Top Curr Chem (Z) (2018) 376:24
1 3
of theory adopted on the resulting 2D map, which is shaped by both TE and TDM
computed values.
Figure 5 summarizes the benchmark study of indole [53] performed with a large
basis set [ANO-L(432,21)-aug], indicating (as in benzene and phenol) that going
beyond a CASSCF treatment with full-π valence active space (by introducing at
least eight extravalence orbitals in the RAS3 subspace) is required in order to converge the energies and dipole moments of the electronic transitions involved and,
thus, the overall aspect of the 2D spectra (see Fig. 5c). Both one-color (UV-pump
and UV-probe, 2DUV–UV) and two-color (UV-pump and Vis-probe, 2DUV–Vis)
2D spectra are reported in Fig. 5c, showing a broad set of ESA and bleaching signals that vary in position and intensity as a function of the level of theory. Overall, in the 2DUV spectra of indole, two signal traces related to the first two (ππ*)
excited states, labeled L b and L a in Platt notation [79], are monitored, as they are
quite close in energy, and their transitions from the GS have similar TDMs. In particular, 2DUV–UV spectra (see Fig. 5) feature off-diagonal bleaching signals (ODB,
see Fig. 1) related to couplings between L b and L a states. It is worth mentioning that
these signals are characteristic 2DUV fingerprints of excited states localized on the
same molecule. They can be seen as strongly coupled “chromophores”, where the
energy of their “bi-exciton” (i.e. the state given by the sum of the two electronic
transitions) is significantly shifted from the sum of the single-exciton energies due
to the large anharmonicity constant, which cannot be computed perturbatively with
exciton Hamiltonians.
Here, we consider the 2D map converged with respect to the levels of theory
when the intensities of all the peaks are maintained and their positions do not vary
more than 1600 cm
−1
, i.e. the expected error in TE for this methodology (0.2 eV). It
is worth mentioning that such type of benchmark study requires a full assignment of
the 2D peaks in order to assess the variations in the 2D maps according to the theoretical level employed. This work is particularly tedious, as it requires labeling of all
relevant transitions (Fig. 5b) according to the molecular orbitals (MO) involved in
the electronic excitations.
Figure 6 shows the benchmark computations for the Ura, Thy and Cyt pyrimidine nucleobases and the effect of the larger active space on the 2DUV–UV spectra [59]. All these canonical nucleobases feature the same π-orbital valence AS,
i.e. CAS(10,8), while the RAS3 spaces involving 4, 8 and 12 virtual orbitals and
the large ANO-L(432,21)-aug basis set have been employed, as for the benchmarks of amino acid side chains (Figs. 4, 6, 5). While some ESA signals show
more AS-dependent fluctuations than others, the overall the 2DUV–UV spectra
converge with the addition of eight extravalence orbitals. Cyt is particularly interesting, as the first excited state is very bright but is not the brightest low-lying
ππ* state (as in the other pyrimidine bases), giving rise to two signal traces along
two relatively close-lying pump frequencies. This occurrence offers the opportunity to monitor the off-diagonal GSB signals related to the two low-lying ππ*
states (as in indole and purine bases) and the performances of the various levels
of theory in producing reliable 2DUV–UV maps.
Adenine and guanine are the canonical purine nucleobases, which feature larger
aromatic heterocyclic structure than pyrimidine bases, with the pyrimidine ring
80
Reprinted from the journal
1 3
of theory adopted on the resulting 2D map, which is shaped by both TE and TDM
computed values.
Figure 5 summarizes the benchmark study of indole [53] performed with a large
basis set [ANO-L(432,21)-aug], indicating (as in benzene and phenol) that going
beyond a CASSCF treatment with full-π valence active space (by introducing at
least eight extravalence orbitals in the RAS3 subspace) is required in order to converge the energies and dipole moments of the electronic transitions involved and,
thus, the overall aspect of the 2D spectra (see Fig. 5c). Both one-color (UV-pump
and UV-probe, 2DUV–UV) and two-color (UV-pump and Vis-probe, 2DUV–Vis)
2D spectra are reported in Fig. 5c, showing a broad set of ESA and bleaching signals that vary in position and intensity as a function of the level of theory. Overall, in the 2DUV spectra of indole, two signal traces related to the first two (ππ*)
excited states, labeled L b and L a in Platt notation [79], are monitored, as they are
quite close in energy, and their transitions from the GS have similar TDMs. In particular, 2DUV–UV spectra (see Fig. 5) feature off-diagonal bleaching signals (ODB,
see Fig. 1) related to couplings between L b and L a states. It is worth mentioning that
these signals are characteristic 2DUV fingerprints of excited states localized on the
same molecule. They can be seen as strongly coupled “chromophores”, where the
energy of their “bi-exciton” (i.e. the state given by the sum of the two electronic
transitions) is significantly shifted from the sum of the single-exciton energies due
to the large anharmonicity constant, which cannot be computed perturbatively with
exciton Hamiltonians.
Here, we consider the 2D map converged with respect to the levels of theory
when the intensities of all the peaks are maintained and their positions do not vary
more than 1600 cm
−1
, i.e. the expected error in TE for this methodology (0.2 eV). It
is worth mentioning that such type of benchmark study requires a full assignment of
the 2D peaks in order to assess the variations in the 2D maps according to the theoretical level employed. This work is particularly tedious, as it requires labeling of all
relevant transitions (Fig. 5b) according to the molecular orbitals (MO) involved in
the electronic excitations.
Figure 6 shows the benchmark computations for the Ura, Thy and Cyt pyrimidine nucleobases and the effect of the larger active space on the 2DUV–UV spectra [59]. All these canonical nucleobases feature the same π-orbital valence AS,
i.e. CAS(10,8), while the RAS3 spaces involving 4, 8 and 12 virtual orbitals and
the large ANO-L(432,21)-aug basis set have been employed, as for the benchmarks of amino acid side chains (Figs. 4, 6, 5). While some ESA signals show
more AS-dependent fluctuations than others, the overall the 2DUV–UV spectra
converge with the addition of eight extravalence orbitals. Cyt is particularly interesting, as the first excited state is very bright but is not the brightest low-lying
ππ* state (as in the other pyrimidine bases), giving rise to two signal traces along
two relatively close-lying pump frequencies. This occurrence offers the opportunity to monitor the off-diagonal GSB signals related to the two low-lying ππ*
states (as in indole and purine bases) and the performances of the various levels
of theory in producing reliable 2DUV–UV maps.
Adenine and guanine are the canonical purine nucleobases, which feature larger
aromatic heterocyclic structure than pyrimidine bases, with the pyrimidine ring
80
Reprinted from the journal