Substituent Effects on Twisted Internal Charge Transfer Excited States

Jan 1, 2012 - 1B ← 1 1A excitations correspond to π(car) → Bpz transitions for all three carbazole molecules. However, their oscillator strengths...
1 downloads 0 Views 2MB Size
Article pubs.acs.org/JPCA

Substituent Effects on Twisted Internal Charge Transfer Excited States of N-Borylated Carbazoles and (Diphenylamino)boranes Jian Wang,† Ying Wang,† Takuhiro Taniguchi,‡ Shigehiro Yamaguchi,‡ and Stephan Irle*,†,‡ †

Institute for Advanced Research, Nagoya University, Nagoya 464-8601, Japan Department of Chemistry, Graduate School of Science, Nagoya University, Nagoya 464-8602, Japan



S Supporting Information *

ABSTRACT: N-Boryl-substituted carbazoles (carBR2) and (diphenylamino)boranes (Ph2NBR2) with R = Mes (mesityl) and FMes [tris(trifluoromethyl)phenyl] substituents on boron exhibit large UV/vis Stokes shifts. To investigate the substituent effect on the magnitude of the Stokes shifts, we studied the electronic structure and spectroscopic properties of carBR2 and Ph2NBR2 with R = H, Mes, and FMes using hybrid density functional theory (B3LYP) and time-dependent density functional theory (TD-B3LYP) for ground and low-lying excited states. The lowest lying excited state with a nonvanishing oscillator strength is a twisted internal charge transfer (TICT) 1A state in the C2 point group, owing to a singleelectron excitation from the nitrogen lone pair to the unoccupied boron pz AO, Nlp → Bpz. Emission from these 1A excited states are predicted to be much brighter than from the energetically close 1B excited states that are not directly related to CT excitation from N to B, due to symmetry. An analysis of geometrical relaxations in the excited state and the state energies relative to the ground state energy of the equilibrium geometry reveals that (a) the carbazole skeleton induces a general red shift in UV/vis spectra, (b) bulky boryl substituents reduce the predicted Stokes shifts of TICT states, and (c) the presence of electron-withdrawing functional groups induces a further general red shift in UV/vis spectra but does not significantly alter Stokes shifts.

1. INTRODUCTION Boron-based π-electron compounds are relevant in various optical applications, such as organic light-emitting diodes,1−5 two-photo absorption materials,6 fluorescent sensors,7−10 and nonlinear optics.11−14 Many experimental investigations have reported the structures, design and synthesis, and spectral properties of such compounds.15−20 An early experimental study on the luminescence and spectral properties of anilinodimesitylboranes from 1974 found the existence of an unusually large Stokes shift,15 where the actual fluorescence frequencies depended on the polarity of the solvent. The authors proposed that an initially twisted >NB< moiety on the ground state surface relaxes in the charge-transfer (CT) excited state to a planar, dipolar >N(+)B(−)< geometry (denoted S1′ in ref 15), resulting in the observed large and solvent-dependent Stokes shift. However, MRD-CI calculations showed more than ten years later that the 1A2(v) CT excited state of the parent compound aminoborane (H2NBH2) possesses a twisted >N(+)B(−)< molecular geometry with an HNBH dihedral angle of 90°.21,22 With the advancement of computer speed, Rettig and co-workers performed in 1999 symmetry-restricted geometry optimizations for ground and lowest excited state of pyrroloborane using a multireference CI level of theory and confirmed that twisting of the NB bond also occurs in the excited state of compounds where the nitrogen is part of a pyrrole moiety.16 This finding suggests that the excited state geometry of anilinodimesitylboranes should © 2012 American Chemical Society

also be nonplanar and that the interpretation of the excited state in ref 15 as involving a planar dipolar >N(+)B(−)< geometry is incorrect. Charge transfer excited states with large conformational changes are nowadays referred to as TICT (twisted internal charge transfer)23−25 and are often found in connection with strongly Stokes-shifted photoemission from excited states. We recently measured the photophysical properties of aminoborane derivatives, namely N-boryl-substituted carbazoles carBR2 (R = Mes, 1b, and R = FMes, 1c), and N-borylsubstituted (diphenylamino)boranes Ph2NBR2 (R = Mes, 2b), shown in Scheme 1.26 It was found that the electronwithdrawing substituent FMes in the case of the carbazole compound 1c causes absorption and emission peaks to strongly red shift in comparison to 1b, and to enhance its already large Stokes shift. It can be expected that two factors, namely the molecular structure of the nitrogen-containing π-system and the effects of the boryl substituents, play an important role in shaping ground and excited state energy profiles. The experimental results indicate that both effects may be exploitable to design a finely tuned Stokes shift for future optical applications, and theoretical investigations of these effects are therefore timely. Received: September 26, 2011 Revised: December 9, 2011 Published: January 1, 2012 1151

dx.doi.org/10.1021/jp209264j | J. Phys. Chem. A 2012, 116, 1151−1158

The Journal of Physical Chemistry A

Article

Scheme 1. Compounds Investigated in This Studya

In this work, we investigated theoretically six N-borylsubstituted carbazoles 1a−c and (diphenylamino)boranes 2a− c, where the boryl substituents are associated with different electronic/steric effects: indifferent/not bulky (R = H, compounds a), moderately electron withdrawing/bulky (R = Mes, compounds b), and strongly electron withdrawing/most bulky (R = FMes, compounds c). Using density functional theory (DFT) and time-dependent DFT (TD-DFT), we computed the molecular structures of the low-lying excited states, and compared the schematic potential energies of the different compounds. The comparison among the six compounds allows a quantitative analysis of the effects of the nitrogen-containing π-system as well as the effects of the boryl substituents influencing the magnitude of the Stokes shift.

charges for ground and excited states. In the case of ground states, we also computed Mulliken and Ahlrich’s MAO32 partial atomic charges charges. It is widely known that CT excited states are not well described by conventional TD-DFT.34−36 We therefore verified the applicability of the TD-B3LYP method for our study by employing the ab initio symmetry adapted cluster/configuration interaction (SAC-CI) method with the SV(P) basis set for the calculation of absorption and emission energies of compounds 1a and 2a based on B3LYP/SV(P) optimized geometries, using the implementation in the GAUSSIAN 03 program.33 The detailed discussion of this benchmark is in the Supporting Information. We found that the performance of TD-B3LYP in these compounds is reasonable, due to the short distance between the localized N and B charge centers in the excited states, which reduces the DFT self-interaction error. The effect of a larger basis set on ground and excited state energies was investigated using the much larger def2-TZVPP28 (abbreviated TZVPP) basis set in DFT and TD-DFT single point calculations at the B3LYP/SV(P) optimized geometries. TD-B3LYP/TZVPP excitation energies are found to be virtually identical to those of TD-B3LYP/SV(P). The latter level of theory is therefore accurate enough to describe the nature of the valence excited states and to capture the relative trends in a series of related compounds. The discussions in the paper will therefore be solely based on (TD-)B3LYP/SV(P) data. The details of the method comparisons can be found in the corresponding section in the Supporting Information.

2. COMPUTATIONAL METHODOLOGY All DFT and TD-DFT calculations were carried out using the TURBOMOLE 5.10 program.27 The molecular structures were optimized in the ground (1 1A, in the remainder of this article denoted “X” for simplicity) and two low-lying excited singlet states (2 1A and 1 1B states under the C2 point group, abbreviated as “1A” and “1B”), using the B3LYP method in combination with Ahlrich’s def2-SV(P) basis set [further denoted SV(P) for simplicity].28 Only in the case of compound 1a do ground as well excited state geometries conform to the C2v point group. For excited states, we employed TD-DFT29−31 using the same functional to compute absorption and emission energies and oscillator strengths in the gas phase. In all calculations we applied the default SCF and geometry optimization threshold parameters of TURBOMOLE. Harmonic vibrational frequencies were computed for structures on the ground state potential energy surface, confirming that the optimized structures correspond to minima. Dipole moments were computed from electrostatic potential derived (ESP)

3. RESULTS AND DISCUSSION Compound 1a possesses C2v symmetry, whereas all other compounds suffer distortion to C2 symmetry caused by the functional groups. Table 1 lists essential geometrical parameters such as the NB bond lengths, ∠CNBH1 or ∠CNBC1 dihedral angles as defined in Scheme 1, and dipole moments for ground and excited states at respective optimized geometries. The full set of Cartesian coordinates for these 18 geometries are given in the appendix of the Supporting Information. 3.A.I. Molecular and Electronic Structure in Ground States. The optimized molecular geometries of 1a−c and 2a− c in their ground states are shown in Figure 1. Table 1 shows that the R = H species 1a and 2a are nearly planar with ∠CNBH dihedral angles close to 0°. Boryl substitutent groups increase the dihedral angles ∠CNBC1 in the order of their bulkiness FMes > Mes up to 36.5° (1c) and 19.9° (2c). The dihedral angles of carbazole compounds are larger by about 16° compared to corresponding (diphenylamino)boranes, indicat-

a

1 = N-boryl-substituted carbazoles carBR2, 2 = N-boryl-substituted (diphenylamino)boranes Ph2NBR2, R = H, Mes (=mesityl), and FMes (=tris(trifluoromethyl)phenyl). Experimentally synthesized were compounds 1b, 1c, and 2b. NB torsional angles are measured by dihedral angles ∠CNBH1 or ∠CNBC1 depending on the nature of substituent R.

Table 1. Optimized NB Bond Lengths, Dihedral Angles ∠CNBC1 (H1) (Definitions in Scheme 1), and Dipole Moments μ from ESP Charges for X Ground State and 1A and 1B Excited States for Compounds 1a−c and 2a−c X

1

A

1

B

r(NB) (Å) ∠CNBC1(H1) (deg) μ (D) r(NB) (Å) ∠CNBC1(H1) (deg) μ (D) r(NB) (Å) ∠CNBC1(H1) (deg) μ (D)

1a

1b

1c

2a

2b

2c

1.413 0.0 0.78 1.577 90.0 6.88 1.538 0.0 6.86

1.454 28.6 0.76 1.598 62.4 7.66 1.580 52.2 9.40

1.435 36.5 0.99 1.560 64.7 7.28 1.540 53.5 10.47

1.411 3.2 1.71 1.586 78.0 6.09 1.439 3.7 0.01

1.441 13.2 0.94 1.623 55.9 5.69 1.495 41.1 3.53

1.420 19.9 3.03 1.557 37.3 7.39 1.526 27.5 12.23

1152

dx.doi.org/10.1021/jp209264j | J. Phys. Chem. A 2012, 116, 1151−1158

The Journal of Physical Chemistry A

Article

istence of a partial NB π−bond. The latter resembles the situation in the parent compound aminoborane H2NBH2, where the 1A1 electronic ground state possesses planar geometry and a very short NB bond length of only 1.393 Å. The electronic nature of the ground state thus corresponds to a significant mixture of >NB< and >N(+)B(−)< Lewis structures, where (+) and (−) indicate formal charges. Mulliken and MAO charges (see Table S3 in the Supporting Information) are consistent with partially negatively charged boron, whereas the situation is less clear in the case of boron ESP charges and the charges on nitrogen. Nevertheless, R = FMes compounds consistently show the highest negative charges on boron. We note that nitrogen as the more electronegative bond partner in all its three bonds receives some excess negative charge, such that formal charges do not properly reflect the charge density distribution. Notwithstanding, molecular dipole moments (Table 1) likewise indicate an increasing weight of the zwitterionic >N(+)B(−)< Lewis structure in going from R = Mes, H to R = FMes. On the basis of the analysis of ground state charge distribution and the shape of the HOMO, the ground state geometrical changes in the >NB< moiety can be rationalized as follows. Without substituents, the >NB< moiety is planar with a very short bond as in the parent compound aminoborane H2NBH2. The nitrogen-containing π-system does not change this tendency much, as indicated by the structural parameters of R = H compounds 1a and 2a. When R becomes bulky, the >NB< unit is forced away from planarity due to steric effects. However, the stronger electron withdrawing R = FMes substituent shortens the NB bond relative to the R = Mes group because the zwitterionic >N(+)B(−)< Lewis structure

Figure 1. Optimized geometries of compounds 1a−c (carBR2, upper panel) and 2a−c (Ph2NBR2, lower panel) in electronic ground states. Black spheres correspond to carbon, white to hydrogen, blue to nitrogen, beige to boron, and green to fluorine atoms.

ing a greater steric repulsion between R functional groups and the carbazole moiety. The NB bond lengths are shortest for R = H compounds with ∼1.41 Å, and significantly elongated up to ∼1.45 Å in compounds with bulkier R substituents. However, the relationship between NB bond length and ∠CNBC1 angle is not linear: R = FMes compounds 1c and 2c possess the largest torsional angles, but their NB bond lengths are intermediate between R = H and R = Mes substituents. Figure 2 shows that the molecular orbitals (MOs) representing the nitrogen lone pair (Nlp) are almost always the highest occupied MO (HOMO), and that these MOs are significantly delocalized across the NB bond and on the nitrogen-containing π-conjugated system, indicating the ex-

Figure 2. (a) Isovalue plots of frontier MOs of 1a−c at the equilibrium geometries of the ground state (left three columns) and 1A (middle three columns) and 1B (right three columns) excited states. Orbital labels indicate the consecutive orbital number and irrep. (b) Isovalue plots of frontier MOs of compounds 2a−c at the equilibrium geometries of the ground state (left three columns) and 1A (middle three columns) and 1B (right three columns) excited states. Orbital labels indicate the consecutive orbital number and irrep. 1153

dx.doi.org/10.1021/jp209264j | J. Phys. Chem. A 2012, 116, 1151−1158

The Journal of Physical Chemistry A

Article

Table 2. Vertical (Franck−Condon) Absorption Energies VAE (eV/nm), Excitation Character, Contribution |ci|2 to TD-DFT Wavefunctions, and Oscillator Strength f for Compounds 1a−c and 2a−c at the B3LYP/SV(P) Level of Theory nature of excited state compound 1a

1b 1c 2a

2b

VAE (eV/nm)

character

MOs

|ci|2

f

1

4.44/279

B2 A 1 B 1 A 1 B 1 A 1 B

4.16/298 3.69/336 3.82/325 2.87/432 2.77/447 4.89/253 4.84/256

1

4.17/297 4.49/276 3.49/355 3.77/329

HOMO−1 → LUMO HOMO−1 → LUMO+1 HOMO → LUMO+2 HOMO → LUMO HOMO → LUMO HOMO-1 → LUMO HOMO−1 → LUMO HOMO → LUMO HOMO → LUMO HOMO → LUMO+1 HOMO−1 → LUMO+2 HOMO → LUMO+2 HOMO → LUMO HOMO−2 → LUMO HOMO → LUMO HOMO → LUMO+1

0.39 0.45 0.13 0.94 0.97 0.99 0.99 0.99 0.96 0.76 0.08 0.07 0.98 0.93 0.99 0.99

0.09

1

Nlp → Bpz Nlp → π*(car) π(car) → π*(car) π(car) → Bpz Nlp → Bpz π(car) → Bpz Nlp → Bpz π(car) → Bpz Nlp → Bpz Nlp → π*(ph) π(ph) → π*(ph) Nlp → π*(ph) Nlp → Bpz π(mes) → Bpz Nlp → Bpz Nlp → π*(mes)

excited state A1

1

A B 1 A 1 B 1

2c

0.00 0.16 0.00 0.11 0.00 0.12 0.03

0.20 0.04 0.17 0.04

that the excited state wave function in this case is strongly of multiconfigurational character, where single-reference methods such as TD-DFT and even standard SAC-CI37 may be expected to perform poorly. The carbazole 1 1B ← 1 1A excitation energies are 4.16 eV (1a), 3.82 eV (1b), and 2.77 eV (1c), and dominant excitations in the 1B states are HOMO → LUMO, HOMO−1 → LUMO, and HOMO → LUMO, respectively. The HOMO−1, HOMO, and LUMO MOs can be interpreted as Nlp, π[car(=carbazole)], and Bpz orbitals in the case of 1a and 1c, whereas π(car) and Nlp switch positions in the case of compound 1b, as shown in Figure 2 for the MO’s of the ground state geometries. We note that the energies of these HOMO and HOMO−1 orbitals are very close: −0.225 and −0.229 eV for 1a, −0.213 and −0.217 eV for 1b, and −0.228 and −0.231 eV for 1c. 1B ← 1 1A excitations correspond to π(car) → Bpz transitions for all three carbazole molecules. However, their oscillator strengths are orders of magnitude smaller than those of excitations to 2 1A with values of only 0.000 (1a), 0.001 (1b), and 0.001 (1c). It can therefore be expected that these excitations do not play a role in experimentally observed UV/ vis absorption spectra. In analogy to the carbazole series, the 2 1A ← 1 1A excitations of the Ph2NBR2 compounds 2a−c correspond to Nlp → Bpz charge transfer excitations (Table 2). The VAEs are 4.89 eV (2a), 4.17 eV (2b), and 3.49 eV (2c), thus roughly 0.5−0.6 eV higher compared to the carbazole series. Their oscillator strengths are 0.12 (2a), 0.20 (2b), and 0.17 (2c). The 1 1B ← 1 1A excitation energies are 4.84 eV (2a), 4.49 eV (2b), and 3.77 eV (2c), and even higher by 0.7−1.0 eV relative to the carbazole series. The nature of the 1B excited states varies depending on the boryl substituent R, containing mixed excitations in the cases of R = H and R = Mes, indicating multiconfigurational character. Their oscillator strengths are also one order of magnitude smaller than those of 2 1A ← 1 1A excitations. As in the carbazole compounds, we expect that 1Btype excitations do not play a role in experimentally observed UV/vis absorption spectra of N-borylated (diphenylamino)boranes. 3.B. Molecular and Electronic Structure in Excited States, Emission Spectra, and Schematic Potential

receives a higher weight as indicated by high negative charges on boron in R = FMes compounds. 3.A.II. Nature of 1A and 1B Excited States in the Franck−Condon Region. Figure 2 displays the main Kohn− Sham frontier MOs involved in the electronic excitations, and Table 2 lists vertical absorption energies VAEs and excitation character with oscillator strengths in the Franck−Condon region. As mentioned earlier, compound 1a possesses C2v symmetry in ground and excited state geometries, whereas all other compounds conform to C2 symmetry. The 1A and 1B excited states of compound 1a are therefore correctly labeled 1 A1 and 1B2 in the planar Franck−Condon geometry. However, to simplify discussions, we will refer to these two excited states of compound 1a by their lower symmetry labels, unless noted otherwise. With two exceptions, the excitation character of 1A and 1B excited states is consistent with a zwitterionic state where boron becomes substantially negatively charged, indicating electronic transfer either from the nitrogen lone pair for all 1A excited states or from occupied π MOs for 1B states, namely from the highest occupied π MO of the carbazole ring in 1a−c or from the highest occupied π MO of the mesityl group in 2b. The two exceptions are the 1B states of compounds 2a and 2c, which correspond to excitation from Nlp to the lowest unoccupied π* MOs of the phenyl (ph), or to the lowest unoccupied π* mesityl (mes) groups, respectively. Nevertheless, all excitations are of CT type. We will now discuss the VAEs that are related to UV/vis absorption spectra. For carBR2 species, the 2 1A←1 1A transitions are 4.44 eV (1a), 3.69 eV (1b), and 2.87 eV (1c), in which the leading excitation configuration of HOMO−1 → LUMO, HOMO → LUMO, and HOMO−1 → LUMO are responsible for the transition, respectively, as shown in Table 2. Oscillator strengths are relatively large with 0.09 (1a), 0.16 (1b), and 0.11 (1c). However, it should be pointed out that in the case of the 2 1A1 state of 1a, the TD-DFT squared CI coefficient corresponding to HOMO−1 → LUMO excitation is substantially smaller with 0.39 than that of the HOMO−1 → LUMO+1 (0.45); hence excitation from Nlp to the lowest π* MO of the carbazole moiety with a-type symmetry is strongly mixed with the Nlp → Bpz transition. It is therefore apparent 1154

dx.doi.org/10.1021/jp209264j | J. Phys. Chem. A 2012, 116, 1151−1158

The Journal of Physical Chemistry A

Article

Table 3. Emission Energies EE (eV/nm) from State1//State2 to X//State2, Excitation Character, Contribution |ci|2 to TD-DFT Wavefunction, and Oscillator Strength f for Compounds 1a−c and 2a−c at the B3LYP/SV(P) Level of Theory nature of excited state1 compound 1a

1b

1c

2a

2b

2c

EE (eV/nm)

contribution

orbital

|ci|2

f

A2(v)// A2(v) 1 B2//1B2 1 B1(v)//1A2(v) 1 A1//1B2 1 A//1A 1 B//1B 1 B//1A 1 A//1B 1 A//1A 1 B//1B 1 B//1A 1 A//1B 1 A//1A 1 B//1B 1 B//1A 1 A//1B 1 A//1A 1 B//1B 1 B//1A

1.76/705 3.18/390 2.45/506 3.80/326 2.28/544 2.57/481 2.87/432 2.62/474 1.36/915 1.48/839 1.65/752 1.83/677 1.72/720 2.95/421 3.72/334 3.70/335 2.26/548 3.37/368 3.61/343 3.17/391 2.21/562 2.68/463 2.89/429 2.65/468

HOMO → LUMO HOMO → LUMO HOMO−1 → LUMO HOMO−1 → LUMO HOMO → LUMO HOMO → LUMO HOMO-1 → LUMO HOMO−1 → LUMO HOMO → LUMO HOMO → LUMO HOMO−1 → LUMO HOMO-1 → LUMO HOMO → LUMO HOMO → LUMO HOMO−2 → LUMO HOMO → LUMO+1 HOMO → LUMO HOMO−1 → LUMO HOMO−1 → LUMO HOMO−2 → LUMO HOMO → LUMO HOMO → LUMO HOMO → LUMO+1 HOMO → LUMO+1 HOMO → LUMO

0.99 0.99 1.00 0.98 0.99 1.00 1.00 0.99 1.00 1.00 1.00 1.00 0.99 0.98 0.94 0.95 0.99 0.99 0.85 0.12 0.99 1.00 0.99 0.99 1.00

0.00 0.00 0.00 0.12 0.02 0.00 0.00 0.05 0.03 0.00 0.00 0.07 0.00 0.03 0.00 0.12 0.06 0.04 0.13

1

Nlp → Bpz π(car) → Bpz π(car) → Bpz Nlp → Bpz Nlp → Bpz π(car) → Bpz π(car) → Bpz Nlp → Bpz Nlp → Bpz π(car) → Bpz π(car) → Bpz Nlp → Bpz Nlp → Bpz Nlp → π*(ph) π(ph) → Bpz Nlp → Bpz Nlp → Bpz π(mes) → Bpz π(mes) → Bpz π(mes)′ → Bpz Nlp → Bpz Nlp → Bpz Nlp → π*(mes) Nlp → π*(mes) Nlp → Bpz

emission from 1

A//1B A//1A 1 B//1B 1 B//1A 1 A//1B 1

1

Energy Surfaces. Optimized excited state molecular geometries (Table 1) are generally consistent with the nature of excited states near the Franck−Condon region, discussed in section 3.A.II. The 1A excited state geometry of compound 1a possesses C2v symmetry; however, as in the H2NBH2 parent compound, the Bpz orbital is now rotated by 90° with respect to the Nlp orbital. Thus, the correct symmetry label for this excited state is now 1A2(v), where (v) indicates the switch in the symmetry label due to the 90° rotation, and the correct symmetry label for the 1B excited state in this geometry is 1 B1(v). In the case of 1A excited states (always Nlp → Bpz), the NB bond lengths increase upon excitation in the range from 0.125 Å (1c) up to 0.182 Å (2b), and in the case of 1B excited states from 0.028 Å (2a, Nlp → π*(ph)) to 0.126 Å (1b). The bond length increase is an expression of the fact that in all these excited states the partial NB double bond in the ground state is weakened either by excitation of an electron away from the nitrogen or by acceptance of an electron in the vacant Bpz orbital, or both. Compounds with R = Mes groups almost always exhibit the largest NB bond stretching. TICT character is clearly visible in the change of dihedral angles ∠CNBC1 (H1) upon excitation: In the case of 1A excited states, these changes are 90.0° (1a) and 74.8° (2a) for R = H, 33.8° (1b) and 42.7° (2b) for R = Mes, and 28.2° (1c) and 17.4° (2c) for R = FMes. As reported earlier for the TICT excited state of aminodiborane H2NBH2,21,22 the Nlp → Bpz transition is accompanied by an increase in the dihedral angle (largest in the case of R = Mes), minimizing the overlap between the two opposite pπ AOs. The excited state wave function maximizes in this way orthogonality to the ground state. ESP-derived charges

0.19 0.13 0.03 0.04 0.19

on boron indicate the presence of a negative charge uptake, except for the 1B excited states of 2a and 2c where the boron center is not involved in the excitation (see Table S2 in the Supporting Information). As chemical intuition suggests, the excited state negative charges on boron are reduced in the case of the strongly electron-withdrawing R = FMes compounds. As can be seen from Table 1, the absolute values of the dipole moments for 1A and 1B excited states are much larger than those of the ground state, with the only exception of the nearly planar 1B excited state of compound 2a. Figure 2 shows that in the optimized geometries of the excited states the MOs involved in the respective excitations (1A//1A and 1B//1B) are more frequently found to be HOMO and LUMO. This is for instance noticeable in the case of the 1A excited state of compounds 1a and 1c, where the Nlp corresponds in the ground state to HOMO−1 but becomes HOMO at the excited state geometry. The above-described geometry changes in the excited state stand in stark contrast to the earlier explanation of the Stokes shift mechanism by Glogowski et al.15 who had proposed planarization of an initially twisted >NB< moiety due to excitation to the CT excited state. Rather, our calculations confirm the TICT mechanism also in the present compounds. The polarity-dependent solvent effect of the Stokes shift reported by Glogowski et al. is right for the wrong reason, as the torsional relaxation argument works in both directions. As Rettig’s16 and our calculations show, however, the ground state adapts a more planar configuration, and internal CT upon excitation forces the molecules to adapt twisted conformations, 1155

dx.doi.org/10.1021/jp209264j | J. Phys. Chem. A 2012, 116, 1151−1158

The Journal of Physical Chemistry A

Article

Figure 3. Schematic potential energy surfaces of theory for ground state and 1A and 1B excited states of carBR2 (upper panel) and Ph2NBR2 (lower panel) compounds.

schematically depicts for all six compounds the state energies at the optimized geometry of the 1A excited state (left), at the Franck−Condon region (middle), and at the optimized geometry of the 1B excited state (right), relative to the ground state global minimum energy. We find that fluorescence from the 1A excited states at respective optimized geometries is energetically most favorable in all cases. However, in the Franck−Condon region, 1B states are sometimes lower than 1A excited states (1a, 1c, 2a), indicating that in these compounds the two excited states of different symmetry have to cross. But because oscillator strengths for the excitations to 1B states are one order of magnitude smaller compared to those for excitations to excited 1A states, we expect that the UV/vis features of these compounds can be well described by considering only the 1A excited states featuring Nlp → Bpz transitions. 3.C. Stokes Shifts and Comparison with Experimental UV/Vis Results. On the basis of the previously described theoretical results, we identify experimental absorption peaks

also predicted earlier by Rettig et al. in the case of smaller model systems.16 In Table 3 we list 2 1A → 1 1A and 1 1B → 1 1A emission energies (EEs) for the optimized geometries of the excited states. We also list emission energies of 1A excited states at the optimized geometry of 1B excited states (denoted “1A//1B”) and vice versa. In Table 3, “1A//1A” and “1B//1B” correspond to “regular” EEs that are observable in UV/vis fluorescence spectra, whereas the “mixed” EEs were included to illustrate the behavior of 1A and 1B excited states upon geometry change beyond the ground state equilibrium geometry. The nature of excited states for respective optimized geometries (“1A//1A” and “1B//1B”) is dominated by a single transition with the leading wave function coefficients always being larger than 0.85, as shown in Table 3. Their character is essentially unchanged from that at the Franck−Condon geometries. To illustrate the potential energy surfaces of the excited and ground states as a function of geometry changes, Figure 3 1156

dx.doi.org/10.1021/jp209264j | J. Phys. Chem. A 2012, 116, 1151−1158

The Journal of Physical Chemistry A

Article

Table 4. Absorption Energies, Fluorescence Emission Energies, and Stokes Shift for 1A Excited States (eV/nm) absorption cal carBR2

Ph2NBR2

a

R R R R R R

= = = = = =

H Mes FMes H Mes FMes

4.44/279 3.69/336 2.87/432 4.89/253 4.17/297 3.49/355

fluorescence expa

cal 1.76/705 2.28/544 1.36/915 1.72/720 2.26/548 2.21/562

3.87/320 3.36/369 4.40/282b

Stokes shift expa

2.82/439 2.06/603 2.72/456b

cal 2.68/426 1.41/208 1.51/483 3.17/467 1.91/251 1.28/207

expa 1.05/119 1.30/234 1.68/174b

From ref 26.; values obtained in cyclohexane. bReference 15.

with 2 1A ← 1 1A excitations, and fluorescent emission as the corresponding 2 1A → 1 1A transition. In Table 4, we list theoretical and experimental wavelengths of absorption, fluorescence, and Stokes shifts. The theoretical values were calculated in vacuum, whereas absorption and emission frequencies were measured in a nonpolar solvent. We find that our calculated absorption frequencies are somewhat redshifted by ca. 0.20 eV for compounds 1b and 2b, and ca. 0.50 eV for compound 1c, relative to the corresponding values measured in experiment. Fluorescence redshifts are often overestimated by TD-DFT, and the error here lies between 0.46 and 0.70 eV. However, relative trends are satisfactorily reproduced by our TD-B3LYP/SV(P) calculations, namely red shifts due to fluorination of the Mes group in the case of the carbazole compound with 0.82 eV (experiment: 0.51 eV) for absorption and 0.92 eV (experiment: 0.76 eV) for emission. Exchanging carbazole by (diphenylamino)borane in the case of R = Mes causes a blue shift by 0.48 eV (experiment: 0.53 eV) in absorption and in emission a red shift by 0.02 eV (experiment: 0.1 eV). Due to the systematic TD-DFT overestimation of fluorescence redshifts, our computed Stokes shifts are systematically overestimated by approximately 0.3 eV. The largest Stokes shifts are theoretically predicted for the R = H compounds, with no experimental data available to verify this prediction. The Stokes shifts in these two compounds are 2.68 eV (1a) and 3.17 eV (2a); however, the emission oscillator strengths are unfortunately practically zero as the 1 1A2(v) → 1 1 A1 nature of these transitions in the perpendicular alignment of the Bpz and Nlp orbitals makes them symmetry-forbidden. The predicted Stokes shifts in R = Mes compounds are much smaller with 1.41 eV (1b) and 1.91 eV (2b), in qualitative agreement with experiment. Absorption and fluorescence frequencies of R = FMes species are substantially red-shifted relative to R = Mes by up to 0.92 eV (fluorescence of 1c), however, not by equal amounts. We note that calculations predict only a 0.05 eV red shift for the corresponding emission from 2c relative to 2b. Absorption frequencies of (diphenyldiamino)borane species are blue-shifted by about 0.5−0.6 eV relative to the carbazole compounds. However, their emission frequencies are nearly equal to 2a−b or blueshifted by 0.85 eV (2c) relative to carbazole compounds. This causes the dependence of the Stokes shifts on the nature of the boron substituent R to be different between carBR2 and Ph2NBR2 species. In particular, compound 2b shows the largest Stokes shift (1.91 eV) among the species with R ≠ H, whereas compound 2c shows the smallest Stokes shift (1.28 eV). Key to understanding this seemingly puzzling array of data is Figure 3. We find there that the excited states of carbazole compounds feature generally a smaller separation from the electronic ground states than diphenylamine compounds, in agreement with the available experimental data. We find further

that the energy gain due to geometrical relaxation in the excited state is almost perfectly mirrored by the energy increase in ground state, with the only exception of the 1B excited state in 2a where the ground state energy rises more steeply by 1.52 eV than the energy of the excited state decreases by 0.38 eV. We conclude that the larger the geometrical changes in the excited state, the larger the anticipated decrease in emission energies and correspondingly the Stokes shift. The largest geometrical changes leading to the largest Stokes shifts take place in the R = H compounds, which, however, happen to begin with excitation energies that are larger than in the case of the other boryl substituents. The effect of the R = Mes group is the reduction of the geometry relaxation in the excited state, because the ground state geometry is not perfectly planar to begin with, due to steric hindrance. On the other hand, the role of the R = FMes group is to significantly lower the energies of both the 1A and 1B excited states; however, the geometry relaxations in the excited states and the anticipated Stokes shifts are less affected by the electronic effect of the R = FMes group and comparable to those of the R = Mes groups.

4. SUMMARY AND CONCLUSIONS The molecular and electronic structures of ground and lowest excited states of six N-boryl-substituted (diarylamino)diarylboranes of type carbazoles carBR 2 (1a−c) and (diphenylamino)boranes Ph2NBR2 (2a−c) (R = H, Mes, and FMes) were investigated using hybrid density functional theory B3LYP and TD-B3LYP methods after benchmarking the R = H compounds by ab initio SAC-CI calculations. We focused on the nature of ground and low-lying excited states and the geometrical relaxations of the latter to clarify the reason for the experimentally observed large Stokes shift in these compounds. The results can be summarized as follows: (1) Electronic ground states show a significant mixture of >NB< and >N(+)B(−)< Lewis structures, causing R = H compounds to exhibit perfectly (1a) and nearly perfectly (2a) planar CNBH dihedral angles in agreement with the previous report by Rettig and coworkers.16 Torsional angles in the ground state equilibrium geometries increase in the order R = H < Mes < FMes, where Ph2NBR2 compounds exhibit smaller ∠CNBC1 angles than carBR2. R = FMes exerts a greater electron withdrawing effect than R = Mes, visible in large negative partial charges on boron and shorter NB bond lengths. (2) The lowest 1A excited states correspond in all cases to intramolecular charge transfer (CT) excitations of Nlp → Bpz type, whereas 1B excited states also correspond to intramolecular CT nature but are more diverse. Oscillator strengths associated with 1A excited states 1157

dx.doi.org/10.1021/jp209264j | J. Phys. Chem. A 2012, 116, 1151−1158

The Journal of Physical Chemistry A

Article

are one order of magnitude larger than those of 1B excited states and can be expected to dominate the experimentally observed UV/vis low energy transitions, although symmetry rules prohibit emission from such states (formally of 1A2 type) in perpendicular arrangements of the Nlp and the Bpz orbitals. All relaxed excited state geometries display significant increase in the CNBH1 (R = H) or ∠CNBC1 (R ≠ H) dihedral angles in the order R = H > Mes > FMes, most significantly for 2 1A relaxed geometries and less for 1 1B geometries, because the latter involve CT either from nitrogen to π* orbitals or from π orbitals to boron. (3) The carbazole skeleton reduces the separation of ground and lowest-lying excited states in comparison to the diphenylamine skeleton. Bulky boryl substituents introduce an increase in ∠CNBC1 dihedral angles already for ground state energies, and in this way reduce the anticipated Stokes shift in the TICT excited state, which follow the same mode in geometry relaxation. The electron-withdrawing influence of the fluorine atoms in the R = FMes compounds results in a general stabilization of TICT excited states but do not by themselves alter the anticipated Stokes shifts. In conclusion, we have shown that steric and electronic effects of the substituent groups in N-boryl-substituted carbazoles and (diphenylamino)boranes can have a conflicting influence on fluorescence energies from TICT states, causing a seemingly erratic dependence of the Stokes shift on the nature of these functional groups. It was demonstrated that these data can be rationalized in a comparative study of ground and excited state energies for series of related compounds.



(4) Jia, W.-L.; Bai, D.-R.; McCormick, T.; Liu, Q.-D.; Motala, M.; Wang, R.-Y.; Seward, C.; Tao, Y.; Wang, S. Chem.Eur. J. 2004, 10, 994−1006. (5) Jia, W. L.; Feng, X. D.; Bai, D. R.; Lu, Z. H.; Wang, S.; Vamvounis, G. Chem. Mater. 2005, 17, 164−170. (6) Liu, Z.-Q.; Feng, Q.; Cao, D.-X.; Wang, D.; Xu, G.-B. Org. Lett. 2004, 6, 2933−2936. (7) Yamaguchi, S.; Akiyama, S.; Tamao, K. J. Am. Chem. Soc. 2001, 123, 11372−11375. (8) Yamaguchi, S.; Shirasaka, T.; Akiyama, S.; Tamao, K. J. Am. Chem. Soc. 2002, 124, 8816−8817. (9) Miyata, M.; Chujo, Y. Polym. J. 2002, 34, 967−969. (10) Sundararaman, A.; Victor, M.; Varughese, R.; Jäkle, F. J. Am. Chem. Soc. 2005, 127, 13748−13749. (11) Branger, C.; Lequan, M.; Lequan, R. M.; Barzoukas, M.; Fort, A. J. Mater. Chem. 1996, 6, 555−558. (12) Branger, C.; Lequan, M.; Lequan, R. M.; Large, M.; Kajzar, F. Chem. Phys. Lett. 1977, 272, 265−270. (13) Yuan, Z.; Collings, J. C.; Taylor, N. J.; Marder, T. B.; Jardin, C.; Halet, J. -F. J. Solid State Chem. 2000, 154, 5−12. (14) Yuan, Z.; Entwistle, C. D.; Collings, J. C.; Albesa-Jové, D.; Batsanoy, A. S.; Howard, J. A. K.; Taylor, N. J.; Kaiser, H. M.; Kaufmann, D. E.; Poon, S.-Y.; Wong, W.-Y.; Jardin, C.; Fathallah, S.; Boucekkine, A.; Halet, J.-F.; Marder, T. B. Chem.Eur. J. 2006, 12, 2758−2771. (15) Glogowski, M. E.; Grisdale, P. J.; Williams, J. L. R.; Costa, L. J. Organomet. Chem. 1974, 74, 175−183. (16) Cornelissen-Gude, C.; Rettig, W. J. Phys. Chem. A. 1999, 103, 4371−4377. (17) Entwistle, C. D.; Marder, T. B. Angw. Chem., Int. Ed. 2002, 41, 2927−2931. (18) Entwistle, C. D.; Marder, T. B. Chem. Mater. 2004, 16, 4574− 4585. (19) Zhao, C.-H.; Wakamiya, A.; Inukai, Y.; Yamaguchi, S. J. Am. Chem. Soc. 2006, 128, 15934−15935. (20) Wakamiya, A.; Mori, K.; Yamaguchi, S. Angw. Chem., Int. Ed. 2007, 46, 4273−4276. (21) Bonačić-Koutecký, V.; Michl, J. J. Am. Chem. Soc. 1985, 107, 1765−1766. (22) Bonačić-Koutecký, V.; Schöffel, K.; Michl, J. J. Am. Chem. Soc. 1989, 111, 6140−6146. (23) Grabowski, Z. R.; Rotkiewicz, K.; Siemiarczuk, A. J. Lumin. 1979, 18/19, 420−424. (24) Retting, W. Angew. Chem., Int. Ed. Engl. 1986, 25, 971−988. (25) Grabowski, Z. R.; Dobkowski, J. Pure Appl. Chem. 1983, 55, 245−252. (26) Taniguchi. T.; Wang, J.; Irle, S.; Yamaguchi, S. Submitted for publication. (27) Ahlrichs, R.; Furche, F.; Hättig, C.; Klopper, W.; Sierka, M.; Weigend, F. TURBOMOLE 5.10, 2008. (28) Weigend, F.; Ahlrichs, R. Phys. Chem. Chem. Phys. 2005, 7, 3297−3305. (29) Casida, M. E.; Jamorski, C.; Casida, K. C.; Salahub, D. R. J. Chem. Phys. 1998, 108, 4439−4449. (30) Stratmann, R. E.; Scuseria, G. E. J. Chem. Phys. 1998, 109, 8218−8224. (31) Matsuzawa, N. N.; Ishitani, A.; Dixon, D. A.; Uda, T. J. Phys. Chem. A. 2001, 105, 4953−4962. (32) Heinzmann, R.; Ahlrichs, R. Theor. Chim. Acta 1976, 42, 33−45. (33) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.;et al. Gaussian 03, Revision C.02. See the Supporting Information for the full citation. (34) Dreuw, A.; Weisman, J. L.; Head-Gordon, M. J. Chem. Phys. 2003, 119, 2943−2946. (35) Dreuw, A.; Head-Gordon, M. J. Am. Chem. Soc. 2004, 126, 4007−4016. (36) Tozer, D. J. J. Chem. Phys. 2003, 119, 12697−12699. (37) Nakatsuji, H. J. Chem. Phys. 1984, 83, 713−722.

ASSOCIATED CONTENT

S Supporting Information *

Extensive SAC-CI and basis set benchmark data, electronic energies, partial atomic charges for N and B atoms in all optimized geometries, full citation of ref 33, and Cartesian coordinates for ground and excited state optimized geometries. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected].



ACKNOWLEDGMENTS We thank a reviewer for helpful comments and acknowledge financial support by the Global Century of Excellence (GCOE) Program in Chemistry at Nagoya University. S.I. also acknowledges support by the Program for Improvement of Research Environment for Young Researchers from Special Coordination Funds for Promoting Science and Technology (SCF) commissioned by the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan. S.Y. acknowledges support by a Grant-in-Aid (19675001) from MEXT.



REFERENCES

(1) Noda, T.; Sirota, Y. J. Am. Chem. Soc. 1998, 120, 9714−9715. (2) Shirota, Y.; Kinoshita, M.; Noda, T.; Okumoto, K.; Ohara, T. J. Am. Chem. Soc. 2000, 122, 11021−11022. (3) Jia, W.-L.; Song, D.; Wang, S. J. Org. Chem. 2003, 68, 701−705. 1158

dx.doi.org/10.1021/jp209264j | J. Phys. Chem. A 2012, 116, 1151−1158