Santoro et al. 10.1073/pnas.0703298104. |
Fig. 6. Schematic drawing of the frontier orbitals of the 9-MA monomer.
Fig. 7. Computed absorption spectra of Ado and 2dA in a B-DNA conformation. Each transition is convoluted by a Gaussian with a FWMH of 0.30 eV.
Fig. 8. Schematic drawing of the adduct between 9-MA and four water molecules of the first solvation shell.
Fig. 9. Schematic drawing of the computed energy minima of a B-DNA-like 9-MA dimer in the ground state (a), in its lowest energy CT excited state (SCT) (b), and in its bright state (SB) (c). Bond lengths are shown in Å.
Fig. 10. Schematic drawing of the frontier orbitals of the 9-MA dimer.
Fig. 11. Computed absorption spectra of the four 9-MA stacks that can be extracted by a 9-MA pentamer in a B-DNA conformation. Each transition is convoluted by a Gaussian with a FWMH of 0.30 eV.
Fig. 12. Schematic drawing and bond distances (in Å) in the ground-state minimum of the cation form (Upper) and the anion form (Lower) of 9-MA.
Fig. 13. Computed absorption spectra of the 9-MA monomer dimer and trimer in a B-DNA conformation. Each transition is convoluted by a Gaussian with a FWMH of 0.30 eV.
Fig. 14. Energy plot of the three lowest energy adiabatic and diabatic excited states of (9-MA)2 in the region of the coordinate space connecting the SCT pseudominimum found in the path between the FC point and the SB minimum (Fig. 5) and the absolute SCT energy minimum.
Fig. 15. Energy plot of the three lowest energy adiabatic excited states of a cytosine dimer for different C5-C5' distance. The blue and magenta curves refer to a fully symmetric arrangement (see text for details). TD/PBE0/6-31G(d) calculations including BSSE by the counterpoise method.
Fig. 16. Schematic drawing of the computed energy minima of a B-DNA-like 9-MA dimer in the second excited state (SDEL).
Fig. 17. Electronic coupling absolute values between the two diabatic states SB and SCT along the path connecting the FC point with the SB minimum reported in Fig. 5.
Fig. 18. Energy plot of the three lowest energy adiabatic excited states of (9-MA)2 (a), their oscillator strengths (b), and the corresponding diabatic states (c) of the absolute values of the diabatic coupling between S1 and S3 (d) in the region of the coordinate space connecting the he SB minimum with the SCT minimum.
Fig. 19. Energy plot of the three lowest energy adiabatic excited states of (9-MA)2 (a) and their oscillator strengths (b) in the region of the coordinate space connecting the FC point with the SB minimum for different intermonomer distances.
Fig. 20. Energy plot of the three lowest energy adiabatic excited states of (9-MA)2 (a) and their oscillator strengths (b) in the region of the coordinate space connecting the SB minimum with the SCT minimum for different intermonomer distances.
SI Text
Computational Details
The effect of the basis set superposition error (BSSE) on the excited-state properties of (9-MA)2 has been estimated by gas-phase TD-PBE0/6-31G(d) test calculations, employing the counterpoise method (1), at the FC point and at the minimum of SCT state. In both cases, the effect on the transition energies is quite modest, with a red-shift of the bright transition by ≈0.01 eV. It is thus important to highlight that the prediction of a blue-shift of the dimer absorption band with respect to that of the monomer is not an artifact resulting from BSSE.
Solvation Model.
Bulk solvent effects on the ground and excited states have been taken into account by means of the latest developments in the polarizable continuum model (PCM), by using the default PCM model in Gaussian (cited in the literature as IEF-PCM or IVC-PCM; refs. 2-4). In this model, the solvent, represented by a homogeneous dielectric, is polarized by the solute, placed within a cavity built as the envelope of spheres centered on the solute atoms. In our study, the cavity was built according to the unilateral administrative order model [united atom topological model (5) using universal force field radii]. When discussing solvent effects on absorption spectra, it is useful to define two limit situations, usually referred to as nonequilibrium and equilibrium time regimes (2-4, 6). In the former case, only solvent electronic polarization (fast solvent degrees of freedom) is in equilibrium with the excited-state electron density of the solute, whereas in the equilibrium regime, nuclear degrees of freedom (and thus both fast and slow solvent degrees of freedom) also are equilibrated with the excited-state electron density. The solvent reaction field in the nonequilibrium regime described above depends in the PCM formalism on the dielectric constant at optical frequency (εopt, usually related to the square of the solvent refractive index n, εopt = n2; for water, εopt = 1.776). PCM equilibrium solvation is instead ruled by the static dielectric constant (ε; for water, ε = 78.39). To calculate the vertical excitation energies (VEEs), nonequilibrium solvation energies are more suitable, whereas the opposite is true for calculating fluorescence energies. Even if a rigorous theoretical procedure for the calculation of state-specific (SS) fluorescence energies would require the SS-TD/PCM approach (7), which is still under development, a sufficient degree of accuracy could be obtained by using the emission energies provided by the standard linear response (LR)-TD/PCM implementation.
Energy Scans and Diabatization Procedure.
As stated in the main text, to investigate the possible pathways followed by the system upon excitation, we have performed several one-dimensional (1D) scans of the excited PES. Although constrained to few dimensions, the motion of the 3N-6 nuclear coordinates involves, of course, some degree of arbitrariness; we think that the selected procedure is reliable enough to point out the main qualitative aspects. In particular, we define collective coordinates by linearly interpolating the internal coordinates of two representative structures (e.g., the FC point and the excited-state minima). Along the scanned paths, two of the lowest excited states, namely S1 and S3, strongly mix their electronic character, as revealed by the remarkable exchange of the oscillator strength with the ground electronic state (Feg). It is therefore convenient to resort to a description in terms of diabatic states that preserves unaltered, as much as possible, their electronic character (described as having a large or small Feg). The diabatic states can be easily defined by determining the rotation of the two excited adiabatic states, which creates a diabatic state, SB, showing the full oscillator strength with the ground-state S0 and another diabatic state, SCT, which is on the contrary completely dark (it is assumed that the transition dipole moment between the two excited adiabatic states is vanishingly small). This procedure is, in principle, only applicable when the two transition dipole moments m01 and m03 from the ground state to the two adiabatic states S1 and S3 are parallel. When this is not the case, it is not possible to set to zero all three of the components of the transition dipole moment by a rotation that only relies on a single variable (the rotation angle). We monitored the values of the angle between m01 and m03 along all of the paths where we performed diabatization. In the region of the paths where both the adiabatic states have a significant transition dipole moment, and it is therefore meaningful to define their angle, we found only a slight deviation from a parallel orientation (average value of the angle ≈ 20°). Outside of this region, the deviation is larger but it is of little significance because one of the two transition dipole vectors is very small. This has two consequences: on the one hand, the determination of the angle it forms with the other vector is subjected to large uncertainty and, on the other hand, when this happens the two adiabatic states are already "diabatic" within the definition we adopted, and our diabatization procedure correctly reproduces this feature, although it assumes that the two transition dipole moments are parallel.
Results
Absorption Spectra of 9-MA and Adenosine
. SI Table 2 reports the VEE of the most significant low-energy electronic transitions computed at the PCM/TD-PBE0//PCM/PBE0/6-31G(d) level in aqueous solution. Our computations predict the presence of two strong absorption peaks at ≈5 and ≈6 eV, in good agreement with the experimental spectra but for a small overestimation of the energy of the red-side peak. Three electronic transitions contribute to the lowest energy absorption band. The most intense transition corresponds essentially to a HOMO→LUMO transition (see SI Fig. 6) (hereafter, we refer to the HOMO and LUMO orbitals as H and L, respectively), it has a π/π* character, and it is usually labeled as π*La. There is also another close-lying, less intense π/π* transition, corresponding mainly to a H→L + 1 transition and usually labeled as π* Lb. Finally, the third transition has a n/π* character (H - 1→L excitation) and a very low oscillator strength. This description essentially agrees with that provided by previous CASPT2 studies on adenine (8-12). Indeed, although the exact energy ordering of the above three states varies according to the basis set, the active space, and the ground-state geometry used (8-12), all of the calculations agree in describing the π*La state as the spectroscopic state carrying most of the absorption intensity.
Replacement of the 9-methyl substituent by the deoxyribose ring (adenosine, see Fig. 1) does not significantly change the absorption spectra (see SI Fig. 7), suggesting that 9-MA can be a good model for the study of the monomer excited-state behavior.
Finally, we have checked the effect on the spectra of the inclusion of explicit water molecules, by performing test calculations on a supermolecule formed by 9-MA and four water molecules (see SI Fig. 8) of the first solvation shell (see SI Table 2). The computed VEEs are similar to those obtained by including only the bulk solvent effect on 9-MA. The most significant outcome of the presence of solute/solvent hydrogen bonds is the destabilization by ≈0.3 eV of the n→π* transitions. Analogous to the condition found in uracil-like molecules, this latter kind of transition is blue-shifted in the presence of hydrogen bonds with the solvent molecules. Together with the results of the excited-state geometry optimizations, this result indicates that n→π* transitions should not play any relevant role in polyadenine (polyA) excited-state dynamics, at least in aqueous solution. The lowest energy states in the oligomers are indeed always those derived from π→π* transitions, and the stability of n→π* excited states is even decreased by the presence of explicit hydrogen bonds, with the solvent or with other nucleobases within a DNA double strand.
It is also noteworthy the π*L
atransitions is red-shifted by ≈0.1 eV, leading to a better agreement with the experimental results. Despite these quantitative differences, inclusion of explicit solvent molecules does not significantly change the qualitative description of the spectra. In the calculations on the oligomers we have thus taken into account bulk solvent effects only by using the PCM.
The nuclear coordinates of the most relevant minima are reported in SI Tables 3-8.
Reliability of TD-DFT Results on the CT States of 9-MA Multimers.
Because TD-DFT calculations have often shown failures in the treatment of long-range CT transitions (13-16), it is important to highlight that several elements support the reliability of TD-DFT calculations in predicting that the lowest energy transition in the stacked dimers and trimers of 9-MA have a CT character.
1. The computed absorption and emission spectra are in good agreement with the experimental results. In regard to the absorption spectra, we have shown that the lowest energy transition can have a partial CT character (for example, in the dimers extracted from the pentamer; see SI Fig. 11), depending on small variation in the monomer geometry. It is thus significant that the energy of this transition does not dramatically depend on the CT nature of the transition (as would happen for spurious CT transition) and, most important of all, the position of this weak transition is in full agreement with the experimental spectra of multimers. Analogously, it is noteworthy that the computed emission energy of SCT is in good agreement with that assigned to the excimer state by experiments, but for a slight overestimation, whereas in the processes for which TD-DFT fails, the CT transition energy is usually dramatically underestimated. (13-16)
2. To check the reliability of our methodological procedure in treating the electronic transitions in stacked nucleobases, we have compared the prediction of TD/PBE0 and CASPT2 calculations for the cytosine dimer studied in the gas phase by Merchan and coworkers (17), which is, to the best of our knowledge, the only system comparable to (9-MA)2 for which CASPT2 calculations are available. We have optimized at the PBE0/6-31G(d) level the ground state of a stacked symmetric cytosine dimer for a ring-ring distance of 4 Å. As with their procedure, we have then characterized the two lowest energy transitions by means of TD-PBE0/6-31G(d) calculations for decreasing intermonomer distances, without any further geometry optimizations. SI Fig. 15 clearly shows that TD-PBE0 provides a picture extremely similar to that of CASPT2 calculations. According to TD-PBE0/6-31G(d) calculations, the S1 state has a minimum for a C5-C5' distance of 3.09 Å, the minimum has a depth of ≈0.5 eV with respect to the separated monomers, and an emission energy of 3.65 eV. CASPT2 calculations provide that the minimum has a C5-C5' distance of 3.076 Å, an emission energy of 3.4 eV, and a binding energy of 0.58 eV. TD/PBE0 calculations provide that the S1 transition has a partial CT character, ≈0.25 arbitrary units. Actually, also in a totally symmetric dimer studied by Merchan et al. (17), according to CASPT2(12,12)/ANO calculations the S1 transition can be described by simultaneous intermonomer charge transfer transitions. In our study, the small asymmetry between the two monomers (originated by the formation of a weak NH2|×|×|×|NH2 interaction) accounts for the partial CT transfer predicted by TD-PBE0 calculations. In fact, when the two monomers are forced to adopt the same geometry (see the blue and magenta curves in SI Fig. 15), TD-PBE0 calculations predict that the S1 transition does not exhibit any CT character, even if the behavior of the S1 state in this latter "symmetric" case is extremely similar to that found in the "asymmetric" dimer. This result shows that in the case of stacked cytosine TD-PBE0, results are not biased by any artificial overstabilization of the CT transitions.
As with CASPT2 calculations, S2 can be described by the combination of intramonomer π→π* transitions, exhibiting a very shallow minimum for a C5-C5' distance of ≈3.8 Ε. The excellent agreement between the picture provided by TD-PBE0 and CASPT2 calculations in the treatment of the lowest energy transitions (including that described as simultaneous intermonomer CT) in stacked nucleobases strongly supports the reliability of the results obtained on (9-MA)2. Actually, the failures of TD-DFT calculations are usually associated with long-range CT between two partners whose molecular orbitals have a vanishing overlap (13-16). This is not, obviously, the case for stacked aromatic molecules with a ring-ring distance of ≈3.5 Å, whose molecular orbitals are significantly mixed, and which can thus be considered a supermolecule. This is also confirmed by the experimental absorption spectra of stacked adenines, which are different (blue-shifted and significantly less intense) than for the isolated monomer. The CT reaction can be seen as an intramolecular CT that, other than in the case of the cytosine dimer described above, has been shown to be correctly described by TD-PBE0 calculations in several systems (18-20). Among these, we can recall the case of coumarins and 4-(N,N-dimethylamino)benzonitrile, textbook examples of intramolecular CT transitions. For example, for C153 the S0→S1 0-0 transition energy computed in the gas phase at the TD/PBE0/ level shows a discrepancy of only 200 cm-1 with respect to the experimental result. In regard to 4-(N,N-dimethylamino)benzonitrile, gas-phase TDPBE0/6-311+G(d,p) calculations overestimate the experimental lowest energy transition by 0.14 eV. For comparison, CASPT2/ANO(DZ) calculations underestimate the same quantity by -0.18 eV.
3. We have performed test PCM/CASSCF/6-31G(d) calculations on the SCT and SB minima, for different numbers of electrons and orbitals in the active space, namely PCM/CASSCF(4,4)/6-31G(d), PCM/CASSCF(6,6)/6-31G(d), PCM/CASSCF(6,8)/6-31G(d), and, finally, PCM/CASSCF(8,8)/6-31G(d). This latter active space contains the two lowest energy occupied π* MO and two highest energy unoccupied π* orbitals of each monomer, and thus it should be adequate to obtain a semiquantitative description of the relative energy of the different adiabatic states. Irrespective of the active space used, the conclusion of the TD-DFT calculations are fully confirmed. Indeed, the lowest energy transition at the SCT minimum exhibits a clear CT character (≈1 arbitrary unit) and is more stable than the lowest energy transition at the SB minimum, which corresponds to a ππ* excitation localized on a single monomer. At the PCM/CASSCF(8,8)/6-31G(d) level, the SCT is more stable than SB by ≈0.46 eV. At its minimum, the emission energy of SCT is 4.26 eV. These results are in good agreement with the results of TD-PBE0 calculations, considering that the effect of dynamical correlation is not taken into account by CASSCF calculations.
4. At the first order, the energy of a CT transition between a donor (D) and acceptor(A) is given by:
IP(D) + EA(A) + Eint(D+/A-), [1]
where IP is ionization potential, EA is electron affinity, and Eint(D+/A-) is the electrostatic interaction between the ion pairs (negative). The sum of IP(D)and EA(A) is thus an upper bound for the energy of the CT transition.
PCM/PBE0 calculations in aqueous solution (neglecting the solvated electron) indicate that the vertical IP of 9-MA is 6.01 eV at the PCM/PBE0/6- 31G(d)//PCM/PBE0/6-31G(d) level and 6.20 eV at the PCM/PBE0/6-311+G(2d,2p)//PCM/PBE0/6-31G(d) level, and the adiabatic values (optimizing the cation geometry and taking into account vibrational contributions) are 5.74 eV and 5.92 eV, respectively. Note that recent experiments in water revealed a monophotonic component in the photoionization of DNA and RNA bases excited by a 266-nm (4.66-eV) laser, whose energy is therefore an upper limit for their adiabatic ionization threshold (21). The same authors cite that the solvation energy of an electron in water is -1.2 or -1.65 eV (according to two different models). When summing these values to the data we computed, we obtain an estimate of 4.09-4.54 eV or 4.27-4.72 eV (depending on the basis set) for the adiabatic ionization threshold of 9-MA, in good agreement with the experiment.
The corresponding values for the vertical EA are 0.67 eV [6-31G(d) basis set] and 1.20 eV [6-311+G(2d,2p) basis set], whereas the corresponding adiabatic values are 1.15 eV and 1.59 eV, respectively.
Therefore, the energetic cost for creating a 9-MA CT dimer (IP-EA) at the geometry of the two neutral monomers is in the range 5.34-5.0 eV, depending on the basis set, whereas the corresponding adiabatic values are in the range 4.59-4.33 eV. The above values are of the same order of magnitude as the lowest energy excited states of the (9-MA)2 dimer (4.96-5.20 eV), supporting the possibility that some of the transitions show a CT character. Furthermore, these estimates represent an upper bound for the energetic cost of creating a CT stacked dimer within a polyA chain, given that electrostatic interaction should stabilize an ion pair between stacked nucleobases and that the IP of adenine is lowered by stacking (22). Those considerations further support the presence of a CT state in the same energy region of the lowest energy 9-MA excited states. Furthermore, an analogous analysis of thymine can shed some light on the marginal role of excimers in polyT (vide infra).
5. Several experimental results suggest that the excimer state has a CT nature (23-25).
Second Excited State.
We locate the minimum structure of the S2 (hereafter SDEL) excited state where it keeps an electronic character very similar to that exhibited in the FC region, with the electronic excitation almost always equally delocalized over the two monomers. In fact, in its minimum energy geometry (see SI Fig. 16), SDEL still corresponds to an antisymmetric combination of H - 1→L and H→L + 1 transitions, and the shapes of the orbitals involved in the transition are similar to those of the ground-state energy minimum (S0) and are shown in SI Fig. 9. As could be expected, the most significant geometry rearrangements are similar to those induced by an H→L transition in a 9-MA monomer (vide infra). However, since the excitation is "shared" between two monomers, the differences with respect to the S0 energy minima are smaller. It is important to highlight that SDEL never corresponds to the lowest energy excited state, at least in the region of the PES explored. In fact, its minimum SDEL is also less stable than SCT. Concerning the possible dynamical role of this S2 state, Fig. 3 indicates that the SB diabatic state crosses S2. It can be thus possible that some population is transferred to SDEL, but it does not seem likely that such a "transferred population" could be removed from the interaction region, so that it would likely be transferred back to SB. Moreover, it is not very likely that some population is trapped on the S2 state for a long time, also because of its possible decay to S1. Beyond recalling the Kasha rule, this is somewhat supported by the fact that optimization of the S2 state has proved to be very troublesome numerically, with a strong tendency to collapse onto the SCT state. This indirectly suggests a coupling between the two states, with SCT being the lower state. Nonetheless, only a full quantum dynamical study will make it possible to arrive at a definitive conclusion on the dynamics of the S2 role. Unfortunately, given the present state of the art, such a study would be difficult to perform with the accurate quantum mechanical methods necessary for a reliable description of the states. Finally, solvent fluctuations and the structural dynamics of the strand makes the existence of long-living delocalized symmetric states less likely. On the balance, our calculations suggest that SDEL should not play a significant role in polyA excited-state dynamics.
Analysis of the 1D and 2D Energy Scans.
Let us start by analyzing in greater detail the path leading from the FC point to the SB minimum (Fig. 5), where it is shown that, along the path, S3 and S1 exchange their electronic character (and, thus SB and SCT exchange their energy ordering). It is noteworthy that in this region, the two adiabatic surfaces do not maintain a negligible energy difference, so that conical intersections are not expected along the selected path (of course, the situation could be different on the full-dimensional PES). At variance with the adiabatic states, the bright SB and dark SCT diabatic states cross along the path and show a substantial coupling up to the minimum of the SB state (see SI Fig. 17, which is a diagnostic of the occurrence of a diabatic state transition). Simple qualitative arguments suggest that, due to the inertia of the motion, a significant part of the excited wavepacket initiated at the FC point should surpass the surface-crossing remaining on the SB bright diabatic state. In any case, the part of the wavepacket transferred to the dark SCT state is expected to be irreversibly accelerated toward the equilibrium structure of this state, reaching a region where it is no longer coupled to the bright state and therefore cannot jump back on it. This can be easily seen in SI Fig. 14, in which the path leading from the SCT "pseudominimum" close to the SB/SCT crossing on the FC→minSB path (point 6 on the abscissae of Fig. 5) toward the SCT minimum is clearly barrierless.
SI Fig. 18 shows the results of the one-dimensional scan between the minima of the SB and SCT states. Confirming the results of the analogous 2D analysis (see the main text), SI Fig. 18b indicates that, along this path, the first excited-state S1 rapidly changes its electronic character, losing all oscillator strength with respect to the ground state. When this happens, the PES shows a very low energy barrier (<0.05 eV) separating the SB minimum from the SCT minimum. As a consequence, the diabatic bright SB and dark SCT states cross very close to the SB minimum. Both the adiabatic and diabatic pictures indicate that, in this region of the coordinate space, the system can easily jump from the bright SB to the dark SCT state, reaching the equilibrium structure of the latter. Note that, by definition, diabatic states should be smooth functions of the nuclear coordinates. The irregularities found at the beginning of the path indicate that an accurate diabatization should also consider the third electronic state, SDEL, which, in our approximate approach, has been left unchanged.
Finally, we have investigated whether small variations in the intermonomer arrangement can qualitatively affect our conclusions, focusing on the effect of the intermonomer distance. SI Fig. 19a shows a 2D scan of the three lowest excited PESs from FC to minSB, and SI Fig. 19b shows the corresponding oscillator strengths with the ground electronic state. The first coordinate of the scan linearly interpolates the internal coordinates of the two monomers between the two structures and is the same used in Fig. 3 a and b. The second coordinate is the distance between the two monomers (defined as the N3-N9' distance; see Fig. 1). Fig. 3 a and b are, respectively, the sections of SI Fig. 19 a and b at N3-N9' = 3.44 Å. It can be clearly seen that the qualitative picture of the dynamics described in Fig. 3 is unaltered by the change in the intermonomer distance (which, in the real polyA, is limited by the existence of the deoxyribose/phosphate backbone). In fact, as discussed for Fig. 3, in the barrierless path between FC and minSB the S3 and S1 states interchange their electronic character (as shown by the oscillator strengths), i.e., the diabatic bright and dark states, SB and SCT, respectively, cross so that SB is the lowest excited-state S1 at minSB. On the other hand, the figure does not suggest any significant FC activity of the intermonomer distance upon optical excitation. At the shortening of the intermonomer distance, the relative stability of the CT state with respect to the ππ* state increases because of the electrostatic contribution to the energy. As a consequence, in the region of the minSB structure (i.e., at the value 11 of the collective coordinate where SB is lower in energy than SCT), the two diabatic states are closer in energy and hence more mixed in the adiabatic states S1 and S3. As shown in SI Fig. 15b, in that region, as a consequence of such increased mixing, the difference between the oscillator strengths with the ground state of S1 and S3 decreases with the shortening of the intermonomer distance. This is, however, a minor effect and, as stated before, it does not alter the qualitative picture of the dynamics upon excitation.
SI Fig. 20a reports a 2D scan of the excited PESs connecting the minSB and minSCT structures, and SI Fig. 20b shows the corresponding oscillator strengths with the ground electronic state. SI Fig. 20 a and b should be compared with SI Fig. 18 a and b, which actually are a section of SI Fig. 20 a and b taken at N3-N9' = 3.44 Å. Also in this scan, the modulation of the intermonomer distance seems to introduce, at most, minor changes in the dynamics without affecting the main finding already discussed in the text: In the path from minSB to minSCT, the S1 state profoundly changes its electronic character (see the oscillator strengths), transforming from SB to SCT. From the viewpoint of energy, the two structures are separated by only a very small energy barrier on the S1 surface, predicting the existence of an effective decay of the bright SB state into the SCT state.
Some Preliminary Considerations Regarding the Excited-State Behavior of Polythymine.
The analysis based on IP and EA values performed in the previous sections for polyA can be applied also to provide a first preliminary interpretation of the causes that make (likely) marginal the role of excimers in polythymine (23, 24). In fact, it is noteworthy that the sum of the experimental gas-phase IP and EA is 8.99 eV (8.45 + 0.54 eV) for adenine and 9.44 eV (9.15 + 0.29 eV) for thymine (26, 27). Therefore, the energetic cost for the creation of a CT pair should be larger for thymine, especially assuming that stacking is also more efficient for polyA than for polythymine (22). Incidentally, these data provide further support to the reliability of our computational approach. In fact, the comparison with experiments of our estimates for the sum IP + AE of adenine (9.13 eV) and thymine (9.38 eV), obtained by PBE0/6-311+G(2d,2p)//PBE0/6-31G(d) calculations in the gas phase, is very satisfactory.
1. Boys SF, Bernardi F (1970) Mol Phys 19:553.
2. Tomasi J, Mennucci B, Cammi R (2005) Chem Rev 105:2999.
3. Cossi M, Scalamani G, Rega N, Barone V (2002) J Chem Phys 117:43.
4. Cossi M, Barone V (2001) J Chem Phys 115:4708.
5. Barone V, Cossi M, Tomasi J (1997) J Chem Phys 107:3210.
6. Cossi M, Barone V (2000) J Phys Chem A 104:10614-10622.
7. Improta R, Barone V, Scalmani G, Frisch MJ (2006) J Chem Phys 125:054103.
8. Blancafort L (2006) J Am Chem Soc 128:210-219.
9. Perun S, Sobolewski AL, Domcke W (2005) J Am Chem Soc 127:6257-6265.
10. Chen H, Li S (2005) J Phys Chem A 109:8443-8446.
11. Serrano-Andres L, Merchan M, Borin AC (2006) Proc Natl Acad Sci USA 103:8691-8696.
12. Serrano-Andres L, Merchan M, Borin AC (2006) Chem Eur J 12:6559-6571.
13. Dreuw A, Head-Gordon M (2005) Chem Rev 105:4009-4037.
14. Wankoi M, Garavelli M, Bernardi F, Niehaus TA, Frauenheim T, Elstner M (2004) J Chem Phys 120:1674.
15. Dreuw A, Weisman JL, Head-Gordon M (2003) J Chem Phys 119: 2943.
16. Dreuw A, Head-Gordon M (2004) J Am Chem Soc 126:4007.
17. Olaso-Gonzalez G, Roca-Sanjuán D, Serrano-Andrés L, Merchan M (2006) J Chem Phys 125:231102.
18. Improta R, Barone V, Santoro F (2007) Angew Chemie Int Ed 46, 405-408.
19. Santoro F, Improta R, Lami A, Bloino J, Barone V (2007) J Chem Phys 127:084509.
20. Santoro F, Barone V, Benzi C, Improta R (2007) Theor Chim Acc DOI:10-1007/s00214-006-0225-y.
21. Crespo-Hernández CE, Arce R (2002) Photochem Photobiol 76:259.
22. Marguet S, Markovitsi D, Talbot F (2006) J Phys Chem B 110:11037-11039.
23. Crespo-Hernández CE, Cohen B, Kohler B (2005) Nature 436:1141-1144.
24. Crespo-Hernández CE, Cohen B, Kohler B (2006) Nature 441:E8.
25. Kwok W-M, Ma C, Phillips DL (2006) J Am Chem Soc 128:11894-11905.
26. Kim SK, Lee W, Herschbach DR (1996) J Phys Chem 100:7933-7937.
27. Aflatooni K, Gallup GA, Burrow PD (1998) J Phys Chem A 102:6205-6207.