Chapter 3 Experimental Part
3.5 Samples
3.5.8 An Anthraquinone-[Ru(bpy) 3 ] 2+ -Oligotriarylamine-
The AQ-RuII-TPA2+-RuII-AQ pentad was synthesized by Dr. Annabell G. Bonn in the Wenger group in 18 steps using standard C-C
and N-C coupling reactions, as described in detail in ref 44.
Chapter 4
Naphthalenediimides *
4.1 Introduction
The result part starts with the excited-state dynamics of small molecules, which are used as building blocks to construct more sophisticated systems that will be discussed in further chapters. Because of this, it is always better to understand the behaviour of the individual unit first before moving to more complex systems. That is why the photophysics of the individual naphthalene diimide (NDI) and its derivatives are discussed in this chapter.
NDI is one of the widely used building blocks for constructing multichromophoric systems for electron transfer. NDIs can be substituted on the diimide nitrogens or on their naphthalene core. The latter attracted particular interest of scientists and is the main target in this project. For the past few years, the core-substituted naphthalenediimides (c-NDIs) have been intensively used as building blocks in the elaboration of sophisticated molecular architectures developed for various applications, such as photovoltaics, artificial photosynthesis, or sensing.45-52 Contrary to the non-core-substituted NDIs that are colorless and barely fluorescent,53-56 c-NDIs absorb in the visible and exhibit fluorescence with quantum yields Φfl that
* This chapter is adapted from the following publication: Yushchenko, O.; Licari, G.;
Mosquera-Vazquez, S.; Sakai, N.; Matile, S.; Vauthey, E. Ultrafast Intersystem-Crossing Dynamics and Breakdown of the Kasha-Vavilov’s Rule of Naphthalenediimides. J. Phys. Chem. Lett. 2015, 6, 2096-2100.
are typically larger than 0.1. This difference is due to the nature of the S1 ← S0 transition, which involves a substantial redistribution of the electronic density from the core substituents to the carbonyl groups. As a consequence, the energy of this state as well as the redox properties of c-NDIs can be tuned by varying the electron-donating ability of the substituents. In this project, we will explore the effect of core substituents on the excited-state dynamics of NDI.
Additionally to that, as it will be discussed below, the S2 ← S0
transition of c-NDIs hardly involves the core substituents and that is why the S2 state resembles the S1 state of the non-core-substituted NDIs. Therefore, in order to better understand the excited-state dynamics of c-NDI of interest, r(ed) NDI, upon S2 ← S0 excitation, we have also investigated that of a NDI without core substituents, p(ale)NDI (Figure 4.1). Although these NDIs are being very intensively used as acceptors in electron-transfer cascades,57-63 their excited-state dynamics, especially those of the ultrafast ISC from the S1
state responsible for the negligible fluorescence, have not been investigated in much detail. The fluorescence lifetime of NDIs with different imide substituents has been estimated to be in the 5-20 ps range.55-56 More recently, Ganesan et al performed femtosecond transient absorption measurements of pNDI in chloroform and reported an ISC time constant of 10 ps.64 In this chapter we will see that the ISC dynamics from the S1 state of pNDI to the triplet manifold is more complex than previously assumed.
Figure 4.1 Structure of pNDI and rNDI.
O
O O
O
O O
HN
O O
Br NH O
O O
O
N
rNDI pNDI
N N
N
4.2 Steady-State Spectroscopy
4.2 Steady-State Spectroscopy
The absorption and fluorescence spectra of pNDI and rNDI in acetonitrile (ACN) are presented in Figure 4.2. The model pNDI exhibits a vibrational progression below 400 nm (Figure 4.2A). This is the characteristic structured band of typical NDIs that is assigned to a ππ* transition, which is localized on the core of NDI. It corresponds to the S1 ← S0 transition.65 Similarly to other NDIs with different imide substituents,53, 55-56, 64, 66
pNDI exhibits a mirror image emission spectrum with fluorescence quantum yield about 10−3. The low quantum yield of amino substituted NDIs is related to the efficient ISC to the triplet manifold.
Figure 4.2 Absorption and fluorescence spectra of (A) pNDI and (B) rNDI in ACN. The fluorescence excitation spectrum of rNDI is also shown in B (dashed line).
The absorption spectrum of rNDI contains a similar band shifted by 940 cm−1 to higher energy, which corresponds to the S2 ← S0 transition (Figure 4.2B). It contains an additional band at 528 nm due to the S1 ← S0
fluorescence / a.u.
700 600
500 400
300
wavelength / nm
absorbance
A
B
absorption emission
absorption emission excitation
transition, which has a charge transfer (CT) character as it involves a displacement of the electron density from the core-substituent to the centre of the rNDI.42, 49 Contrary to pNDI, rNDI shows intense fluorescence around 570 nm with Φfl = 0.22 in ACN upon S1 ← S0 excitation. However, as illustrated by its excitation spectrum (Figure 4.2B), the fluorescence quantum yield of rNDI decreases by a factor of 1.9 upon S2 ← S0 excitation.
This is an unusual and strong deviation from the Kasha-Vavilov’s rule (section 2.2.2).
Additionally to that, the time-resolved fluorescence measurements performed with rNDI, measured with TCSPC, lead to the same lifetime of 5.5 ns independently of whether S2 ← S0 or S1 ← S0 excitation. One can conclude that the decrease of Φfl should arise from the existence of a new relaxation pathway of the S2 state that competes with internal conversion to the S1 state.
Figure 4.3. Fluorescence dynamics measured at 570 nm with rNDI in ACN upon S2 ← S0 and S1 ← S0 excitation.
4.3 Transient Absorption
4.3.1 rNDI
Further insight into the origin of this breaking of the Kasha-Vavilov’s rule was obtained by comparing the transient absorption (TA) spectra measured with rNDI upon S1 ← S0 excitation at 532 nm (Figure 4.4)
10-3 10-2 10-1 100
fluorescence
60 50 40 30 20 10 0
time / ns
upon S1 S0exc upon S2 S0exc IRF
τ = 5.5 ns
4.3 Transient Absorption
and S2 ← S0 excitation at 385 and 355 nm (Figure 4.6) in ACN and in dichloromethane (DCM, Figure 4.5 and 4.7).
TA measurements were performed with two pump-probe setups, which have been described in detail in section 3.4, namely the femtosecond TA setup used to record spectra up to 1.7 ns and the sub-nanosecond TA setup, used to record spectra up to 1 ms. The agreement between the data recorded with the two different TA setups was checked by comparing the spectra measured in the 0.5-1 ns region. A single, wavelength-independent, scaling factor was sufficient to match the spectra measured with the two setups. As the excited-state dynamics of the investigated molecules occurs on timescales either shorter than 100 ps or longer than 1 ns, the data recorded with each setup were analyzed separately.
4.3.1.1 S1 ← S0 Excitation
The TA spectra recorded with rNDI within the first nanosecond after 532 nm excitation are dominated by a positive band extending from 370 to 510 nm and by negative bands at 365 nm and in the 540-650 nm region (Figure 4.4A and 4.5A). The negative bands at 365 nm and in the 540-650 nm region can be assigned to ground state bleach (GSB) and stimulated emission (SE) by comparison with the steady-state absorption and emission spectra, respectively. The positive broad band decreases on a similar time scale as SE. It can therefore be ascribed to the absorption of the S1 state (Sn ← S1 absorption).67
During the next ∼10 ns, the positive band transforms into a band peaking at 440 nm, and the SE is replaced by a positive absorption band. In ACN (Figure 4.4A), this spectrum decays on the ∼1 µs time scale to a weak residual spectrum with two bands around 450 and 600 nm, which itself decays to zero within 10 µs. In DCM (Figure 4.5A), the spectrum with the 440 nm band decays entirely to zero, and no residual spectrum is observed.
Global analysis using the sum of three exponential functions or target analysis assuming three sequential steps (A → B → C → D) could successfully reproduce the TA data in ACN with the time constants and species-associated difference spectra (SADS) shown in Figure 4.4B.36 The first step with a time constant identical to the fluorescence lifetime can be assigned to the decay of the S1 state population to the ground state and to the
triplet state (species B) by ISC. The second step (B → C) can be interpreted as the decay of the triplet-state population to the ground state and to another state (species C), which is only present in ACN. Assignment of B to the triplet state of rNDI is comforted by a shortening of its lifetime in aerated solutions. As the SADS of C resembles strongly the absorption spectrum of the radical anion of an analogue of rNDI with a Cl atom instead of the Br substituent,67 C is interpreted as rNDI·−. This ion most probably results from triplet-triplet annihilation, as already reported for several organic ketones in polar solvents.68 The cation rNDI·+ should also be generated during this process. However, the Cl analogue of rNDI·+ has been shown to have either a very similar absorption spectrum as that of the anion or a weak absorption in the visible.67 Previous observations of photoinduced symmetry-breaking charge separation (SB-CS) between other c-NDIs support the formation of ions by triplet−triplet annihilation in polar solvents.50, 67, 69
Figure 4.4 (A) Transient absorption spectra recorded at several time delays after S1 ← S0 rNDI in ACN; (B) species-associated difference spectra obtained from target analysis assuming the scheme in the inset.
15
4.3 Transient Absorption
Figure 4.5 (A) Transient absorption spectra recorded at various time delays after S1 ← S0 excitation of rNDI in DCM; (B) species-associated difference spectra obtained from target analysis assuming the scheme in the inset.
To calculate the driving force of the SB-CS we assume the following process:
3rNDI* + 3rNDI* → 1rNDI* + rNDI → 2rNDI·+ + 2rNDI·¯
The driving force for the SB-CS is given by:13
∆𝐺!" =−𝐸!!+𝑒 𝐸!" 𝐷 −𝐸!"# 𝐴 +𝐶+𝑆 (4.1)
where 𝐸!! is the energy of rNDI S1 state (𝐸!!= 2.25 eV), 𝐸!" and 𝐸!"# are the oxidation and reduction potentials of rNDI (𝐸!" = 1.5 V and 𝐸!"# = -0.55 vs. SCE).49 𝐶=𝑒!/(4𝜋𝜀!𝜀!𝑑!") is a correction term that represents the Coulombic interaction between the ions at a distance 𝑑!" and screened by a solvent of dielectric constant, 𝜀!. Finally, 𝑆 is a factor accounting for a
dielectric constant of the solvent different from that used for the determination of the redox potentials, 𝜀!!:
𝑆=− 𝑒! 8𝜋𝜀!
1 𝑟!+1
𝑟! 1 𝜀!!+1
𝜀! (4.2) According to the above equations with 𝑟!= 𝑟!= 4 Å and neglecting the Coulombic factor, ∆𝐺!" should amount to -0.2 eV in ACN and +0.15 eV in DCM. Thus, SB-CS should be energetically feasible in ACN, but not in DCM, in agreement with our observation.
4.3.1.2 S2 ← S0 Excitation
The TA spectra recorded with rNDI in ACN during the first few nanoseconds after S2 ← S0 excitation show several differences (Figure 4.6A): (1) the 440 nm band dominates at much earlier time and (2) the SE band is barely visible. On the other hand, the spectra recorded after ∼10 ns are mostly identical to those obtained upon 532 nm excitation and can be explained likewise.
The same target analysis yields similar time constants and SADS for B and C. However, as shown in Figure 4.6B, the SADS obtained for A contains features of both S1 and T1 spectra, indicating that the T1 state contributes to this SADS as well. Target analysis assuming that both the S1
and T1 states are initially equally populated (i.e., A(0) = B(0) = 0.5) yields SADS, which are identical to those obtained above from the 532 nm TA data (compare Figure 4.4B with Figure 4.6C). Thus, these results are fully consistent with the reduced fluorescence quantum yield measured upon S2 ← S0 excitation and points to the occurrence of an ultrafast ISC process from the S2 state to the triplet manifold that competes with internal conversion. According to the TA data, the decay of the S2 state takes place within the instrument response, that is, around 200 fs (section 3.4).
4.3 Transient Absorption
Figure 4.6 (A) Transient absorption spectra recorded at several time delays after S2 ← S0 rNDI in ACN; (B) and (C) species-associated difference spectra obtained from target analysis assuming the scheme in the inset with different initial conditions.
The early TA spectra measured with rNDI in DCM show the same dependence on the excitation wavelength as those were found with rNDI in ACN (Figure 4.7). Additionally, which is not surprising, the radical anion
Figure 4.7 (A) Transient absorption spectra recorded at various time delays after S2 ← S0 excitation of rNDI in DCM; (B) and (C) species-associated difference spectra obtained from target analysis assuming the scheme in the inset with different initial conditions.
4.3.2 pNDI
To better understand how core substitution affects the ISC dynamics of rNDI S2 state, TA measurements have also been performed with pNDI in the same two solvents (Figure 4.8 and Figure 4.9). The TA spectra are first dominated by a band at 597 nm, which loses half of its intensity after only
4.3 Transient Absorption
2 ps, whereas another weaker band rises at 460 nm. Afterward, both bands decay simultaneously within a few tens of picoseconds and a structured band with the maximum at 483 nm rises. In DCM (Figure 4.9A), this band decreases entirely to zero on the microsecond time scale. In ACN (Figure 4.8A), it transforms into a spectrum with bands at 473 and 608 nm, which lives for hundreds of microseconds.
Figure 4.8 (A) Transient absorption spectra recorded at several time delays after S1 ← S0 excitation of pNDI in ACN; (B) and (C) species associated difference spectra obtained from target analysis assuming the reaction scheme in the inset.
Figure 4.9 (A) Transient absorption spectra recorded at different time delays after S1 ← S0 excitation of pNDI in DCM; (B) Species-associated difference absorption spectra obtained from target analysis assuming the scheme in the inset.
The temporal evolution of the TA spectra in ACN was analyzed by target analysis assuming sequential steps and yielded the time constants and SADS shown in Figure 4.8B. Species C can be undoubtedly assigned to the T1 state of pNDI,64, 70 whereas species D is attributed to pNDI·− because its spectrum is essentially the same as that of the radical anion of a similar NDI.71 pNDI·− is most probably generated by SB-CS upon triplet-triplet annihilation as discussed above with rNDI. Whereas species A can be safely assigned to the S1 state of pNDI, the interpretation of SADS B is not evident as it also contains the 597 nm band.
In order to better understand the origin of the intermediate state B that is related to the S1 state the time resolved fluorescence measurements have been performed. The fluorescence decay of the pNDI is too fast to be resolved with TCSPC that is why fluorescence up-conversion has been used to monitor the fast dynamics. The fluorescence decays of pNDI measured at
50
4.4 Quantum Chemical Calculations
several wavelengths between 420 and 450 nm are biexponential with the same time constants, that is, 1 and 10 ps, as found for the A → B → C steps (Figure 4.10). It was found that their associated amplitudes do not depend on the emission wavelength, therefore these time constants are not due to solvent/vibrational relaxation or to two different emitting states.
Consequently, the B SADS (Figure 4.8B) suggests the decay of two different states/species with the same time constant. All this can be accounted for by an equilibrium between the S1 state and species B. The result of a target analysis of the TA data with a scheme including such an equilibrium is shown in Figure 4.8C. The SADS of A and B are now totally different and thus arise from distinct species/states. Because of this equilibrium, the decay of the S1 state is now biexponential, in agreement with the fluorescence measurements. The same behaviour of pNDI is revealed in DCM (Figure 4.9B).
Figure 4.10. Fluorescence dynamics at 440 nm measured with pNDI in ACN upon 385 nm excitation and best biexponential fit.
4.4 Quantum Chemical Calculations
To get more insight into the possible processes occurring from the S2 and S1 states in rNDI and pNDI, respectively, quantum chemical calculations have been carried out. These calculations were performed on analogues of pNDI and rNDI, with methyl N substituents (Figure 4.11). Two types of calculations were performed: (1) ground-state gas-phase geometry
150 100 50 0
fluorescence
30 25 20 15 10 5 0
time delay / ps
1.2 ps (0.57) + 9.8 ps (0.43)
optimisation performed at the density functional level of theory (DFT) using the B3LYP functional, and (2) electronic vertical transition energies calculated with time-dependent density functional theory (TD-DFT) using the long-range corrected hybrid CAM-B3LYP functional.72 The details on how the calculations were performed by Giuseppe Licari are discussed in the article.39 It was found that the results obtained with m-pNDI are in good agreements with the calculations of the lowest singlet excited state of a similar NDI.73
Figure 4.11. m-pNDI (left) and m-rNDI (right).
Table 4.1. TDDFT calculations of m-pNDI
State Type Energy / eV
(nm)
Osc.
strength
Major MO contribution
S3 A2 n → π* 4.000 (310) 0.000 74 → 77
S2 A1 π → π* 3.945 (314) 0.043 75 → 77 T5 B2 π → π* 3.842 (323) 0 73 → 77 S1 B2 π → π* 3.663 (338) 0.368 76 → 77
T4 A2 n → π* 3.660 (339) 0 74 → 77
T3 B2 π → π* 3.428 (362) 0 68 → 77 T2 A1 π → π* 3.148 (394) 0 75 → 77 T1 B2 π → π* 1.896 (654) 0 76 → 77
O O
O
O N
N N
N O O
O O
HN
Br
4.4 Quantum Chemical Calculations
77 (LUMO) 76 (HOMO) 75
74 73 68
Figure 4.12. Molecular orbitals associated with the first electronic transitions of m-pNDI.
Table 4.2. TDDFT calculations of m-rNDI
State Type Energy / eV
(nm)
Osc.
strength
Major MO contribution S3 π → π* 3.913 (317) 0.180 108 → 110
S2 n → π* 3.818 (325) 0.001 107 → 110
T5 π → π* 3.757 (330) 0 106 → 110
T4 n → π* 3.511 (353) 0 107 → 110
T3 π → π* 3.280 (378) 0 109 → 111 S1 π → π* 2.896 (428) 0.222 109 → 110 T2 π → π* 2.413 (514) 0 108 → 110 T1 π → π* 1.807 (686) 0 109 → 110
111 110 (LUMO) 109 (HOMO)
108 107 106
Figure 4.13. Molecular orbitals associated with the first electronic transitions of m-rNDI.
4.5 Discussion
Previous investigations on related NDIs to pNDI pointed to the presence of an upper excited triplet state with nπ* character close to the S1
ππ* state.56 This is confirmed by our quantum chemistry calculations that predict a triplet state, T4, with nπ* character at the same energy as the S1
state (Table 4.1). This suggests that species B is this T4 state, and that back-ISC to the S1 state competes with internal conversion to lower triplet states (Figure 4.14, right). The latter process requires the release of a substantial amount of energy into the environment, a process that occurs on a few picoseconds time scale,74-75 in good agreement with the 3-5 ps time constant for the T4 → T1 process obtained from target analysis.
4.5 Discussion
Concerning rNDI, the TA results reveal that ISC to the triplet manifold occurs with a 2 ps time constant in pNDI. As a consequence, the much faster ISC from the S2 state measured with rNDI should be related to the core substituents. According to quantum chemistry calculations, the only available triplet state of nπ* character is T4 and is located 0.4 eV below S2
(Table 4.2). This larger singlet-triplet energy gap compared to pNDI explains the absence of back-ISC with rNDI, but should not accelerate S2 → T4 ISC.
The key factor responsible for the faster ISC of rNDI and, thus, the breakdown of the Kasha−Vavilov’s rule is the presence of the Br core substituent, which further increases the spin-orbit coupling associated with the S2(ππ*) → T4(nπ*) transition76 and makes ISC as fast as S2 → S1 internal conversion (Figure 4.14). This is confirmed by fluorescence excitation spectra measured with other c-NDIs that do not have heavy core substituents and do not exhibit any deviation from the Kasha-Vavilov’s rule (Figure 4.15). This sub-200-fs ISC from rNDI is among the fastest reported for an organic molecule.
Figure 4.14. Energy level scheme of rNDI and pNDI with the most relevant relaxation pathways and time scale (unless specified, the excited states have a ππ* character, intermolecular processes have been omitted).
Figure 4.15 Absorption (solid lines) and fluorescence excitation (dashed lines) and emission spectra measured with y(ellow)NDI, v(iolet)NDI and b(lue)NDI in DCM.
4.6 Conclusion
In this project, we have investigated the excited-state dynamics of a red core-substituted naphthalenediimide dye (rNDI) that exhibits strong deviation from the Kasha-Vavilov’s rule. Thanks to the time-resolved spectroscopic measurements it has been revealed that this deviation is due to an ultrafast, < 200 fs, intersystem-crossing (ISC) from the S2 state to the triplet manifold, due to the ππ* → nπ* character of the transition and to the
absorbance
4.6 Conclusion
presence of the heavy Br atom. Additionally, a new insight into the excited state dynamics of non-core substituted NDI was obtained. It was found that in non-core substituted NDI, ISC occurs in ~2 ps and is reversible on a time scale shorter than that of vibrational cooling.
Whereas c-NDIs are already valued for the high tunability of their optical and electrochemical properties upon modest chemical modification, the results presented here reveal a new facet of their versatility, namely the possibility to influence their excited-state dynamics by varying the magnitude of the spin-orbit coupling. This allows the design of dyes like rNDI, whose fluorescence and triplet quantum yields would be substantially changed by a simple variation of the excitation wavelength. As an example, the excited-state dynamics of such systems will also be presented further in this thesis.
Chapter 5
Naphthalenediimide Triad in Solution and of Solid Films *
5.1 Introduction
In this chapter, we are going to continue our discussion about naphtalenediimides, which have been described in the previous chapter.
However, here we will explore how the excited-state dynamics of these units
However, here we will explore how the excited-state dynamics of these units