DOI: 10.1002/chem.201600077
Full Paper
& Electrochemiluminescence
Photophysical Properties and Efficient, Stable, Electrogenerated Chemiluminescence of Donor–Acceptor Molecules Exhibiting Thermal Spin Upconversion Ryoichi Ishimatsu,*[a] Tomohiko Edura,[b] Chihaya Adachi,*[a, b, c] Koji Nakano,[a] and Toshihiko Imato*[a] Abstract: The photophysical properties and electrogenerated chemiluminescence (ECL) of three donor–acceptor molecules composed of dicyanobenzene and methyl-, tert-butyl-, and phenyl-substituted carbazolyl groups, 1,2,3,5-tetrakis(3,6-disubstituted-carbazol-9-yl)-4,6-dicyanobenzene (4CzIPN-Me, 4CzIPN-tBu, and 4CzIPN-Ph, respectively) are described. These molecules show delayed fluorescence as a result of thermal spin upconversion from the lowest triplet state to the lowest singlet state at room temperature. The three molecules showed yellow to yellowish–red ECL. Re-
Introduction
markably, the ECL efficiencies of 4CzIPN-tBu in dichloromethane reached almost 40 %. Moreover, stable ECL was emitted from 4CzIPN-tBu and 4CzIPN-Ph. In case of 4CzIPN-Me, the ECL intensity decreased during voltage cycles because of polymerization. Quantum chemical calculations revealed that polymerization was inhibited by the steric hindrance of the bulky tert-butyl and phenyl groups on the carbazolyl moieties and lowered the spin density on the carbazolyl groups through electron conjugation for 4CzIPN-Ph.
RC ¢ þ RC þ ! n1 R* þ ð1¢nÞ3 R* þ R ðexcited-state formation by ion annihilationÞ
Electrogenerated chemiluminescence (ECL) can be emitted from molecules at excited states that are formed by electron transfer between a radical anion and cation produced through an electrode reaction as shown in Equations (1)–(4):[1–3] R þ e¢ ! R C ¢ ðradical anion formationÞ R¢e¢ ! RC þ ðradical cation formationÞ [a] Dr. R. Ishimatsu, Prof. C. Adachi, Prof. K. Nakano, Prof. T. Imato Department of Applied Chemistry, Graduate School of Engineering Kyushu University, 744 Motooka, Nishi-ku Fukuoka, 819-0395 (Japan) E-mail: [email protected] [email protected] [email protected] [b] Dr. T. Edura, Prof. C. Adachi Center for Organic Photonics and Electronics Research Kyushu University 744 Motooka, Nishi-ku Fukuoka, 819-0395 (Japan) [c] Prof. C. Adachi JST, ERATO, Adachi Molecular Exciton Engineering Project c/o Center for Organic Photonics and Electronics Research, and International Institute for Carbon Neutral Energy Research (WPI-I2CNER), Kyushu University 744 Motooka, Nishi-ku, Fukuoka, 819-0395 (Japan) Supporting information for this article can be found under http:// dx.doi.org/10.1002/chem.201600077. Chem. Eur. J. 2016, 22, 4889 – 4898
ð1Þ
ð2Þ
1
R* ! R þ hn
ð3Þ
ð4Þ
ðECL emissionÞ
in which R represents a fluorescent molecule, and 1R* and 3R* are the lowest excited singlet (S1) and triplet (T1) states of R, respectively. Usually, the S1 and T1 states of most fluorescent molecules relax through radiative and nonradiative transitions, respectively. According to spin statistics,[4, 5] that is, 25 % S1 and 75 % T1 states will be produced by charge recombination processes, such as hole–electron recombination and ion annihilation, n in Equation (3) is considered to be 0.25 for an energysufficient ECL system, as illustrated in Figure 1. To populate S1 and T1 with the ratio following spin statistics, the energy level of radical species should be large enough and the energy splitting between S1 and T1 states should be small. When the energy of a pair of radical species is insufficient to produce the S1 state, only the T1 state will be formed. The ground state may directly be formed without forming excited states for an energy-deficient system as shown in Equation (5):[1, 6] R C ¢ þ RC þ ! 2 R ðground state formationÞ
ð5Þ
For an energy-sufficient ECL system, the pathway to produce the ground state directly [Eq. (5)] occurs in the Marcus inverted region, and thus, is an unlikely process.[6, 7] 4889
Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper
Figure 1. Schematic representation of electronic transitions of thermally activated delayed fluorescence (TADF) molecules. Excited states are formed by light absorption (left) and ion annihilation in an ECL system (right). Reverse intersystem crossing (RISC) increases the yield of “delayed” luminescence. ISC: intersystem crossing, FL: fluorescence.
Because the kinetics of electronic transitions from excited states, including light emission, should be similar for ECL and photoluminescence (PL) processes, “pure” spin statistics predicts that the maximum ECL efficiency (fECL) of fluorescent molecules is the value of their quantum efficiency of PL (fPL) multiplied by a factor of 0.25 (i.e., 0.25 Õ fPL) without spin upconversion processes, such as triplet–triplet (T–T) annihilation.[1, 8] Generally, fluorescent molecules that exhibit high fPL and produce stable radical ions through electrode reactions are favorable to obtain efficient, stable ECL. This is because more stable radical ions can eliminate additional parasitic processes, such as polymerization and reactions with solvents, which result in a larger generation of excited states than less stable ones. Based on the guidelines of high fPL and stable radical ions, recently, some donor–acceptor (D–A) molecules have been used as ECL systems to attain high fECL.[9–18] The ECL mechanism of D–A molecules has also been studied. Electrondonating and -accepting groups in D–A molecules can be oxidized and reduced, respectively, through individual electrode reactions, and then the D–A molecules can form excited states through ion annihilation. In addition, by changing the D and A groups in D–A molecules, it is possible to adjust the energy levels of molecular orbitals, such as the HOMO and LUMO. This, in turn, allows the absorption, PL, and ECL spectra of D–A molecules to be tuned.[11, 14] D–A molecules with multiple D and A groups that show multistep electrode reactions have been used to study ECL behavior, including electron-transfer mechanisms between multivalent ions, such as divalent ones.[10, 12, 18] An interesting feature of D–A molecules is that the energy gap between S1 and T1 states (DEst) can be decreased by using appropriate combinations of D and A groups[9] to achieve thermal spin upconversion from T1 to S1 states (RISC), as shown by Equation (6) 3
Results and Discussion Photophysical properties
R* þ D ! 1 R*
ð6Þ
ðthermal spin upconversionÞ
Quantum chemical calculations have enabled researchers to design D–A molecules with small DEst, the value of which markedly influences the kinetics of thermal spin upconversion. Chem. Eur. J. 2016, 22, 4889 – 4898
Small DEst values can be achieved by overlapping HOMO and LUMO wave functions effectively, whereby the HOMO and LUMO wave functions of the designed D–A molecules are spatially separated and mainly distributed on D and A groups, respectively.[19, 20] Because the rate constant of thermal spin upconversion, depending on DEst, is much lower than that of the radiative transition, upon light absorption, both prompt FL and TADF[21] can be observed. Based on results of quantum chemical calculations, several D–A molecules showing delayed FL as a result of thermal spin upconversion have been synthesized and used in organic light-emitting diodes (OLEDs).[22–34] Owing to small DEst values of the synthesized TADF molecules, the ratio of S1 and T1 states generated through charge recombination can approach the value expected by spin statistics. OLEDs containing TADF molecules are able to convert electrical power into light emission very efficiently; the conversion efficiencies of current into luminescence are similar to those of fPL of TADF molecules, which is a breakthrough in the theoretical limitation imposed by spin statistics. Recently, we reported the photophysical,[35] electrochemical, and ECL properties[16] of some TADF molecules composed of dicyanobenzene (A) and carbazolyl (D) groups, which were synthesized for use in OLEDs, and showed that TADF molecules were very useful to obtain fECL values higher than 0.25 Õ fPL because of thermal spin upconversion, which led to “delayed ECL” described in the same manner as delayed PL. However, the ECL intensity of these systems decayed quickly relative to stable ECL systems because their unstable radical cations induced polymerization. Therefore, TADF molecules that form stable radical ions are needed to realize an efficient, stable ECL system. Establishing an efficient and stable ECL system will be beneficial for light-emitting liquid displays.[36–38] Furthermore, knowledge of TADF molecules that exhibit stable ECL can be extended to OLED systems to understand their degradation processes and establish stable, efficient OLEDs by using TADF molecules. Although the stability of radical ions is crucial for ECL, few systematic studies focused on enhancing the stability of radical ions for ECL have been carried out.[15, 39, 40] Herein, we report the synthesis, electrochemistry, and photophysical and ECL properties of three D–A molecules composed of dicyanobenzene and methyl-, tert-butyl-, or phenyl-substituted carbazolyl moieties (Figure 2) that exhibit TADF. We systematically examine the effects of the substituents on the electrochemical and photophysical properties of the materials. Through both TADF and the formation of stable radical ions, an efficient, stable ECL system is realized.
www.chemeurj.org
To elucidate the photophysical properties of the synthesized molecules in solution, UV/Vis absorption and PL spectra, as well as PL decay curves, were measured in four solvents with different polarity: toluene, chloroform, dichloromethane, and acetonitrile. Figure 3 shows the UV/Vis absorption and PL spectra of the molecules in the four different solvents. Relatively
4890
Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper similar. The greater redshift of labs of 4CzIPN-Ph than those of 4CzIPN and 4CzIPN-tBu indicates that the energy gaps between molecular orbitals related to transitions induced by light absorption are narrowed by the presence of phenyl substituents, which is probably caused by electron conjugation. To confirm that conjugation affected the molecular orbitals of the materials, DFT calculations were carried out. The estimated HOMO and LUMO and energy levels of the compounds are illustrated in Figure 4. The smaller HOMO–LUMO gap for
Figure 2. Molecular structures of TADF molecules composed of substituted carbazolyl and dicyanobenzene moieties considered in this study.
Figure 4. HOMO and LUMO distributions and energy levels estimated at the B3LYP/6-31G(d) level for the optimized structures of the three TADF molecules.
Figure 3. UV/Vis absorption (dashed lines) and normalized PL (solid lines) spectra of the TADF materials in four different solvents. Concentration for PL spectra: 15 mm.
large Stokes shifts were observed for all three molecules. The Stokes shifts caused small overlaps between the absorption and PL spectra at l 500 nm, and the overlap diminished as the solvent polarity increased. The UV/Vis absorption maxima (labs) and PL maxima (lPL) of the three molecules, together with those of unsubstituted 4CzIPN, are summarized in Table 1. The substituents on carbazolyl moieties shifted labs to longer wavelengths by about 10–20 nm compared with that of unsubstituted 4CzIPN. The redshift caused by phenyl groups is greater than that induced by methyl and tert-butyl groups in all of the solvents. The effect of the methyl and tert-butyl substituents on the UV/Vis absorption spectra of the material is Chem. Eur. J. 2016, 22, 4889 – 4898
www.chemeurj.org
4CzIPN-Ph by about 0.2 eV compared with those for 4CzIPNMe and 4CzIPN-tBu is consistent with the trend observed in the UV/Vis absorption spectra. The HOMOs and LUMOs of each molecule are spatially separated; HOMOs are mainly distributed over the carbazolyl (D) moieties, and LUMOs are localized over the dicyanobenzene (A) moiety. This spatial separation means that HOMO–LUMO transitions of the molecules will have charge-transfer characteristics. It should be noted that the HOMO of 4CzIPN-Ph is partially distributed over the phenyl groups of carbazolyl moieties because of conjugation, whereas the HOMOs of 4CzIPN-Me and 4CzIPN-tBu are hardly distributed over the methyl and tert-butyl substituents. With increasing solvent polarity, labs at longer wavelength, such as the bands at l 400 nm and shoulders at l 470 nm, exhibited slight blueshifts, whereas labs at shorter wavelength related to transitions from inner molecular orbitals were almost independent of solvent polarity. TD-DFT calculations were performed to identify the main transitions between molecular orbitals affecting labs. The results of TD-DFT calculations for 10 low-lying singlet excited states (S1–S10) upon light absorption under vacuum, and also considering the solvent effect by using the polarized continuum model (PCM), are summarized in Tables S1 and S2 in the Supporting Information. For excitation to S1–S10, transitions from HOMO-m (m = 0, 1, 2, 3, 4, and 6) to LUMO or LUMO + 1 were mainly dominant. The largest oscillator strength (f) in the solvents was found for the transition to S3 or S4 for 4CzIPN-Me (f = 0.40–0.46), S4 for 4CzIPN-tBu (f = 0.40–0.49), and S3 or S4 for 4CzIPN-Ph (f = 0.55– 0.62), and these energy levels of S3 and S4 were very similar with differences of less than 0.02 eV. Therefore, labs 400 nm is mainly caused by the transitions to these states. From f values
4891
Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper Table 1. Photophysical properties of the TADF molecules in different solvents. 4CzIPN-X
Solvent
labs [nm]
lPL [nm]
FWHM [nm]
kp¢1 [ns]
kd¢1 [ms]
kr (Õ 108 s¢1)
fPL [%][a]
fp [%]
fd [%]
X = Me (0.307 eV)[b]
toluene chloroform CH2Cl2 MeCN toluene chloroform CH2Cl2 MeCN toluene chloroform CH2Cl2 MeCN toluene CH2Cl2 MeCN
325, 338, 357sh, 375, 395, 467sh 325, 338, 394, 478sh 324, 338, 391, 472sh 324sh, 339, 379, 456sh 321, 334, 391, 468sh 321, 334, 396, 478sh 321, 334, 393, 470sh 320sh, 335, 380, 459sh 330, 345, 404, 486sh 330, 346, 404, 491sh 331, 345, 404, 487sh 331, 348, 390, 467sh 312, 326, 375, 452sh 312, 326, 378, 455sh 312, 325, 365, 438sh
525 568 577 603 521 554 565 588 547 575 588 611 507 536 551
88 116 128 145 96 107 116 146 101 115 128 147 83 107 125
13 20 15 6 15 23 20 10 22 19 13 4 14 25 19
2.8 2.1 1.7 0.99 2.9 2.6 2.4 1.4 1.7 1.6 1.5 0.79 1.8 2.0 1.4
18 11 8.7 5.3 18 7.8 13 7.1 16 14 10 8.5 18 9.2 5.3
76 38 18 3.7 86 56 44 8.5 72 38 17 3.7 94 54 18
23 22 13 3.2 27 18 25 7.1 36 27 13 3.4 25 23 10
53 16 5.0 0.5 59 38 19 1.4 36 11 4.0 0.3 69 31 8.0
tBu (0.308 eV)[b]
Ph (0.214 eV)[b]
H[c] (0.345 eV)[b]
[a] Estimated by using an integrating sphere. sh: shoulder. FWHM: full width at half maximum of PL spectra. [b] DEST values were estimated by time-dependent (TD) DFT at the M06-2X/6-31 + G(d) level. [c] Unsubstituted 4CzIPN. Values are from ref. [35].
(Table S2 in the Supporting Information), other characteristic labs, such as those at l 460 (shoulder) and 340 nm, are derived from transitions to S1 and S8, respectively. These calculations also confirmed that all molecular orbitals involved in the transitions to S1–S10 had charge-transfer characteristics, that is, HOMO-m, and LUMO and LUMO + 1 were localized on carbazolyl and dicyanobenzene groups, respectively. Intramolecular charge transfer from carbazolyl to dicyanobenzene groups at these excited states was induced through light absorption. PL spectra of the molecules exhibited lPL 520–610 nm. Shifts of PL spectra caused by the substituents and solvents were clearly seen; lPL was shifted to longer wavelength by the substituents in the order of tert-butyl < methyl < phenyl; this reflected the calculated HOMO–LUMO gaps. The lPL values of these molecules showed relatively large redshifts of about 60– 80 nm when the solvent polarity increased upon changing from toluene to MeCN. Such large redshifts of lPL induced by increasing solvent polarity are typical behavior of D–A molecules. This is because, upon light adsorption, D–A molecules form an intramolecular charge-transfer excited state with a larger dipole moment that can be stabilized by surrounding polar solvent molecules, resulting in emission at longer wavelengths.[11, 41–43] To estimate the PL lifetimes of the molecules, PL decay curves were measured in four different solvents (Figure 5). Importantly, both “prompt” and “delayed” PL components were observed for the three molecules at room temperature in the four solvents. This confirms that spin upconversion from T1 to S1 states occurs in the solutions at room temperature. It is notable that the lifetimes of both prompt and delayed components depend on the substituents and both shorten with increasing solvent polarity. Experimental decay curves of PL intensity (I(t)) can be expressed by using the biexponential function given in Equation (7):[35, 44] IðtÞ ¼ Ip expð¢kp tÞ þ Id expð¢kd tÞ þ C Chem. Eur. J. 2016, 22, 4889 – 4898
www.chemeurj.org
ð7Þ
Figure 5. PL decay curves for the three TADF molecules in four different solvents at room temperature. Insets: Decay curves for the prompt components. Concentration: 15 mm.
4892
Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper in which Ip and Id represent the PL intensities of the prompt and delayed components at t = 0, respectively; C is a constant; and kp and kd are the decay constants for prompt and delayed components, respectively. We estimated Ip, Id, kp, and kd values by fitting the experimental curves with Equation (7) and calculated the contributions of the quantum efficiencies of the prompt (fp) and delayed (fd) components to fPL. The calculated values and values of fPL are listed in Table 1. The lifetimes of the prompt (kp¢1) and delayed (kd¢1) components related to the kinetics of radiative and nonradiative transitions, including RISC [see Eqs. (S1) and (S2) in the Supporting Information] were in the order of 10¢8 and 10¢6 s, respectively, and lifetimes of both prompt and delayed components were at least halved upon increasing the solvent polarity upon switching from toluene to MeCN. The rate constants of radiative transition (kr) calculated from kp and fp by using Equation (S6) in the Supporting Information decreased with increasing solvent polarity (Table 1). In less polar solvents, such as toluene, the TADF molecules exhibit high fPL (76, 86, and 72 % for 4CzIPN-Me, 4CzIPN-tBu, and 4CzIPN-Ph in toluene, respectively) with a large contribution from delayed PL (fd = 53, 59, and 36 %, for 4CzIPN-Me, 4CzIPN-tBu, and 4CzIPN-Ph in toluene, respectively), although these values are somewhat smaller than those of 4CzIPN (fPL = 94 %, and fd = 69 % in toluene).[35] Both fp and fd decreased dramatically with increasing solvent polarity, resulting in lower fPL. This means that the rate constants of thermal deactivation increase in polar solvents. Because DEst is an important factor in delayed FL, TD-DFT calculations were performed for S1 and T1 states of the molecules to estimate DEst, and the obtained values are listed in Table 1. The DEst values of 4CzIPN-Me (0.307 eV) and 4CzIPNtBu (0.308 eV) are almost identical and slightly smaller than that of unsubstituted 4CzIPN (0.345 eV), whereas the DEst value of 4CzIPN-Ph (0.214 eV) is much smaller than those of 4CzIPNMe and 4CzIPN-tBu. The markedly smaller DEst for 4CzIPN-Ph may be explained by the spatial distribution and more effective overlap of its HOMO and LUMO induced by extended conjugation compared with that in 4CzIPN-Me and 4CzIPN-tBu, as calculated for some other TADF molecules.[32] Thus, introducing substituents of this kind, especially phenyl groups, is favorable to decrease DEst (the DFT calculations with M06-2X/6-31 + G(d) tend to overestimate DEst by about 0.1–0.2 eV[24]). It should be mentioned that DEst decreases with increasing solvent polarity, which may be described as a function of fp, fd, kp, kd, and kISC, in which kISC is the rate constant of intersystem crossing from S1 to T1 states[35] [Eqs. (S3) and (S4) in the Supporting Information]. The smaller DEst value, that is, the higher rate constant of upconversion and shorter lifetime of the T1 state for the three molecules compared with those of unsubstituted 4CzIPN, could be useful to reduce interactions with quenchers. Indeed, introducing such substituents is not deemed to be suitable to achieve high fPL, but they are able to stabilize the radical cations, leading to stable ECL (as discussed below).
Chem. Eur. J. 2016, 22, 4889 – 4898
www.chemeurj.org
Electrochemical properties of TADF molecules The redox behavior of the TADF molecules was observed with cyclic voltammetry. Figure 6 shows cyclic voltammograms of 4CzIPN-tBu and 4CzIPN-Ph in MeCN, and that of 4CzIPN-Me in a mixture of CH2Cl2 and MeCN (50:50 vol %). Because 4CzIPNMe displayed poor solubility in MeCN, a mixture of CH2Cl2 and MeCN was used as a solvent.
Figure 6. Cyclic voltammograms obtained for 1 mm solutions of TADF molecules containing 0.1 m tetrabutylammonium hexafluorophosphate (TBAPF6). Solvent: MeCN for 4CzIPN-tBu and 4CzIPN-Ph, and a mixture of CH2Cl2 and MeCN (50:50 vol %) for 4CzIPN-Me. Scan rate: 100 mV s¢1. The arrows indicate the scan direction.
In the cathodic scans, reduction waves of dicyanobenzene (A) groups were observed for the three molecules, and peak separations of the reduction waves were about 70 mV at a scan rate of 100 mV s¢1. In the anodic scans, the TADF molecules showed oxidation waves of carbazolyl (D) groups, and the peak separations for 4CzIPN-tBu and 4CzIPN-Ph were about 65 mV at 100 mV s¢1, which were close to 59 mV for the reversible one-electron reaction. It should be emphasized that rate constants of the electrode reactions for both oxidation and reduction seem to be relatively large because these peak separations are close to that of the reversible one, and it is difficult to estimate the values accurately because of uncompensated for resistance (see the Experimental Section). For 4CzIPNMe, an irreversible oxidation wave was observed; this means that the electrode reaction (E) was followed by a chemical reaction (C; EC mechanism). The irreversibility is because the high reactivity of the radical cation of 4CzIPN-Me generated through the electrode reaction leads to polymerization. The polymerization of molecules containing unsubstituted carbazolyl groups[11, 16, 45] has been reported, and is mainly attributable to the high spin density on the C(3), C(6), and N(9) atoms of the radical cation of carbazole.[45] Overall, the radical anions and cations of 4CzIPN-tBu and 4CzIPN-Ph are stable, and the radical anion and cation of 4CzIPN-Me are stable and unstable, respectively, because of polymerization. Both reduction and oxidation waves were fitted with simulated curves of the one-electron reaction by using simulation software based on the Butler–Volmer model[46] to estimate standard redox potentials, E0(R/RC¢) and E0(R/RC + ), related to HOMO and LUMO levels, respectively. The estimated values are summarized in Table 2. The value of E0(R/RC¢) for 4CzIPN-Ph
4893
Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper Table 2. Redox potentials of the TADF molecules. 4CzIPN-X
E0(R/RC¢) [V][a]
E0(R/RC + ) [V][a]
EL-H [eV][b]
X = Me tBu Ph H[c]
¢1.23 ¢1.21 ¢1.15 ¢1.21
N.D. 1.42 1.41 N.D.
2.99 3.00 2.82 2.91
[a] Values are given versus a standard calomel electrode (SCE). N.D.: not determined. [b] HOMO–LUMO energy gaps were estimated by DFT at the B3LYP/6-31G(d) level. [c] Unsubstituted 4CzIPN. Values are taken from ref. [35].
(¢1.15 V vs. SCE) was slightly shifted to the positive direction compared with those of 4CzIPN-Me (¢1.23 V vs. SCE) and 4CzIPN-tBu (¢1.21 V vs. SCE), whereas the values of E0(R/RC + ) were similar for 4CzIPN-tBu (1.42 V vs. SCE) and 4CzIPN-Ph (1.41 V vs. SCE). Overall, the trend of the difference of E0(R/RC¢) and E0(R/RC + ) for the molecules is consistent with that of the energy gaps between HOMO and LUMO calculated by the DFT method (Table 2). Spin density of radical cations We performed DFT calculations to estimate the spin density of radical cations of the TADF molecules in the doublet state related to the radical stability observed in the cyclic voltammograms. Figure 7 depicts equivalent surfaces for a spin density at 0.004 obtained by DFT calculations. The spin densities of all atoms in the radical cations of the TADF molecules are listed in Table S3 in the Supporting Information. High spin densities corresponding to charge distribution caused by the unpaired electron were mainly distributed on the C(1), C(3), C(6), C(8), and N(9) atoms in the carbazolyl moieties of the TADF molecules. Figure 7 reveals that the spin densities of carbazolyl groups located at the para position with respect to cyano
Figure 7. Atom numbering for carbazole. Equivalent surfaces of the radical cations of the TADF molecules at a spin density of 0.004 calculated at the B3LYP/6-31 + G(d) level for the optimized structure of each doublet state. Chem. Eur. J. 2016, 22, 4889 – 4898
www.chemeurj.org
groups are smaller than those of the other two carbazole groups. The spin densities on the atoms tended to decrease slightly in the presence of methyl and tert-butyl substituents, and were substantially lowered by phenyl groups, which could delocalize electrons through conjugation. The carbazolyl moieties in the TADF molecules are bound to dicyanobenzene, so it is likely that the steric hindrance of this molecular structure can lower the reactivity of the C(1), C(8), and N(9) atoms of the radical ions of the substituted carbazolyl groups. Thus, high spin densities on the C(3) and C(6) atoms are crucial for inducing polymerization. The spin densities estimated by DFT calculations and the stability of the radical cations observed in the anodic scan of cyclic voltammograms indicate that polymerization initiated by the C(3) and C(6) atoms on the carbazolyl moieties was sufficiently inhibited in 4CzIPN-tBu because of steric hindrance arising from the bulky tert-butyl substituents (kinetically), and for 4CzIPN-Ph by steric hindrance and conjugation (thermodynamically) because of its phenyl substituents. Conversely, polymerization was insufficiently inhibited for 4CzIPN-Me. Indeed, the reactivity of the radical cation of 4CzIPN-Me was somewhat lower than that of unsubstituted 4CzIPN, but stabilization was weaker than that of 4CzIPN-tBu and 4CzIPN-Ph, which limited its ability to obtain stable ECL (see the next section).
ECL properties of TADF molecules ECL spectra of the molecules acquired by applying a squarewave voltage (50 Hz) are presented in Figure 8. CH2Cl2 and MeCN, which can dissolve the supporting electrolytes well, were used as solvents for ECL of the TADF molecules to facilitate comparison with the photophysical properties. Yellow to yellowish–red ECL emissions with maximum wavelengths of ECL (lECL) ranging from l = 570 to 610 nm were observed. The ECL spectra were redshifted in polar MeCN compared with their location in CH2Cl2, as observed in the PL spectra. In MeCN, the lECL values for the three molecules agree well with their corresponding lPL values. This agreement indicates that an inner filter effect caused by self-absorption at high concentration ( 1 mm for ECL measurements) is minimal because of the small overlaps between the UV/Vis and PL spectra of the
Figure 8. ECL spectra of the TADF molecules in CH2Cl2 and MeCN generated by applying a square-wave voltage (50 Hz). ECL spectra for 4CzIPN-Me were measured in CH2Cl2 and a mixture of CH2Cl2 and MeCN (50:50 vol %).
4894
Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper molecules (Figure 3). In CH2Cl2, a slight redshift of lECL of about 10 nm compared with the corresponding lPL value was observed for the three TADF molecules. It should be noted that the presence of a supporting electrolyte can shift the PL spectra of D–A molecules in less polar media, such as CH2Cl2.[16, 43] The PL spectra in CH2Cl2 containing 0.1 m TBAPF6 were redshifted and agreed well with the ECL spectra (Figure S2 and Table S4 in the Supporting Information), which indicated that S1 states of the molecules in CH2Cl2 were stabilized by the supporting electrolyte. Meanwhile, in MeCN, no significant shift in the PL spectra by the presence of 0.1 m TBAPF6 was observed. It is important to confirm that ECL spectra coincide with PL spectra because their agreement indicates that there is no ECL emission from other species, such as excimers, exciplexes, or polymers such as di- and trimers. Although polymerization of 4CzIPN-Me was observed in cycling potentials, the agreement between its PL and ECL spectra is reasonable because of the high frequency of the applied switching potentials (50 Hz). Therefore, we can conclude that the emission processes in PL and ECL of the TADF molecules are the same. ECL intensities of the three TADF molecules were compared with that of [Ru(bpy)3]2 + (bpy: 2,2’-bipyridine; fECL = 5 %[47]) as a standard substrate to estimate fECL by applying a train pulse voltage (5 Hz), considering total ECL intensity and charge during the potential step [Eq. (S7) in the Supporting Information]. Estimated fECL values are listed in Table 3. As for fPL, the fECL values depend on solvent polarity, that is, solutions in CH2Cl2 exhibited higher fECL than those in MeCN. It should be emphasized that the fECL values of the three TADF molecules were higher than those of the corresponding 0.25 Õ fPL values, which were considered to be the limit of fECL according to spin statistics. The TADF molecules have small DEst values and the total energies of a pair of radical species ( 2.6 eV estimated with the redox potentials) are energetically high enough to produce S1 (2.0–2.2 eV calculated with lPL) and T1 states; this leads to the generation of S1 and T1 states in a ratio similar to that of the value expected by spin statistics. Thus, these results confirm that thermal spin upconversion from T1 to S1 states occurs in these ECL systems and resulted in delayed ECL through thermal spin upconversion. Among the three TADF molecules, 4CzIPN-tBu showed the highest fECL ; remarkably, the fECL of 4CzIPN-tBu in CH2Cl2 was close to 40 %. Although
fECL should be the same as that of fPL for the TADF molecules because of spin upconversion, it seems that the fECL values for the three molecules are smaller than the corresponding fPL values. A possible explanation for this is quenching by the supporting electrolyte in the ECL system. We confirmed that the PL intensity decreased in the presence of 0.1 m TBAPF6, and the quenching effect depended on the solvents, that is, the decrease in PL intensity induced by 0.1 m TBAPF6 was about 15–30 % in CH2Cl2 and less than 10 % in MeCN. The same trend was observed in the shift of lPL caused by the presence of the supporting electrolyte in the solutions. It was also confirmed that the PL lifetimes of both prompt and delayed components were shortened in the presence of supporting electrolyte, and the effect on PL lifetime was large in CH2Cl2 and small in MeCN (Table S4 in the Supporting Information). Despite some quenching, sufficient delayed PL was observed when each 1 mm solution of TADF contained 0.1 m TBAPF6 (Figure S3 in the Supporting Information). In addition to the high fECL values, especially for 4CzIPN-tBu, the ECL intensities were stable for 4CzIPN-tBu and 4CzIPN-Ph, as shown in Figure 9 (typical ECL intensities and transient ECL responses are shown in Figure S4 in the Supporting Information). Upon applying voltage for more than 50 pulse voltage cycles, the change of ECL intensity was less than 10 % for 4CzIPN-tBu and 4CzIPN-Ph; this change may be ascribed to convective transport. In contrast, the ECL intensity of 4CzIPNMe decreased by 30–40 % during cycling, decreasing quickly during the first several pulse voltages cycles and then decreasing gradually. When a polymer film formed on the Pt electrode during the initial potential cycles, electron transfer between the electrode and 4CzIPN-Me was blocked, so fewer radical ions were produced, decreasing ECL intensity and polymer formation. Thus, after several potential cycles, the polymer film developed only gradually during successive voltages cycles, gradually decreasing the ECL intensity. The stability of ECL intensities is consistent with the stability of the radical cations produced by the electrode reaction (Figure 6). It should be noted that the ECL stability of 4CzIPN-Me was enhanced compared with that of unsubstituted 4CzIPN, the ECL intensity of which decayed by about 80 % under the same cycling conditions (Figure S5 in the Supporting Information).
Table 3. ECL properties of the TADF molecules in different solvents. 4CzIPN-X
Solvent
lECL [nm]
FWHM [nm]
fECL [%][a]
0.25 Õ fPL [%]
X = Me
CH2Cl2 MeCN[b] CH2Cl2 MeCN CH2Cl2 MeCN CH2Cl2 MeCN
585 595 572 587 595 610 555 565
124 132 110 130 121 139 107 122
13 1.5 7.5 0.6 35 2.8 9.2 2.2 13 1.7 3.2 0.5 47 6.0 15 1.6
4.5 3.0 11 2.1 4.3 0.9 14 4.5
tBu Ph H[c]
[a] Determined by using [Ru(bpy)3]2 + in MeCN (fECL = 5 %). FWHM: full width at half maximum of ECL spectra. [b] 50:50 vol % mixture with CH2Cl2. [c] Unsubstituted 4CzIPN from refs. [16] and [35]. Chem. Eur. J. 2016, 22, 4889 – 4898
www.chemeurj.org
4895
Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper
Figure 9. Stability of the ECL intensity of TADF molecules. ECL was generated by applying square-wave voltages 0.1–0.2 V larger than the peak potentials. The potential was first set to generate radical anions for 0.1 s, and then stepped to generate radical cations for 0.1 s. A potential of 0 V was applied for 1 s after each anodic step.
Conclusion We demonstrated ECL with TADF molecules composed of dicyanobenzene and methyl-, tert-butyl-, and phenyl-substituted carbazolyl groups, and examined the electrochemical and photophysical properties of these TADF molecules systematically. In ECL systems, because excited states of the TADF molecules are generated through ion annihilation between their radical ions produced by the electrode reaction directly, not by energy transfer from host molecules to TADF molecules doped in the host molecules, like in common OLED systems, the stability of radical ions is crucial to obtain stable ECL. Herein, stable and efficient ECL was realized by enhancing both the stability of the radical cations of TADF molecules and thermal spin upconversion. In the series of TADF molecules, that with the tert-butyl substituent showed the best performance in terms of both stability and efficiency of ECL because 4CzIPNtBu exhibited a higher fPL value than those of the other molecules, and the bulky tert-butyl substituents inhibited the radical reaction of 4CzIPN-tBu through steric hindrance. Quantum chemical calculations based on DFT and TD-DFT methods revealed that the HOMO–LUMO gaps, excitation energies, and DEst values were strongly affected by phenyl substituents beChem. Eur. J. 2016, 22, 4889 – 4898
www.chemeurj.org
cause of electron conjugation, but less so by methyl and tertbutyl substituents. Overall, electrochemical and photophysical measurements, coupled with theoretical calculations, are very useful to understand the properties of TADF molecules. We believe that designing TADF molecules based on quantum chemical calculations combined with experimental evidence, regarding the stability of their radical ions, is enormously useful for not only ECL systems, but also OLEDs, to establish efficient and stable systems.
Experimental Section Synthesis of TADF molecules Three TADF molecules were synthesized by using a slightly modified literature method (Scheme 1)[24] First, 60 % NaH in oil (0.25 (6.2), 0.40 (10), and 0.40 g (10 mmol) for methyl-, tert-butyl-, and phenyl-substituted carbazoles, respectively) was rinsed with hexane (2 Õ 20 mL) to remove the oil under N2 in a round-bottomed flask. Then, hexane was evaporated under reduced pressure and anhydrous THF (50 mL) was introduced into the flask. 3,6-Dimethylcarbazole (Cz-Me; 1.0 g, 5.1 mmol), 3,6-di-tert-butylcarbazole (Cz-tBu; 2.2 g, 7.9 mmol), or 3,6-diphenylcarbazole (Cz-Ph; 2.5 g, 7.8 mmol) was added to the solution of NaH, which was subse-
4896
Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper Photonics K. K., Japan). PL decay curves were obtained with a time-resolved fluorophotometer (Quantaurus Tau, Hamamatsu Photonics K. K., Hamamatsu, Japan) by irradiation with a lightemitting diode (LED; l = 365 nm). Irradiation light was removed with an optical filter. Prior to measurement of the PL decay curves and fPL values, solutions were deoxygenated by bubbling N2 through them. Scheme 1. Synthesis of TADF molecules.
Quantum chemical calculations quently stirred for 1 h at room temperature. Tetrafluoroisophthalonitrile (0.21 (1.1), 0.33 (1.7), and 0.33 g (1.7 mmol) for methyl-, tertbutyl-, and phenyl-substituted carbazoles, respectively) was then gradually added to the solution at 0 8C. The solution was stirred overnight at room temperature and then for 6 h at 60 8C. A mixture of methanol and water was poured into the flask to quench any remaining NaH, and then further methanol was added to precipitate the target compound. The precipitate was filtered and dried in a desiccator to yield a fluorescent powder. The residue was purified by column chromatography on silica gel with a mixture of hexane and CH2Cl2 as the eluent (50:50!0:100 vol %). The compounds synthesized with Cz-Me, Cz-tBu and Cz-Ph were obtained as fluorescent powders that were yellow, greenish–yellow, and yellowish– red, respectively, after purification. The desired products, 1,2,3,5tetrakis(3,6-dimethylcarbazol-9-yl)-4,6-dicyanobenzene (4CzIP-Me), 1,2,3,5-tetrakis(3,6-di-tert-butylcarbazol-9-yl)-4,6-dicyanobenzene (4CzIPN-tBu), and 1,2,3,5-tetrakis(3,6-diphenylcarbazol-9-yl)-4,6-dicyanobenzene (4CzIPN-Ph) were isolated in yields of 76, 54, and 60 %, respectively. 1
4CzIPN-Me: H NMR (400 MHz, (CD3)2CO): d = 8.02 (s, 2 H), 7.83 (d, J = 8.42 Hz, 2 H), 7.60 (s, 4 H), 7.51 (d, J = 8.42 Hz, 4 H), 7.42 (m, 4 H), 7.23 (s, 2 H), 6.94 (dd, J = 7.32, 1.10 Hz, 4 H), 6.59 (dd, J = 6.59, 1.60 Hz, 2 H), 2.54 (s, 6 H), 2.33 (s, 12 H), 2.17 ppm (s, 6 H); 13C NMR (400 MHz, CDCl3): d = 145.8, 145.0, 144.6, 144.2, 143.4, 138.5, 137.0, 135.5, 134.0, 124.9, 124.4, 124.0, 123.2, 122.0, 117.5, 116.1, 115.3, 114.9, 112.2, 109.7, 108.9, 34.93, 34.54, 34.25 ppm; m/z: 902.45 [M + H] + . 4CzIPN-tBu: 1H NMR (400 MHz, (CD3)2CO): d = 8.40 (d, J = 1.83 Hz, 2 H), 7.94 (d, J = 8.79 Hz, 2 H), 7.87 (d, J = 1.83 Hz, 4 H), 7.53, (dd, J = 10.62, 2.20 Hz, 2 H), 7.43 (m, 6 H), 7.12 (m, 6 H), 6.63 (dd, J = 10.62, 2.20 Hz, 2 H), 1.49 (s, 18 H), 1.31 (s, 36 H), 1.22 ppm (s, 18 H); 13 C NMR (400 MHz, CDCl3): d = 145.8, 145.0, 144.6, 144.3, 143.4, 138.5, 137.0, 135.5, 134.0, 124.9, 124.4, 124.0, 123.2, 122.0, 117.5, 116.1, 115.3, 114.9, 112.2, 109.7, 108.9, 34.93, 34.54, 34.25, 32.00, 37.72 ppm, m/z: 1238.48 [M + H] + . 4CzIPN-Ph: 1H NMR (400 MHz, (CD3)2CO): d = 8.81 (s, 2 H), 8.26, (m, 6 H), 8.09, (dd, J = 10.25, 1.83 Hz, 2 H), 7.91 (d, J = 8.79 Hz, 8 H), 7.82 (d, J = 1.46 Hz, 2 H), 7.60 (m, 18 H), 7.35 (m, 24 H), 7.11 ppm (dd, J = 10.25, 1.83 Hz, 2 H); 13C NMR (400 MHz, CDCl3): d = 141.0, 140.9, 138.5, 136.4, 135.7, 134.8, 128.9, 128.8, 127.6, 127.3, 127.2, 127.1, 127.0, 125.4, 120.1, 119.2, 119.1, 118.4, 116.4, 110.5, 110.3, 110.0 ppm, m/z: 1397.56 [M + H] + .
Characterization and photophysical measurements H and 13C NMR spectra of the synthesized molecules were acquired with a JMN-EPC400 spectrometer (JEOL Ltd., Tokyo, Japan) and MALDI-TOF-MS were measured on an Autoflex III mass spectrometer (Bruker Co., MA). Absorption and PL spectra were collected by using a UV/Vis spectrophotometer (V-560, JASCO, Tokyo, Japan) and spectrofluorometer (FluoroMax-4, Horiba, Kyoto, Japan), respectively. fPL values for the TADF solutions were estimated by using an integrating sphere (Quantaurus QY, Hamamatsu
Quantum chemical calculations based on DFT were performed by using the Gaussian 09 program package.[48] All structures were optimized by using the 6-31G(d) basis set before performing subsequent calculations. Molecular orbitals at the ground state were estimated at the B3LYP/6-31G(d) level, whereas the spin density of radical cations of TADF molecules was calculated by considering a diffuse function at the doublet state by using B3LYP/6-31 + G(d). To estimate DEst, TD-DFT calculations at the M06–2X/6-31 + G(d) level for the optimized structures of the S1 and T1 states were performed by using the same functional. This functional could reasonably explain the absorption spectra of D–A molecules composed of carbazole and dicyanobenzene groups.[24] Computations were mainly performed by using computer facilities (Fujitsu Primergy CX400) at the Research Institute for Information Technology, Kyushu University.
Electrochemical measurements A conventional three-electrode system with a Pt disk (diameter d = 1.6 and 3.0 mm for cyclic voltammetry and ECL measurements, respectively), coiled Pt wire, and Ag wire as working, counter, and reference electrodes, respectively, was employed. The Pt disk was polished by using an alumina suspension (d = 50 nm) before electrochemical measurements. The electrodes were placed in a glass vial containing 1 mm solutions of the TADF molecules with 0.1 m TBAPF6 as a supporting electrolyte. An optical fiber (core diameter: 1 mm) connected to an optical photometer (Maya2000 pro, Ocean Optics Inc., FL) or a photodetector connected to a power meter (model 1918-C, Newport Co., CA) was positioned under the glass vial. To acquire ECL spectra, a square-wave voltage (50 Hz) was applied to the Pt disk electrode at 0.1–0.2 V over the peak potentials observed by cyclic voltammetry. Transient ECL intensities at the maximum ECL wavelength were collected by the photodetector using a triple-potential-step method.[1] For the first 100 ms, the applied potential was set to generate anion radicals, and then the potential was switched to positive for 100 ms (5 Hz) to oxidize the TADF molecules. The external voltage to the electrochemical cell was controlled with an electrochemical workstation (Electrochemical analyzer, 715AN, BAS Inc., Tokyo, Japan). Cyclic voltammograms were recorded by using a positive feedback for IR compensation. The typical solution resistance was about 500 W and uncompensated for resistance was about 5 W. Solutions were deoxygenated by bubbling with N2 before all electrochemical experiments, and electrochemical measurements were carried out under a N2 atmosphere. Solvents were dehydrated by using molecular sieve before preparing solutions.
1
Chem. Eur. J. 2016, 22, 4889 – 4898
www.chemeurj.org
Acknowledgements This research was supported in part by the Japan Society for the Promotion of Science through its “Funding Program for World-Leading Innovative R&D on Science and Technology”. R.I. acknowledges Hiroshi Miyazaki and Hiroko Nomura
4897
Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper (Kyushu University) for valuable discussions about synthesis of the TADF molecules. Keywords: donor–acceptor systems · fluorescence electrochemiluminescence · lifetimes · solvent effects
·
[1] Electrogenerated Chemiluminescence (Ed.: A. J. Bard), Marcel Dekker, New York, 2004. [2] M. M. Richter, Chem. Rev. 2004, 104, 3003 – 3036. [3] W. Miao, Chem. Rev. 2008, 108, 2506 – 2553. [4] K. M. Maness, R. M. Wightman, J. Electroanal. Chem. 1995, 396, 85 – 95. [5] N. R. Armstrong, R. M. Wightman, E. M. Gross, Annu. Rev. Phys. Chem. 2001, 52, 391 – 422. [6] A. Kapturkiewicz, J. Electroanal. Chem. 1991, 302, 131 – 144. [7] A. Kapturkiewicz, Chem. Phys. 1992, 166, 259 – 273. [8] S. W. Feldberg, J. Phys. Chem. 1966, 70, 3928 – 3930. [9] A. Kapturkiewicz, J. Herbich, J. Nowacki, Chem. Phys. Lett. 1997, 275, 355 – 362. [10] M. M. Sartin, C. Shu, A. J. Bard, J. Am. Chem. Soc. 2008, 130, 5354 – 5360. [11] K. M. Omer, S.-Y. Ku, Y.-C. Chen, K.-T. Wong, A. J. Bard, J. Am. Chem. Soc. 2010, 132, 10944 – 10952. [12] M. Shen, J. Rodrguez-Lûpez, J. Huang, Q. Liu, X. H. Zhu, A. J. Bard, J. Am. Chem. Soc. 2010, 132, 13453 – 13461. [13] J. Suk, J. Cheng, K. Wong, A. J. Bard, J. Phys. Chem. C 2011, 115, 14960 – 14968. [14] F. Polo, F. Rizzo, M. Veiga-Gutierrez, L. De Cola, S. Quici, J. Am. Chem. Soc. 2012, 134, 15402 – 15409. [15] H. Qi, Y. H. Chen, C. H. Cheng, A. J. Bard, J. Am. Chem. Soc. 2013, 135, 9041 – 9049. [16] R. Ishimatsu, S. Matsunami, T. Kasahara, J. Mizuno, T. Edura, C. Adachi, K. Nakano, T. Imato, Angew. Chem. Int. Ed. 2014, 53, 6993 – 6996; Angew. Chem. 2014, 126, 7113 – 7116. [17] Z. Mohsan, A. L. Kanibolotsky, A. J. Stewart, A. R. Inigo, L. Dennany, P. J. Skabara, J. Mater. Chem. C 2015, 3, 1166 – 1171. [18] S. Rashidnadimi, T. H. Hung, K.-T. Wong, A. J. Bard, J. Am. Chem. Soc. 2008, 130, 634 – 639. [19] K. Shizu, H. Tanaka, M. Uejima, T. Sato, K. Tanaka, H. Kaji, C. Adachi, J. Phys. Chem. C 2015, 119, 1291 – 1297. [20] K. Shizu, M. Uejima, H. Nomura, T. Sato, K. Tanaka, H. Kaji, C. Adachi, Phys. Rev. Appl. 2015, 3, 014001-1-7. [21] C. A. Parker, Photoluminescence of Solutions, Elsevier, Amsterdam, 1968. [22] T. Nakagawa, S.-Y. Ku, K.-T. Wong, C. Adachi, Chem. Commun. 2012, 48, 9580 – 9582. [23] H. Tanaka, K. Shizu, H. Miyazaki, C. Adachi, Chem. Commun. 2012, 48, 11392 – 11394. [24] H. Uoyama, K. Goushi, K. Shizu, H. Nomura, C. Adachi, Nature 2012, 492, 234 – 238. [25] S.-Y. Lee, T. Yasuda, H. Nomura, C. Adachi, Appl. Phys. Lett. 2012, 101, 093306-1-4. [26] Q. Zhang, J. Li, K. Shizu, S. Huang, S. Hirata, H. Miyazaki, C. Adachi, J. Am. Chem. Soc. 2012, 134, 14706 – 14709. [27] F. B. Dias, K. N. Bourdakos, V. Jankus, K. C. Moss, K. T. Kamtekar, V. Bhalla, J. Santos, M. R. Bryce, A. P. Monkman, Adv. Mater. 2013, 25, 3707 – 3714.
Chem. Eur. J. 2016, 22, 4889 – 4898
www.chemeurj.org
[28] H. Nakanotani, K. Masui, J. Nishide, T. Shibata, C. Adachi, Sci. Rep. 2013, 3, 2127-1-5. [29] K. Sato, K. Shizu, K. Yoshimura, A. Kawada, H. Miyazaki, C. Adachi, Phys. Rev. Lett. 2013, 110, 247401-1-5. [30] Y. Im, J. Y. Lee, Chem. Mater. 2014, 26, 1413 – 1419. [31] H. Wang, L. Xie, Q. Peng, L. Meng, Y. Wang, Y. Yi, P. Wang, Adv. Mater. 2014, 26, 5198 – 5204. [32] Q. Zhang, H. Kuwabara, W. J. Potscavage, S. Huang, Y. Hatae, T. Shibata, C. Adachi, J. Am. Chem. Soc. 2014, 136, 18070 – 18081. [33] J. Lee, K. Shizu, H. Tanaka, H. Nakanotani, T. Yasuda, H. Kaji, C. Adachi, J. Mater. Chem. C 2015, 3, 2175 – 2181. [34] Y. Sakai, Y. Sagara, H. Nomura, N. Nakamura, Y. Suzuki, H. Miyazaki, C. Adachi, Chem. Commun. 2015, 51, 3181 – 3184. [35] R. Ishimatsu, S. Matsunami, K. Shizu, C. Adachi, K. Nakano, T. Imato, J. Phys. Chem. A 2013, 117, 5607 – 5612. [36] H. C. Moon, T. P. Lodge, C. D. Frisbie, J. Am. Chem. Soc. 2014, 136, 3705 – 3712. [37] T. Kasahara, S. Matsunami, T. Edura, R. Ishimatsu, J. Oshima, M. Tsuwaki, T. Imato, S. Shoji, C. Adachi, J. Mizuno, Sens. Actuators B 2015, 207, 481 – 489. [38] T. Nobeshima, K. Nakamura, N. Kobayashi, J. Photopolym. Sci. Technol. 2013, 26, 397 – 402. [39] X. Jiang, X. Yang, C. Zhao, K. Jin, L. Sun, J. Phys. Chem. C 2007, 111, 9595 – 9602. [40] J. W. Oh, Y. O. Lee, T. H. Kim, K. C. Ko, J. Y. Lee, H. Kim, J. S. Kim, Angew. Chem. Int. Ed. 2009, 48, 2522 – 2524; Angew. Chem. 2009, 121, 2560 – 2562. [41] A. Kapturkiewicz, J. Herbich, J. Karpiuk, J. Nawacki, J. Phys. Chem. A 1997, 101, 2332 – 2344. [42] W. Rettig, M. Zander, Chem. Phys. Lett. 1982, 87, 229 – 234. [43] P. Natarajan, M. Schmittel, J. Org. Chem. 2013, 78, 10383 – 10394. [44] K. Goushi, K. Yoshida, K. Sato, C. Adachi, Nat. Photonics 2012, 6, 253 – 258. [45] P. Marrec, C. Dano, N. Gueguen-Simonet, J. Simonet, Synth. Mater. 1997, 89, 171 – 179. [46] A. J. Bard, L. R. Faulkner, Electrochemical Methods: Fundamentals and Applications, 2nd ed.,Wiley, Hoboken, 2001. [47] W. L. Wallace, A. J. Bard, J. Phys. Chem. 1979, 83, 1350 – 1357. [48] Gaussian 09, Revision C.01, M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery, Jr., J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, T. Keith, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, A. D. Daniels, O. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski, D. J. Fox, Gaussian, Wallingford CT, 2010. Received: January 8, 2016 Published online on February 24, 2016
4898
Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper
& Electrochemiluminescence
Photophysical Properties and Efficient, Stable, Electrogenerated Chemiluminescence of Donor–Acceptor Molecules Exhibiting Thermal Spin Upconversion Ryoichi Ishimatsu,*[a] Tomohiko Edura,[b] Chihaya Adachi,*[a, b, c] Koji Nakano,[a] and Toshihiko Imato*[a] Abstract: The photophysical properties and electrogenerated chemiluminescence (ECL) of three donor–acceptor molecules composed of dicyanobenzene and methyl-, tert-butyl-, and phenyl-substituted carbazolyl groups, 1,2,3,5-tetrakis(3,6-disubstituted-carbazol-9-yl)-4,6-dicyanobenzene (4CzIPN-Me, 4CzIPN-tBu, and 4CzIPN-Ph, respectively) are described. These molecules show delayed fluorescence as a result of thermal spin upconversion from the lowest triplet state to the lowest singlet state at room temperature. The three molecules showed yellow to yellowish–red ECL. Re-
Introduction
markably, the ECL efficiencies of 4CzIPN-tBu in dichloromethane reached almost 40 %. Moreover, stable ECL was emitted from 4CzIPN-tBu and 4CzIPN-Ph. In case of 4CzIPN-Me, the ECL intensity decreased during voltage cycles because of polymerization. Quantum chemical calculations revealed that polymerization was inhibited by the steric hindrance of the bulky tert-butyl and phenyl groups on the carbazolyl moieties and lowered the spin density on the carbazolyl groups through electron conjugation for 4CzIPN-Ph.
RC ¢ þ RC þ ! n1 R* þ ð1¢nÞ3 R* þ R ðexcited-state formation by ion annihilationÞ
Electrogenerated chemiluminescence (ECL) can be emitted from molecules at excited states that are formed by electron transfer between a radical anion and cation produced through an electrode reaction as shown in Equations (1)–(4):[1–3] R þ e¢ ! R C ¢ ðradical anion formationÞ R¢e¢ ! RC þ ðradical cation formationÞ [a] Dr. R. Ishimatsu, Prof. C. Adachi, Prof. K. Nakano, Prof. T. Imato Department of Applied Chemistry, Graduate School of Engineering Kyushu University, 744 Motooka, Nishi-ku Fukuoka, 819-0395 (Japan) E-mail: [email protected] [email protected] [email protected] [b] Dr. T. Edura, Prof. C. Adachi Center for Organic Photonics and Electronics Research Kyushu University 744 Motooka, Nishi-ku Fukuoka, 819-0395 (Japan) [c] Prof. C. Adachi JST, ERATO, Adachi Molecular Exciton Engineering Project c/o Center for Organic Photonics and Electronics Research, and International Institute for Carbon Neutral Energy Research (WPI-I2CNER), Kyushu University 744 Motooka, Nishi-ku, Fukuoka, 819-0395 (Japan) Supporting information for this article can be found under http:// dx.doi.org/10.1002/chem.201600077. Chem. Eur. J. 2016, 22, 4889 – 4898
ð1Þ
ð2Þ
1
R* ! R þ hn
ð3Þ
ð4Þ
ðECL emissionÞ
in which R represents a fluorescent molecule, and 1R* and 3R* are the lowest excited singlet (S1) and triplet (T1) states of R, respectively. Usually, the S1 and T1 states of most fluorescent molecules relax through radiative and nonradiative transitions, respectively. According to spin statistics,[4, 5] that is, 25 % S1 and 75 % T1 states will be produced by charge recombination processes, such as hole–electron recombination and ion annihilation, n in Equation (3) is considered to be 0.25 for an energysufficient ECL system, as illustrated in Figure 1. To populate S1 and T1 with the ratio following spin statistics, the energy level of radical species should be large enough and the energy splitting between S1 and T1 states should be small. When the energy of a pair of radical species is insufficient to produce the S1 state, only the T1 state will be formed. The ground state may directly be formed without forming excited states for an energy-deficient system as shown in Equation (5):[1, 6] R C ¢ þ RC þ ! 2 R ðground state formationÞ
ð5Þ
For an energy-sufficient ECL system, the pathway to produce the ground state directly [Eq. (5)] occurs in the Marcus inverted region, and thus, is an unlikely process.[6, 7] 4889
Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper
Figure 1. Schematic representation of electronic transitions of thermally activated delayed fluorescence (TADF) molecules. Excited states are formed by light absorption (left) and ion annihilation in an ECL system (right). Reverse intersystem crossing (RISC) increases the yield of “delayed” luminescence. ISC: intersystem crossing, FL: fluorescence.
Because the kinetics of electronic transitions from excited states, including light emission, should be similar for ECL and photoluminescence (PL) processes, “pure” spin statistics predicts that the maximum ECL efficiency (fECL) of fluorescent molecules is the value of their quantum efficiency of PL (fPL) multiplied by a factor of 0.25 (i.e., 0.25 Õ fPL) without spin upconversion processes, such as triplet–triplet (T–T) annihilation.[1, 8] Generally, fluorescent molecules that exhibit high fPL and produce stable radical ions through electrode reactions are favorable to obtain efficient, stable ECL. This is because more stable radical ions can eliminate additional parasitic processes, such as polymerization and reactions with solvents, which result in a larger generation of excited states than less stable ones. Based on the guidelines of high fPL and stable radical ions, recently, some donor–acceptor (D–A) molecules have been used as ECL systems to attain high fECL.[9–18] The ECL mechanism of D–A molecules has also been studied. Electrondonating and -accepting groups in D–A molecules can be oxidized and reduced, respectively, through individual electrode reactions, and then the D–A molecules can form excited states through ion annihilation. In addition, by changing the D and A groups in D–A molecules, it is possible to adjust the energy levels of molecular orbitals, such as the HOMO and LUMO. This, in turn, allows the absorption, PL, and ECL spectra of D–A molecules to be tuned.[11, 14] D–A molecules with multiple D and A groups that show multistep electrode reactions have been used to study ECL behavior, including electron-transfer mechanisms between multivalent ions, such as divalent ones.[10, 12, 18] An interesting feature of D–A molecules is that the energy gap between S1 and T1 states (DEst) can be decreased by using appropriate combinations of D and A groups[9] to achieve thermal spin upconversion from T1 to S1 states (RISC), as shown by Equation (6) 3
Results and Discussion Photophysical properties
R* þ D ! 1 R*
ð6Þ
ðthermal spin upconversionÞ
Quantum chemical calculations have enabled researchers to design D–A molecules with small DEst, the value of which markedly influences the kinetics of thermal spin upconversion. Chem. Eur. J. 2016, 22, 4889 – 4898
Small DEst values can be achieved by overlapping HOMO and LUMO wave functions effectively, whereby the HOMO and LUMO wave functions of the designed D–A molecules are spatially separated and mainly distributed on D and A groups, respectively.[19, 20] Because the rate constant of thermal spin upconversion, depending on DEst, is much lower than that of the radiative transition, upon light absorption, both prompt FL and TADF[21] can be observed. Based on results of quantum chemical calculations, several D–A molecules showing delayed FL as a result of thermal spin upconversion have been synthesized and used in organic light-emitting diodes (OLEDs).[22–34] Owing to small DEst values of the synthesized TADF molecules, the ratio of S1 and T1 states generated through charge recombination can approach the value expected by spin statistics. OLEDs containing TADF molecules are able to convert electrical power into light emission very efficiently; the conversion efficiencies of current into luminescence are similar to those of fPL of TADF molecules, which is a breakthrough in the theoretical limitation imposed by spin statistics. Recently, we reported the photophysical,[35] electrochemical, and ECL properties[16] of some TADF molecules composed of dicyanobenzene (A) and carbazolyl (D) groups, which were synthesized for use in OLEDs, and showed that TADF molecules were very useful to obtain fECL values higher than 0.25 Õ fPL because of thermal spin upconversion, which led to “delayed ECL” described in the same manner as delayed PL. However, the ECL intensity of these systems decayed quickly relative to stable ECL systems because their unstable radical cations induced polymerization. Therefore, TADF molecules that form stable radical ions are needed to realize an efficient, stable ECL system. Establishing an efficient and stable ECL system will be beneficial for light-emitting liquid displays.[36–38] Furthermore, knowledge of TADF molecules that exhibit stable ECL can be extended to OLED systems to understand their degradation processes and establish stable, efficient OLEDs by using TADF molecules. Although the stability of radical ions is crucial for ECL, few systematic studies focused on enhancing the stability of radical ions for ECL have been carried out.[15, 39, 40] Herein, we report the synthesis, electrochemistry, and photophysical and ECL properties of three D–A molecules composed of dicyanobenzene and methyl-, tert-butyl-, or phenyl-substituted carbazolyl moieties (Figure 2) that exhibit TADF. We systematically examine the effects of the substituents on the electrochemical and photophysical properties of the materials. Through both TADF and the formation of stable radical ions, an efficient, stable ECL system is realized.
www.chemeurj.org
To elucidate the photophysical properties of the synthesized molecules in solution, UV/Vis absorption and PL spectra, as well as PL decay curves, were measured in four solvents with different polarity: toluene, chloroform, dichloromethane, and acetonitrile. Figure 3 shows the UV/Vis absorption and PL spectra of the molecules in the four different solvents. Relatively
4890
Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper similar. The greater redshift of labs of 4CzIPN-Ph than those of 4CzIPN and 4CzIPN-tBu indicates that the energy gaps between molecular orbitals related to transitions induced by light absorption are narrowed by the presence of phenyl substituents, which is probably caused by electron conjugation. To confirm that conjugation affected the molecular orbitals of the materials, DFT calculations were carried out. The estimated HOMO and LUMO and energy levels of the compounds are illustrated in Figure 4. The smaller HOMO–LUMO gap for
Figure 2. Molecular structures of TADF molecules composed of substituted carbazolyl and dicyanobenzene moieties considered in this study.
Figure 4. HOMO and LUMO distributions and energy levels estimated at the B3LYP/6-31G(d) level for the optimized structures of the three TADF molecules.
Figure 3. UV/Vis absorption (dashed lines) and normalized PL (solid lines) spectra of the TADF materials in four different solvents. Concentration for PL spectra: 15 mm.
large Stokes shifts were observed for all three molecules. The Stokes shifts caused small overlaps between the absorption and PL spectra at l 500 nm, and the overlap diminished as the solvent polarity increased. The UV/Vis absorption maxima (labs) and PL maxima (lPL) of the three molecules, together with those of unsubstituted 4CzIPN, are summarized in Table 1. The substituents on carbazolyl moieties shifted labs to longer wavelengths by about 10–20 nm compared with that of unsubstituted 4CzIPN. The redshift caused by phenyl groups is greater than that induced by methyl and tert-butyl groups in all of the solvents. The effect of the methyl and tert-butyl substituents on the UV/Vis absorption spectra of the material is Chem. Eur. J. 2016, 22, 4889 – 4898
www.chemeurj.org
4CzIPN-Ph by about 0.2 eV compared with those for 4CzIPNMe and 4CzIPN-tBu is consistent with the trend observed in the UV/Vis absorption spectra. The HOMOs and LUMOs of each molecule are spatially separated; HOMOs are mainly distributed over the carbazolyl (D) moieties, and LUMOs are localized over the dicyanobenzene (A) moiety. This spatial separation means that HOMO–LUMO transitions of the molecules will have charge-transfer characteristics. It should be noted that the HOMO of 4CzIPN-Ph is partially distributed over the phenyl groups of carbazolyl moieties because of conjugation, whereas the HOMOs of 4CzIPN-Me and 4CzIPN-tBu are hardly distributed over the methyl and tert-butyl substituents. With increasing solvent polarity, labs at longer wavelength, such as the bands at l 400 nm and shoulders at l 470 nm, exhibited slight blueshifts, whereas labs at shorter wavelength related to transitions from inner molecular orbitals were almost independent of solvent polarity. TD-DFT calculations were performed to identify the main transitions between molecular orbitals affecting labs. The results of TD-DFT calculations for 10 low-lying singlet excited states (S1–S10) upon light absorption under vacuum, and also considering the solvent effect by using the polarized continuum model (PCM), are summarized in Tables S1 and S2 in the Supporting Information. For excitation to S1–S10, transitions from HOMO-m (m = 0, 1, 2, 3, 4, and 6) to LUMO or LUMO + 1 were mainly dominant. The largest oscillator strength (f) in the solvents was found for the transition to S3 or S4 for 4CzIPN-Me (f = 0.40–0.46), S4 for 4CzIPN-tBu (f = 0.40–0.49), and S3 or S4 for 4CzIPN-Ph (f = 0.55– 0.62), and these energy levels of S3 and S4 were very similar with differences of less than 0.02 eV. Therefore, labs 400 nm is mainly caused by the transitions to these states. From f values
4891
Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper Table 1. Photophysical properties of the TADF molecules in different solvents. 4CzIPN-X
Solvent
labs [nm]
lPL [nm]
FWHM [nm]
kp¢1 [ns]
kd¢1 [ms]
kr (Õ 108 s¢1)
fPL [%][a]
fp [%]
fd [%]
X = Me (0.307 eV)[b]
toluene chloroform CH2Cl2 MeCN toluene chloroform CH2Cl2 MeCN toluene chloroform CH2Cl2 MeCN toluene CH2Cl2 MeCN
325, 338, 357sh, 375, 395, 467sh 325, 338, 394, 478sh 324, 338, 391, 472sh 324sh, 339, 379, 456sh 321, 334, 391, 468sh 321, 334, 396, 478sh 321, 334, 393, 470sh 320sh, 335, 380, 459sh 330, 345, 404, 486sh 330, 346, 404, 491sh 331, 345, 404, 487sh 331, 348, 390, 467sh 312, 326, 375, 452sh 312, 326, 378, 455sh 312, 325, 365, 438sh
525 568 577 603 521 554 565 588 547 575 588 611 507 536 551
88 116 128 145 96 107 116 146 101 115 128 147 83 107 125
13 20 15 6 15 23 20 10 22 19 13 4 14 25 19
2.8 2.1 1.7 0.99 2.9 2.6 2.4 1.4 1.7 1.6 1.5 0.79 1.8 2.0 1.4
18 11 8.7 5.3 18 7.8 13 7.1 16 14 10 8.5 18 9.2 5.3
76 38 18 3.7 86 56 44 8.5 72 38 17 3.7 94 54 18
23 22 13 3.2 27 18 25 7.1 36 27 13 3.4 25 23 10
53 16 5.0 0.5 59 38 19 1.4 36 11 4.0 0.3 69 31 8.0
tBu (0.308 eV)[b]
Ph (0.214 eV)[b]
H[c] (0.345 eV)[b]
[a] Estimated by using an integrating sphere. sh: shoulder. FWHM: full width at half maximum of PL spectra. [b] DEST values were estimated by time-dependent (TD) DFT at the M06-2X/6-31 + G(d) level. [c] Unsubstituted 4CzIPN. Values are from ref. [35].
(Table S2 in the Supporting Information), other characteristic labs, such as those at l 460 (shoulder) and 340 nm, are derived from transitions to S1 and S8, respectively. These calculations also confirmed that all molecular orbitals involved in the transitions to S1–S10 had charge-transfer characteristics, that is, HOMO-m, and LUMO and LUMO + 1 were localized on carbazolyl and dicyanobenzene groups, respectively. Intramolecular charge transfer from carbazolyl to dicyanobenzene groups at these excited states was induced through light absorption. PL spectra of the molecules exhibited lPL 520–610 nm. Shifts of PL spectra caused by the substituents and solvents were clearly seen; lPL was shifted to longer wavelength by the substituents in the order of tert-butyl < methyl < phenyl; this reflected the calculated HOMO–LUMO gaps. The lPL values of these molecules showed relatively large redshifts of about 60– 80 nm when the solvent polarity increased upon changing from toluene to MeCN. Such large redshifts of lPL induced by increasing solvent polarity are typical behavior of D–A molecules. This is because, upon light adsorption, D–A molecules form an intramolecular charge-transfer excited state with a larger dipole moment that can be stabilized by surrounding polar solvent molecules, resulting in emission at longer wavelengths.[11, 41–43] To estimate the PL lifetimes of the molecules, PL decay curves were measured in four different solvents (Figure 5). Importantly, both “prompt” and “delayed” PL components were observed for the three molecules at room temperature in the four solvents. This confirms that spin upconversion from T1 to S1 states occurs in the solutions at room temperature. It is notable that the lifetimes of both prompt and delayed components depend on the substituents and both shorten with increasing solvent polarity. Experimental decay curves of PL intensity (I(t)) can be expressed by using the biexponential function given in Equation (7):[35, 44] IðtÞ ¼ Ip expð¢kp tÞ þ Id expð¢kd tÞ þ C Chem. Eur. J. 2016, 22, 4889 – 4898
www.chemeurj.org
ð7Þ
Figure 5. PL decay curves for the three TADF molecules in four different solvents at room temperature. Insets: Decay curves for the prompt components. Concentration: 15 mm.
4892
Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper in which Ip and Id represent the PL intensities of the prompt and delayed components at t = 0, respectively; C is a constant; and kp and kd are the decay constants for prompt and delayed components, respectively. We estimated Ip, Id, kp, and kd values by fitting the experimental curves with Equation (7) and calculated the contributions of the quantum efficiencies of the prompt (fp) and delayed (fd) components to fPL. The calculated values and values of fPL are listed in Table 1. The lifetimes of the prompt (kp¢1) and delayed (kd¢1) components related to the kinetics of radiative and nonradiative transitions, including RISC [see Eqs. (S1) and (S2) in the Supporting Information] were in the order of 10¢8 and 10¢6 s, respectively, and lifetimes of both prompt and delayed components were at least halved upon increasing the solvent polarity upon switching from toluene to MeCN. The rate constants of radiative transition (kr) calculated from kp and fp by using Equation (S6) in the Supporting Information decreased with increasing solvent polarity (Table 1). In less polar solvents, such as toluene, the TADF molecules exhibit high fPL (76, 86, and 72 % for 4CzIPN-Me, 4CzIPN-tBu, and 4CzIPN-Ph in toluene, respectively) with a large contribution from delayed PL (fd = 53, 59, and 36 %, for 4CzIPN-Me, 4CzIPN-tBu, and 4CzIPN-Ph in toluene, respectively), although these values are somewhat smaller than those of 4CzIPN (fPL = 94 %, and fd = 69 % in toluene).[35] Both fp and fd decreased dramatically with increasing solvent polarity, resulting in lower fPL. This means that the rate constants of thermal deactivation increase in polar solvents. Because DEst is an important factor in delayed FL, TD-DFT calculations were performed for S1 and T1 states of the molecules to estimate DEst, and the obtained values are listed in Table 1. The DEst values of 4CzIPN-Me (0.307 eV) and 4CzIPNtBu (0.308 eV) are almost identical and slightly smaller than that of unsubstituted 4CzIPN (0.345 eV), whereas the DEst value of 4CzIPN-Ph (0.214 eV) is much smaller than those of 4CzIPNMe and 4CzIPN-tBu. The markedly smaller DEst for 4CzIPN-Ph may be explained by the spatial distribution and more effective overlap of its HOMO and LUMO induced by extended conjugation compared with that in 4CzIPN-Me and 4CzIPN-tBu, as calculated for some other TADF molecules.[32] Thus, introducing substituents of this kind, especially phenyl groups, is favorable to decrease DEst (the DFT calculations with M06-2X/6-31 + G(d) tend to overestimate DEst by about 0.1–0.2 eV[24]). It should be mentioned that DEst decreases with increasing solvent polarity, which may be described as a function of fp, fd, kp, kd, and kISC, in which kISC is the rate constant of intersystem crossing from S1 to T1 states[35] [Eqs. (S3) and (S4) in the Supporting Information]. The smaller DEst value, that is, the higher rate constant of upconversion and shorter lifetime of the T1 state for the three molecules compared with those of unsubstituted 4CzIPN, could be useful to reduce interactions with quenchers. Indeed, introducing such substituents is not deemed to be suitable to achieve high fPL, but they are able to stabilize the radical cations, leading to stable ECL (as discussed below).
Chem. Eur. J. 2016, 22, 4889 – 4898
www.chemeurj.org
Electrochemical properties of TADF molecules The redox behavior of the TADF molecules was observed with cyclic voltammetry. Figure 6 shows cyclic voltammograms of 4CzIPN-tBu and 4CzIPN-Ph in MeCN, and that of 4CzIPN-Me in a mixture of CH2Cl2 and MeCN (50:50 vol %). Because 4CzIPNMe displayed poor solubility in MeCN, a mixture of CH2Cl2 and MeCN was used as a solvent.
Figure 6. Cyclic voltammograms obtained for 1 mm solutions of TADF molecules containing 0.1 m tetrabutylammonium hexafluorophosphate (TBAPF6). Solvent: MeCN for 4CzIPN-tBu and 4CzIPN-Ph, and a mixture of CH2Cl2 and MeCN (50:50 vol %) for 4CzIPN-Me. Scan rate: 100 mV s¢1. The arrows indicate the scan direction.
In the cathodic scans, reduction waves of dicyanobenzene (A) groups were observed for the three molecules, and peak separations of the reduction waves were about 70 mV at a scan rate of 100 mV s¢1. In the anodic scans, the TADF molecules showed oxidation waves of carbazolyl (D) groups, and the peak separations for 4CzIPN-tBu and 4CzIPN-Ph were about 65 mV at 100 mV s¢1, which were close to 59 mV for the reversible one-electron reaction. It should be emphasized that rate constants of the electrode reactions for both oxidation and reduction seem to be relatively large because these peak separations are close to that of the reversible one, and it is difficult to estimate the values accurately because of uncompensated for resistance (see the Experimental Section). For 4CzIPNMe, an irreversible oxidation wave was observed; this means that the electrode reaction (E) was followed by a chemical reaction (C; EC mechanism). The irreversibility is because the high reactivity of the radical cation of 4CzIPN-Me generated through the electrode reaction leads to polymerization. The polymerization of molecules containing unsubstituted carbazolyl groups[11, 16, 45] has been reported, and is mainly attributable to the high spin density on the C(3), C(6), and N(9) atoms of the radical cation of carbazole.[45] Overall, the radical anions and cations of 4CzIPN-tBu and 4CzIPN-Ph are stable, and the radical anion and cation of 4CzIPN-Me are stable and unstable, respectively, because of polymerization. Both reduction and oxidation waves were fitted with simulated curves of the one-electron reaction by using simulation software based on the Butler–Volmer model[46] to estimate standard redox potentials, E0(R/RC¢) and E0(R/RC + ), related to HOMO and LUMO levels, respectively. The estimated values are summarized in Table 2. The value of E0(R/RC¢) for 4CzIPN-Ph
4893
Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper Table 2. Redox potentials of the TADF molecules. 4CzIPN-X
E0(R/RC¢) [V][a]
E0(R/RC + ) [V][a]
EL-H [eV][b]
X = Me tBu Ph H[c]
¢1.23 ¢1.21 ¢1.15 ¢1.21
N.D. 1.42 1.41 N.D.
2.99 3.00 2.82 2.91
[a] Values are given versus a standard calomel electrode (SCE). N.D.: not determined. [b] HOMO–LUMO energy gaps were estimated by DFT at the B3LYP/6-31G(d) level. [c] Unsubstituted 4CzIPN. Values are taken from ref. [35].
(¢1.15 V vs. SCE) was slightly shifted to the positive direction compared with those of 4CzIPN-Me (¢1.23 V vs. SCE) and 4CzIPN-tBu (¢1.21 V vs. SCE), whereas the values of E0(R/RC + ) were similar for 4CzIPN-tBu (1.42 V vs. SCE) and 4CzIPN-Ph (1.41 V vs. SCE). Overall, the trend of the difference of E0(R/RC¢) and E0(R/RC + ) for the molecules is consistent with that of the energy gaps between HOMO and LUMO calculated by the DFT method (Table 2). Spin density of radical cations We performed DFT calculations to estimate the spin density of radical cations of the TADF molecules in the doublet state related to the radical stability observed in the cyclic voltammograms. Figure 7 depicts equivalent surfaces for a spin density at 0.004 obtained by DFT calculations. The spin densities of all atoms in the radical cations of the TADF molecules are listed in Table S3 in the Supporting Information. High spin densities corresponding to charge distribution caused by the unpaired electron were mainly distributed on the C(1), C(3), C(6), C(8), and N(9) atoms in the carbazolyl moieties of the TADF molecules. Figure 7 reveals that the spin densities of carbazolyl groups located at the para position with respect to cyano
Figure 7. Atom numbering for carbazole. Equivalent surfaces of the radical cations of the TADF molecules at a spin density of 0.004 calculated at the B3LYP/6-31 + G(d) level for the optimized structure of each doublet state. Chem. Eur. J. 2016, 22, 4889 – 4898
www.chemeurj.org
groups are smaller than those of the other two carbazole groups. The spin densities on the atoms tended to decrease slightly in the presence of methyl and tert-butyl substituents, and were substantially lowered by phenyl groups, which could delocalize electrons through conjugation. The carbazolyl moieties in the TADF molecules are bound to dicyanobenzene, so it is likely that the steric hindrance of this molecular structure can lower the reactivity of the C(1), C(8), and N(9) atoms of the radical ions of the substituted carbazolyl groups. Thus, high spin densities on the C(3) and C(6) atoms are crucial for inducing polymerization. The spin densities estimated by DFT calculations and the stability of the radical cations observed in the anodic scan of cyclic voltammograms indicate that polymerization initiated by the C(3) and C(6) atoms on the carbazolyl moieties was sufficiently inhibited in 4CzIPN-tBu because of steric hindrance arising from the bulky tert-butyl substituents (kinetically), and for 4CzIPN-Ph by steric hindrance and conjugation (thermodynamically) because of its phenyl substituents. Conversely, polymerization was insufficiently inhibited for 4CzIPN-Me. Indeed, the reactivity of the radical cation of 4CzIPN-Me was somewhat lower than that of unsubstituted 4CzIPN, but stabilization was weaker than that of 4CzIPN-tBu and 4CzIPN-Ph, which limited its ability to obtain stable ECL (see the next section).
ECL properties of TADF molecules ECL spectra of the molecules acquired by applying a squarewave voltage (50 Hz) are presented in Figure 8. CH2Cl2 and MeCN, which can dissolve the supporting electrolytes well, were used as solvents for ECL of the TADF molecules to facilitate comparison with the photophysical properties. Yellow to yellowish–red ECL emissions with maximum wavelengths of ECL (lECL) ranging from l = 570 to 610 nm were observed. The ECL spectra were redshifted in polar MeCN compared with their location in CH2Cl2, as observed in the PL spectra. In MeCN, the lECL values for the three molecules agree well with their corresponding lPL values. This agreement indicates that an inner filter effect caused by self-absorption at high concentration ( 1 mm for ECL measurements) is minimal because of the small overlaps between the UV/Vis and PL spectra of the
Figure 8. ECL spectra of the TADF molecules in CH2Cl2 and MeCN generated by applying a square-wave voltage (50 Hz). ECL spectra for 4CzIPN-Me were measured in CH2Cl2 and a mixture of CH2Cl2 and MeCN (50:50 vol %).
4894
Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper molecules (Figure 3). In CH2Cl2, a slight redshift of lECL of about 10 nm compared with the corresponding lPL value was observed for the three TADF molecules. It should be noted that the presence of a supporting electrolyte can shift the PL spectra of D–A molecules in less polar media, such as CH2Cl2.[16, 43] The PL spectra in CH2Cl2 containing 0.1 m TBAPF6 were redshifted and agreed well with the ECL spectra (Figure S2 and Table S4 in the Supporting Information), which indicated that S1 states of the molecules in CH2Cl2 were stabilized by the supporting electrolyte. Meanwhile, in MeCN, no significant shift in the PL spectra by the presence of 0.1 m TBAPF6 was observed. It is important to confirm that ECL spectra coincide with PL spectra because their agreement indicates that there is no ECL emission from other species, such as excimers, exciplexes, or polymers such as di- and trimers. Although polymerization of 4CzIPN-Me was observed in cycling potentials, the agreement between its PL and ECL spectra is reasonable because of the high frequency of the applied switching potentials (50 Hz). Therefore, we can conclude that the emission processes in PL and ECL of the TADF molecules are the same. ECL intensities of the three TADF molecules were compared with that of [Ru(bpy)3]2 + (bpy: 2,2’-bipyridine; fECL = 5 %[47]) as a standard substrate to estimate fECL by applying a train pulse voltage (5 Hz), considering total ECL intensity and charge during the potential step [Eq. (S7) in the Supporting Information]. Estimated fECL values are listed in Table 3. As for fPL, the fECL values depend on solvent polarity, that is, solutions in CH2Cl2 exhibited higher fECL than those in MeCN. It should be emphasized that the fECL values of the three TADF molecules were higher than those of the corresponding 0.25 Õ fPL values, which were considered to be the limit of fECL according to spin statistics. The TADF molecules have small DEst values and the total energies of a pair of radical species ( 2.6 eV estimated with the redox potentials) are energetically high enough to produce S1 (2.0–2.2 eV calculated with lPL) and T1 states; this leads to the generation of S1 and T1 states in a ratio similar to that of the value expected by spin statistics. Thus, these results confirm that thermal spin upconversion from T1 to S1 states occurs in these ECL systems and resulted in delayed ECL through thermal spin upconversion. Among the three TADF molecules, 4CzIPN-tBu showed the highest fECL ; remarkably, the fECL of 4CzIPN-tBu in CH2Cl2 was close to 40 %. Although
fECL should be the same as that of fPL for the TADF molecules because of spin upconversion, it seems that the fECL values for the three molecules are smaller than the corresponding fPL values. A possible explanation for this is quenching by the supporting electrolyte in the ECL system. We confirmed that the PL intensity decreased in the presence of 0.1 m TBAPF6, and the quenching effect depended on the solvents, that is, the decrease in PL intensity induced by 0.1 m TBAPF6 was about 15–30 % in CH2Cl2 and less than 10 % in MeCN. The same trend was observed in the shift of lPL caused by the presence of the supporting electrolyte in the solutions. It was also confirmed that the PL lifetimes of both prompt and delayed components were shortened in the presence of supporting electrolyte, and the effect on PL lifetime was large in CH2Cl2 and small in MeCN (Table S4 in the Supporting Information). Despite some quenching, sufficient delayed PL was observed when each 1 mm solution of TADF contained 0.1 m TBAPF6 (Figure S3 in the Supporting Information). In addition to the high fECL values, especially for 4CzIPN-tBu, the ECL intensities were stable for 4CzIPN-tBu and 4CzIPN-Ph, as shown in Figure 9 (typical ECL intensities and transient ECL responses are shown in Figure S4 in the Supporting Information). Upon applying voltage for more than 50 pulse voltage cycles, the change of ECL intensity was less than 10 % for 4CzIPN-tBu and 4CzIPN-Ph; this change may be ascribed to convective transport. In contrast, the ECL intensity of 4CzIPNMe decreased by 30–40 % during cycling, decreasing quickly during the first several pulse voltages cycles and then decreasing gradually. When a polymer film formed on the Pt electrode during the initial potential cycles, electron transfer between the electrode and 4CzIPN-Me was blocked, so fewer radical ions were produced, decreasing ECL intensity and polymer formation. Thus, after several potential cycles, the polymer film developed only gradually during successive voltages cycles, gradually decreasing the ECL intensity. The stability of ECL intensities is consistent with the stability of the radical cations produced by the electrode reaction (Figure 6). It should be noted that the ECL stability of 4CzIPN-Me was enhanced compared with that of unsubstituted 4CzIPN, the ECL intensity of which decayed by about 80 % under the same cycling conditions (Figure S5 in the Supporting Information).
Table 3. ECL properties of the TADF molecules in different solvents. 4CzIPN-X
Solvent
lECL [nm]
FWHM [nm]
fECL [%][a]
0.25 Õ fPL [%]
X = Me
CH2Cl2 MeCN[b] CH2Cl2 MeCN CH2Cl2 MeCN CH2Cl2 MeCN
585 595 572 587 595 610 555 565
124 132 110 130 121 139 107 122
13 1.5 7.5 0.6 35 2.8 9.2 2.2 13 1.7 3.2 0.5 47 6.0 15 1.6
4.5 3.0 11 2.1 4.3 0.9 14 4.5
tBu Ph H[c]
[a] Determined by using [Ru(bpy)3]2 + in MeCN (fECL = 5 %). FWHM: full width at half maximum of ECL spectra. [b] 50:50 vol % mixture with CH2Cl2. [c] Unsubstituted 4CzIPN from refs. [16] and [35]. Chem. Eur. J. 2016, 22, 4889 – 4898
www.chemeurj.org
4895
Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper
Figure 9. Stability of the ECL intensity of TADF molecules. ECL was generated by applying square-wave voltages 0.1–0.2 V larger than the peak potentials. The potential was first set to generate radical anions for 0.1 s, and then stepped to generate radical cations for 0.1 s. A potential of 0 V was applied for 1 s after each anodic step.
Conclusion We demonstrated ECL with TADF molecules composed of dicyanobenzene and methyl-, tert-butyl-, and phenyl-substituted carbazolyl groups, and examined the electrochemical and photophysical properties of these TADF molecules systematically. In ECL systems, because excited states of the TADF molecules are generated through ion annihilation between their radical ions produced by the electrode reaction directly, not by energy transfer from host molecules to TADF molecules doped in the host molecules, like in common OLED systems, the stability of radical ions is crucial to obtain stable ECL. Herein, stable and efficient ECL was realized by enhancing both the stability of the radical cations of TADF molecules and thermal spin upconversion. In the series of TADF molecules, that with the tert-butyl substituent showed the best performance in terms of both stability and efficiency of ECL because 4CzIPNtBu exhibited a higher fPL value than those of the other molecules, and the bulky tert-butyl substituents inhibited the radical reaction of 4CzIPN-tBu through steric hindrance. Quantum chemical calculations based on DFT and TD-DFT methods revealed that the HOMO–LUMO gaps, excitation energies, and DEst values were strongly affected by phenyl substituents beChem. Eur. J. 2016, 22, 4889 – 4898
www.chemeurj.org
cause of electron conjugation, but less so by methyl and tertbutyl substituents. Overall, electrochemical and photophysical measurements, coupled with theoretical calculations, are very useful to understand the properties of TADF molecules. We believe that designing TADF molecules based on quantum chemical calculations combined with experimental evidence, regarding the stability of their radical ions, is enormously useful for not only ECL systems, but also OLEDs, to establish efficient and stable systems.
Experimental Section Synthesis of TADF molecules Three TADF molecules were synthesized by using a slightly modified literature method (Scheme 1)[24] First, 60 % NaH in oil (0.25 (6.2), 0.40 (10), and 0.40 g (10 mmol) for methyl-, tert-butyl-, and phenyl-substituted carbazoles, respectively) was rinsed with hexane (2 Õ 20 mL) to remove the oil under N2 in a round-bottomed flask. Then, hexane was evaporated under reduced pressure and anhydrous THF (50 mL) was introduced into the flask. 3,6-Dimethylcarbazole (Cz-Me; 1.0 g, 5.1 mmol), 3,6-di-tert-butylcarbazole (Cz-tBu; 2.2 g, 7.9 mmol), or 3,6-diphenylcarbazole (Cz-Ph; 2.5 g, 7.8 mmol) was added to the solution of NaH, which was subse-
4896
Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper Photonics K. K., Japan). PL decay curves were obtained with a time-resolved fluorophotometer (Quantaurus Tau, Hamamatsu Photonics K. K., Hamamatsu, Japan) by irradiation with a lightemitting diode (LED; l = 365 nm). Irradiation light was removed with an optical filter. Prior to measurement of the PL decay curves and fPL values, solutions were deoxygenated by bubbling N2 through them. Scheme 1. Synthesis of TADF molecules.
Quantum chemical calculations quently stirred for 1 h at room temperature. Tetrafluoroisophthalonitrile (0.21 (1.1), 0.33 (1.7), and 0.33 g (1.7 mmol) for methyl-, tertbutyl-, and phenyl-substituted carbazoles, respectively) was then gradually added to the solution at 0 8C. The solution was stirred overnight at room temperature and then for 6 h at 60 8C. A mixture of methanol and water was poured into the flask to quench any remaining NaH, and then further methanol was added to precipitate the target compound. The precipitate was filtered and dried in a desiccator to yield a fluorescent powder. The residue was purified by column chromatography on silica gel with a mixture of hexane and CH2Cl2 as the eluent (50:50!0:100 vol %). The compounds synthesized with Cz-Me, Cz-tBu and Cz-Ph were obtained as fluorescent powders that were yellow, greenish–yellow, and yellowish– red, respectively, after purification. The desired products, 1,2,3,5tetrakis(3,6-dimethylcarbazol-9-yl)-4,6-dicyanobenzene (4CzIP-Me), 1,2,3,5-tetrakis(3,6-di-tert-butylcarbazol-9-yl)-4,6-dicyanobenzene (4CzIPN-tBu), and 1,2,3,5-tetrakis(3,6-diphenylcarbazol-9-yl)-4,6-dicyanobenzene (4CzIPN-Ph) were isolated in yields of 76, 54, and 60 %, respectively. 1
4CzIPN-Me: H NMR (400 MHz, (CD3)2CO): d = 8.02 (s, 2 H), 7.83 (d, J = 8.42 Hz, 2 H), 7.60 (s, 4 H), 7.51 (d, J = 8.42 Hz, 4 H), 7.42 (m, 4 H), 7.23 (s, 2 H), 6.94 (dd, J = 7.32, 1.10 Hz, 4 H), 6.59 (dd, J = 6.59, 1.60 Hz, 2 H), 2.54 (s, 6 H), 2.33 (s, 12 H), 2.17 ppm (s, 6 H); 13C NMR (400 MHz, CDCl3): d = 145.8, 145.0, 144.6, 144.2, 143.4, 138.5, 137.0, 135.5, 134.0, 124.9, 124.4, 124.0, 123.2, 122.0, 117.5, 116.1, 115.3, 114.9, 112.2, 109.7, 108.9, 34.93, 34.54, 34.25 ppm; m/z: 902.45 [M + H] + . 4CzIPN-tBu: 1H NMR (400 MHz, (CD3)2CO): d = 8.40 (d, J = 1.83 Hz, 2 H), 7.94 (d, J = 8.79 Hz, 2 H), 7.87 (d, J = 1.83 Hz, 4 H), 7.53, (dd, J = 10.62, 2.20 Hz, 2 H), 7.43 (m, 6 H), 7.12 (m, 6 H), 6.63 (dd, J = 10.62, 2.20 Hz, 2 H), 1.49 (s, 18 H), 1.31 (s, 36 H), 1.22 ppm (s, 18 H); 13 C NMR (400 MHz, CDCl3): d = 145.8, 145.0, 144.6, 144.3, 143.4, 138.5, 137.0, 135.5, 134.0, 124.9, 124.4, 124.0, 123.2, 122.0, 117.5, 116.1, 115.3, 114.9, 112.2, 109.7, 108.9, 34.93, 34.54, 34.25, 32.00, 37.72 ppm, m/z: 1238.48 [M + H] + . 4CzIPN-Ph: 1H NMR (400 MHz, (CD3)2CO): d = 8.81 (s, 2 H), 8.26, (m, 6 H), 8.09, (dd, J = 10.25, 1.83 Hz, 2 H), 7.91 (d, J = 8.79 Hz, 8 H), 7.82 (d, J = 1.46 Hz, 2 H), 7.60 (m, 18 H), 7.35 (m, 24 H), 7.11 ppm (dd, J = 10.25, 1.83 Hz, 2 H); 13C NMR (400 MHz, CDCl3): d = 141.0, 140.9, 138.5, 136.4, 135.7, 134.8, 128.9, 128.8, 127.6, 127.3, 127.2, 127.1, 127.0, 125.4, 120.1, 119.2, 119.1, 118.4, 116.4, 110.5, 110.3, 110.0 ppm, m/z: 1397.56 [M + H] + .
Characterization and photophysical measurements H and 13C NMR spectra of the synthesized molecules were acquired with a JMN-EPC400 spectrometer (JEOL Ltd., Tokyo, Japan) and MALDI-TOF-MS were measured on an Autoflex III mass spectrometer (Bruker Co., MA). Absorption and PL spectra were collected by using a UV/Vis spectrophotometer (V-560, JASCO, Tokyo, Japan) and spectrofluorometer (FluoroMax-4, Horiba, Kyoto, Japan), respectively. fPL values for the TADF solutions were estimated by using an integrating sphere (Quantaurus QY, Hamamatsu
Quantum chemical calculations based on DFT were performed by using the Gaussian 09 program package.[48] All structures were optimized by using the 6-31G(d) basis set before performing subsequent calculations. Molecular orbitals at the ground state were estimated at the B3LYP/6-31G(d) level, whereas the spin density of radical cations of TADF molecules was calculated by considering a diffuse function at the doublet state by using B3LYP/6-31 + G(d). To estimate DEst, TD-DFT calculations at the M06–2X/6-31 + G(d) level for the optimized structures of the S1 and T1 states were performed by using the same functional. This functional could reasonably explain the absorption spectra of D–A molecules composed of carbazole and dicyanobenzene groups.[24] Computations were mainly performed by using computer facilities (Fujitsu Primergy CX400) at the Research Institute for Information Technology, Kyushu University.
Electrochemical measurements A conventional three-electrode system with a Pt disk (diameter d = 1.6 and 3.0 mm for cyclic voltammetry and ECL measurements, respectively), coiled Pt wire, and Ag wire as working, counter, and reference electrodes, respectively, was employed. The Pt disk was polished by using an alumina suspension (d = 50 nm) before electrochemical measurements. The electrodes were placed in a glass vial containing 1 mm solutions of the TADF molecules with 0.1 m TBAPF6 as a supporting electrolyte. An optical fiber (core diameter: 1 mm) connected to an optical photometer (Maya2000 pro, Ocean Optics Inc., FL) or a photodetector connected to a power meter (model 1918-C, Newport Co., CA) was positioned under the glass vial. To acquire ECL spectra, a square-wave voltage (50 Hz) was applied to the Pt disk electrode at 0.1–0.2 V over the peak potentials observed by cyclic voltammetry. Transient ECL intensities at the maximum ECL wavelength were collected by the photodetector using a triple-potential-step method.[1] For the first 100 ms, the applied potential was set to generate anion radicals, and then the potential was switched to positive for 100 ms (5 Hz) to oxidize the TADF molecules. The external voltage to the electrochemical cell was controlled with an electrochemical workstation (Electrochemical analyzer, 715AN, BAS Inc., Tokyo, Japan). Cyclic voltammograms were recorded by using a positive feedback for IR compensation. The typical solution resistance was about 500 W and uncompensated for resistance was about 5 W. Solutions were deoxygenated by bubbling with N2 before all electrochemical experiments, and electrochemical measurements were carried out under a N2 atmosphere. Solvents were dehydrated by using molecular sieve before preparing solutions.
1
Chem. Eur. J. 2016, 22, 4889 – 4898
www.chemeurj.org
Acknowledgements This research was supported in part by the Japan Society for the Promotion of Science through its “Funding Program for World-Leading Innovative R&D on Science and Technology”. R.I. acknowledges Hiroshi Miyazaki and Hiroko Nomura
4897
Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper (Kyushu University) for valuable discussions about synthesis of the TADF molecules. Keywords: donor–acceptor systems · fluorescence electrochemiluminescence · lifetimes · solvent effects
·
[1] Electrogenerated Chemiluminescence (Ed.: A. J. Bard), Marcel Dekker, New York, 2004. [2] M. M. Richter, Chem. Rev. 2004, 104, 3003 – 3036. [3] W. Miao, Chem. Rev. 2008, 108, 2506 – 2553. [4] K. M. Maness, R. M. Wightman, J. Electroanal. Chem. 1995, 396, 85 – 95. [5] N. R. Armstrong, R. M. Wightman, E. M. Gross, Annu. Rev. Phys. Chem. 2001, 52, 391 – 422. [6] A. Kapturkiewicz, J. Electroanal. Chem. 1991, 302, 131 – 144. [7] A. Kapturkiewicz, Chem. Phys. 1992, 166, 259 – 273. [8] S. W. Feldberg, J. Phys. Chem. 1966, 70, 3928 – 3930. [9] A. Kapturkiewicz, J. Herbich, J. Nowacki, Chem. Phys. Lett. 1997, 275, 355 – 362. [10] M. M. Sartin, C. Shu, A. J. Bard, J. Am. Chem. Soc. 2008, 130, 5354 – 5360. [11] K. M. Omer, S.-Y. Ku, Y.-C. Chen, K.-T. Wong, A. J. Bard, J. Am. Chem. Soc. 2010, 132, 10944 – 10952. [12] M. Shen, J. Rodrguez-Lûpez, J. Huang, Q. Liu, X. H. Zhu, A. J. Bard, J. Am. Chem. Soc. 2010, 132, 13453 – 13461. [13] J. Suk, J. Cheng, K. Wong, A. J. Bard, J. Phys. Chem. C 2011, 115, 14960 – 14968. [14] F. Polo, F. Rizzo, M. Veiga-Gutierrez, L. De Cola, S. Quici, J. Am. Chem. Soc. 2012, 134, 15402 – 15409. [15] H. Qi, Y. H. Chen, C. H. Cheng, A. J. Bard, J. Am. Chem. Soc. 2013, 135, 9041 – 9049. [16] R. Ishimatsu, S. Matsunami, T. Kasahara, J. Mizuno, T. Edura, C. Adachi, K. Nakano, T. Imato, Angew. Chem. Int. Ed. 2014, 53, 6993 – 6996; Angew. Chem. 2014, 126, 7113 – 7116. [17] Z. Mohsan, A. L. Kanibolotsky, A. J. Stewart, A. R. Inigo, L. Dennany, P. J. Skabara, J. Mater. Chem. C 2015, 3, 1166 – 1171. [18] S. Rashidnadimi, T. H. Hung, K.-T. Wong, A. J. Bard, J. Am. Chem. Soc. 2008, 130, 634 – 639. [19] K. Shizu, H. Tanaka, M. Uejima, T. Sato, K. Tanaka, H. Kaji, C. Adachi, J. Phys. Chem. C 2015, 119, 1291 – 1297. [20] K. Shizu, M. Uejima, H. Nomura, T. Sato, K. Tanaka, H. Kaji, C. Adachi, Phys. Rev. Appl. 2015, 3, 014001-1-7. [21] C. A. Parker, Photoluminescence of Solutions, Elsevier, Amsterdam, 1968. [22] T. Nakagawa, S.-Y. Ku, K.-T. Wong, C. Adachi, Chem. Commun. 2012, 48, 9580 – 9582. [23] H. Tanaka, K. Shizu, H. Miyazaki, C. Adachi, Chem. Commun. 2012, 48, 11392 – 11394. [24] H. Uoyama, K. Goushi, K. Shizu, H. Nomura, C. Adachi, Nature 2012, 492, 234 – 238. [25] S.-Y. Lee, T. Yasuda, H. Nomura, C. Adachi, Appl. Phys. Lett. 2012, 101, 093306-1-4. [26] Q. Zhang, J. Li, K. Shizu, S. Huang, S. Hirata, H. Miyazaki, C. Adachi, J. Am. Chem. Soc. 2012, 134, 14706 – 14709. [27] F. B. Dias, K. N. Bourdakos, V. Jankus, K. C. Moss, K. T. Kamtekar, V. Bhalla, J. Santos, M. R. Bryce, A. P. Monkman, Adv. Mater. 2013, 25, 3707 – 3714.
Chem. Eur. J. 2016, 22, 4889 – 4898
www.chemeurj.org
[28] H. Nakanotani, K. Masui, J. Nishide, T. Shibata, C. Adachi, Sci. Rep. 2013, 3, 2127-1-5. [29] K. Sato, K. Shizu, K. Yoshimura, A. Kawada, H. Miyazaki, C. Adachi, Phys. Rev. Lett. 2013, 110, 247401-1-5. [30] Y. Im, J. Y. Lee, Chem. Mater. 2014, 26, 1413 – 1419. [31] H. Wang, L. Xie, Q. Peng, L. Meng, Y. Wang, Y. Yi, P. Wang, Adv. Mater. 2014, 26, 5198 – 5204. [32] Q. Zhang, H. Kuwabara, W. J. Potscavage, S. Huang, Y. Hatae, T. Shibata, C. Adachi, J. Am. Chem. Soc. 2014, 136, 18070 – 18081. [33] J. Lee, K. Shizu, H. Tanaka, H. Nakanotani, T. Yasuda, H. Kaji, C. Adachi, J. Mater. Chem. C 2015, 3, 2175 – 2181. [34] Y. Sakai, Y. Sagara, H. Nomura, N. Nakamura, Y. Suzuki, H. Miyazaki, C. Adachi, Chem. Commun. 2015, 51, 3181 – 3184. [35] R. Ishimatsu, S. Matsunami, K. Shizu, C. Adachi, K. Nakano, T. Imato, J. Phys. Chem. A 2013, 117, 5607 – 5612. [36] H. C. Moon, T. P. Lodge, C. D. Frisbie, J. Am. Chem. Soc. 2014, 136, 3705 – 3712. [37] T. Kasahara, S. Matsunami, T. Edura, R. Ishimatsu, J. Oshima, M. Tsuwaki, T. Imato, S. Shoji, C. Adachi, J. Mizuno, Sens. Actuators B 2015, 207, 481 – 489. [38] T. Nobeshima, K. Nakamura, N. Kobayashi, J. Photopolym. Sci. Technol. 2013, 26, 397 – 402. [39] X. Jiang, X. Yang, C. Zhao, K. Jin, L. Sun, J. Phys. Chem. C 2007, 111, 9595 – 9602. [40] J. W. Oh, Y. O. Lee, T. H. Kim, K. C. Ko, J. Y. Lee, H. Kim, J. S. Kim, Angew. Chem. Int. Ed. 2009, 48, 2522 – 2524; Angew. Chem. 2009, 121, 2560 – 2562. [41] A. Kapturkiewicz, J. Herbich, J. Karpiuk, J. Nawacki, J. Phys. Chem. A 1997, 101, 2332 – 2344. [42] W. Rettig, M. Zander, Chem. Phys. Lett. 1982, 87, 229 – 234. [43] P. Natarajan, M. Schmittel, J. Org. Chem. 2013, 78, 10383 – 10394. [44] K. Goushi, K. Yoshida, K. Sato, C. Adachi, Nat. Photonics 2012, 6, 253 – 258. [45] P. Marrec, C. Dano, N. Gueguen-Simonet, J. Simonet, Synth. Mater. 1997, 89, 171 – 179. [46] A. J. Bard, L. R. Faulkner, Electrochemical Methods: Fundamentals and Applications, 2nd ed.,Wiley, Hoboken, 2001. [47] W. L. Wallace, A. J. Bard, J. Phys. Chem. 1979, 83, 1350 – 1357. [48] Gaussian 09, Revision C.01, M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery, Jr., J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, T. Keith, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, A. D. Daniels, O. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski, D. J. Fox, Gaussian, Wallingford CT, 2010. Received: January 8, 2016 Published online on February 24, 2016
4898
Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
