Article pubs.acs.org/JPCB
Intermolecular and Intramolecular Electron Transfer Processes from Excited Naphthalene Diimide Radical Anions Mamoru Fujitsuka,*,† Sung Sik Kim,†,‡ Chao Lu,† Sachiko Tojo,† and Tetsuro Majima*,† †
The Institute of Scientific and Industrial Research (SANKEN), Osaka University, Mihogaoka 8-1, Ibaraki, Osaka 567-0047, Japan Department of Chemistry, Chonbuk National University, Jeonju 561-756, Korea
‡
S Supporting Information *
ABSTRACT: Excited radical ions are interesting reactive intermediates owing to powerful redox reactivities, which are applicable to various reactions. Although their reactivities have been examined for many years, their dynamics are not well-defined. In this study, we examined intermolecular and intramolecular electron transfer (ET) processes from excited radical anions of naphthalene-1,4,5,8-tetracarboxydiimide (NDI•−*). Intermolecular ET processes between NDI•−* and various electron acceptors were confirmed by transient absorption measurements during laser flash photolysis of NDI•− generated by pulse radiolysis. Although three different imide compounds were employed as acceptors for NDI•−*, the bimolecular ET rate constants were similar in each acceptor, indicating that ET is not the rate-determining step. Intramolecular ET processes were examined by applying femtosecond laser flash photolysis to two series of dyad compounds, where NDI was selectively reduced chemically. The distance dependence of the ET rate constants was described by a β value of 0.3 Å−1, which is similar or slightly smaller than the reported values for donor−acceptor dyads with phenylene spacers. Furthermore, by applying the Marcus theory to the driving force dependence of the ET rate constants, the electronic coupling for the present ET processes was determined.
■
INTRODUCTION Excited radical ions are interesting reactive intermediates owing to powerful redox reactivities, which are applicable to various reactions. The reactivity of excited radical ions can be explained on the basis of the schematic diagram indicated in Scheme 1;
acceptor ability of A is not sufficient. Excitation of intermediates is attractive because it can lead to various useful reactions, as in the case of the “Z-scheme” of natural photosynthesis.1 This concept is currently being applied to various reactions, including water-splitting by photocatalysis.2 The reactivities of excited radical ions have previously been examined by product analysis3−7 and time-resolved spectroscopy.8−15 Although these studies revealed reactivities of the excited radical ions, knowledge of the dynamics of excited radical ions (such as their absorption and lifetimes) is limited. For example, lifetimes of excited radical ions have been determined by indirect method,10−12 and direct determination using ultrafast laser spectroscopy is limited to only a few cases.14,15 Direct observation of excited radical anions was reported by Wasielewski and co-workers.14 Transient absorption spectra of the excited state electrochemically reduced imide compounds were observed. The study confirmed that some of the imide radical anions exhibited excited-state lifetimes in the order of hundreds of picoseconds, which were long enough for various photoinduced reactions. Furthermore, they reported ET from the excited state of naphthalene-1,4,5,8-tetracarboxydiimide radical anions (NDI•−*) to pyromellitimide (PI) and 1,8-
Scheme 1
the diagram shows the electron transfer (ET) process from excited radical anion as a representative case. In this scheme, a radical anion of the donor (D•−) can be generated by photoinduced ET, electron beam radiation, chemical reduction, or electrochemical reduction. Further excitation by photon irradiation generates excited radical anion (D•−*) with a more negative excited state oxidation potential, which is applicable to the reduction of an acceptor (A), even when the electron © XXXX American Chemical Society
Special Issue: John R. Miller and Marshall D. Newton Festschrift Received: October 29, 2014 Revised: November 14, 2014
A
dx.doi.org/10.1021/jp510850z | J. Phys. Chem. B XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry B
Article
Figure 1. Molecular structures of imides and dyads.
naphthalimide (NI).16−18 Their experiments showed that NDI•− was generated by intramolecular charge separation (CS) with the donor after excitation by the first laser. The subsequent excitation to NDI•−* was accomplished by a second laser. In the study, regeneration of the donor excited state upon the second laser excitation was included as a competitive process of ET. In spite of this, characteristics of the ET process, such as the distance and driving force dependences of ET rate, have not been reported for excited radical anions. In this study, intermolecular and intramolecular ET processes from NDI•−* were investigated by time-resolved spectroscopic methods, because imide compounds used in this study have been widely employed in various molecular systems aiming at light energy conversion, biochemical applications, and so on.19−24 Intermolecular ET processes were studied by a combination of pulse radiolysis and laser flash photolysis. PI, NI, and phthalimide (Ph) (Figure 1) were used as electron acceptors for NDI•−*, as they possess reduction potentials more negative than that of NDI. To gain an understanding of ET dynamics from NDI•−*, two series of dyad compounds were synthesized. The first series was dyads of NDI and PI with varied phenylene spacers. The other dyads were NDI and acceptors (PI, perylene-3,4-dicarboximide (PMI), NI, and Ph)
connected by a phenyl ring in the m-position (Figure 1). The former dyads were used to study the distance dependence of intramolecular ET from NDI•−*, while the latter dyads were used to study the driving force dependence of the ET rate constants. The observed ET dynamics were analyzed on the basis of ET theory.
■
EXPERIMENTAL SECTION Materials. Imide compounds (NDI, PI, NI, and Ph) were synthesized by the reactions between corresponding carboxylic anhydrides and 2-ethyl-1-hexylamine. Dyads were synthesized as summarized in the Supporting Information. Formations of the compounds were confirmed by 1H NMR and mass spectroscopy. In the present study, N,N-dimethylformamide (DMF) was used as a solvent of spectroscopies. Tetrakis(dimethylamino)ethylene (TDAE) was purchased from Tokyo Chemical Industry. Apparatus. Pulse radiolysis−laser flash photolysis was carried out using a linear accelerator at Osaka University (28 MeV, 8 ns fwhm, 0.3 kGy per pulse) synchronized with a nanosecond Nd:YAG laser (Quantel, Brilliant, 532 nm, 5 ns fwhm, ∼80 mJ per pulse). The laser pulse was irradiated to the B
dx.doi.org/10.1021/jp510850z | J. Phys. Chem. B XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry B
Article
sample at 3.0 μs after the electron pulse. The probe light from a 450 W Xe lamp (Ushio, UXL-451-0) was detected with a multichannel spectrometer (UNISOKU, TSP-1000) or a photomultiplier (Hamamatsu Photonics, R2949) connected to a digital oscilloscope (Tektronix, TDS580D).25 Transient absorption spectra during the femtosecond laser flash photolysis were measured as described in the previous paper.26 In the present study, the sample was excited with 475 or 605 nm femtosecond laser pulse (∼130 fs fwhm, ∼5 μJ per pulse). Steady state absorption spectra were measured using Shimadzu UV-3100PC. Theoretical Calculation. Optimized structures of imides and dyads were estimated by density functional theory (DFT) at the (U)B3LYP/6-31G(d) level using the Gaussian 09 package.27 For the simplicity of the calculations, alkyl groups of the compounds were reduced to methyl groups. Absence of imaginary frequency was confirmed for the optimized structures.
■
Figure 2. (a) Transient absorption spectra of NDI (1.3 mM) in DMF in the presence of PI (100 mM) at 2.5 and 3.1 μs after electron pulse irradiation during the pulse radiolysis−laser flash photolysis. The 532 nm laser was irradiated to the sample at 3.0 μs after the pulse irradiation. (b) Kinetic traces of ΔO.D. at 470 and 710 nm.
RESULTS AND DISCUSSION Intermolecular ET from the NDI•−*. First, the electron donor ability of the excited radical anion to electron acceptors in solution was examined by the pulse radiolysis−laser flash photolysis method. It is well established that pulse radiolysis of substrate (S) in DMF effectively generates a radical anion of S (S•−) according to the following reaction schemes (eqs 1−3) DMF ⇝ DMF•+ + e−
(1)
S + e− → S•−
(2)
DMF•+ + DMF → DMF( −H+)• + DMF( +H+)+
(3)
concomitant with the fast rise and decay in the kinetic trace of ΔO.D. at 710 nm. The transient absorption spectrum at 3.1 μs after the electron pulse (0.1 μs after the laser pulse) exhibited a clear peak at 710 nm, indicating the generation of PI•−. Thus, the bleach (470 nm) and rise (710 nm) can be attributed to the generation of PI•− by ET from NDI•−* (eq 5) kInterET
NDI•−* + PI ⎯⎯⎯⎯⎯→ NDI + PI•−
In addition, recovery (470 nm) and decay (710 nm) can be attributed to regeneration of NDI•− and PI according to eq 4. This is supported by the fact that the recovery rate (2.0 × 106 s−1) is equivalent to that observed with the initial rise. The generation of PI•− by the ET from NDI•−* (eq 5) was also supported by a sufficiently negative free energy change (ΔGET) of −1.38 eV, which was calculated using eq 6, assuming that the ET occurred from the D1 state
Laser pulse irradiation, so that the wavelength is absorbed by S•− within the lifetime, generates the excited state of S•− (S•−*). Thus, by using the pulse radiolysis−laser flash photolysis method, the transient phenomena of S•−* can be followed. Figure 2 shows the transient absorption spectra and kinetic traces during the pulse radiolysis−laser flash photolysis of NDI (1.3 mM) and PI (100 mM) in DMF. Due to the higher concentration of PI than NDI, PI•− was initially generated according to eq 2. This was evident from a kinetic trace at 710 nm, at which PI•− exhibits an absorption peak. The PI•− showed decay over the initial ∼2 μs, which was concomitant with a rise at 470 nm, indicating the generation of NDI•− because of ET from PI•− (eq 4) PI•− + NDI → PI + NDI•−
(5)
ΔG ET = E(NDI•−/NDI) − E(PI•−/PI) − E D1(NDI•−*) (6)
where ED1(NDI•−*) is the D1 state energy of NDI•−*, and it is indicated that the absorption band of NDI•− at 769 nm is due to the D1 ← D0 transition;14 thus, ED1(NDI•−*) = 1.61 eV can be applied. Similar pulse radiolysis−laser flash photolysis experiments were performed by varying the concentration of PI (Figure 3). When the sample solution did not contain PI, generation of NDI•− was attributable to eq 2 (not eq 4), as evident in the initial rise that was faster than hundreds of nanoseconds. In this case, laser irradiation at 3.0 μs did not promote any change in the kinetic trace of ΔO.D. at 470 nm because the D1 state lifetime of NDI•− is shorter than the resolution time of the instrument (discussed in a later section). In the presence of PI, laser irradiation at 3.0 μs caused a bleach and rise in the kinetic traces of ΔO.D. at 470 and 710 nm, respectively, as indicated in Figure 2b. It should be noted that the extent of the bleach (470 nm) and rise (710 nm) became larger with an increase in the concentration of PI (Figure 3a). For the bimolecular ET from the excited state, the ratio |ΔΔO.D.|/ΔO.D.0 (≡Y), where ΔO.D.0 is the ΔO.D. value just before the laser excitation and
(4)
The apparent rise rate of the kinetic trace of ΔO.D. at 470 nm was 2.0 × 106 s−1, indicating that the bimolecular rate constant is in the order of 109 M−1 s−1, which is close to the diffusioncontrolled rate. The efficient ET can be attributed to the reduction potential of NDI (E(NDI•−/NDI) = −0.48 V vs SCE) being less negative than that of PI (E(PI•−/PI) = −0.71 V),14 providing a negative free energy change for eq 4 (−0.23 eV). The transient absorption spectrum at 2.5 μs after the pulse irradiation (Figure 2a) exhibited the spectral shape of NDI•− without the contribution of PI•−. At 3.0 μs after the electron pulse, the sample was irradiated with a 532 nm nanosecond laser pulse. Upon irradiation of the laser pulse, which is absorbed by NDI•− generating NDI•−*, the kinetic trace of ΔO.D. at 470 nm showed the bleach and recovery, which were C
dx.doi.org/10.1021/jp510850z | J. Phys. Chem. B XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry B
Article
× 1011 M−1 s−1. Although Ph•− does not exhibit clear absorption bands in the studied spectral range, ET to Ph was confirmed by an increase in bleaching of the NDI•− absorption band upon laser excitation with an increase in concentration of Ph. Reduction of Ph by NDI•−* is reasonable because the ΔGET value for ET between Ph and NDI•−* (−0.69 eV) is almost the same as that between NI and NDI•−* (−0.72 eV). From the relation between [Ph] and the bleaching, the kInterET value was estimated to be 6.3 × 1011 M−1 s−1. Therefore, it was revealed that PI, NI, and Ph act as an electron acceptor for NDI•−*, in spite of smaller −ΔGET values for NI and Ph. The almost identical kInterET values indicate that ET is not the ratedetermining step in spite of variation in the ΔGET values. For discussion on the distance and driving force dependences of the ET process, intramolecular ET processes have to be examined as in the following section. Intramolecular ET from the NDI•−*. In this study, two series of dyads were investigated. In the first series, NDI and PI were connected by three kinds of phenylene spacers, namely, m-phenyl (NDI-m-PI), p-phenyl (NDI-p-PI), and biphenyl (NDI-b-PI) (Figure 1). For NDI-p-PI and NDI-b-PI, methyl groups were introduced to the phenylene spacers from a synthetic reason. Without methyl groups, the yield of the coupling between imide-substituted phenylamine and carboxylic anhydride became quite low, probably due to lower electron density at the p-amine group after the substitution. In the case of m-phenylenediamine, coupling products were obtained without methyl groups. To study the intramolecular ET from NDI•−*, femtosecond laser flash photolysis was applied to the dyads that were chemically reduced by TDAE.33 Figure 4 shows absorption
Figure 3. (a) Kinetic traces of ΔO.D. at 470 nm during the pulse radiolysis−laser flash photolysis of NDI (1.3 mM) in DMF in the presence of PI (0−100 mM). The 532 nm laser was irradiated to the sample at 3.0 μs after the pulse irradiation. (b) Plot of Y470−1 against [PI]−1 (see eq 7 in the text).
ΔΔO.D. is the change in ΔO.D. caused by the laser excitation, represents the yield of ET. By using the bimolecular ET rate constant (kInterET), deactivation rate of excited state (kd), and concentration of acceptor ([A]), Y can be expressed as eq 710−12 Y = I0(kInterET[A]/(kd + kInterET[A]))
(7) −1
−1
where I0 is a constant. Thus, the plot of Y against [A] is expected to show a linear relation, of which the intercept and slope provide the kInterET value. In Figure 3b, Y−1 at 470 nm (Y470−1) was plotted against [PI]−1. From the slope, intercept, and deactivation rate (evaluated in a later section), the kInterET value was estimated to be 5.6 × 1011 M−1 s−1, which is larger than the diffusion controlled rate in DMF (7.0 × 109 M−1 s−1).28 The larger kInterET value than the diffusion controlled rate can be explained on the basis of a static mechanism.15,29−31 This mechanism contributes significantly when the concentration of the quencher becomes larger, as present in this case. The estimated kInterET value is similar to the rate constants of the intermolecular processes of the short-lived precursor excited state such as the excited radical cation and higher triplet state.15,31 Furthermore, the quantum yield for the present ET reaction ([PI] = 100 mM) was estimated to be 0.84 by using Zn tetraphenylporphyrin in cyclohexane as a standard of the actinometry.32 The efficient ET reaction can be attributed to the relatively long lifetime of the excited state and the larger free energy change for ET as indicated above. Intermolecular ET processes were also examined by using NI and Ph as acceptors for NDI•−*. In the case of ET to NI, the formation of NI•−, which exhibits an absorption peak around 420 nm, was confirmed upon laser excitation during pulse radiolysis (Figure S1 in the Supporting Information). The kInterET value for ET from NDI•−* to NI was estimated to be 6.3
Figure 4. Absorption spectra of NDI-p-PI (30 μM) with varied concentration of TDAE (0−35 μM) in DMF. Inset: Absorbance at 474 nm with varied concentrations of TDAE.
spectra of NDI-p-PI in DMF with varied concentrations of TDAE. With an increase in the concentration of TDAE, absorption bands attributable to NDI•− appeared. As shown in the inset of Figure 4, absorbance due to NDI•− reached a plateau when an equivalent amount of TDAE was added. The ratio of maximal absorbance of NDI•− to that of neutral NDI was almost close to a value previously reported,14 indicating a quantitative reduction of NDI by TDAE. It should be noted that the absorption bands due to PI•− were not observed in Figure 4, indicating that the NDI in the dyad was selectively reduced by TDAE. Notably, the absorption bands due to NDI•− kept their intensity for hours when molecular oxygen D
dx.doi.org/10.1021/jp510850z | J. Phys. Chem. B XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry B
Article
taking the kd value (=1/(130 ps)) into account. The quantum yield of the intramolecular ET (ΦIntraET) was 0.94, indicating efficient ET. Similar intramolecular ET processes were observed with NDI-m-PI and NDI-b-PI (Figure 5b). The absence of the reduced spacer was also confirmed, indicating single step ET. The kIntraET value of NDI-m-PI was faster than that of NDI-pPI, while a slower kIntraET value was obtained with NDI-b-PI, as shown in Table 1. The quantum yields of the ET in NDI-m-PI
was removed from DMF by Ar bubbling. This indicated that NDI•− was stable under these conditions. The excited state properties of monomeric NDI•− were estimated by measuring the transient absorption spectrum of NDI•−, which was generated by the chemical reduction of NDI by TDAE, during laser flash photolysis using a 475 nm femtosecond laser. Upon excitation, a transient absorption spectrum with maxima at 570, 649, and 721 nm and minima at 760 nm was observed (Figure S2, Supporting Information). The observed maxima and minima corresponded with the minima and maxima of the ground state absorption spectrum of NDI•−, respectively. The observed absorption band showed decay, of which profile was analyzed by a single exponential function with a lifetime of 130 ps. This resembled the reported value for alkyl-substituted NDI•−* by Wasielewski et al.14 Figure 5a shows the transient absorption spectra of NDI-p-PI in DMF in the presence of TDAE during laser flash photolysis
Table 1. Rate Constant and Quantum Yield of Forward ET (kIntraET and ΦIntraET, Respectively), BET Rate Constant (kIntraBET), and Distance between NDI and PI (rcc and rb) in the Dyads kIntraETa (s−1) NDI-m-PI NDI-p-PI NDI-b-PI
3.1 × 10 1.4 × 1011 3.9 × 1010 11
ΦIntraET
kIntraBETa (s−1)
rccb (Å)
rbc (Å)
0.97 0.94 0.82
8.7 × 10
