Letter pubs.acs.org/JPCL
Self-Assembly of Supramolecular Light-Harvesting Arrays from Symmetric Perylene-3,4-dicarboximide Trefoils Kelly M. Lefler, Chul Hoon Kim, Yi-Lin Wu, and Michael R. Wasielewski* Department of Chemistry and Argonne-Northwestern Solar Energy Research (ANSER) Center, Northwestern University, 2145 North Sheridan Road, Evanston, Illinois 60208-3113, United States S Supporting Information *
ABSTRACT: Unlike the widely studied perylene-3,4:9,10-bis(dicarboximide) (PDI) dyes, self-assembly of the corresponding perylene-3,4-dicarboximide (PMI) dyes into large arrays and studies of their excited state properties have received far less attention. Two symmetric PMI trefoils were synthesized by connecting the 9-position of the perylene core either directly (1) or through a phenylene linker (2) to the 1,3,5-positions of a central benzene ring. Synchrotron-based small- and wide-angle X-ray scattering measurements in methylcyclohexane show that trefoil 1 self-assembles into cofacial trimers (13) on average, while trefoil 2 forms much larger assemblies that are tridecamers (213) on average. Their photophysics were characterized using steady-state as well as transient absorption and emission spectroscopy. Time-resolved spectroscopy reveals that both 13 and 213 initially form excitonically coupled excited states that subsequently relax to excimer states having 20 and 8.4 ns lifetimes, respectively, which decay to ground-state primarily nonradiatively. The data are consistent with stronger electronic coupling between the PMI molecules in 213 relative to 13. SECTION: Spectroscopy, Photochemistry, and Excited States
I
The self-assembly characteristics and resulting photophysical behavior of a number of simple trefoil chromophore arrays have been explored. Specifically, three-fold symmetric molecules with a 1,3,5-triphenylbenzene core have been used to study energy and electron transfer in PDIs31−33 as well as chlorophylls.34−36 In the case of the PDI-based system, solution phase small- and wide-angle X-ray scattering (SAXS/WAXS) experiments in toluene showed that the PDI trefoil monomers self-assemble into π-stacked dimers.31 Single-molecule spectroscopy of the monomer shows that individual PDIs within the molecule are excitonically coupled to the other PDIs within the monomer but that this interaction is highly influenced by the local environment.33 Here we present data on two symmetric 2,5-di-n-dodecylPMI trefoils (Scheme 1), in which PMI is attached through the nine-position to either the 1,3,5-positions of a benzene ring (1) or the 4-positions of the phenyl rings in 1,3,5-triphenylbenzene (2). We examine monomeric 1 and 2 in THF and the selfassembly of these molecules in methylcyclohexane (MCH) and compare the excited singlet-state properties of the monomers and the supramolecular assemblies. N-(2-Ethylhexyl)PMI (P1, Scheme 1) was synthesized via one-pot imide condensation and monodecarboxylation of perylene-3,4:9,10-bis(dicarboxyanhydride) with 2-ethylhexylamine, zinc(II) acetate monohydrate, imidazole, and water in a high-pressure autoclave.9 Molecule P1 was subjected to catalytic C−H insertion of 1-dodecene using
mitation of the economic design seen in photosynthetic proteins is a key strategy for solar energy conversion, where well-ordered chromophore assemblies carry out the initial steps of photosynthesis, including efficient light harvesting as well as energy and charge transfer over large distances.1 Understanding the energy and electron-transfer properties of multichromophore systems through photophysical studies of covalent and noncovalent self-assembled systems is necessary to inform the rational design of molecular building blocks for light harvesting in artificial photosynthetic systems.2−4 Perylene-3,4-dicarboximide (PMI) has generated significant interest as a sensitizer in dye-sensitized solar cells.5−7 It is thermally and photochemically stable,8 absorbs light strongly in the visible region,9 and has redox potentials (EOX = 1.4 V vs SCE and ERED = −0.9 V vs SCE) that make it an attractive electron donor or acceptor.10 In addition, it can be selectively functionalized at the 9,11 1,6,12,13 and 2,5 positions14−16 as well as the imide position17 to tune its electronic properties. Yet, in contrast with the more widely studied perylene-3,4:9,10bis(dicarboximide) (PDI),18−20 there have been relatively few studies on the electronic interactions between PMI chromophores in self-assembled systems,21−23 even though PMI has been incorporated into several covalent donor−acceptor systems for energy and electron-transfer studies.24−30 PMI may offer interesting self-assembly opportunities unavailable with PDI derivatives because PMI has a 6.4 D ground-state dipole moment directed along the long axis of the molecule, which could, in a suitably aligned self-assembled PMI nanostructure, enhance the bulk dipolar nature of the structure.23 © 2014 American Chemical Society
Received: March 31, 2014 Accepted: April 21, 2014 Published: April 21, 2014 1608
dx.doi.org/10.1021/jz500626g | J. Phys. Chem. Lett. 2014, 5, 1608−1615
The Journal of Physical Chemistry Letters
Letter
Scheme 1. Synthesis of 1 and 2a
a Reagents and conditions: a. 2-ethyl-1-hexylamine, imidazole, H2O, Zn(OAc)2·H2O, 190 °C, in a sealed autoclave, 24 h, 49%; b. 1-dodecene, Ru(H2)CO(PPh3)3, mesitylene, reflux, 2 d, 91%; c. Br2, CH2Cl2, reflux, 2 h, 81%; d. bis(pinacolato)diboron, potassium acetate, Pd(dppf)Cl2· CH2Cl2, 1,4-dioxane, reflux, 14 h, 85%; e. Na2CO3, Pd(PPh3)4, H2O, EtOH, toluene, 80 °C, 2 d, 71% for 1 and 34% for 2.
in Figure 1 show semilogarithmic plots of scattering intensity versus q2 that are fit using the Guinier relationship43
carbonyldihydridotris(triphenylphosphine)ruthenium in mesitylene to give P2.14,15 Selective bromination at the nineposition with bromine in refluxing dichloromethane afforded P3.12 P3 was used in a three-fold Suzuki−Miyaura reaction with 1,3,5-tris(pinacolatoboron)benzene37 using tetrakis(triphenylphosphine)palladium(0) and sodium carbonate in a toluene, ethanol, and water mixture to give 1.38 P3 was also converted to its boronic ester, P4, with bis(pinacolato)diboron, [1,1′-bis(diphenylphosphino)ferrocene]dichloropalladium(II), and potassium acetate in dioxane.12 P4 was used in a three-fold Suzuki-Miyaura cross-coupling reaction with 1,3,5-tris(4bromophenyl)benzene39 using the same conditions previously described to afford 2.38 SAXS/WAXS measurements in solution using a synchrotron X-ray source are a powerful tool for studying self-assembling chromophoric structures.2,23,40−42 Scattering data for 1 and 2 in MCH are shown in Figure 1A,B, respectively, as semilogarithmic plots of scattering intensity versus q. The insets
I(q) = Io exp( −q2R g 2/3)
(1)
where I0 is the forward scattering amplitude, Rg is the radius of gyration, and q = (4π/λ) sin θ, where λ is the X-ray scattering wavelength and 2θ is the scattering angle. Fits of the scattering data in MCH show that 1 and 2 each show a linear Guinier relationship, which indicates that the size dispersity of the assembly is fairly narrow.43,44 The radii of gyration obtained from these fits for 1 and 2 are Rg = 12.9 ± 0.2 and 21.9 ± 0.2 Å, respectively. Given that linear Guinier regions were found for each molecule, Fourier transforms were performed on the scattering data to generate pair distribution functions (PDFs).45 Energyminimized models of hypothetical supramolecular structures were generated using MM+ force-field calculations.40,41 PDFs 1609
dx.doi.org/10.1021/jz500626g | J. Phys. Chem. Lett. 2014, 5, 1608−1615
The Journal of Physical Chemistry Letters
Letter
Figure 1. (A,B): Scattering data for 1 and 2, respectively, in MCH (10−4 to 10−3 M). Insets: low-q region with Guinier fit (red). (C,D): Pair distribution functions (PDFs) for 1 and 2, respectively.
Figure 2. Structures of best-fit models to SAXS/WAXS data. Top (A) and side (B) view of cofacial trimer of 1. Top (C) and side (D) view of cofacial tridecamer of 2. Hydrogens are omitted for clarity.
1610
dx.doi.org/10.1021/jz500626g | J. Phys. Chem. Lett. 2014, 5, 1608−1615
The Journal of Physical Chemistry Letters
Letter
Figure 3. Steady state absorption and emission spectra of 10−4−10−3 M (A)1 and (B) 2 in THF (black) and MCH (red). Solid lines: UV−vis spectra. Dashed lines: fluorescence spectra using 470 nm excitation.
Table 1. Photophysical Data at 295 K steady-state data molecule
solvent
λmax,abs (nm)
λmax em (nm)
1 2 13 213
THF THF MCH MCH
513 486 482 462
557 567 572 573
ΦF
τ1 (ns)
τ2 (ns)
τ3 (ns)
τ4 (ns)
0.90 0.78 0.04 0.05
5.0 ± 0.4b,c 5.6 ± 0.1b,c 0.045 ± 0.010b 0.014 ± 0.002b
3.7 ± 0.2c (λ = 555 nm) 3.7 ± 0.2c (λ = 555 nm)
23 ± 1c (λ = 610 nm) 0.62 ± 0.11b
8.4 ± 0.2c (λ = 610 nm)
Fluorescence spectra following 470 nm excitation. Determined from fsTA using λexc = 532 nm, except for 2 in THF, where λexc = 525 nm. Emission lifetimes determined following 390 nm, 25 fs excitation.
a c
excited-state kinetics a
b
and its slightly weaker (0,1) vibronic band absorbs at 486 nm, while for 13 in MCH the intensities of these bands invert and shift slightly to 517 and 482 nm, respectively. The absorption spectrum of monomeric 2 in THF (Figure 3B) is very similar to that of monomeric 1 and displays a 509 nm (0,0) band and a slightly enhanced (0,1) band at 486 nm. In contrast, the spectrum of 213 in MCH shows a dramatic overall blue shift with the band maximum occurring at 462 nm and a much less intense peak at 522 nm. The ground-state spectra of monomeric 1 and 2 in THF are red-shifted relative to that of PMI itself,9,10,30 due in part to the 9-phenyl substitution on the PMI extending the conjugation. In contrast, the 2,5-di-n-dodecyl substituents on PMI have little effect on the electronic spectra of the chromophore.15 However, it is well known that the excited-state properties of π-stacked chromophore assemblies, such as 13 and 213, depend strongly on the relative orientations of the stacked chromophores.47−49 The inversion of vibronic band intensities for 13 and 213 relative to those of monomeric 1 and 2 results from the excitonic coupling of the transition dipole moments of neighboring PMI chromophores, splitting the lowest excited singlet state into an upper and a lower exciton state.50 Because the transition dipole moment of PMI for the S0 → S1 transition lies parallel the long molecular axis,23 when the π systems of two or more PMIs are cofacial with parallel transition dipole moments the transition to the upper exciton state is allowed, corresponding to an apparent inversion of the vibronic band intensities (H-aggregate formation). The energy difference between the (0,0) transition to the upper exciton state and the (0,1) transition to the forbidden lower exciton state (the lower energy vibronic peak present in these four spectra) reveals information about the strength of the exciton interaction in these assemblies.23 The energy difference between these transitions in 213 (∼2400 cm−1) is nearly double that of 13
of these model structures were calculated and compared with the experimental PDFs for each molecule. Figure 1C,D compares the experimental PDFs with those for model structures that provide the best fit to each set of experimental data. The experimental PDF of 1 is best fit with a cofacial trimer (13) structure (Figure 2), while that of 2 is best fit with a large cofacial tridecameric (213). Several alternative linear structures were modeled (Figures S1 and S2 in the Supporting Information), but the computed PDFs are inconsistent with the experimental data. The SAXS/WAXS data for these molecules demonstrate several effects of monomer structure on self-assembly. While the experimental PDFs of 13 and 213 are modeled reasonably well by the structures shown in Figure 2, it is important to note that in the absence of rigid linkers the association of these molecules in solution is a dynamic process;46 equilibrium likely exists between the best fitting structure and a narrow distribution of similarly sized structures as well as a small amount of monomer. The structural models show that the lack of a phenyl spacer between the PMIs and the central benzene in 1 forces the PMIs to be canted out of plane relative to the benzene core by ∼60°, compared with ∼15° for 2. Thus, the “pin-wheel” structure of 1 prevents the PMI π systems from adopting the extended π-stacking demonstrated by 2 in MCH. The dodecyl side groups at the 2,5-positions are angled away from the center of the PMI so that their potential steric effects are mitigated. Because the SAXS/WAXS experiments show that 13 and 213 form in MCH at concentrations of 10−4 to 10−3 M, the optical experiments were performed at similar concentrations, so that their photophysical behavior can be directly related to their supramolecular structures. Figure 3A compares the UV−vis spectra of monomeric 1 in THF and 13 in MCH. The spectrum of 1 in THF shows that the PMI (0,0) band absorbs at 513 nm 1611
dx.doi.org/10.1021/jz500626g | J. Phys. Chem. Lett. 2014, 5, 1608−1615
The Journal of Physical Chemistry Letters
Letter
Figure 4. Femtosecond transient absorption spectra following 1 μJ/pulse, 110 fs pulse excitation: 1 (A) in THF (1 × 10−4 M) and (B) in MCH (6 × 10−4 M), 532 nm; 2 (C) in THF (1 × 10−4 M), 525 nm and (D) MCH (1 × 10−3 M), 532 nm. Residual laser scatter was removed from the spectra at 514−536 (C) and 518−550 nm (D). Insets: Principal components of the spectra from SVD-global analysis.
(∼1300 cm−1), which suggests significantly increased exciton coupling in 213. The higher ratios of the allowed to the forbidden exciton transitions in 213 compared with 13 also indicate that the overall parallel alignment of the transition dipole moments is better in 213 than in 13. Thus, the observed spectra of 13 and 213 in MCH are consistent with the SAXS/ WAXS data, showing that the PMI molecules of the trefoil πstack upon those of a neighboring trefoil, adopting an approximately cofacial geometry. The corresponding fluorescence emission spectra of 1 and 2 in THF and 13 and 213 in MCH are also shown in Figure 3, and their fluorescence quantum yields are presented in Table 1. In THF, monomeric 1 and 2 display similar spectra, with emission maxima at 557 and 567 nm and lower energy vibronic shoulders at 585 and 596 nm, respectively. The fluorescence quantum yields are all quenched somewhat relative to that of PMI itself (Table 1) but nevertheless remain large. In MCH, the fluorescence spectra of 13 and 213 are broadened, and their maxima are both red-shifted to 573 nm, while their fluorescence quantum yields are both strongly quenched (ΦF < 0.10). These changes in the steady-state emission spectra are indicative of excimer formation,47 which will be further elucidated using time-resolved spectroscopy. Molecules 1 and 2 were photoexcited with 525 or 532 nm, 1 μJ, 110 fs laser pulses. Transient absorption spectra of 1 and 2 in THF and 13 and 213 in MCH are presented in Figure 4. The spectra of monomeric 1 and 2 in THF show similar features and temporal evolution (Figure 4A,C) comprising an initial ground-state bleach at 450−525 nm, stimulated emission at 525−600 nm, and excited singlet state absorption at 600−800 nm. These transient spectral features decay with time constants of τ = 5.0 and 5.6 ns for 1 and 2, as determined by SVD and
global analysis of the 3-D ΔA versus time and wavelength data set (Figure 4, insets). The excited-state decay kinetics are laserpower-independent. In MCH, the transient spectra of 13 show no stimulated emission, and the initial 1*PMI absorption at 719 nm blue shifts to 668 nm in τ = 45 ± 10 ps (Figure 4B). This new excited-state absorption is broader than the corresponding absorption of 1 in THF and does not decay within the maximum pump−probe delay time (∼6 ns) of our fsTA apparatus. The 51 nm shift following photoexcitation is attributed to a structural change resulting in excimer formation. The corresponding transient spectra of 213 in MCH show no stimulated emission and display a 1*PMI excited-state absorption that appears initially at 728 nm, then blue shifts first to 720 nm in τ1 = 14 ± 2 ps and finally to 645 nm in τ2 = 620 ± 110 ps (Figure 4D). The 645 nm absorption band remains for τ ≫ 6 ns. Once again, the excited-state decay kinetics are laser-power-independent. The initial 8 nm spectral shift most likely results from vibrational or structural relaxation, while the second 75 nm blue shift is attributed to relative slow excimer-state formation. Time-resolved fluorescence (TRF) spectroscopy provides an additional check on the excited-state lifetimes determined by transient absorption measurements. The fluorescence lifetimes of monomeric of 1 and 2 in THF agree well with those measured by fsTA spectroscopy, yielding τ = 5.0 ± 0.4 and 5.6 ± 0.1 ns, respectively. In MCH, 13 and 213 both display wavelength-dependent biexponential decays, with the shorter of the two components, τ = 3.7 ± 0.2 ns for both 13 and 213, corresponding to the monomer excited singlet-state decay, while the longer decay occurs in τ = 23 ± 1 ns for 13 and τ = 8.4 ± 0.2 ns for 213. The transient fluorescence spectra at early 1612
dx.doi.org/10.1021/jz500626g | J. Phys. Chem. Lett. 2014, 5, 1608−1615
The Journal of Physical Chemistry Letters
Letter
times are consistent with the steady-state fluorescence spectra of monomeric 1 and 2 in THF (Figures S3 to S4 in the Supporting Information). It is likely that this emission results from the small, highly fluorescent monomeric population of 1 and 2 that is in equilibrium with the aggregates. In contrast, the spectra at long times show broad, featureless, red-shifted emission consistent with an excimer state.47,48,51 Because the instrument response of the TRF instrument using the 100 ns time window is 1.5 ns, the excimer formation dynamics were not observed. Excimer-state lifetimes in excess of 20 ns like those exhibited by 213 have been observed previously in PDI Haggregates and related covalent dimers and trimers.47 Thus, the excimer-state lifetimes of 13 and 213 align well with previously studied PDI molecules that have favorable conformations for excimer formation. The excimer quantum yield of 213 is much smaller than that of 13, which is likely due to the significant initial excited-state population (∼33% for both) that decays by vibrational or structural relaxation. However, this observation does not conflict with the conclusion that 213 shows increased exciton coupling compared with 13 because exciton coupling is a dipole−dipole interaction, while excimer-state formation depends on orbital overlap. In summary, we have synthesized two symmetric PMI-based trefoils around a benzene core. SAXS/WAXS experiments show that steric constraints resulting from the direct attachment of the PMI chromophores to the benzene core in 1 limit its πstacked assembly on average to a trimer in MCH. Lengthening the distance between the PMI chromophores and the benzene core by adding a phenyl spacer, which also allows the PMI chromophores to adopt a more planar conformation relative to one another in 2, results in the formation of on average a tridecamer in MCH. Steady-state optical absorption spectra show that the more planar PMI−PMI conformation in 2 results in stronger exciton coupling in 213 than in 13. Long-lived excimer states are produced in both aggregates and decay largely nonradiatively. Future studies will explore charge transport through these different types of π-stacked PMI assemblies.
ments were made with 470 nm excitation on a PTI QuantaMaster 1 single-photon-counting fluorimeter. Measurements at 10−5 to 10−4 M in THF were performed in a right angle configuration with a 10 mm quartz cuvette, and measurements in MCH were performed at 10−4 to 10−3 M in a 2 mm cuvette using a front-face detection geometry. Quantum yield measurements were performed with a Rhodamine 6G standard in EtOH in a 10 mm cuvette and a 2 mm cuvette for comparison with the THF and MCH samples, respectively. Femtosecond transient absorption (fsTA) measurements for 1 and 2 were made using a regeneratively amplified Ti:sapphire laser system operated at a 2 kHz repetition rate previously described.53 The 110 fs, 1 μJ, 525 or 532 nm pump pulses were focused to a 0.5 mm diameter spot and overlapped with the white-light continuum probe pulse. The total instrument response was 150 fs. Five-second averaging was used to collect transient spectra. In MCH, samples of 1 and 2 had an absorbance of ∼0.4 to 0.6 at the excitation wavelength in 1 mm quartz cuvettes. In THF, all samples had an absorbance of 0.3 to 0.5 in a 2 mm quartz cuvette. Single-wavelength kinetic analyses were performed at several wavelengths using a nonlinear least-squares fit to a sum of exponentials convoluted with a Gaussian instrument response function (IRF). Singular value decomposition (SVD) of the 3-D data set was carried out using Surface Xplorer 2.0 (Ultrafast Systems).54 Picosecond TRF measurements were made using a lab-built, frequencydoubled, cavity-dumped Ti:sapphire laser system that produced 390 nm, 25 fs laser pulses at an 820 kHz repetition rate.55 The samples were excited with
Self-Assembly of Supramolecular Light-Harvesting Arrays from Symmetric Perylene-3,4-dicarboximide Trefoils Kelly M. Lefler, Chul Hoon Kim, Yi-Lin Wu, and Michael R. Wasielewski* Department of Chemistry and Argonne-Northwestern Solar Energy Research (ANSER) Center, Northwestern University, 2145 North Sheridan Road, Evanston, Illinois 60208-3113, United States S Supporting Information *
ABSTRACT: Unlike the widely studied perylene-3,4:9,10-bis(dicarboximide) (PDI) dyes, self-assembly of the corresponding perylene-3,4-dicarboximide (PMI) dyes into large arrays and studies of their excited state properties have received far less attention. Two symmetric PMI trefoils were synthesized by connecting the 9-position of the perylene core either directly (1) or through a phenylene linker (2) to the 1,3,5-positions of a central benzene ring. Synchrotron-based small- and wide-angle X-ray scattering measurements in methylcyclohexane show that trefoil 1 self-assembles into cofacial trimers (13) on average, while trefoil 2 forms much larger assemblies that are tridecamers (213) on average. Their photophysics were characterized using steady-state as well as transient absorption and emission spectroscopy. Time-resolved spectroscopy reveals that both 13 and 213 initially form excitonically coupled excited states that subsequently relax to excimer states having 20 and 8.4 ns lifetimes, respectively, which decay to ground-state primarily nonradiatively. The data are consistent with stronger electronic coupling between the PMI molecules in 213 relative to 13. SECTION: Spectroscopy, Photochemistry, and Excited States
I
The self-assembly characteristics and resulting photophysical behavior of a number of simple trefoil chromophore arrays have been explored. Specifically, three-fold symmetric molecules with a 1,3,5-triphenylbenzene core have been used to study energy and electron transfer in PDIs31−33 as well as chlorophylls.34−36 In the case of the PDI-based system, solution phase small- and wide-angle X-ray scattering (SAXS/WAXS) experiments in toluene showed that the PDI trefoil monomers self-assemble into π-stacked dimers.31 Single-molecule spectroscopy of the monomer shows that individual PDIs within the molecule are excitonically coupled to the other PDIs within the monomer but that this interaction is highly influenced by the local environment.33 Here we present data on two symmetric 2,5-di-n-dodecylPMI trefoils (Scheme 1), in which PMI is attached through the nine-position to either the 1,3,5-positions of a benzene ring (1) or the 4-positions of the phenyl rings in 1,3,5-triphenylbenzene (2). We examine monomeric 1 and 2 in THF and the selfassembly of these molecules in methylcyclohexane (MCH) and compare the excited singlet-state properties of the monomers and the supramolecular assemblies. N-(2-Ethylhexyl)PMI (P1, Scheme 1) was synthesized via one-pot imide condensation and monodecarboxylation of perylene-3,4:9,10-bis(dicarboxyanhydride) with 2-ethylhexylamine, zinc(II) acetate monohydrate, imidazole, and water in a high-pressure autoclave.9 Molecule P1 was subjected to catalytic C−H insertion of 1-dodecene using
mitation of the economic design seen in photosynthetic proteins is a key strategy for solar energy conversion, where well-ordered chromophore assemblies carry out the initial steps of photosynthesis, including efficient light harvesting as well as energy and charge transfer over large distances.1 Understanding the energy and electron-transfer properties of multichromophore systems through photophysical studies of covalent and noncovalent self-assembled systems is necessary to inform the rational design of molecular building blocks for light harvesting in artificial photosynthetic systems.2−4 Perylene-3,4-dicarboximide (PMI) has generated significant interest as a sensitizer in dye-sensitized solar cells.5−7 It is thermally and photochemically stable,8 absorbs light strongly in the visible region,9 and has redox potentials (EOX = 1.4 V vs SCE and ERED = −0.9 V vs SCE) that make it an attractive electron donor or acceptor.10 In addition, it can be selectively functionalized at the 9,11 1,6,12,13 and 2,5 positions14−16 as well as the imide position17 to tune its electronic properties. Yet, in contrast with the more widely studied perylene-3,4:9,10bis(dicarboximide) (PDI),18−20 there have been relatively few studies on the electronic interactions between PMI chromophores in self-assembled systems,21−23 even though PMI has been incorporated into several covalent donor−acceptor systems for energy and electron-transfer studies.24−30 PMI may offer interesting self-assembly opportunities unavailable with PDI derivatives because PMI has a 6.4 D ground-state dipole moment directed along the long axis of the molecule, which could, in a suitably aligned self-assembled PMI nanostructure, enhance the bulk dipolar nature of the structure.23 © 2014 American Chemical Society
Received: March 31, 2014 Accepted: April 21, 2014 Published: April 21, 2014 1608
dx.doi.org/10.1021/jz500626g | J. Phys. Chem. Lett. 2014, 5, 1608−1615
The Journal of Physical Chemistry Letters
Letter
Scheme 1. Synthesis of 1 and 2a
a Reagents and conditions: a. 2-ethyl-1-hexylamine, imidazole, H2O, Zn(OAc)2·H2O, 190 °C, in a sealed autoclave, 24 h, 49%; b. 1-dodecene, Ru(H2)CO(PPh3)3, mesitylene, reflux, 2 d, 91%; c. Br2, CH2Cl2, reflux, 2 h, 81%; d. bis(pinacolato)diboron, potassium acetate, Pd(dppf)Cl2· CH2Cl2, 1,4-dioxane, reflux, 14 h, 85%; e. Na2CO3, Pd(PPh3)4, H2O, EtOH, toluene, 80 °C, 2 d, 71% for 1 and 34% for 2.
in Figure 1 show semilogarithmic plots of scattering intensity versus q2 that are fit using the Guinier relationship43
carbonyldihydridotris(triphenylphosphine)ruthenium in mesitylene to give P2.14,15 Selective bromination at the nineposition with bromine in refluxing dichloromethane afforded P3.12 P3 was used in a three-fold Suzuki−Miyaura reaction with 1,3,5-tris(pinacolatoboron)benzene37 using tetrakis(triphenylphosphine)palladium(0) and sodium carbonate in a toluene, ethanol, and water mixture to give 1.38 P3 was also converted to its boronic ester, P4, with bis(pinacolato)diboron, [1,1′-bis(diphenylphosphino)ferrocene]dichloropalladium(II), and potassium acetate in dioxane.12 P4 was used in a three-fold Suzuki-Miyaura cross-coupling reaction with 1,3,5-tris(4bromophenyl)benzene39 using the same conditions previously described to afford 2.38 SAXS/WAXS measurements in solution using a synchrotron X-ray source are a powerful tool for studying self-assembling chromophoric structures.2,23,40−42 Scattering data for 1 and 2 in MCH are shown in Figure 1A,B, respectively, as semilogarithmic plots of scattering intensity versus q. The insets
I(q) = Io exp( −q2R g 2/3)
(1)
where I0 is the forward scattering amplitude, Rg is the radius of gyration, and q = (4π/λ) sin θ, where λ is the X-ray scattering wavelength and 2θ is the scattering angle. Fits of the scattering data in MCH show that 1 and 2 each show a linear Guinier relationship, which indicates that the size dispersity of the assembly is fairly narrow.43,44 The radii of gyration obtained from these fits for 1 and 2 are Rg = 12.9 ± 0.2 and 21.9 ± 0.2 Å, respectively. Given that linear Guinier regions were found for each molecule, Fourier transforms were performed on the scattering data to generate pair distribution functions (PDFs).45 Energyminimized models of hypothetical supramolecular structures were generated using MM+ force-field calculations.40,41 PDFs 1609
dx.doi.org/10.1021/jz500626g | J. Phys. Chem. Lett. 2014, 5, 1608−1615
The Journal of Physical Chemistry Letters
Letter
Figure 1. (A,B): Scattering data for 1 and 2, respectively, in MCH (10−4 to 10−3 M). Insets: low-q region with Guinier fit (red). (C,D): Pair distribution functions (PDFs) for 1 and 2, respectively.
Figure 2. Structures of best-fit models to SAXS/WAXS data. Top (A) and side (B) view of cofacial trimer of 1. Top (C) and side (D) view of cofacial tridecamer of 2. Hydrogens are omitted for clarity.
1610
dx.doi.org/10.1021/jz500626g | J. Phys. Chem. Lett. 2014, 5, 1608−1615
The Journal of Physical Chemistry Letters
Letter
Figure 3. Steady state absorption and emission spectra of 10−4−10−3 M (A)1 and (B) 2 in THF (black) and MCH (red). Solid lines: UV−vis spectra. Dashed lines: fluorescence spectra using 470 nm excitation.
Table 1. Photophysical Data at 295 K steady-state data molecule
solvent
λmax,abs (nm)
λmax em (nm)
1 2 13 213
THF THF MCH MCH
513 486 482 462
557 567 572 573
ΦF
τ1 (ns)
τ2 (ns)
τ3 (ns)
τ4 (ns)
0.90 0.78 0.04 0.05
5.0 ± 0.4b,c 5.6 ± 0.1b,c 0.045 ± 0.010b 0.014 ± 0.002b
3.7 ± 0.2c (λ = 555 nm) 3.7 ± 0.2c (λ = 555 nm)
23 ± 1c (λ = 610 nm) 0.62 ± 0.11b
8.4 ± 0.2c (λ = 610 nm)
Fluorescence spectra following 470 nm excitation. Determined from fsTA using λexc = 532 nm, except for 2 in THF, where λexc = 525 nm. Emission lifetimes determined following 390 nm, 25 fs excitation.
a c
excited-state kinetics a
b
and its slightly weaker (0,1) vibronic band absorbs at 486 nm, while for 13 in MCH the intensities of these bands invert and shift slightly to 517 and 482 nm, respectively. The absorption spectrum of monomeric 2 in THF (Figure 3B) is very similar to that of monomeric 1 and displays a 509 nm (0,0) band and a slightly enhanced (0,1) band at 486 nm. In contrast, the spectrum of 213 in MCH shows a dramatic overall blue shift with the band maximum occurring at 462 nm and a much less intense peak at 522 nm. The ground-state spectra of monomeric 1 and 2 in THF are red-shifted relative to that of PMI itself,9,10,30 due in part to the 9-phenyl substitution on the PMI extending the conjugation. In contrast, the 2,5-di-n-dodecyl substituents on PMI have little effect on the electronic spectra of the chromophore.15 However, it is well known that the excited-state properties of π-stacked chromophore assemblies, such as 13 and 213, depend strongly on the relative orientations of the stacked chromophores.47−49 The inversion of vibronic band intensities for 13 and 213 relative to those of monomeric 1 and 2 results from the excitonic coupling of the transition dipole moments of neighboring PMI chromophores, splitting the lowest excited singlet state into an upper and a lower exciton state.50 Because the transition dipole moment of PMI for the S0 → S1 transition lies parallel the long molecular axis,23 when the π systems of two or more PMIs are cofacial with parallel transition dipole moments the transition to the upper exciton state is allowed, corresponding to an apparent inversion of the vibronic band intensities (H-aggregate formation). The energy difference between the (0,0) transition to the upper exciton state and the (0,1) transition to the forbidden lower exciton state (the lower energy vibronic peak present in these four spectra) reveals information about the strength of the exciton interaction in these assemblies.23 The energy difference between these transitions in 213 (∼2400 cm−1) is nearly double that of 13
of these model structures were calculated and compared with the experimental PDFs for each molecule. Figure 1C,D compares the experimental PDFs with those for model structures that provide the best fit to each set of experimental data. The experimental PDF of 1 is best fit with a cofacial trimer (13) structure (Figure 2), while that of 2 is best fit with a large cofacial tridecameric (213). Several alternative linear structures were modeled (Figures S1 and S2 in the Supporting Information), but the computed PDFs are inconsistent with the experimental data. The SAXS/WAXS data for these molecules demonstrate several effects of monomer structure on self-assembly. While the experimental PDFs of 13 and 213 are modeled reasonably well by the structures shown in Figure 2, it is important to note that in the absence of rigid linkers the association of these molecules in solution is a dynamic process;46 equilibrium likely exists between the best fitting structure and a narrow distribution of similarly sized structures as well as a small amount of monomer. The structural models show that the lack of a phenyl spacer between the PMIs and the central benzene in 1 forces the PMIs to be canted out of plane relative to the benzene core by ∼60°, compared with ∼15° for 2. Thus, the “pin-wheel” structure of 1 prevents the PMI π systems from adopting the extended π-stacking demonstrated by 2 in MCH. The dodecyl side groups at the 2,5-positions are angled away from the center of the PMI so that their potential steric effects are mitigated. Because the SAXS/WAXS experiments show that 13 and 213 form in MCH at concentrations of 10−4 to 10−3 M, the optical experiments were performed at similar concentrations, so that their photophysical behavior can be directly related to their supramolecular structures. Figure 3A compares the UV−vis spectra of monomeric 1 in THF and 13 in MCH. The spectrum of 1 in THF shows that the PMI (0,0) band absorbs at 513 nm 1611
dx.doi.org/10.1021/jz500626g | J. Phys. Chem. Lett. 2014, 5, 1608−1615
The Journal of Physical Chemistry Letters
Letter
Figure 4. Femtosecond transient absorption spectra following 1 μJ/pulse, 110 fs pulse excitation: 1 (A) in THF (1 × 10−4 M) and (B) in MCH (6 × 10−4 M), 532 nm; 2 (C) in THF (1 × 10−4 M), 525 nm and (D) MCH (1 × 10−3 M), 532 nm. Residual laser scatter was removed from the spectra at 514−536 (C) and 518−550 nm (D). Insets: Principal components of the spectra from SVD-global analysis.
(∼1300 cm−1), which suggests significantly increased exciton coupling in 213. The higher ratios of the allowed to the forbidden exciton transitions in 213 compared with 13 also indicate that the overall parallel alignment of the transition dipole moments is better in 213 than in 13. Thus, the observed spectra of 13 and 213 in MCH are consistent with the SAXS/ WAXS data, showing that the PMI molecules of the trefoil πstack upon those of a neighboring trefoil, adopting an approximately cofacial geometry. The corresponding fluorescence emission spectra of 1 and 2 in THF and 13 and 213 in MCH are also shown in Figure 3, and their fluorescence quantum yields are presented in Table 1. In THF, monomeric 1 and 2 display similar spectra, with emission maxima at 557 and 567 nm and lower energy vibronic shoulders at 585 and 596 nm, respectively. The fluorescence quantum yields are all quenched somewhat relative to that of PMI itself (Table 1) but nevertheless remain large. In MCH, the fluorescence spectra of 13 and 213 are broadened, and their maxima are both red-shifted to 573 nm, while their fluorescence quantum yields are both strongly quenched (ΦF < 0.10). These changes in the steady-state emission spectra are indicative of excimer formation,47 which will be further elucidated using time-resolved spectroscopy. Molecules 1 and 2 were photoexcited with 525 or 532 nm, 1 μJ, 110 fs laser pulses. Transient absorption spectra of 1 and 2 in THF and 13 and 213 in MCH are presented in Figure 4. The spectra of monomeric 1 and 2 in THF show similar features and temporal evolution (Figure 4A,C) comprising an initial ground-state bleach at 450−525 nm, stimulated emission at 525−600 nm, and excited singlet state absorption at 600−800 nm. These transient spectral features decay with time constants of τ = 5.0 and 5.6 ns for 1 and 2, as determined by SVD and
global analysis of the 3-D ΔA versus time and wavelength data set (Figure 4, insets). The excited-state decay kinetics are laserpower-independent. In MCH, the transient spectra of 13 show no stimulated emission, and the initial 1*PMI absorption at 719 nm blue shifts to 668 nm in τ = 45 ± 10 ps (Figure 4B). This new excited-state absorption is broader than the corresponding absorption of 1 in THF and does not decay within the maximum pump−probe delay time (∼6 ns) of our fsTA apparatus. The 51 nm shift following photoexcitation is attributed to a structural change resulting in excimer formation. The corresponding transient spectra of 213 in MCH show no stimulated emission and display a 1*PMI excited-state absorption that appears initially at 728 nm, then blue shifts first to 720 nm in τ1 = 14 ± 2 ps and finally to 645 nm in τ2 = 620 ± 110 ps (Figure 4D). The 645 nm absorption band remains for τ ≫ 6 ns. Once again, the excited-state decay kinetics are laser-power-independent. The initial 8 nm spectral shift most likely results from vibrational or structural relaxation, while the second 75 nm blue shift is attributed to relative slow excimer-state formation. Time-resolved fluorescence (TRF) spectroscopy provides an additional check on the excited-state lifetimes determined by transient absorption measurements. The fluorescence lifetimes of monomeric of 1 and 2 in THF agree well with those measured by fsTA spectroscopy, yielding τ = 5.0 ± 0.4 and 5.6 ± 0.1 ns, respectively. In MCH, 13 and 213 both display wavelength-dependent biexponential decays, with the shorter of the two components, τ = 3.7 ± 0.2 ns for both 13 and 213, corresponding to the monomer excited singlet-state decay, while the longer decay occurs in τ = 23 ± 1 ns for 13 and τ = 8.4 ± 0.2 ns for 213. The transient fluorescence spectra at early 1612
dx.doi.org/10.1021/jz500626g | J. Phys. Chem. Lett. 2014, 5, 1608−1615
The Journal of Physical Chemistry Letters
Letter
times are consistent with the steady-state fluorescence spectra of monomeric 1 and 2 in THF (Figures S3 to S4 in the Supporting Information). It is likely that this emission results from the small, highly fluorescent monomeric population of 1 and 2 that is in equilibrium with the aggregates. In contrast, the spectra at long times show broad, featureless, red-shifted emission consistent with an excimer state.47,48,51 Because the instrument response of the TRF instrument using the 100 ns time window is 1.5 ns, the excimer formation dynamics were not observed. Excimer-state lifetimes in excess of 20 ns like those exhibited by 213 have been observed previously in PDI Haggregates and related covalent dimers and trimers.47 Thus, the excimer-state lifetimes of 13 and 213 align well with previously studied PDI molecules that have favorable conformations for excimer formation. The excimer quantum yield of 213 is much smaller than that of 13, which is likely due to the significant initial excited-state population (∼33% for both) that decays by vibrational or structural relaxation. However, this observation does not conflict with the conclusion that 213 shows increased exciton coupling compared with 13 because exciton coupling is a dipole−dipole interaction, while excimer-state formation depends on orbital overlap. In summary, we have synthesized two symmetric PMI-based trefoils around a benzene core. SAXS/WAXS experiments show that steric constraints resulting from the direct attachment of the PMI chromophores to the benzene core in 1 limit its πstacked assembly on average to a trimer in MCH. Lengthening the distance between the PMI chromophores and the benzene core by adding a phenyl spacer, which also allows the PMI chromophores to adopt a more planar conformation relative to one another in 2, results in the formation of on average a tridecamer in MCH. Steady-state optical absorption spectra show that the more planar PMI−PMI conformation in 2 results in stronger exciton coupling in 213 than in 13. Long-lived excimer states are produced in both aggregates and decay largely nonradiatively. Future studies will explore charge transport through these different types of π-stacked PMI assemblies.
ments were made with 470 nm excitation on a PTI QuantaMaster 1 single-photon-counting fluorimeter. Measurements at 10−5 to 10−4 M in THF were performed in a right angle configuration with a 10 mm quartz cuvette, and measurements in MCH were performed at 10−4 to 10−3 M in a 2 mm cuvette using a front-face detection geometry. Quantum yield measurements were performed with a Rhodamine 6G standard in EtOH in a 10 mm cuvette and a 2 mm cuvette for comparison with the THF and MCH samples, respectively. Femtosecond transient absorption (fsTA) measurements for 1 and 2 were made using a regeneratively amplified Ti:sapphire laser system operated at a 2 kHz repetition rate previously described.53 The 110 fs, 1 μJ, 525 or 532 nm pump pulses were focused to a 0.5 mm diameter spot and overlapped with the white-light continuum probe pulse. The total instrument response was 150 fs. Five-second averaging was used to collect transient spectra. In MCH, samples of 1 and 2 had an absorbance of ∼0.4 to 0.6 at the excitation wavelength in 1 mm quartz cuvettes. In THF, all samples had an absorbance of 0.3 to 0.5 in a 2 mm quartz cuvette. Single-wavelength kinetic analyses were performed at several wavelengths using a nonlinear least-squares fit to a sum of exponentials convoluted with a Gaussian instrument response function (IRF). Singular value decomposition (SVD) of the 3-D data set was carried out using Surface Xplorer 2.0 (Ultrafast Systems).54 Picosecond TRF measurements were made using a lab-built, frequencydoubled, cavity-dumped Ti:sapphire laser system that produced 390 nm, 25 fs laser pulses at an 820 kHz repetition rate.55 The samples were excited with
