO is largely localized in the whole anthraquinone group. The difference in electron cloud distribution between HOMO and LUMO also shows the ICT process. The significant change of C=O and C–N bond from the S₀ state geometry to the S₁ state geometry originates from the strong donating nature of the amino group, which induces large charge localization in the amino group. The electron density distribution also suggests that the S₁ state is a strong ICT state, and such an ICT state induces a strong absorption in the visible spectrum.^[40]

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. (color online) Optimized conformations of 1-NH2-AQ in (a) ground state and (b) the first excited singlet state in the gas phase. Vectors of dipole moments are displayed by the arrows: (a) × 2 and (b) × 0.7, respectively. Charge quantities of nitrogen atom in S0 and S1 states are 0.906 e− and 0.650 e−, respectively.

Yoshihara et al. pointed out that the absorption and fluorescence spectra are dependent on strong solvent polarity.^[40,41] The calculated steady absorption spectra of 1-NH₂-AQ in three solvents with different polarities at the B3LYP/6-311G level are depicted in Fig. 2. The results show that the calculated absorption spectra are well consistent with the experimental absorptions in the visible region,^[48] and are dependent on the solvent polarity. The strong solvent polarity dependence leads to the charge transfer character of the S₁ state.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. (color online) Absorption spectra of 1-NH2-AQ, calculated at the B3LYP/6-311G level in gas, cyclohexane, dioxane, and ethanol.

Table 3.

Table 3.

Table 3. Values of dipole moment μ (D) of the S0 and S1 states, oscillator strength f of S1 state of 1-NH2-AQ for different values of twisted angle (°) in gas, cyclohexane, dioxane, and ethanol using (TD)DFT/B3LYP/6-311G. . Angle 10 20 30 40 50 60 70 80 90 Gas μ(S0) 2.382 2.371 2.338 2.171 2.156 1.983 1.751 1.485 1.246 1.143 μ(S1) 6.757 6.560 6.214 6.004 5.947 5.997 6.183 6.613 7.333 7.808 f(S1) 0.124 0.113 0.089 0.065 0.047 0.033 0.021 0.012 0.004 0.00 Cyclohexane μ(S0) 2.802 2.788 2.742 2.654 2.506 2.287 2.000 1.671 1.377 1.251 μ(S1) 7.865 7.792 7.600 7.393 7.270 7.266 7.430 7.884 8.687 9.223 f(S1) 0.169 0.160 0.136 0.104 0.075 0.052 0.034 0.018 0.005 0.000 Dioxane μ(S0) 2.861 2.846 2.799 2.708 2.555 2.331 2.035 1.698 1.396 1.266 μ(S1) 8.015 7.950 7.774 7.574 7.447 7.437 7.597 8.053 8.865 9.410 f(S1) 0.171 0.162 0.138 0.106 0.076 0.053 0.034 0.019 0.006 0.000 Ethanol μ(S0) 3.382 3.362 3.300 3.182 2.990 2.711 2.347 1.932 1.561 1.401 μ(S1) 9.337 9.317 9.246 9.117 8.990 8.947 9.083 9.556 10.45 11.05 f(S1) 0.167 0.160 0.138 0.109 0.080 0.057 0.037 0.020 0.006 0.000

Table 3. Values of dipole moment μ (D) of the S0 and S1 states, oscillator strength f of S1 state of 1-NH2-AQ for different values of twisted angle (°) in gas, cyclohexane, dioxane, and ethanol using (TD)DFT/B3LYP/6-311G. .

As mentioned above, the conformation of 1-NH₂-AQ maintains a coplanar structure in the Franck–Condon region after excitation and relaxes to a twisted structure on the potential surface of the S₁ state. The parameters including the dipole moments of the S₀ and S₁ states and the oscillator strengths of the S₁ state as a function of twisted angle between the amino and anthraquinone group for the gas and three solvents are calculated using the B3LYP method with the 6-311G basis set, and the results are listed in Table 3. There is similarity that the oscillator strengths all decrease to zero as the twisted angle increases to 90°. This obviously suggests that the fluorescence just radiates from an initial LE state, and no fluorescence radiates from a relaxed excited state. This fits the experimental phenomenon of no dual fluorescence.^[38,39,47] The dipole moments of the S₀ state decrease as the twisted angle increases, whereas the dipole moments of the S₁ state have the opposite trend. The dipole difference between the S₀ state and the S₁ state increases as the twisted angle increases. These results suggest that the TICT character, and the conformational relaxation on the potential surface of the S₁ state in the gas phase were manifested in detail by the dipole and charge change.^[48] The TICT character is also easily shown by the dipole change trend in solvents.

Table 4.

Table 4.

Table 4. Energies (in unit eV) of S0 state, and vertical excitation energies (in unit eV) of T2 and S1 states of 1-NH2-AQ for different values of twisted angle (in unit degree) in different solvents using B3LYP/6-311G. . Angle Cyclohexane Dioxane Ethanol S0 T2 S1 S0 T2 S1 S0 T2 S1 0 0 2.5066 2.5766 0 2.5126 2.5699 0 2.5600 2.5311 10 0.0340 2.5044 2.5661 0.0503 2.5104 2.5596 0.0336 2.5557 2.521 20 0.1301 2.497 2.5299 0.1462 2.5023 2.5239 0.1284 2.5412 2.4863 30 0.2733 2.4836 2.4611 0.2893 2.4876 2.4558 0.2706 2.5148 2.4202 40 0.4454 2.4718 2.3604 0.4615 2.474 2.3557 0.4435 2.4862 2.3219 50 0.6285 2.4788 2.2333 0.6448 2.4789 2.2291 0.6292 2.4755 2.1961 60 0.8065 2.5144 2.0837 0.8232 2.5127 2.0797 0.8118 2.4949 2.0457 70 0.9612 2.5696 1.9155 0.9785 2.5679 1.9114 0.9718 2.5437 1.874 80 1.0706 2.5715 1.7546 1.0884 2.5737 1.7500 1.0857 2.5751 1.7071 90 1.1108 2.5729 1.6802 1.1288 2.5737 1.6752 1.1277 2.5747 1.6296

Table 4. Energies (in unit eV) of S0 state, and vertical excitation energies (in unit eV) of T2 and S1 states of 1-NH2-AQ for different values of twisted angle (in unit degree) in different solvents using B3LYP/6-311G. .

In the gas phase, the energies of the S₀ and T₂ states increase as the twisted angle increases, whereas the energies of the S₁ and T₁ states decrease. At the B3LYP/6-311G level, the energy gap between the S₁ and S₀ states in the Franck-Condon region is 2.6757 eV, and the energy gap between both states is only 0.6049 eV for the optimized geometry of the S₁ state. It shows that the energy gap decreases from 2.6757 eV to 0.6049 eV along the twisting conformation relaxation coordinate. The obviously decreasing trend of the energy gap is benefitial to an ultrafast IC process from the S₁ state to the S₀ state. Srivatsavoy et al. also pointed out that the S₁ state is mainly deactivated through IC to the ground state experimentally.^[40] The ISC process was presented by the energy inversion between the S₁ and T₂ states nearby the twisted angle of 40° along the twisting relaxation.^[48] For different twisted angles, the energies of the S₀ state, and vertical excitation energies of the T₂ and S₁ states of 1-NH₂-AQ are listed in Table 4 at the B3LYP/6-311G level in three solvents. Here, the energies are calculated just by changing the twisted angle but not for partial optimization for simplicity. In solvents, the vertical excitation energies of the S₁ state also decrease as the twisted angle increases as in gas. The results show that the energy gap between the S₀ and S₁ states decreases with increasing solvent polarity. This provides further evidence for the experimental result that the rate constant of IC increases considerably in a polar solvent, and the fluorescence yield decreases gradually as cyclohexane changes into a more-polar solvent.^[40,47]

There are energy intersections between the T₂ and S₁ states in cyclohexane and dioxane solvents but in the ethanol solvent, which is easily shown in Fig. 3(a). It is clear that the potential energy surfaces of the S₁ and T₂ states become isoenergetic near the twisted angle of 20°–30° in cyclohexane and dioxane solvents. This could explain the experimental result that the rate constant of ISC decreases in polar solvents. The rate constant of ISC was determined to be only 3.0 × 10⁷ s⁻¹ in methanol by picosecond fluorescence spectra.^[40] The rate constant of IC is about 8 times larger than that of ISC, and the rate constant of ISC is about 3 times larger than that of the radiation.^[40,47] Our calculations well support the reported large radiationless rate constant.^[40,41,47]

Fig. 3.

	Figure Option ViewDownloadNew Window
	Fig. 3. (color online) Energies of the T2 and S1 states of 1-NH2-AQ varying with twisted angle in cyclohexane (blue), dioxane (green), and ethanol (red) using B3LYP/6-311G, showing (a) vertical excitation energies of the excited states and (b) energies of the S1 state.

The energies of the S₁ state at different twisted angles in three solvents are shown in Fig. 3(b). The energy barriers of 1-NH₂-AQ are 0.3330, 0.3264, and 0.3136 eV in dioxane, ethanol, and cyclohexane respectively. These barriers appear nearby the twisted angle of 50°, which is similar to the experimental result that the –NH₂ plane is tilted 38° to the ring.^[49] The N–H stretching frequencies in 1-NH₂-AQ^[50] are very close to those in the typical free amino groups.^[51] It suggests a very weak hydrogen bond, which opposes C–N bond rotation. So the –NH₂ twisting happens easily once the excitation energy exceeds the barriers. The energy barrier decreases with solvent viscosity decreasing, and is higher than 0.2595 eV in the gas phase. The timescale of TICT was usually measured to be a few hundred femtoseconds in some barrierless systems.^[17,18] The time of TICT of 1-NH₂-AQ is demonstrated to be about 5 ps in ethanol by using ultrafast transient absorption spectroscopy. The longer time of 5 ps is attributed to the energy barrier existing on the amino twisting potential surface.^[48] From these results, we can infer that the timescales of TICT of 1-NH₂-AQ in cyclohexane and ethanol are shorter than that in dioxane because of the smaller energy barrier.

The twisted intramolecular charge transfer and following radiationless dynamics of the excited states of 1-NH₂-AQ in different solvents have been investigated using ab initio calculations. The optimized structure of the ground state is confirmed to be a planar conformation, whereas the structure of the S₁ state is a twisted conformation with perpendicular amino group to the anthraquinone plane. The difference in dipole moment between the S₀ and S₁ states manifests a charge transfer process in the molecule upon excitation. Furthermore, it is elucidated that the dipole moment in the S₁ state increases with the twisted angle of –NH₂ increasing, i.e., the TICT is associated with the conformational relaxation on the potential surface of the S₁ state. Quantum chemical calculations indicate that the IC and ISC process via the twisting of the amino group, and the two relaxation processes, have been established and are responsible for the ultrafast deactivation of the S₁ state. The rate constants of IC and ISC processes relate to the solvent polarity, and the TICT time is predicted to be positively dependent on solvent viscosity.