Photoinduced Electron Transfer Dynamics in Triarylamine

Article pubs.acs.org/JPCC

Photoinduced Electron Transfer Dynamics in Triarylamine− Naphthalene Diimide Cascades Fabian Zieschang,† Maximilian H. Schreck,† Alexander Schmiedel,† Marco Holzapfel,† Johannes H. Klein,† Christof Walter,‡ Bernd Engels,‡ and Christoph Lambert*,† †

Wilhelm Conrad Röntgen Research Center for Complex Material Systems, Center for Nanosystems Chemistry, Institut für Organische Chemie, Universität Würzburg, Am Hubland, 97074 Würzburg, Germany ‡ Institut für Physikalische und Theoretische Chemie, Universität Würzburg, Am Hubland, 97074 Würzburg, Germany S Supporting Information *

ABSTRACT: A series of dyads and triads of the A-D and A-D1-D2 type, respectively, containing triarylamine (TAA) donors and naphthalene diimide (NDI) acceptors, which are linked via triazole (Tz) heterocycles, were synthesized by Cu(I)-catalyzed azide alkyne cycloaddition (CuAAC). Upon photoexcitation, these systems undergo charge separation leading to long-lived charge-separated (CS) states. The population of these CS states was monitored using femtosecond and nanosecond transient absorption spectroscopy. The transient signals of the CS states of all triads and dyads feature biexponential decays in the nanosecond time regime with a short and a long component. These biexponential decays are the result of an ISC from the primarily populated 1CS state into the 3CS, from which charge recombination to the S0 state is forbidden by spin conservation rules. The existence of 3CS states in the triads was confirmed by strong magnetic field dependent transient absorption kinetics, while for the dyads no effect could be observed due to a much larger singlet−triplet splitting. Thus, although charge recombination from the 1CS state in the triads is slowed down compared to the dyads, the lifetime of the 3CS states is clearly longer in the dyads. This is the result of the larger singlet− triplet splitting in the dyads which leads to lifetimes of several microseconds.

■

the population of triplet CS states (3CS) can significantly enhance the CS state lifetime.5 Charge recombination (CR) from the 3CS to the electronic ground state is forbidden by spin selection rules, and in the absence of, e.g., heavy atoms, there is not sufficient spin−orbit coupling to overcome this rule. For the cascades given in Chart 1, we anticipated the CS step to be in the Marcus normal region while CR should occur in the inverted region. In this case, small reorganization energies will lead to fast CS and concurrently slow down CR.6 Because TAAs are known for their relatively small reorganization energies, we used them as strong electron donors in our cascades.7−9 Furthermore, para-substituted TAAs form very stable radical cations with characteristic strong and sharp absorption bands between ca. 15 000 cm−1 (670 nm) and 13 000 cm−1 (770 nm).10−12 These absorption bands are generally not overlapped by other transitions (e.g., the absorption of neutral TAA) which leads to very specific spectroscopic signatures of TAA+ and makes TAAs ideal candidates for the investigation of photoinduced electron transfer processes. Moreover, their redox potential can easily be tuned by different para-substituents, which opens the prospect to generate a downhill-directed redox gradient in the cascades.11,13−15 In all cascades, we employed a 1,4,5,8-naphthalene diimide (NDI) as a strong electron acceptor, which can reversibly be

INTRODUCTION One of the most important challenges is the artificial conversion of (solar) photon energy into chemical energy mimicking natural photosynthesis.1 In order to do this, many new organic materials were designed ranging from complex and large multichromophoric to small donor−acceptor systems. In such systems, charges are separated by a photoinduced electron transfer from an electron donor to a suitable electron acceptor, leading to a charge-separated (CS) state.2,3 For practical applications, the lifetime of this CS state should be on the order of at least 100 ns. One approach to enhance the lifetime of CS states is the use of redox cascades in which a number of subsequent, relatively fast, electron-transfer processes leads to large overall charge separation, while charge recombination (CR) is slowed down by the very weak electronic coupling between the CS state and the electronic ground state.4 In this study, we present the synthesis and photophysical investigations of a series of cascade molecules, in particular dyads and triads of A-D and A-D1-D2 type, respectively. These consist of triarylamine (TAA) donors and naphthalene diimide (NDI) acceptors, which are linked via triazole (Tz) heterocycles (see Chart 1). Such systems are expected to undergo charge separation (CS) upon photoexcitation leading to longlived CS states. The photoinduced electron transfer dynamics are strongly influenced by relative state energies, reorganization energies, and the electronic coupling between the corresponding states. Very recently, we reported our investigations of compact cyclophane-bridged TAA-NDI dyads, where we showed that © 2014 American Chemical Society

Received: August 22, 2014 Revised: October 28, 2014 Published: November 6, 2014 27698

dx.doi.org/10.1021/jp5085058 | J. Phys. Chem. C 2014, 118, 27698−27714

The Journal of Physical Chemistry C

Article

Chart 1. Triazole-Bridged Dyads and Triads with Triarylamine Donors and Naphthalene Diimide Acceptors and Reference Compounds

Figure 1. Schematic representation of the three possible excited state pathways in the triads. The star marks the locally excited group.

reduced at ca. −1.1 V and ca. −1.5 V.16 N-Alkyl-substituted NDIs are well-known to experience fast ISC into the locally excited triplet state (1/k ≈ 10 ps in chloroform).17 In contrast N-aryl-substituted NDIs rapidly populate a CT state where the positive charge is located at the aryl substituent and the negative charge at the NDI core (1/k ≈ 500 fs in chloroform).5,17,18 For our purpose, N-aryl-substituted NDIs are excellent complementary chromophores to the TAAs because their radical anions exhibit characteristic strong and intense absorption bands in the visible and near-infrared but otherwise do not overlap with the TAA radical cation band.16,19 The triazoles in the cascades come into play because they allow easy linkage between the redox chromophores by copper(I) catalyzed “click reaction” between a terminal alkyne and an azide.20−22 Although controversially discussed in the literature,23−25 we found triazoles to be short conjugated bridges which provide weak electronic coupling in electrontransfer reactions, a point which may enhance the CS state lifetimes in the cascades. Additionally, N-aryl-substituted NDIs show weak electronic interaction when linked via the nitrogen atom to another chromophore for two reasons. First, NDIs

exhibit a nodal plane along the molecular long axis going through the nitrogen atoms in the π-LUMO as well as in the πHOMO. Second, the aryl substituent is twisted out-of-plane by almost 90°, which reduces orbital interaction with the NDI core.26−28 Thus, the above outlined properties set the basis to achieve long-lived charge-separated states in our cascades. Upon excitation of the triads, three photoinduced processes may be conceived as presented in Figure 1. Accordingly, upon excitation of the NDI or the TAA1 (cases 1 and 2 in Figure 1), we expect a stepwise electron transfer (ET), first leading to the CS state (CS1) where the negative charge is localized on the NDI and TAA1 is oxidized (NDI−TAA1+-TAA2). Then, the positive charge is transferred to the terminal TAA2 and the fully CS state (CS2) is populated (NDI−-TAA1-TAA2+). Consequently, the charges are much more distant in the CS2 state of the triads than in the CS state of the dyads. The latter consist of a NDI and a single, methoxysubstituted TAA. Furthermore, the CS2 state of the triads is energetically slightly lower than the CS state of the dyads although the charged chromophores are identical in both 27699



Article

Figure 2. Upper left: absorption spectra of the dyads Da and Db in MeCN and toluene. Upper right: normalized absorption spectra of Da, Ref 1, Ref 2 and the sum of Ref 1 and Ref 2 in MeCN. Lower left: normalized absorption spectra of Ref 1 and Ref 2 in MeCN and toluene.

Figure 3. Absorption spectra of T-Me, T-CN, and T-Cl in MeCN (left) and toluene (right).

characteristics (see Figure 2, top left). In MeCN, both compounds exhibit a sharp absorption band at 26 500 cm−1 (375 nm) with nearly identical extinction coefficients (35 200 M−1 cm−1). The main absorption band is at 27 800 cm−1 (360 nm) and is more intense for Db (ε = 40 900 M−1 cm−1) compared to Da (ε = 37 200 M−1 cm−1). At the higher energy side, there are two small bands at 29 200 cm−1 (340 nm) and at 31 000 cm−1 (320 nm) with decreasing intensity. Both the spectra of Da and of Db are considerably different in MeCN and toluene. While in MeCN many bands are sharp and resolved, they become broad and featureless in toluene. As apparent from Figure 2, bottom left, the absorption spectrum of Ref 2 is sharper in toluene than in MeCN but is sharper in MeCN than in toluene for Ref 1. This indicates that the broadening for Da and Db is caused by the NDI, probably due to solvent-specific effects such as exciplex formation with aromatic solvents.29,30 The onset of the absorption (estimated by applying a tangent on the flank of the lowest energy transition) of Da and Db in toluene (24 800 cm−1) compared to that MeCN (25 800 cm−1) indicates lower 00-energies for both compounds in toluene.

species. This is a result of the larger distance of the opposite charges in the CS2 state of the triads. A smaller rate for the recombination process is expected for the triads compared to the dyads due to a weaker electronic communication between the terminal TAA and the NDI. The generation of the CS2 state in the triads from the CS1 state requires a downhill progression of states whose dynamics were examined in this paper by femtosecond and nanosecond time-resolved transient absorption spectroscopy. A different scenario is presented by case 3 (in Figure 1). The terminal TAA2 is excited but subsequent electron transfer to the TAA1 is prevented. However, we will show in the following that in fact predominantly case 1 is followed upon photoexcitation of the triads.

■

RESULTS Steady-State Absorption Spectroscopy. UV/vis steadystate absorption spectra of all cascades were recorded in polar MeCN and nonpolar toluene to investigate potential groundstate interactions between the TAA donor and the NDI acceptor. The two dyads Da and Db show very similar 27700



Article

found to be chemically fully reversible, as was proved by multiple thin-layer measurements.

As shown for Da in MeCN, the absorption spectra of the dyads are roughly the sum of the absorption spectra of the TAA and the NDI fragments (see Figure 2, top right), which indicates little, if at all, electronic interaction between the NDI and the TAA in the electronic ground state. A complete assignment of the absorption bands to either the NDI or the TAA chromophore is complicated; however, the sharp bands are certainly caused by the π−π* transitions of the NDI moiety with its typical vibronic features. It is obvious that the strong overlap of the NDI and the TAA absorption precludes an exclusive excitation of either the NDI or the TAA. We estimate in MeCN at 28 200 cm−1 (355 nm) ca. 50% and at 26 300 cm−1 (380 nm) ca. 10% TAA excitation. The absorption spectra of the triads are even much broader and more diffuse compared to the spectra of the dyads (see Figure 3). This finding can be explained by the second TAA donor in the triads, since TAAs are characterized by relatively broad absorption bands. In MeCN, all triads possess three absorption bands at 26 500 cm−1 (375 nm), 27 800 cm−1 (360 nm), and 29 200 cm−1 (340 nm) with a tiny shoulder at ca. 30 300 cm−1 (330 nm). The intensities of the three bands vary between the triads, which can be explained by the different absorption characteristics of the intermediate TAA1 in the triads. The absorption bands of T-Me, T-CN, and T-Cl in toluene are very broad and present only a single maximum. For T-Me this maximum at 27 800 cm−1 (360 nm) is in agreement with the most intense absorption band in MeCN. For T-Cl and T‑CN the maxima are at 28 400 cm−1 (350 nm) and, thus, are in between the two absorption bands at 27 800 cm−1 (360 nm) and 29 200 cm−1 (340 nm) found in MeCN. Similar to the dyads, a shift of ca. 210−230 cm−1 is observed in the spectra changing the solvent from MeCN to toluene. The onsets of absorption are similar for all compounds in MeCN (ca. 25 600 cm−1) and toluene (ca. 24 900 cm−1), and indicate very similar 00-energies as found for the dyads. As for the dyads, neither the NDI acceptor nor the TAA donors of the triads can selectively be excited. Cyclic Voltammetry. We performed cyclic voltammetry measurements for the cascades in order to determine the redox potentials of the building blocks and to estimate the energies of various charge-separated states upon photoexcitation (half-wave redox potentials (E1/2) are listed in Table 1). In these measurements, the NDI acceptor is reduced and up to two TAA donors are oxidized (see Figure 4). All processes were

Figure 4. Cyclic voltammograms in DCM/TBAHFP (∼0.2 M) of Ref 1, Ref 2, Ref 3, the dyads (scan rate = 250 mV s−1), and the triads (scan rate = 1000 mV s−1) vs Fc/Fc+.

The triazoles influence the oxidation of the adjacent triarylamines in an ambivalent way, inasmuch as triazoles connected via the carbon to the TAA act as an electrondonating group (EDG) while nitrogen-linked triazoles act as an electron-withdrawing group (EWG). Thus, in the cyclic voltammetry experiment Ref 2 and Ref 3 show reversible oxidation waves at 330 and 240 mV, respectively. This EWG behavior of N-linked triazole might be a result of the delocalization of the nitrogen lone pair into the triazole πsystem, which reduces the +M-effect and, thus, only the -I-effect due to the nitrogen electronegativity remains active. As a result, in Da the oxidation of the MeO-substituted TAA connected via the nitrogen to the triazole is observed at 320 mV, almost at the same position as in Ref 2 (E1/2 = 330 mV). Analogously, in Db the oxidation potential of the MeO-substituted TAA connected via the carbon to the triazole is found at 250 mV, which is very similar to the value found for Ref 3 (E1/2 = 240 mV). As apparent from Figure 4, the reduction potentials of the NDI in Da and Db are rather similar to the value found for Ref 1 (ca. −1000 mV) and, thus, are only marginally influenced by the connection mode to the triazole which suggests electronic decoupling of the NDI from the N-aryl substituents. Consequently, in the triads the NDIs are reduced at ca. −1000 mV and the oxidation of the MeO-substituted TAAs (TAA2) corresponds to the lowest oxidation wave at ca. 250 mV, which is similar to the values found for Db. All triads exhibit an additional oxidation wave at more positive potential caused by the oxidation of the intermediate TAA1. These oxidation processes are governed by the substituent (Me < Cl < CN) at TAA1. Thus, the CV experiment proved a downhilldirected redox gradient between TAA1 and TAA2 in all triads. Transient Absorption Spectroscopy. All cascades do not exhibit any fluorescence due to the quenching of the excited state by photoinduced electron transfer. In order to elucidate the dynamics of electron-transfer processes in the excited state and of other nonradiative processes upon photoexcitation, transient absorption spectroscopy measurements were per-

Table 1. Half-Wave Potentials (E1/2) of Dyads, Triads and Model Compounds. All E1/2 Values Were Estimated by Cyclic Voltammetry in DCM/TBAHFP (∼0.2 M) at Room Temperature and Referenced vs Fc/Fc+ Ered 1/2/mV Ref 2a Ref 3a Ref 1a Daa Dba T-Meb T-Clb T-CNb a

Eox1 1/2/mV

Eox2 1/2/mV

330 240 −1000 −1000 −980 −1000 −970 −980

320 250 240 250 240

530 640 780

Scan rate =250 mV s−1. bScan rate =1000 mV s−1. 27701



Article

Figure 5. Femtosecond transient absorption spectra corrected for chirp and scattered light of Db in MeCN (upper left) and toluene (upper right) at 26 300 cm−1 (380 nm) pump energy and related transient absorption decays at 21 100 cm−1 (470 nm) in MeCN (lower left) and toluene (lower right). Early spectra are given in dark blue, later spectra in red.

For Db in toluene, right after excitation the rise of a broad band is found extending from 12 500 cm−1 (800 nm) up to 25 000 cm−1 (400 nm) with a pronounced maximum at 21 100 cm−1 (470 nm). After ca. 1.5 ps, the spectra sharpen and the characteristic bands of the NDI radical anion16,31 at ca. 20 800 cm−1 (480 nm) and ca. 16 400 cm−1 (610 nm) emerge. A third weak and broad band at ca. 13 500 cm−1 (740 nm) might be partly caused by the NDI radical anion but also by the absorption of a TAA radical cation.10 All these bands rise within the first 10 ps and from then on they decrease. In MeCN Db behaves rather differently. Initially, an intensive band at ca. 16 900 cm−1 (590 nm) rises. This band has its maximum intensity at ∼1.3 ps and is caused by the 1NDI state. This band is replaced by an even stronger one at 16 400 cm−1 (610 nm) together with a very intense band at 21 100 cm−1 (470 nm). Similar to the pump−probe measurements in toluene, these two bands are assigned to the NDI radical anion. In parallel, a broad band around 14 300−12 500 cm−1 (700− 800 nm) rises, which is caused by the TAA radical cation. Thus, a CS state is populated which decays within ∼100 ps and gives rise to weak and broad ESA around 22 200 cm−1 (450 nm), which we assign to a 3NDI state.17 The spectra of T-CN in toluene develop quite identically to Db in toluene except that all bands are marginally broader for T-CN. Also in MeCN the evolution of the spectra of T-CN and

formed in the nanosecond and in the femtosecond time regime. Therefore, two different pump−probe setups were used. In the latter, the delay stage allows examinations of processes up to 8 ns and, in the former, deconvolution with the instrument response function (IRF ca. 9 ns) allows investigations from 2 ns to several milliseconds. In the following, the dynamics induced by photoexcitation will be described step by step until relaxation to the electronic ground state. Therefore, the results of the femtosecond measurements will be discussed first, which comprise the charge-separation processes, and an analysis of the nanosecond measurements will then describe the charge recombination and ISC processes. Femtosecond Transient Absorption. To shed light on the possible CT and CS state population processes in the cascades, femtosecond measurements in toluene and MeCN were performed for Db and T-CN and for T-Cl in MeCN only. All cascades were excited with 140 fs pulses at 26 300 cm−1 (380 nm) to prevent population of higher excited states. At this wavelength, ca. 90% NDI are excited for Da and 95% for T-Cl. The optical changes were probed by a white light continuum (140 fs) between 12 500 cm−1 (800 nm) and 25 000 cm−1 (400 nm). The transient absorption spectra corrected for white light dispersion (chirp) and scattered light are given in Figure 5 and in Figures S7−S9 (in the Supporting Information) and consist almost exclusively of excited-state absorption (ESA). 27702



Article

Figure 6. Normalized nanosecond transient absorption spectra in MeCN and toluene of Db and T-CN at 28 200 cm−1 (355 nm) pump energy. Early spectra are given in dark blue, later spectra in red.

Figure 7. Biexponential tail fits of the nanosecond transient absorption kinetics of Db and T-CN in MeCN and toluene pumped at 28 200 cm−1 (355 nm) and probed at 21 000 cm−1 (475 nm). The arrows mark the delay times after which the normalized transient absorption spectra in Figure 12 and Figure S13 (in the Supporting Information) are taken.

dyads and triads in MeCN and toluene, the ESA rise with the instrument response (ca. 9 ns) and decrease rapidly (see Figure 6 and Figure S10 in the Supporting Information). The spectra are dominated by a sharp and intensive band at about 21 000 cm−1 (475 nm) together with a smaller and less pronounced band at ca. 16 400 cm−1 (610 nm) which both are related to the NDI radical anion.16,31 The broad absorption band at ca.

T-Cl are very similar to the ones of Db in MeCN (see Figures S8 and S9 in the Supporting Information). Nanosecond Transient Absorption. In these measurements, the cascades were excited with a 5 ns laser pulse at 28 200 cm−1 (355 nm) and the excited-state dynamics were probed by a xenon flashlamp white light continuum between 12 500 cm−1 (800 nm) and 25 000 cm−1 (400 nm). For all 27703



Article

13 200 cm−1 (760 nm) is assigned to a TAA radical cation but may also feature some NDI radical anion absorption.10 According to the femtosecond measurements in MeCN, Db shows 3NDI state characteristics at the end of the time window (8 ns) in this femtosecond experiment. Due to signal-to-noise ratio limitations of the nanosecond measurements, this cannot be fully elucidated but it seems reasonable to consider that the dynamics of Db in the nanosecond measurement in MeCN consist of the CS and the 3NDI state since the spectra at least at the beginning clearly show the NDI anion and the TAA cation signatures. This might be caused by the long instrument response function of the nanosecond setup. Thus, the lifetimes evaluated by the femtosecond measurements are considered to be more accurate. To compare the results from the femtosecond transient absorption measurements with those from the nanosecond transient absorption measurements, the normalized transient absorption spectra at t = 0 and after longer delay times, where only the long-lived species persist, are presented in Figure 12 and Figure S13 in the Supporting Information. The lifetimes extracted from the biexponential tail fits and reconvolution fits (= deconvoluted with the measured instrument response function) at 21 000 cm−1 (475 nm) (see Figure 7 and Figure S11 in the Supporting Information) are listed in Table 2 and Table S1 in the Supporting Information. In

singlet and triplet species, which will be discussed below. For the triads in toluene, both lifetimes are in the nanosecond time regime with τshort ∼ 20−50 ns and τlong ∼ 300−600 ns. In MeCN both lifetimes are more similar to a slightly shorter lifetime of ca. 100−300 ns and a longer lifetime of ca. 350−800 ns. No concentration dependence of the lifetimes was found for the dyads in MeCN and triads in MeCN and toluene; however, the sample concentration could not be significantly varied due to the rather modest signal-to-noise ratio of the nanosecond transient absorption setup used. Much in contrast, for the dyads in toluene a decrease of τlong up to ∼35% with increasing concentration (2.1 × 10−6 to 1.0 × 10−5 M) is observed (see Table 2 and Table S1 in the Supporting Information). A reasonable explanation for this finding is a bimolecular deactivation process. Thus, the apparent τlong consists of two rate constants, kdi (rate constant for infinitely diluted concentration) and kda (rate constant for bimolecular deactivation). Indeed, a plot of 1/τlong as a function of concentration leads to a linear correlation. The slope of the linear regression gives kda and the intersection with the 1/τlong axis yields kdi ((kdi(Da) = 1.0 × 107 s−1, kdi(Db) = 7.7 × 108 s−1) while the values of kda (kda(Da) = 4.7 × 109 s−1, kda(Db) = 4.3 × 109 s−1) lie in the typical range for a diffusion-controlled bimolecular deactivation process (see Figure S12 in the Supporting Information).5 As indicated by the O2 sensitivity of the excited-state lifetimes, the relaxation dynamics of the cascades are strongly influenced by singlet−triplet transitions. Therefore, we performed magnetic field dependent transient absorption measurements for Db and T-CN in toluene. Upon stepwise increase of the magnetic field from 0 to ca. 2000 mT, no significant effect could be observed for Db. However, the transient absorption decay of T-CN is strongly magnetic field dependent (see Figure 8). From 0 to 2000 mT all transient absorption decays show biexponential kinetics but the corresponding lifetimes τshort and τlong change with increasing magnetic field. With an applied magnetic field of 1 mT, τshort becomes significantly and τlong slightly longer than the zero-field lifetimes. From then on, τshort starts to decrease while τlong simultaneously starts to increase with increasing field until changes of the decay become insignificant at higher fields (>30 mT).

Table 2. Lifetimes and Ratio of Amplitudes of Nanosecond Transient Absorption Experiments of Da, Db, T-Me, T-Cl, and T-CN in Toluene at the Given Concentration and Laser Pulse Energy at 28 200 cm−1 (355 nm) Assuming Independent Biexponential Decay at 21 000 cm−1 (475 nm) E/ mJ

c/M Da Db T-Me T-Cl T-CN

2.9 2.1 4.6 4.5 5.1

× × × × ×

−6a

10 10−6a 10−6 10−6 10−6

1.2 1.2 0.8 0.8 0.8

along/ ashort

τlong/ns

0.032 0.057 0.86 0.56 0.58

± ± ± ± ±

(8.6 (12 700 540 270

0.4) × 10 1) × 103 9 1 10

τshort/ns 3

19 7.1 50 36 22

± ± ± ± ±

1 0.5 1 1 1

a Used for the simulation with TENUA32 to obtain the rate constants kS and kST.

■

toluene, all the transient absorption bands decay biexponentially with almost identical time constants for both dyads, a long one (τlong) in the microsecond time regime, which is strongly oxygen sensitive, and a short one (τshort) in the nanosecond time regime, which is only slightly influenced by remaining oxygen in the sample. This might be a hint to the presence of

DISCUSSION From the results presented in the preceding section, it is obvious that upon photoexcitation of the cascades an electron is transferred from the TAA to the NDI. This leads to the

Figure 8. Left: Normalized biexponential reconvolution fits of the nanosecond transient absorption decays of T-CN in toluene at increasing magnetic field. Right: Experimental decays superposed by the biexponential reconvolution fits for selected magnetic fields. 27704



Article

population of CS states with lifetimes of several nanoseconds to microseconds. Before we can discuss the diverse photophysical processes in the dyads and triads, an estimation of the relative state energies is required, as these data will determine the ET dynamics to a large extent. The energies of the CS states populated in the two dyads are expected to be somewhat different, as the redox potentials of the TAA in Da and Db differ, which in turn leads to different ΔG0CS values. These ΔG0CS values of the CS states were calculated using the redox potentials obtained from cyclic voltammetry experiment (Table 1) as the input in the Weller equation (eq 1)33 0 ΔGCS =

NAze ox NAe 2 red (E1/2 − E1/2 )− 1000 1000·4πε0 ⎛⎛ 1 1 ⎞⎛ 1 1⎞ 1 ⎞ ⎟⎟ × ⎜⎜⎜ + ⎟⎜ − ⎟ + 2rA ⎠⎝ εr εs ⎠ εsdDA ⎠ ⎝⎝ 2rD

(1)

where NA is Avogadro’s constant, e is the elementary charge, ε0 is the is vacuum permittivity, εs is the permittivity of the solvent used for spectroscopic measurements, εr is the permittivity of the solvent used in the electrochemical measurements, z is the number of transferred electrons, rD and rA are the radii of the donor and the acceptor, respectively, and dDA is the electrontransfer distance.The radii of the donor rD (=4.81 Å) and the acceptor rA (=3.97 Å) were estimated from the Connolly molecular surfaces volume of the respective moieties calculated with MM2 using the ChemBio3D Ultra34 software. For the electron-transfer distances dDA, the distances between the chromophore centers were used, which was determined by calculating the geometries of the cascades at B3LYP/6-31G* level of theory (see Figure 9). The triads may populate two different CS states, in which the NDI is reduced and either TAA1 (leading to CS1) or the terminal TAA2 (leading to CS2) is oxidized. As a result of the very similar E1/2 values for the reduction of the NDI and for the oxidation of the MeO-TAA, the CS2 states of all triads possess almost the same free energy (ΔG0CS). Because the CS2 energies are strongly distance dependent and since the triads are flexible in solution, the CS2 energies were calculated assuming two different conformers (see Figure 9). One of these conformers is highly bent and has a chromophore distance of 22.5 Å and the other is elongated with a distance of 28.0 Å. The two structures may serve to estimate the smallest and the largest possible distance and, thus, the range of possible CS2 energies.35 Much in contrast, the free energies of the CS1 states differ for the triads according to the different substituents (Me, Cl, CN) attached to the TAA1. For estimating the ΔG0CS values of the CS1 states, we used the distance obtained for the dyads (dAD = 16.7 Å). The corresponding free energies of all states are listed in Table 3. In the following, we will discuss the photophysical processes monitored by femtosecond transient absorption spectroscopy. For this reason, the number of individual spectral components n were determined from the transient (wavenumber × time) maps by singular value decomposition. Subsequently, the timeresolved spectra were globally fitted with GLOTARAN36 employing the target models presented in Figure 10, the IRF was treated as a Gaussian-shaped function (ca. 150 fs), corrected for the white light dispersion (chirp) and the coherent artifact (the models used have the time characteristics

Figure 9. Optimized structure of Db and T-CN in an elongated and a strongly bent conformation using DFT at B3LYP/6-31G* level of theory.

Table 3. Free Energies ΔG0CS of the CS States Estimated by Weller Equation and the E1/2 Values Given in Table 1 ΔG0CS/eV Da Db T-Me

T-Cl

T-CN

CS1 CS2 CS1 CS2 CS1 CS2

bent elongated bent elongated bent elongated

toluene

MeCN

1.97 1.88 2.18 1.99 2.04 2.26 1.97 2.02 2.41 1.97 2.02

1.01 0.93 1.23 0.94 0.95 1.30 0.92 0.93 1.46 0.92 0.93

of the IRF) at time zero to give species associated difference spectra (SADS) and their related time constants (Figure 11 and Table 4). In this target model, the efficiencies of individual processes were adjusted in order to obtain equal absorption intensity of the NDI radical anion at 21 000 cm−1 (475 nm). Before we discuss the photophysics of the triads, first the processes occurring in Db need to be elucidated to provide the basis for a better understanding of the processes taking place in the triads. In MeCN Db gives five SADS corresponding to five lifetimes. The first SADS with a lifetime of 130 fs exhibits the typical features of the NDI singlet excited state (1NDI), a distinct maximum at 16 900 cm−1 (590 nm) with a shoulder at 18 200 cm−1 (550 nm) and a broad band at ca. 21 750 cm−1 (460 nm).17 Thus, although both NDI and TAA absorb to some extend at the pump energy, only the NDI is excited, 27705



Article

Figure 10. State diagrams and relaxation dynamics of the photophysical processes of Db, T-Cl, and T-CN in MeCN and of Db and T-CN in toluene upon excitation at 26 300 cm−1 (380 nm) from femtosecond and nanosecond time-resolved pump probe measurements. The gray horizontal bars represent states. Lifetimes and state energies are indicated as well as percentages for specific decay pathways.a Taken from the rising edge of the absorption band of Ref 1 in MeCN or toluene.b Taken from the rising edge of the high-energy side of the fluorescence band of NDI.18 c State energies estimated by the Weller approach (Table 3).d Taken as the average of the triplet energy of NDI in in EtOH at 77 K (2.03 eV)37 and in 2‑methyltetrahydrofuran at 77 K (2.05 eV).38 e Lifetimes estimated as 1/ks, extracted from TENUA32 fits of the nanosecond measurements.

lasting SADS with an (in the femtosecond experiment) infinitely long lifetime is related to the triplet NDI state, with its typical double absorption maxima at 22 700 cm−1 (440 nm) and 21 750 cm−1 (460 nm), populated via an additional pathway directly starting from the 1NDI state.18 According to this target model, the CS state is populated in only ca. 25% quantum yield. In toluene Db possesses three SADS. The first SADS features a broad absorption band which extends up to 12 500 cm−1 (800 nm) but exhibits a pronounced maximum at 21 100 cm−1 (470 nm). This SADS can be assigned to the localized excited NDI, which appears to have a different spectral characteristic in toluene compared to MeCN. With τ = 1.8 ps the 1NDI is transformed into a CT state, as indicated by the rise of the typical NDI radical anion band at 21 000 cm−1 (475 nm) and the weaker absorption at 16 400 cm−1 (610 nm). This CT state has a lifetime of only 13 ps, then the charge is transferred to the TAA and a CS state with a lifetime of 4.6 ns is populated. This interpretation is supported by the small shift of the weak NDI radical anion band from 16 400 cm−1 (610 nm) to 16 600 cm−1

which probably is a consequence of ultrafast energy transfer from the TAA to the NDI within the instrument response time. The 1NDI spectrum is followed by a SADS presenting an intense absorption at 21 100 cm−1 (470 nm) and a weaker one at 16 400 cm−1 (610 nm), which decays within τ = 1.1 ps into a very similar SADS but with somewhat sharper peaks. These signals are characteristic of the NDI radical anion16,31 and we assign those SADS to a hot charge transfer (hotCT) state where the negative charge is located on the NDI and the positive charge on one of the phenyl groups attached to the nitrogen atoms. This vibrationally hot state relaxes with τ = 1.1 ps into the vibrational CT ground state (coolCT).18 We stress at this point that, while there is a strong evidence for the NDI radical anion formation, the exact position of the corresponding positive charge in the CT states is unclear as no specific spectroscopic features are visible. Wherever the positive charge is, it is then transferred with τ = 13 ps to the TAA populating a charge-separated state. This process is indicated by the appearance of a broad TAA radical cation band at ca. 13 300 cm−1 (750 nm), well-described in the literature.10 From this CS state charge recombination occurs with τ = 43 ps. The longest 27706



Article

Figure 11. Species-associated difference spectra (SADS) of Db, T-Cl, and T-CN in MeCN and of Db and T-CN in toluene at 26 300 cm−1 (380 nm) pump energy. The gray bars hide stray light of the laser fundamental at 12 500 cm−1 (800 nm) and the second-order signal from the pump pulse at 13 200 cm−1 (760 nm).

Table 4. Lifetimesa Associated with the SADS Obtained by Global Analysis of the Femtosecond Transient Absorption Map Db T-Cl T-CN a

MeCN toluene MeCN MeCN toluene

τ1(1NDI)/ps

τ2(hotCT)/ps

τ3(coolCT)/ps

τ4(CS1)/ps

τ5(CS2)/ps

0.13 1.8 0.11 0.092 1.8

1.1 − 0.97 0.96 11

13 13 13 12 120

43 4600 480 170 4000

∞b − ∞ ∞ ∞

The standard errors of the fits were below 1% in all cases.

b3

NDI state.

Subsequent electron transfer with τ = 170 ps from the terminal TAA2 to TAA1 yields the SADS of the CS2 which lives up to the nanosecond time regime. The development of SADS spectra of T-Cl in MeCN is essentially the same as that of T-CN, besides the lifetime for the CS1 state, which is with 480 ps much longer than in T-CN (τ = 170 ps). The redox potentials of the TAA1s are considerably different for T-CN and T-Cl and so are the differences of the free energies between the CS1 and CS2 states (Table 3). Accordingly, the CS1−CS2 transition in T-CN possesses a ca. 0.16 eV higher driving force than in T-Cl, resulting in a much

(600 nm) and by the increased intensity of the broad band at ca. 13 000 cm−1 (770 nm). With this information at hand, we can now discuss the processes occurring in the triads. Similar to Db, the first SADS for T-CN in MeCN is assigned to the 1NDI state which decays with a lifetime of 92 fs into a hotCT state followed by relaxation (τ = 960 fs) to the vibrational CT ground state. This cooled CT state has a lifetime of 12 ps and goes into the CS1 state by transferring the positive charge from the phenyl group to TAA1. This process can be monitored by the rise of the TAA radical cation absorption at ca. 13 300 cm−1 (750 nm). 27707



Article

Figure 12. Normalized nanosecond transient absorption spectra of Db in MeCN and toluene at t = 0 and after longer time, at which the short component only insignificantly contributes to the spectra (indicated in Figure 7 and Figure S11 in the Supporting Information). The femtosecond transient absorption spectra at infinite time are shown for comparison.

is spin-forbidden and, thus, charge recombination should only take place via the 1CS state which requires again a spin flip from the 3CS to the 1CS state. At zero magnetic field, this 1,3CS equilibrium is mediated by the isotropic hyperfine coupling interaction (ihfc) which gives rates between the triplet and the singlet states and between all degenerate triplet levels on the order of 108 s−1 for organic radical pairs.39 Using the hyperfine coupling constants (aik) extracted from ESR experiments of N,N′-disulfonylphenyl-NDI radical anion40 and dianisylphenylamine radical cation41 and eq 2 (Iik = total spin quantum number), we evaluate an effective hyperfine field42 of B1/2 = 2.5 mT which corresponds via eq 343,44 to a spin-flip rate of 1/τ = 8.3 × 107 s−1 at applied zero magnetic field.

shorter lifetime of the CS1 state for T-CN compared to T-Cl because these processes are located in the Marcus normal region. Exciting T-CN in toluene with 26 300 cm−1 (380 nm) pump energy first populates the 1NDI state similar to the situation in Db. This 1NDI state has a lifetime of 1.8 ps and decays into a hot CT, which then relaxes with τ = 11 ps into the coolCT state. From there the CS1 state is populated with τ = 120 ps and finally the CS1 state decays with τ = 4.0 ns into the CS2 state, which has an infinitely long lifetime in the femtosecond experiment. From the data given in Table 4 it is obvious that the lifetimes of the particular species of T-CN in toluene are significantly larger (by ca. one order of magnitude) compared to those in MeCN. A comparison of the results from the femtosecond transient absorption measurements with the normalized transient absorption spectra at t = 0 and those after longer delay times from the nanosecond transient absorption measurements revealed that the first spectra of the triads in MeCN and toluene and of the dyads in toluene show very similar characteristics as the femtosecond transient absorption spectra at infinite time (see Figure 12 and Figure S13 in the Supporting Information). These features are a sharp and intense absorption at ca. 20 800 cm−1 (480 nm) and a smaller band at 16 400 cm−1 (610 nm), which are related to the NDI radical anion, and a much broader band at ca. 13 200 cm−1 (760 nm), related to the TAA radical cation. Thus, the t = 0 spectra of the nanosecond measurements pertain to the lowest energy CS states in the cascades. The femtosecond measurements of Db in MeCN show a lifetime of 43 ps for the CS state while in the nanosecond measurement a biexponential decay of the CS state of 9 and 500 ns is observed. We assume that the shorter component with τ = 9 ns is caused by convolution of the 48 ps with the instrument response (τ ∼ 9 ns) and the larger component (τ = 500 ns) is invisible in the femtosecond measurements because of its small amplitude. The bad signal-to-noise ratios of the spectra at longer delay times complicate an analysis. But it seems obvious that, even after delay times, when the short component contributes only insignificantly to the observed spectra, the spectra show similar bands to the spectra at t = 0 and are, therefore, also related to a CS state species. As a working hypothesis, we assume that primarily only the 1 CS state is populated, which then undergoes ISC into the 3CS state. Charge recombination to the S0 state from the 3CS state

B1/2 =

3(B12 + B2 2 )

with Bi =

∑ aik 2Iik(Iik + 1) i

τ=

30 ns = 12 ns B1/2 /mT

(2)

(3)

However, at magnetic fields B > 0 the equilibrium is expected to be strongly influenced since the triplet sublevels experience a Zeeman splitting (Figure 13). In this case, only the spin

Figure 13. Illustration of the singlet−triplet dynamics at zero magnetic field and at B > 0.45

exchange between the 1CS and the 3CS0 state is a coherent process caused by ihfc while all other processes (e.g., between 1 CS and 3CS+) are slowed down and become incoherent relaxation processes mediated by the anisotropic hfc at low field. This general description holds as long as the energy difference between 1CS and the 3CS0 is negligibly small at B = 0. Below, we will refine this picture somewhat. 27708



Article

For the analysis, we assume that the spectra of the 1CS and CS are indistinguishable and the two states are practically degenerated because of the large distance between the two radical centers. Applying the kinetic scheme of Figure 13 for B = 0, the deconvoluted decays of all dyads and triads were simulated using the TENUA32 software and the mechanism given in the Supporting Information to obtain the rate constants for charge recombination kS and ISC (1CS → 3CS) kST. Herein we set kST = 3kTS to account for spin statistics of degenerate states.33 Furthermore, it was assumed that the 1CS state is initially exclusively populated. The resulting data are listed in Table 5. 3

Table 5. Rate Constants of Charge Recombination kS and Intersystem Crossing kST Obtained from Fitting the Nanosecond Transient Absorption Decays with the Kinetic Model of Figure 13 at B = 0 toluene kS/s−1 a

Da Dba T-Meb T-Clb T-CNb

5.3 1.5 1.2 1.7 2.9

× × × × ×

107 108 107 107 107

Figure 14. Singlet and triplet energy levels of the triads as a function of the magnetic field B.

suppressed from this state. At even higher fields, ihfc is no longer effective and other incoherent processes (mainly anisotropic hyperfine interaction (ahfc)) are operating for the interconversion of 1CS with either 3CS− or 3CS+.53 Applying the kinetic scheme from Figure 13 in a global analysis of the biexponential reconvolution fits (see Figure 8) essentially support the supposed mechanism for 1CS−3CS interconversion and the assumption that 1CS is primarily populated. This can be seen in Figure 15 where the fitted decay

MeCN kST/s−1

3.5 2.5 7.2 8.0 2.0

× × × × ×

105 105 106 106 107

kS/s−1 − 2.3 1.5 1.2 8.7

× × × ×

1010c 107 107 106

kST/s−1 − − 6.7 × 107 6.2 × 107 4.7 × 107

a

Measurement at lowest concentration, see Table 2. bAverage value of the corresponding fits. ckS for Db in MeCN is calculated as 1/τ4, extracted from the femtosecond measurement.

It becomes obvious that kS is up to one order of magnitude smaller in the triads than in the dyads in toluene (three orders of magnitude in MeCN) whereas kST in toluene is two orders of magnitude higher. While for Db the kS rate constant differs strongly between MeCN and toluene, for the triads the kS rates are very similar in both solvents. This might be fortuitous because charge recombination in toluene takes place in the Marcus inverted region and in MeCN in the Marcus normal region (vide infra). The much slower kTS of the dyads compared to the triads is expected to arise from singlet−triplet splittings (ΔST) being larger in the dyads than in the triads (vide infra). The supposed kinetic model of Figure 13 at B = 0 leads to the observed biexponential decay if either 1CS is primarily populated (this refers to Hayashi’s so-called “relaxation mechanism”)39,45−48 or if 3CS is populated in parallel to 1CS but not if 3CS is populated primarily. In order to support this mechanism, magnetic field dependent measurements may help.49−51 In general, at zero magnetic field, the three triplet states are degenerate as long as zero field splitting of organic radicals is neglected. The triplet states are higher or lower in energy by twice the exchange interaction J compared to the 1CS state.4 In the present case, the singlet state is lower as was estimated by TD-DFT computations (vide infra). With increasing magnetic field, the 3CS0 state is unaffected while the 3CS+ and 3CS− states start to split in energy with ΔE = gβB, where g is the electronic g-factor, β Bohr’s magneton, and B the magnetic field (see Figure 14).52 At medium field, the 3CS− comes close to the 1CS in energy and thus kT−,S increases caused by coherent spin interconversion mediated by the isotropic hyperfine interaction (ihfc) (=level crossing), and simultaneously 3CS+ rises in energy and the coherent spin interconversion becomes

Figure 15. Experimental nanosecond transient absorption decays of T‑CN in toluene and fits of the reconvolution fits (Figure 8) for selected magnetic fields applying the kinetic scheme from Figure 13 in a global analysis.

curves are overlaid onto the experimental curves. While the general trend with increasing magnetic field is followed very well, there are deviations at early decay times. However, we stress that there are intensity fluctuations in the noisy experimental data which are inherent to the relatively weak transient signals compared to other measurements of similar compounds with the same setup. In the global fits we assume that, because the 3CS0 state energy remains unchanged by an applied magnetic field, the rate constant for the 3CS0−1CS transition (kT0,S) turns out to be essentially unaffected by the magnetic field. The same is true for the rate constant kS, the rate constant for charge recombination from 1CS to the electronic ground state. Due to spin conservation rules, the rate constant for charge recombination from the 3CS state was set to zero. Furthermore, it was assumed that kT0,S ≫ kS, which means that this equilibrium is held during all stages of charge recombination. This allows to combine the particular rate constants kT+,S, kT−,S, kT+,T0, and kT−,T0 to a single effective rate kr (=kT+,S + kT−,S + kT+,T0 + kT−,T0) as the rate constant considering processes including 3CS+ and 3CS−.45 The global 27709



Article

and ΔST = 0.026 cm−1 for T-CN and also show that the 3CS state is higher in energy than the 1CS state. Thus, the level crossing is expected at 27 mT for the triads and at 7700 mT for the dyads. Although the first value is clearly an order of magnitude higher than the experimental estimate, the latter is clearly much higher than the applied magnetic field (B < 2000 mT). Thus, we assume that ΔST of the triads is small enough to lie within the effective hyperfine field of ca. 2.5 mT (see above) which promotes fast coherent ISC at applied zero magnetic field while for the dyads ΔST is so large that only incoherent processes are at work even at zero field and the level-crossing field is out of experimental reach. Consequently, the smaller rate constants for ISC between the CS states (kST) of the dyads compared to the triads reflect the larger ΔST values of the dyads. Furthermore, the ΔST of the triads’ CS states seem to be larger in toluene than in MeCN, as indicated by the smaller kST values in toluene compared to MeCN; see Table 5. This solvent dependence needs explanation: according to the perturbation approximation (eq 6) given by Anderson et al. and applied by many others, the ΔST is developed as a sum over states5,54−58 where ES and ET are the energies of the 1CS and 3CS states, respectively, V1,3CS‑Nn are the couplings between the 1,3CS states and the surrounding states n and ΔG1,3CS‑Nn are the energy gaps between the 1,3CS states and the states to which they are coupled.

analysis revealed a strong magnetic field dependency for kr, which proceeds exactly as expected from Figure 14; that is, starting from the zero field situation kr increases at an applied field of 1 mT due to a level crossing of 3CS− with 1CS. From then on kr decreases asymptotically with increasing magnetic field due to an increasing Zeeman splitting (see Figure 16).

Figure 16. Plot of the rate constant kr for T-CN in toluene (obtained from a global analysis of the biexponential reconvoluted fits of the experimental decays (see Figure 8) vs applied magnetic field and fit of the curves by eq 4. The kr value for B = 0 is indicated by an arrow.

Thus, a plot of the rate constant kr follows eq 4,42 where |Vafhc|2 is the squared value of the anisotropic hyperfine coupling matrix element between the 1CS and the 3CS+ and 3CS− state, respectively, ω0 is the angular frequency, and τc is the orientational correlation time. Equation 4 describes the anisotropic hyperfine interaction and thus the influence of the magnetic field on the 1CS−3CS mixing.45 From this fit, we obtain τc = 3.53 × 10−10 s, which is in reasonable agreement with the value (τc = 5.9 × 10−10 s, with r = 9.0 Å (determined from the Connolly molecular surface of T-CN with ChemBio3D Ultra34) estimated by the Debye equation (eq 5)39 with r being the radius of the molecule in meters, k being the Boltzmann constant (Js), T being the temperature, and η being the solvent viscosity (η = 0.6 m·Pa·s) of toluene. 2τc 2 kahfc = 2 |Vahfc|2 1 + ω02τc2 ℏ

⎡ ⎡ V12CS−N ⎤ V32CS−N ⎤ n ⎥ n ⎥ ⎢ ⎢ ΔST = ES − E T = ∑ − ∑ ⎢ n ΔG1 ⎥ ⎢ n ΔG3 ⎥ ⎣ ⎣ CS − Nn ⎦ CS − Nn ⎦ S T (6)

(5)

The solvent dependence of the ΔST might be the result of a solvent dependence of the mixing of the 1,3CS states with other states. As apparent from eq 6, those states that are next in energy to the 1,3CS states have the strongest influence on the ΔST. The energetic position of the 3NDI state (2.04 eV) (taken as the average of the triplet energy of NDI in EtOH at 77 K (2.03 eV)37 and in 2-MeTHF at 77 K (2.05 eV)38) seems to be the most important factor because it is next to the 3CS state in energy (for an approximate energy diagram, see ref 5). As apparent from the values given in Figure 10, and Table 3, the 3CS state is closer to the 3NDI state in toluene than in MeCN which results in a stronger mixing of these states. This in turn causes a larger ΔST in toluene compared to in MeCN.

As apparent from the discussion above, the transient absorption decays of T-CN show a strong magnetic field dependence at rather low magnetic fields while higher fields result in insignificant changes of the lifetimes. Thus, in the case of the triad, it can be assumed that the ΔST splitting is very small and the level-crossing occurs at very low magnetic field. From the plot in Figure 16 one can see a deviation of the fit at B = 3 mT. This deviation could indeed be caused by the levelcrossing effect which would place ΔST on the order of ca. 1−3 mT. On the other side, Db does not exhibit any change of either the lifetimes or the amplitudes of the biexponential transient absorption decay during magnetic field dependent measurements. For Db the ΔST is probably so large that the necessary field to accomplish the level crossing is not reached and therefore no magnetic field effect is observed. This hypothesis was supported by TD-DFT computations of Db and T-CN at CAM-B3LYP/6-31G* level of theory in the gas phase.5 These computations yielded ΔST = 7.2 cm−1 for Db

CONCLUSION The investigated dyads and triads exhibit charge separation upon photoexcitation. As apparent from the femtosecond and nanosecond transient absorption measurements, the locally excited triplet NDI state is only populated in Db in MeCN. These measurements and the CV measurements reveal a redox gradient and a downhill oriented progression of CS states in the triads. In these triads, first a CS state is populated in which the intermediate TAA1 is oxidized and the NDI is reduced followed by an additional electron transfer to populate the fully charge separated CS2 state. The signals of the CS states of all triads and dyads feature biexponential decays with a short and a long component. These biexponential decays are the result of an ISC from the primarily populated 1CS state into the 3CS, from which charge recombination to the S0 state is forbidden by spin conservation rules. Charge recombination into local triplet states is prevented as the 3CS state is the lowest-lying

with ω0 = 1.76 × 1011 s−1 × B /mT

(4)

3

τc =

4000 πr η 3 kT

■

27710



Article

electrode, and either a platinum pseudoreference or a Ag/AgCl Leak Free reference electrode (Warner Instruments, Hamden, CT) were used in a sealed glass flask purged with argon. Chemical and electrochemical reversibility of the redox processes were tested by multi-thin-layer measurements and all potentials were measured against the ferrocene/ferrocenium (Fc/Fc+) redox couple as reference. Femtosecond Transient Absorption Spectroscopy. These pump−probe experiments were performed in a 2 mm quartz cuvette (Spectrocell Inc.) at RT. The sample was dissolved in the same solvent as used for the nanosecond measurements (OD ≈ 0.3 at the excitation energy), degassed with argon for 30 min prior to the measurement, and sustained stirred during the measurement. The setup consists of a commercial Helios transient spectrometer from Ultrafast Systems and is driven by a Solstice Ti:sapphire oscillator amplifier from Newport-Spectra-Physics (pulse duration of 100 fs) with a fundamental wavenumber of 12 500 cm−1 (800 nm) and a repetition rate of 1 kHz. The output beam gained from the Solstice amplifier was split into two parts. One part was used to pump an optical parameter amplifier (TOPAS-C) from Light Conversion as the source for the pump pulses with a pulse length of 140 fs and a wavenumber of 26 300 cm−1 (380 nm) with an attenuated energy of 150−210 nJ. The other, quite small part of the Solstice amplifier output beam was focused into a moving CaF2 plate to generate a white light continuum between 12 500 cm−1 (800 nm) and 25 000 cm−1 (400 nm) which was used as the probe pulse. The excitation pulse was collimated to a spot, which was at least 2 times larger than the diameter of the spatially overlapping probe pulse. The excitation pulse polarization was set to a magic angle toward the probe pulse polarization. Detection of the probe pulses was achieved using a CMOS sensor (Ultrafast Systems, Helios) with 1.5 nm intrinsic resolution and 350−800 nm sensitivity range. A part of the probe light was used to correct for intensity fluctuations of the white light continuum. A mechanical chopper (working at 500 Hz) blocked every second pump pulse, in order to measure I and I0. The photoinduced change of the optical density can be recorded by comparing the transmitted spectral intensity of consecutive pulses (I(λ,τ),I0(λ)) by eq 7:

triplet state. This assumption was supported by a strong magnetic field effect found in a magnetic field dependent nanosecond transient absorption measurement of T-CN in toluene. The charge recombination rate ks for the recombination of the 1CS state is prolonged in the triads compared to the dyads due to an increased spatial distance between the two charges in the fully CS states in the triads. In all cascades, the total reorganization energy (λv + λo) is smaller than the |ΔG0CS| in toluene and larger in MeCN. Thus, charge recombination in toluene takes place in the Marcus inverted region and in MeCN in the Marcus normal region. Indeed, Db exhibits a strong inverted region effect that places the lifetime of the CS state in MeCN in the ps and in toluene in the nanosecond time regime. However, for the triads no inverted region effect can be observed which is probably because the charge recombination processes in MeCN (Marcus normal region) and toluene (Marcus inverted region) are by accident similar. Although rate constants allow a more detailed analysis of the processes in the cascades, for practical applications the overall lifetime of the specific CS state is much more important. Thus, although the charge recombination (ks) from the 1CS state in the triads is slowed down compared to the dyads, the lifetime of the 3CS states is clearly longer in the dyads. This is a result of the larger ΔST splitting in the dyads. A lifetime of tens of nanoseconds is a reasonable value for small donor−acceptor dyads but a lifetime of several microseconds is remarkable for such a small distance separating the two charges. From the ISC rate constants kST it can be seen that the ΔST in the triads is strongly solvent dependent. In toluene the coupling of the 3CS state to the 3NDI state is more effective than in MeCN which leads to a larger ΔST. Therefore, concerning the design of species with a high 1,3CS state energy, it should not only be guaranteed that local excited triplet states are higher in energy than the 3CS state in order to prevent triplet charge recombination but also that higher lying local triplet states may influence the ΔST splitting and in this way the 1,3CS spinflip dynamics. Furthermore, the fully CS state is much faster populated in the dyads because in the triads a second electron transfer must occur from the intermediate CS1 state, which also exhibits a certain lifetime. The reasons given above prove that the control of spin correlation and of the ΔST splitting is a promising key parameter to achieve long-lived CS states in small donor− acceptor systems.

⎛ I (λ , τ ) ⎞ ΔOD = − log⎜ ⎟ ⎝ I0(λ) ⎠

■

(7)

The relative temporal delay between pump and probe pulses was varied over a maximum range of 8 ns with a motorized, computer-controlled linear stage. While the delay interval between two consecutive data points was 13.3 fs for small delay times, it increased up to 200 ps for very large delay times. Steady-state absorption measurements of the sample before and after the pump−probe measurements were performed to verify the stability of the sample. The obtained time-resolved spectra were corrected for stray light and the white light dispersion (chirp). The chirp was corrected by fitting a polynomial to the cross phase modulation signal of the pure solvent under otherwise experimental conditions. Further analyses of the spectra using GLOTARAN36 is described above. Nanosecond Transient Absorption Spectroscopy. For these experiments, the substances were dissolved in the same solvent as used for steady-state experiments filled in 1 cm sealed quartz cuvettes (Starna, Pfungstadt, Germany) and the solutions were degassed by bubbling argon through the

EXPERIMENTAL SECTION Steady-State Absorption Spectroscopy. The absorption spectra of all cascades were measured in 1 cm quartz cuvettes using a JASCO V-570 UV/vis/NIR spectrometer and with pure solvent as reference. The solvents were Uvasol solvent purchased from Merck (MeCN) or Spectronorm purchased from VWR (toluene). All compounds were investigated in a concentration range from ca. 1.0 × 10−6 to ca. 5.0 × 10−5 M to exclude aggregation effects. Cyclic Voltammetry. Electrochemical experiments were either performed with a BAS CV-50W electrochemical workstation or with a computer-controlled Gamry Instruments (Reference 600) potentiostat (Warminster, PA) under argon atmosphere in dry oxygen-free DCM with n-tetrabutylammonium hexafluorophosphate (TBAHFP) (ca. 0.2 M) as supporting electrolyte. A three-electrode setup with a platinum disc working electrode (Ø = 1 mm), a platinum wire counter 27711



Article

Figure 17. Illustration of the magnetic field dependent nanosecond transient absorption setup.

■

solution for 60 min prior to measurements. These pump− probe experiments were performed with an Edinburgh LP 920 Laser Flash spectrometer consisting of a Nd:YAG laser (Continuum, Minilite II) operating at 10 Hz, probe light source (xenon arc lamp), sample chamber and detector (photomultiplier tube). As pump pulse we used the third harmonic at 28 200 cm−1 (355 nm) and varied the laser pulse energy between 0.2 mJ and 2.0 mJ with a pulse length of 5 ns. The instrument response function (IRF) (∼9 ns) was determined by measuring the scattered light using a LUDOX AS-30 colloidal silica suspension in water. The measured transient signal intensity was corrected by the fluorescence intensity obtained from measurements with low-intensity probe light. Decay traces with lifetimes shorter than 100 ns were deconvoluted with the IRF and decay curves with longer lifetimes were tail-fitted using the corresponding spectrometer software. Residuals and autocorrelation function (without any significant structure) served as the main criteria in the evaluation of the fits. Decay curves recorded at the threeband maxima all present identical biexponential kinetics for a given compound. Excitation with different laser pulse energies had no effect on the dynamics and the transient spectra and thus two-photon processes are excluded.37 For the magnetic field dependent measurements the pump beam was directed through bores in the poles of an electromagnet (model 5403FG, GMW Associates) which was powered by a Sorensen (DLM-E 3 kW) power supplier. The pulse and the probe beam were oriented perpendicular and met each other in the sample cuvette (see Figure 17). The magnetic field was measured by a Hall sensor (Single-Axis Magnetic Field Transductor YM12-25-5T, SENIS GmbH) and varied in 1 mT steps (up to 20 mT), 4 mT steps (up to 100 mT), 10 mT steps (up to 500 mT), 50 mT steps (up to 1000 mT), and 100 mT steps (up to 2000 mT).

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Phone: +49 931-31-85318. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by the Deutsche Forschungsgemeinschaft.

■

REFERENCES

(1) Balzani, V.; Credi, A.; Venturi, M. Photochemical Conversion of Solar Energy. ChemSusChem 2008, 1, 26−58. (2) Wasielewski, M. R. Energy, Charge, and Spin Transport in Molecules and Self-Assembled Nanostructures Inspired by Photosynthesis. J. Org. Chem. 2006, 71, 5051−5066. (3) Adams, D. M.; Brus, L.; Chidsey, C. E. D.; Creager, S.; Creutz, C.; Kagan, C. R.; Kamat, P. V.; Lieberman, M.; Lindsay, S.; Marcus, R. A.; Metzger, R. M.; Michel-Beyerle, M. E.; Miller, J. R.; Newton, M. D.; Rolison, D. R.; Sankey, O.; Schanze, K. S.; Yardley, J.; Zhu, X. Charge Transfer on the Nanoscale: Current Status. J. Phys. Chem. B 2003, 107, 6668−6697. (4) Verhoeven, J. W. On the Role of Spin Correlation in the Formation, Decay, and Detection of Long-Lived, Intramolecular Charge-Transfer States. J. Photochem. Photobiol. C: Photochem. Rev. 2006, 7, 40−60. (5) Kaiser, C.; Schmiedel, A.; Holzapfel, M.; Lambert, C. Long-Lived Singlet and Triplet Charge Separated States in Small CyclophaneBridged Triarylamine−Naphthalene Diimide Dyads. J. Phys. Chem. C 2012, 116, 15265−15280. (6) Verhoeven, J. W.; van Ramesdonk, H. J.; Groeneveld, M. M.; Benniston, A. C.; Harriman, A. Long-Lived Charge-Transfer States in Compact Donor−Acceptor Dyads. ChemPhysChem 2005, 6, 2251− 2260. (7) Lambert, C.; Schelter, J.; Fiebig, T.; Mank, D.; Trifonov, A. Photoinduced Charge Transfer Processes along Triarylamine Redox Cascades. J. Am. Chem. Soc. 2005, 127, 10600−10610. (8) Lin, B. C.; Cheng, C. P.; Lao, Z. P. M. Reorganization Energies in the Transports of Holes and Electrons in Organic Amines in Organic Electroluminescence Studied by Density Functional Theory. J. Phys. Chem. A 2003, 107, 5241−5251.

ASSOCIATED CONTENT

S Supporting Information *

Experimental details including synthesis, femtosecond and nanosecond transient absorption spectra and data, details of the TENUA fitting procedure, computational details, and a discussion of electronic couplings. This material is available free of charge via the Internet at http://pubs.acs.org. 27712



Article

(9) Cias, P.; Slugovc, C.; Gescheidt, G. Hole Transport in Triphenylamine Based OLED Devices: From Theoretical Modeling to Properties Prediction. J. Phys. Chem. A 2011, 115, 14519−14525. (10) Strehmel, B.; Amthor, S.; Schelter, J.; Lambert, C. Two-Photon Absorption of Bis[4-(N,N-diphenylamino)phenylethynyl]arenes. ChemPhysChem 2005, 6, 893−896. (11) Amthor, S.; Noller, B.; Lambert, C. UV/Vis/NIR Spectral Properties of Triarylamines and Their Corresponding Radical Cations. Chem. Phys. 2005, 316, 141−152. (12) Zhao, H.; Tanjutco, C.; Thayumanavan, S. Design and Synthesis of Stable Triarylamines for Hole-Transport Applications. Tetrahedron Lett. 2001, 42, 4421−4424. (13) Holzapfel, M.; Lambert, C. Photoinduced Charge Separation and Recombination in Acridine−Triarylamine-Based Redox Cascades. J. Phys. Chem. C 2007, 112, 1227−1243. (14) Dapperheld, S.; Steckhan, E.; Brinkhaus, K.-H. G.; Esch, T. Organic Electron Transfer Systems, II Substituted Triarylamine Cation-Radical Redox Systems − Synthesis, Electrochemical and Spectroscopic Properties, Hammet Behavior, and Suitability as Redox Catalysts. Chem. Ber. 1991, 124, 2557−2567. (15) Schmidt, W.; Steckhan, E. Ü ber Organische Elektronenüberträgersysteme, I. Elektrochemische und Spektroskopische Untersuchung Bromsubstituierter Triarylamin-Redoxsysteme. Chem. Ber. 1980, 113, 577−585. (16) Andric, G.; Boas, J. F.; Bond, A. M.; Fallon, G. D.; Ghiggino, K. P.; Hogan, C. F.; Hutchison, J. A.; Lee, M. A. P.; Langford, S. J.; Pilbrow, J. R.; Troup, G. J.; Woodward, C. P. Spectroscopy of Naphthalene Diimides and Their Anion Radicals. Aust. J. Chem. 2004, 57, 1011−1019. (17) Ganesan, P.; Baggerman, J.; Zhang, H.; Sudhölter, E. J. R.; Zuilhof, H. Femtosecond Time-Resolved Photophysics of 1,4,5,8Naphthalene Diimides. J. Phys. Chem. A 2007, 111, 6151−6156. (18) Klein, J. H.; Sunderland, T. L.; Kaufmann, C.; Holzapfel, M.; Schmiedel, A.; Lambert, C. Stepwise Versus Pseudo-Concerted TwoElectron-Transfer in a Triarylamine-Iridium Dipyrrin-Naphthalene Diimide Triad. Phys. Chem. Chem. Phys. 2013, 15, 16024−16030. (19) Miller, S. E.; Lukas, A. S.; Marsh, E.; Bushard, P.; Wasielewski, M. R. Photoinduced Charge Separation Involving an Unusual Double Electron Transfer Mechanism in a Donor−Bridge−Acceptor Molecule. J. Am. Chem. Soc. 2000, 122, 7802−7810. (20) Zhang, Q.; Ning, Z. J.; Tian, H. ’Click’ Synthesis of Starburst Triphenylamine as Potential Emitting Material. Dyes Pigm. 2009, 81, 80−84. (21) Rostovtsev, V. V.; Green, L. G.; Fokin, V. V.; Sharpless, K. B. A Stepwise Huisgen Cycloaddition Process: Copper(I)-Catalyzed Regioselective “Ligation” of Azides and Terminal Alkynes. Angew. Chem., Int. Ed. 2002, 41, 2596−2599. (22) Tornøe, C. W.; Christensen, C.; Meldal, M. Peptidotriazoles on Solid Phase: [1,2,3]-Triazoles by Regiospecific Copper(I)-Catalyzed 1,3-Dipolar Cycloadditions of Terminal Alkynes to Azides. J. Org. Chem. 2002, 67, 3057−3064. (23) Harriman, A.; Elliott, K. J.; Alamiry, M. A. H.; Pleux, L. L.; Séverac, M.; Pellegrin, Y.; Blart, E.; Fosse, C.; Cannizzo, C.; Mayer, C. R.; Odobel, F. Intramolecular Electron Transfer Reactions Observed for Dawson-Type Polyoxometalates Covalently Linked to Porphyrin Residues. J. Phys. Chem. C 2009, 113, 5834−5842. (24) de Miguel, G.; Wielopolski, M.; Schuster, D. I.; Fazio, M. A.; Lee, O. P.; Haley, C. K.; Ortiz, A. L.; Echegoyen, L.; Clark, T.; Guldi, D. M. Triazole Bridges as Versatile Linkers in Electron Donor− Acceptor Conjugates. J. Am. Chem. Soc. 2011, 133, 13036−13054. (25) Natali, M.; Ravaglia, M.; Scandola, F.; Boixel, J.; Pellegrin, Y.; Blart, E.; Odobel, F. Long-Range Charge Separation in a Ferrocene− (Zinc Porphyrin)−Naphthalenediimide Triad. Asymmetric Role of 1,2,3-Triazole Linkers. J. Phys. Chem. C 2013, 117, 19334−19345. (26) Geng, Y.; Wu, S.-X.; Li, H.-B.; Tang, X.-D.; Wu, Y.; Su, Z.-M.; Liao, Y. A Theoretical Discussion on the Relationships Among Molecular Packings, Intermolecular Interactions, and Electron Transport Properties for Naphthalene Tetracarboxylic Diimide Derivatives. J. Mater. Chem. 2011, 21, 15558−15566.

(27) Gawrys, P.; Djurado, D.; Rimarčík, J. n.; Kornet, A.; Boudinet, D.; Verilhac, J.-M.; Lukeš, V. r.; Wielgus, I.; Zagorska, M.; Pron, A. Effect of N-Substituents on Redox, Optical, and Electronic Properties of Naphthalene Bisimides Used for Field-Effect Transistors Fabrication. J. Phys. Chem. B 2010, 114, 1803−1809. (28) Würthner, F.; Ahmed, S.; Thalacker, C.; Debaerdemaeker, T. Core-Substituted Naphthalene Bisimides: New Fluorophors with Tunable Emission Wavelength for FRET Studies. Chem.Eur. J. 2002, 8, 4742−4750. (29) Miyasaka, H.; Masuhara, H.; Mataga, N. Picosecond Ultraviolet Multiphoton Laser Photolysis and Transient Absorption Spectroscopy of Liquid Benzenes. J. Phys. Chem. 1985, 89, 1631−1636. (30) Miyasaka, H.; Masuhara, H.; Mataga, N. Picosecond 266-nm Multiphoton Laser Photolysis Studies on Liquid Alkane Solution of Aromatic Hydrocarbons: Ultrafast Solute Triplet Formation. J. Phys. Chem. 1990, 94, 3577−3582. (31) Barros, T. C.; Brochsztain, S.; Toscano, V. G.; Berci, P.; Politi, M. J. Photophysical Characterization of a 1,4,5,8-Naphthalenediimide Derivative. J. Photochem. Photobiol., A 1997, 111, 97−104. (32) Wachsstock, D. TENUA 2.0, 2005. (33) Weller, A. Photoinduced Electron-Transfer in Solution Exciplex and Radical Ion-Pair Formation Free Enthalpies and their Solvent Dependence. Z. Phys. Chem. 1982, 133, 93−98. (34) ChemBio3D Ultra 12.0.2, CambridgeSoft, 2010. (35) Different conformers may also affect the photoinduced dynamics of all dyads and triads. Because we have no information about these conformers, we treat all kinetic data as average values over the conformational space. (36) Van Stokkum, I. H. M.; Larsen, D. S.; Van Grondelle, R. Global and Target Analysis of Time-Resolved Spectra. Biochim. Biophys. Acta, Bioenerg. 2004, 1657, 82−104. (37) Rogers, J. E.; Kelly, L. A. Nucleic Acid Oxidation Mediated by Naphthalene and Benzophenone Imide and Diimide Derivatives: Consequences for DNA Redox Chemistry. J. Am. Chem. Soc. 1999, 121, 3854−3861. (38) Green, S.; Fox, M. A. Intramolecular Photoinduced Electron Transfer from Nitroxyl Radicals. J. Phys. Chem. 1995, 99, 14752− 14757. (39) Klumpp, T.; Linsenmann, M.; Larson, S. L.; Limoges, B. R.; Bürssner, D.; Krissinel, E. B.; Elliott, C. M.; Steiner, U. E. Spin Chemical Control of Photoinduced Electron-Transfer Processes in Ruthenium(II)-Trisbipyridine-Based Supramolecular Triads. J. Am. Chem. Soc. 1999, 121, 1076−1087. (40) Reszka, K. J.; Takayama, M.; Slk, R. H.; Chignell, C. F.; Salto, I. Photochemistry of Naphthalene Diimides: EPR Study of Free Radical Formation via Photoredox Process. Photochem. Photobiol. 2005, 81, 573−580. (41) Kattnig, D. R.; Mladenova, B.; Grampp, G.; Kaiser, C.; Heckmann, A.; Lambert, C. Electron Paramagnetic Resonance Spectroscopy of Bis(triarylamine) Paracyclophanes as Model Compounds for the Intermolecular Charge-Transfer in Solid State Materials for Optoelectronic Applications. J. Phys. Chem. C 2009, 113, 2983−2995. (42) Steiner, U. E.; Wolff, H. J. In Photochemisty and Photophysics; CRC Press: Boca Raton, FL, 1991; Vol. IV, pp 1−130. (43) Schulten, K.; Wolynes, P. G. Semiclassical Description of Electron Spin Motion in Radicals Including the Effect of Electron Hopping. J. Chem. Phys. 1978, 68, 3292−3297. (44) Knapp, E.-W.; Schulten, K. Magnetic Field Effect on the Hyperfine-Induced Electron Spin Motion in Radicals Undergoing Diamagnetic−Paramagnetic Exchange. J. Chem. Phys. 1979, 71, 1878− 1883. (45) Rawls, M. T.; Kollmannsberger, G.; Elliott, C. M.; Steiner, U. E. Spin Chemical Control of Photoinduced Electron-Transfer Processes in Ruthenium(II)-Trisbipyridine-Based Supramolecular Triads: 2. The Effect of Oxygen, Sulfur, and Selenium as Heteroatom in the Azine Donor. J. Phys. Chem. A 2007, 111, 3485−3496. 27713



Article

(46) Hayashi, H.; Nagakura, S. Theoretical Study of Relaxation Mechanism in Magnetic Field Effects on Chemical Reactions. Bull. Chem. Soc. Jpn. 1984, 57, 322−328. (47) Mori, Y.; Sakaguchi, Y.; Hayashi, H. Spin Effects on Decay Dynamics of Charge-Separated States Generated by Photoinduced Electron Transfer in Zinc Porphyrin−Naphthalenediimide Dyads. J. Phys. Chem. A 2002, 106, 4453−4467. (48) Werner, U.; Sakaguchi, Y.; Hayashi, H.; Nohya, G.; Yoneshima, R.; Nakajima, S.; Osuka, A. Magnetic Field Effects in the Radical Ion Pair Recombination of Fixed-Distance Triads Consisting of Porphyrins and an Electron Acceptor. J. Phys. Chem. 1995, 99, 13930−13937. (49) Colvin, M. T.; Ricks, A. B.; Scott, A. M.; Co, D. T.; Wasielewski, M. R. Intersystem Crossing Involving Strongly Spin ExchangeCoupled Radical Ion Pairs in Donor−bridge−Acceptor Molecules. J. Phys. Chem. A 2012, 116, 1923−1930. (50) Dance, Z. E. X.; Mi, Q.; McCamant, D. W.; Ahrens, M. J.; Ratner, M. A.; Wasielewski, M. R. Time-Resolved EPR Studies of Photogenerated Radical Ion Pairs Separated by p-Phenylene Oligomers and of Triplet States Resulting from Charge Recombination. J. Phys. Chem. B 2006, 110, 25163−25173. (51) Scott, A. M.; Wasielewski, M. R. Temperature Dependence of Spin-Selective Charge Transfer Pathways in Donor−Bridge−Acceptor Molecules with Oligomeric Fluorenone and p-Phenylethynylene Bridges. J. Am. Chem. Soc. 2011, 133, 3005−3013. (52) Colvin, M. T.; Ricks, A. B.; Scott, A. M.; Smeigh, A. L.; Carmieli, R.; Miura, T.; Wasielewski, M. R. Magnetic Field-Induced Switching of the Radical-Pair Intersystem Crossing Mechanism in a Donor− Bridge−Acceptor Molecule for Artificial Photosynthesis. J. Am. Chem. Soc. 2010, 133, 1240−1243. (53) Steiner, U. E.; Ulrich, T. Magnetic Field Effects in Chemical Kinetics and Related Phenomena. Chem. Rev. 1989, 89, 51−147. (54) Weiss, E. A.; Ahrens, M. J.; Sinks, L. E.; Ratner, M. A.; Wasielewski, M. R. Solvent Control of Spin-Dependent Charge Recombination Mechanisms within Donor−Conjugated Bridge− Acceptor Molecules. J. Am. Chem. Soc. 2004, 126, 9510−9511. (55) Scott, A. M.; Miura, T.; Ricks, A. B.; Dance, Z. E. X.; Giacobbe, E. M.; Colvin, M. T.; Wasielewski, M. R. Spin-Selective Charge Transport Pathways through p-Oligophenylene-Linked Donor− Bridge−Acceptor Molecules. J. Am. Chem. Soc. 2009, 131, 17655− 17666. (56) Anderson, P. W. New approach to the theory of superexchange interactions. Phys. Rev. 1959, 115, 2−13. (57) Weiss, E. A.; Ratner, M. A.; Wasielewski, M. R. Direct Measurement of Singlet-Triplet Splitting within Rodlike Photogenerated Radical Ion Pairs Using Magnetic Field Effects: Estimation of the Electronic Coupling for Charge Recombination. J. Phys. Chem. A 2003, 107, 3639−3647. (58) Miura, T.; Scott, A. M.; Wasielewski, M. R. Electron Spin Dynamics as a Controlling Factor for Spin-Selective Charge Recombination in Donor-Bridge-Acceptor Molecules. J. Phys. Chem. C 2010, 114, 20370−20379.

27714


Photoinduced Electron Transfer Dynamics in Triarylamine

Recommend Documents