Characterization of Excimer Relaxation via Femtosecond Shortwave

Subscriber access provided by GAZI UNIV

Article

Characterization of Excimer Relaxation via Femtosecond Shortwave- and Mid-Infrared Spectroscopy Catherine M. Mauck, Ryan M. Young, and Michael R. Wasielewski J. Phys. Chem. A, Just Accepted Manuscript • DOI: 10.1021/acs.jpca.6b11388 • Publication Date (Web): 03 Jan 2017 Downloaded from http://pubs.acs.org on January 5, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

The Journal of Physical Chemistry A is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Characterization of Excimer Relaxation via Femtosecond Shortwave- and Mid-Infrared Spectroscopy Catherine M. Mauck, Ryan M. Young, Michael R. Wasielewski* Department of Chemistry and Argonne-Northwestern Solar Energy Research (ANSER) Center, Northwestern University, Evanston, Illinois 60208-3113 *Email: [email protected] ABSTRACT Excimer formation plays a significant role in trapping excitons within organic molecular solids. Covalent dimers of perylene-3,4:9,10-bis(dicarboximide) (PDI) are useful model systems for studying these processes as their intermolecular geometries can be precisely tuned. Using femtosecond visible-pump infrared-probe (fsIR) spectroscopy in the shortwave- and mid-infrared regions, we characterize two PDI dimers with a cofacial and a slip-stacked geometry that are coupled through a triptycene bridge. In the mid-infrared region, fsIR spectra for the strongly coupled dimers are highly blue-shifted compared to monomeric

1*

PDI. The perylene core

stretching modes provide a directly observable probe of excimer relaxation, as they are particularly sensitive to this process, which is associated with a small blue shift of these modes in both dimers. The broad Frenkel-to-CT state electronic transition of the excimer, the edge of which has previously been detected in the NIR region, is now fully resolved to be much broader and to extend well into the shortwave infrared region for both dimers, and is likely a generic feature of π-extended aromatic excimers.

1 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 31

INTRODUCTION Exciton trapping in organic molecular solids can occur through the formation of excimer states, in which excitation energy is delocalized over two or more chromophores, and subsequently relax to a lower energy state.1-4 Strong intermolecular coupling in films or aggregates can create multiple competing, and often parasitic, relaxation pathways. Organic electronic device performance frequently decreases when excimer formation is competitive with desirable processes such as charge transfer (CT) or exciton diffusion because excitons become trapped in the low-lying excimer state, where they decay instead of separating into free charge carriers.5-8 The excimer trapping process has been observed to be quite rapid, as fast as a few hundreds of fs.9 Excimer states are known to have partial Frenkel exciton (FE) and CT character resulting from a stabilizing admixture of charge resonance and exciton resonance interactions.1,

10

The

resulting excimer to ion pair electronic transition, i.e. 1(M+–M–) ← 1*(M–M), has been observed in the edge of the near-infrared (NIR) in various organic excimers, in particular for perylene and its derivatives, where this band is spectrally distinct from other excited state features and allows direct reporting on excimer electronic dynamics.1, 3, 10-12 Perylene dyes have been widely studied for the development of organic electronics and artificial photosynthetic systems. Perylene-3,4:9,10-bis(dicarboximide) (PDI) is of particular interest because it absorbs strongly in the visible region and possesses high photochemical stability in addition to its high electron mobility and synthetic tunability.2,

13-15

These planar

aromatic chromophores can self-assemble through strong π-π interactions, where the packing geometry in crystals and films is highly dependent on the nature of the substituents on the perylene core and its imide nitrogens.16 In the solid state, PDIs often form π-stacked structures in which adjacent chromophores are electronically coupled to one another, and the dominant


Page 3 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


exciton decay pathway is usually determined by this coupling. The interchromophore coupling includes dipole-dipole coupling of the molecular transition dipole moments17-18 as well as orbital overlap, both of which depend the interchromophore geometry and produce shifted and broadened electronic spectra. Transition dipole coupling in molecular aggregates has typically been classified on basis of the angle θ between the transition dipoles and the line connecting the centroids of the chromophore π systems, with J- and H-aggregates representing the θ = 0 and π/2 extremes, respectively.19 The absorption spectra of the dimers depend on the intermolecular orientation, with an increasing ratio of the 1←0 vibronic absorption band to the 0←0 band as the π-π interaction between the monomer units becomes stronger. Recent work in our group and others has focused on molecular analogues of larger aggregates and film structures in the form of covalently linked dimers that have a fixed geometry between chromophores imposed by a rigid bridging scaffold.11,

20-24

The synthetic tunability of covalent dimers allows for the direct

manipulation of the degree of interchromophore interaction, enabling detailed exploration of the role of electronic coupling in both the ground and excited states. Accordingly, a study of two PDI dimers using a redox-inactive triptycene bridge demonstrated that excimer dynamics were slowed when the PDIs were in a slip-stacked orientation 2J (τ = 12 ± 2 ps) as opposed to a cofacial arrangement 2H (τ = 2.2 ± 0.2 ps), based on the rise time of a band in the near infrared (NIR) assigned to the 1(M+–M–) ← 1*(M–M) transition.21 The structures of these compounds are given in Scheme 1. The observed dynamics were originally ascribed to formation and subsequent relaxation of the PDI excimer. However, a recent study by Kim and co-workers9 on selfassembled PDI H-aggregated dimers was able to directly probe the LE-to-excimer population transfer process using high-temporal resolution broadband fluorescence up-conversion



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 31

spectroscopy, and showed the excimer formation process occurs in ~200 fs. If their results are generally applicable to other PDI and aromatic aggregates, then the ~ps dynamics we observed in our previous study arise from the excimer relaxation process instead. In that case, the excimer relaxation dynamics themselves are quite complex and require more detailed information to construct a molecular-level picture of the relaxation process. However, time-resolved electronic spectroscopies often lack structural insight owing to significant spectral congestion, and thus can obscure which intra- and intermolecular structural changes may accompany the population dynamics. Time-resolved vibrational spectroscopy, alternatively, is ideally suited to obtain such information about excited states, owing to the naturally narrow linewidths of these transitions, while providing data complementary to electronic spectroscopies. For example, IR spectroscopy has been established as a method to characterize aggregates and delocalized states in proteins,2528

and has also been extended to surface-bound Re(bpy)(CO)3Cl

29-30

and Mn2(CO)10

aggregates.31-32 Pump-probe femtosecond transient infrared (fsIR) spectroscopy can yield additional information about photoinduced excited states and their associated changes in nuclear configuration. Time-resolved IR techniques have been successfully used to study electron transfer reactions,33-36 charge transport in bulk heterojunction films,37-38 and electron injection from dyes into TiO2.39-40 Recently, time-resolved IR spectroscopy has also been applied to study the formation of bimolecular exciplexes,41-42 as well as the effect of electron delocalization on electron transfer in aryl oligomers,43 underscoring the sensitivity of this technique to delocalized states. Yet to our knowledge fsIR has not been used to study pure excimer dynamics in organic chromophores. As the C=C and C=O modes of PDI are strongly IR active, fsIR offers a means of


Page 5 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


directly probing the structure and relaxation of the PDI excimer, while identifying participating modes in excimer evolution and possible associated structural changes. Here, we investigate the excited state dynamics of two covalent dimers using fsIR spectroscopy, building upon our original work to gain insight into the structural dynamics of excimers in PDI aggregates. While the excimer formation time is likely too fast to observe with fsIR spectroscopy, we directly probe the structural details of the nascent excimer and its subsequent relaxation. We take advantage of the imide carbonyl and perylene core modes of PDI to directly observe the ensuing structural changes sensed by those modes, which differ both spectrally and kinetically in the cofacial (2H) and slip-stacked (2J) geometries. As our fsIR experimental apparatus allows for detection of wavelengths in the shortwave IR (SWIR, 20004000 nm) region, we are able to observe the rise of the 1(M+–M–) ← 1*(M–M) transition as the excimer relaxes.10,

44

Recent studies have investigated the PDI excimer out to ~1600 nm,11-12

including the present molecular dimers,21 with band maxima for PDI excimers typically extrapolated from Gaussian fits to the NIR spectra. As the maxima determined previously were outside the range of detection, full resolution of the band was not possible. Using SWIR detection, we are able to observe the entire PDI excimer 1(M+–M–) ← 1*(M–M) transition. The rise of the SWIR absorption temporally corresponds to a shift to higher frequencies of the excited state C=C modes in the mid-infrared spectra for both dimers, a change that is absent in monomeric PDI. This shift is consistent with greater electronic coupling between the PDI cores in the excimer geometry.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Scheme 1. Chemical structures of PDI monomer 1 and dimers 2H and 2J. R = C8H17.

EXPERIMENTAL SECTION fsIR measurements. Molecules 1, 2H, and 2J were synthesized previously.21 Steady state electronic absorption spectra were acquired with a Shimadzu UV-1800 spectrophotometer. FTIR spectra were collected by drop-casting solutions of each compound on CaF2 (2 mm) slides, which were then dried and measured using a Shimadzu IRAffinity-1 spectrophotometer in transmission mode (2 cm-1 resolution). FsIR spectra were collected using a commercial Ti:sapphire 3.5 W, 1 kHz oscillator/amplifier (Solstice, Spectra-Physics) to pump two optical parametric amplifiers (TOPAS-C, Light Conversion) to generate a 100 fs visible excitation pulse tuned to 492 nm or 530 nm (2 µJ/pulse) and a 100 fs mid-infrared probe pulse between 55606670 nm (1500-1800 cm-1) for mid-infrared experiments. To detect in the SWIR region, the probe pulse could also be tuned to 1200-3500 nm. The measured instrument response was 310 fs. Data were acquired using a customized Helios-IR spectrometer (Ultrafast Systems, LLC). The wavelengths of fsIR spectra were calibrated using the ground state FTIR spectra. Further details about the experimental set-up are provided in the Supporting Information (SI). Samples were purified via HPLC (dichloromethane/isopropanol) prior to analysis, then prepared in dry deuterated dichloromethane (Aldrich, 99.9%) with a maximum optical density of 0.7 in a


Page 6 of 31

Page 7 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


demountable liquid cell (Harrick Scientific) with a 500 µm Teflon spacer and 2 mm-thick CaF2 windows. During data acquisition, the cell was mounted and rastered on a motorized stage to prevent sample degradation. Computational Details. Optimized ground state geometries and frequency calculations were performed using density functional theory (DFT) at the B3LYP/6-31G* level using QChem (version 4.0). A scaling factor of 0.96 was applied to all frequency calculations,45 and Lorentzian lineshapes (FWHM 5 cm-1) were used to model FTIR spectra. For all molecules, structures were modified to reduce computational cost by substitution with a methyl group in place of the octyl chains. RESULTS Ground State Infrared Spectroscopy. The FTIR spectra of 1, 2H, and 2J deposited on CaF2 slides are given in Figure 1 alongside the computed ground state IR frequencies for 2H, 2J, and monomeric PDI. The measured C=O and C=C stretching frequencies are given in Table 1. We note that different relative peak intensities have been observed in the infrared spectrum of PDI depending on the type of substrate used.46-48 The main contributions to the spectra come from imide carbonyl and perylene core stretches, as has been widely characterized in the literature.48-49 Both the symmetric and antisymmetric C=O ground state stretches are split in the dimers, and exhibit broad absorptions centered at 1707, 1698, 1659, and 1632 cm-1 for 2H and at 1699, 1658, and 1632 cm-1 for 2J. This can be seen clearly for the calculated spectra in Figure 1, where the asymmetry of the slip-stacked orientation leads to split C=O modes and the cofacial arrangement has split modes of more equal intensity. The branching of these modes is similar to what has been seen in Re(bpy)(CO)3Cl aggregates,30 with peaks that can be represented by linear



b

a

2H

νas,C=O νs,C=O

νC=C

Calculated Intensity

Normalized Absorption

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 31

2J

2H 2J

PDI

1 1750

1700

1650

1750

1600

1700

1650

1600

1550

-1

-1

Wavenumber (cm )

Wavenumber (cm )

Figure 1. (a) Ground state FTIR spectra for 1 (black), 2J (blue) and 2H (red). (b) Calculated ground state IR frequencies for optimized dimer geometries, compared to PDI.

Table 1. Frequencies (cm-1) of ground state FTIR peaks for monomer 1 and dimers 2H and 2J. νs.C=O

ν1,C=C ν2,C=C

νas,C=O

1

1692

--

1657

--

1592

1580

2J

1699

--

1658

1632

1594

1578

2H

1707

1698

1659

1632

1595

1579

combinations of monomeric stretches. Splitting of C=O in perylenetetracarboxylic dianhydride films on Si(111) has also been observed and attributed to Davydov splitting.46 The C=C ground state peaks do not shift dramatically going from the monomer to the dimers, but occur at slightly higher frequencies by 2-3 cm-1. This trend is similar to PDI nanostructures with strong π-π interactions, which demonstrated C=C stretch shifts of +8-30 cm-1 in FTIR spectra compared to a powder.50 In the monomer and the dimers, the most intense C=C peak around 1592-1595 cm-1 can be assigned to the same ring core stretching motion, the displacement vectors of which are given in Figure 2.


Page 9 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 2. Visualization of displacement vectors for the main C=C core stretch in calculated ground state -1 frequencies for 1 (left; 1579 cm ), 2H (middle; 1582 -1 -1 cm ), and 2J (right; 1581 cm ).

fsIR Spectroscopy. FsIR spectra were acquired following excitation at 530 nm (0←0 vibronic transition) in CD2Cl2. The fsIR excited state spectra at selected times for molecules 1, 2H and 2J are presented in Figure 3. The excited state spectrum for PDI monomer 1 is similar to the IR excited state spectra of other rylene derivatives.51-52 We note that triptycene has no strong IR bands above 1451 cm-1, although weak contributions may appear at 1589 and 1598 cm-1.53 As triptycene does not participate in the excited state dynamics of the molecules studied here, it should have negligible contribution to the fsIR spectra.21 The two excited state C=O stretches in 1 appear at 1645 and 1616 cm-1, which, as expected, are lower frequencies relative to those of the ground state. In the C=C ring stretch region, three sharp excited state peaks appear at 1538, 1495, and 1433 cm-1 with the 1495 cm-1 having the highest intensity. The excited state of 1 decays in 3.9 ± 0.1 ns.



20

∆ A (x 10-3)

15

1

10 5 0 1 ps 5 ps 50 ps

-5

500 ps 2 ns 5 ns

-10 1700 1650 1600 1550 1500 1450

Wavenumber (cm-1)

20 15

2H

∆ A (x 10-3)

10 5 0

1 ps 5 ps 10 ps 25 ps 100 ps

-5 -10 1700

15

1650

1600

500 ps 2 ns 4 ns 7.6 ns

1550

1500

Wavenumber (cm-1)

20

∆ A (x 10-3)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 31

2J

10 5 0

1 ps 5 ps 10 ps 25 ps 100 ps

-5 -10 1700

1650

1600

500 ps 2 ns 4 ns 7.6 ns

1550

1500

Wavenumber (cm-1)

Figure 3. fsIR spectra at selected time delays for 1, 2H, and 2J, following excitation at 530 nm.

To make assignments for the excited state spectrum of 1, TD-DFT frequency calculations were performed on the optimized excited state geometry of a modified PDI with methyl groups substituted at both imide positions (1*PDI) in vacuum. The resulting calculated peaks are given in Figure S5 and agree well with experiment, yielding two sharp modes from the symmetric and antisymmetric C=O stretches, as well as three in-plane ring stretching modes spaced by 35-40 cm-1 from one another. There is little difference in intensity between the three modes. The


Page 11 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


displacement vectors for each of the C=C modes are given in Figure S6. We use the calculated 1*

PDI spectra to assign the experimental fsIR spectra of 1, which can then be used as a guide to

analyze the excimer spectra, i.e. where C=O and C=C stretches appear. We avoid, however, making specific mode assignments for the excimer due to the computational complexity and cost associated with accurately modeling the excimer state. Figure 3 gives the excited state spectra at selected time traces for dimers 2H and 2J. Both dimers were excited at the 0←0 vibronic band, although the intensity of this band differs depending on the interchromophore coupling.18,

21

In contrast to 1, the excited state C=O

absorption for both dimers is broadened in both the ground and excited states. This effect is consistent with the calculated ground state IR spectra for 2H and 2J in Figure 1. In the excited state, 2H has positive absorption spanning from 1600-1700 cm-1, with peaks at 1653 and 1686 cm-1, and 2J has similar positive absorption with peaks at 1640 and 1688 cm-1. Resolution of specific bands is obscured by the negative signal from the ground state C=O bleach around 16571660 cm-1 overlaid on top of the broad positive absorption. Additionally, as C=O stretches are highly sensitive to their electronic environment and each PDI in the dimer interacts with another PDI as well as solvent, inhomogeneous spectral broadening is likely. Following photoexcitation, the broad C=O absorption decreases in intensity between 1600-1650 cm-1 for both dimers, although the absorption above 1650 cm-1 remains equal in intensity. Due to the strong overlap of the ground state in this region, a shift of the C=O modes to higher frequency may be masked, resulting in an intensity decrease on the lower energy edge of this absorption. This effect is more pronounced in 2H, where the PDIs are most closely associated with one another. The dimer C=C core stretches in the excited state spectra occur at higher frequencies than in 1, and are more closely spaced. In 2H, the mode at 1530 cm-1 is more intense than the other C=C



a

b 1.0

Normalized ∆ A

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

25 20

2J

2H *

Page 12 of 31

15

*

10

2J τExc = 19 ± 2 ps

5 0

0.5

*

15 10

*

2H τExc = 1.3 ± 0.2 ps

5 0

0.0 1000 1500 2000 2500 3000 3500 4000

-20

0

100

1000

Time (ps)

Wavelength (nm)

Figure 4. (a) NIR-to-shortwave-IR spectra of the excimer absorption band for 2H (red, at 50 ps) and 2J (blue, at 100 ps). Asterisks denote areas of strong water or CO2 absorption. (b) Single-wavelength kinetic traces (at 2138 nm for 2H and 2873 nm for 2J) for shortwave edge of each excimer species, with corresponding kinetic fits given for the rise of the absorption feature.

modes by a factor of three, with a higher ∆A than even the C=O stretches at each delay time. A similar if less severe trend is seen in 2J, where the most intense C=C mode appears at 1508 cm1

, although the peak at 1538 cm-1 is of comparable intensity. Notably, the C=C stretches in the

excited state spectrum of 2H are much narrower than in 2J, although they appear broader than those in 1. Given the calculated ground state C=C stretches for 2J (Figure 1), it is possible that multiple modes contribute to this spectral region in the excited state, or that inhomogeneous broadening contributes to the spectral shape more significantly in 2J than in 2H. A small shift to higher frequencies occurs at early times for both dimers. Finally, the photoinduced absorption in 2J and 2H persists longer than the timescale of the experiment (8 ns), in agreement with the previously measured excimer lifetimes of τ = 12.0 ± 0.3 ns for 2H and τ = 24.1 ± 0.4 ns for 2J.21 In addition to the mid-infrared absorption features, we observe a continuous broad absorption band extending from the NIR into the shortwave IR region for both dimers. The spectra and kinetics for each are given in Figure 4. The rise times of this absorption are in agreement with the previous NIR fsTA data,21 which were analyzed using single representative wavelengths; 12 ACS Paragon Plus Environment

Page 13 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


these data were reacquired here with additional filtering to suppress the second order diffraction of the 827 nm fundamental to decrease the scatter around 1500-1600 nm, and are provided in Figure S3. The resulting NIR spectra below 1500 nm were merged with the shortwave IR spectra to yield the selected spectra for 2H at 50 ps and for 2J at 100 ps shown in Figure 4. Absorption in this region from carbon dioxide and water cannot be completely suppressed, and contributes to the noise of the resulting spectra. Nonetheless, peak maxima can be resolved for the excimer absorption, and are clearly centered in the shortwave region. Peak maxima were assigned by assuming a Gaussian lineshape, and fit to 0.44 eV for 2H and 0.62 eV nm for 2J. These fits are provided in Figure S2. The SWIR band appears within the IRF and continues to rise with 1.3 ± 0.2 and 19 ± 2 ps, respectively. This data is consistent with the results of Kim and co-workers on self-assembled PDI dimers, which show that a similar rise results from relaxation of the hot excimer state.9 DISCUSSION Excimer-to-Ion Pair Transition in the Shortwave IR Region. As discussed above, the excimer formation time is very close to our respective fsTA and fsIR instrument response functions, so we are unable to directly observe the initial formation step. However, the broad NIR 1(PDI+– PDI–) ←

1*

(PDI–PDI) absorption11 observed in our previous study is a direct probe of the

excimer state, as the band appears within the instrument response. The oscillator strength of that transition appears to increase as the excimer relaxes, directly implying a change in wavefunction overlap over time; such a change would be expected to accompany nuclear rearrangement (vide infra). The position of this band also appears quite sensitive to molecular conformation, and so gives more details of the local energy landscape. Previously, in 2H, the peak maximum was estimated to be at 0.75 eV, which was out of the range of the detector. In 2J, a weakly concave



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 31

absorption peaked at 0.93 eV was assumed to be the maximum for this excimer transition. By extending the range of detection, we are now able to observe the entire transition and conclude that in fact the NIR fsTA spectra for the excimer in both 2H and 2J represent the onset of a much broader band, which is peaked in the SWIR region of the spectrum at 0.62 eV for 2H and 0.44 eV for 2J. The SWIR region is unaffected by the perturbed free induction decay at time zero observed in the mid-IR region because the electronic dephasing time T2 for these electronic transitions should be shorter than the fs pulse duration,54 and so we can assign the excimer relaxation time based on the rise of the spectrally distinct 1(PDI+–PDI–) ←

1*

(PDI–PDI)

transition in the SWIR region. The asymmetry in the band for 2J is greater than in 2H, consistent with the more flexible geometry of the slip-stacked dimer, which can sample a larger range of small conformational changes within the scale of typical molecule-solvent fluctuations. Interestingly, the energy of this transition is quite close to the calculated energy difference ~0.5-0.6 eV between the lower Frenkel and upper CT states in a theoretical treatment of rotationally-displaced PDI dimers.55 In agreement with the PDI excimer study by Brown et al,11 the energy of the excimer band tracks with increased coupling, as the cofacial orientation of PDIs stabilizes the Frenkel state

1*

(PDI–PDI) leading to a higher energy transition, effectively

resolving the discrepancy between the coupling trends of the PDI-triptycene dimers and the PDIxanthene/PDI-cyclophane dimers previously reported.11, 21 Global Analysis of fsIR Spectra. Structural dynamics can be observed in the mid-IR transient spectra. To analyze the spectral changes associated with excimer relaxation, a global analysis was performed on the fsIR data for 2H and 2J by simultaneously fitting the dataset to the solution of the coupled differential equations that describe a specific kinetic model. By selecting multiple frequencies at which to perform these fitting procedures, this analysis gave the kinetic


Page 15 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


fits and species-associated spectra in Figure 5. This procedure has been described in more detail elsewhere.56 We utilize the so-called “hot excimer” model in which the localized excitation forms a vibrationally excited state prior to the relaxed excimer geometry.3 Since the formation rate is too fast to resolve, we treat the initial state as the hot/unrelaxed excimer, [*Exc], decaying with τ1 to the relaxed state [Exc], which subsequently decays with τ2. The model is given by eqns. 1 and 2:

d [* Exc ] 1 = − [* Exc ] dt τ1

(1)

d [Exc ] 1 1 = [* Exc ] − [Exc ] . dt τ1 τ2

(2)

Previously determined fsTA lifetimes were derived from single-wavelength kinetic fits to a multi-exponential decay;21 we now apply a global analysis that allows us to impose a kinetic model to derive associated vibrational spectra for each relevant species. In addition, an initial component was fixed at 0.5 ps to filter out the nonlinear coherence artifacts that occur around time zero in time-resolved IR spectra.54, 57 Despite the complexity of observing coherence-free spectra at times earlier than 0.5 ps in fsIR spectra, and the lower signal-to-noise ratio, the fits obtained with these kinetic models are in agreement with the lifetimes obtained from fsTA as well as time-resolved fluorescence.21 For 2H, the fsIR decay lifetimes are τ1 = 5 ± 1 ps and τ2 = 12.4 ± 0.4 ns. For 2J, the resulting lifetimes for each species fits to τ1 = 16 ± 2 ps and τ2 = 19.8 ± 0.9 ns.

Comparison of Dimer Species-Associated Spectra. A comparison of the shifts or changes in intensity of the IR absorptions between the hot excimer state (*Exc) and the relaxed excimer state (Exc) can be made from the species-associated spectra in Figure 5. Excited state IR



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 31

Figure 5. Species-associated spectra from global analysis of fsIR data for 1, 2H, and 2J excited at 530 nm. Ground state FTIR spectra are given below.

frequencies for *Exc and Exc for 2H and 2J are listed in Table 2. The broad C=O region and blue-shifted C=C peaks for *Exc suggest that the PDIs are coupled in the initially observed state of 2H and 2J, as expected from a hot excimer This is also consistent with the ground state IR

Table 2. Observed C=C frequencies (cm-1) for the initial excited state (S1) of 1 and the initially formed state (*Exc) and excimer (Exc) of 2H and 2J from a global fit of fsIR data. S1(ν1)

S1(ν2)

S1(ν3)

1538

--

1495

--

1433

--

*Exc (ν1)

Exc (ν1)

*Exc (ν2)

Exc (ν2)

*Exc (ν3)

Exc (ν3)

2H

1572

1573

1551

1553

1529

1530

2J

1572

1576

1533

1538

1503

1508

1


Page 17 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


spectra, which also demonstrate strong coupling between chromophores, causing broadened and split peaks in the C=O region. The effect on the ground state C=C peaks is much smaller with no significant broadening, yet in the excited state the C=C modes are markedly different in their position, spacing, and relative intensity. In the *Exc spectra, the three broad modes span ~20 cm1

in 2H and ~30 cm-1 in 2J, whereas 1 has three sharp peaks spanning 100 cm-1. For clarity we refer to the three main C=C peaks as ν1, ν2, and ν3 to compare the dimers to the

monomer, although we acknowledge that the relevant vibrational modes in each dimer and the monomer do not necessarily originate from equivalent motions, though all are likely perylene core stretches. Although there is no shift in ν1-3 for 1 over the timescale of the experiment, a blue shift in the spectra for 2H and 2J occurs with τ1. The sensitivity of the C=C modes to intermolecular orientation is to be expected, as the dimer π-π interactions are driven by the degree of core-core overlap. To measure peak shifts more accurately, scans were acquired at a higher spectral resolution (150 lines/mm grating, ~1 cm-1 per pixel). These data are provided in Figure S3. Consistent with the more flexible geometry of dimer 2J, excimer relaxation is associated with larger frequency shifts in 2J than in 2H. For 2H, the frequency shifts from *Exc to Exc (∆ν) are on the order of +1-2 cm-1 for all C=C modes. For 2J, ∆ν is +4-5 cm-1 for all C=C modes. Spectra acquired with higher energy excitation (492 nm, 1←0 vibronic transition, Figure S4) show that these shifts do not originate from vibrational relaxation on the excimer surface. Interestingly, the modes seen here do not resemble either the PDI radical cation or anion, which appear at ~1551 cm-1 (weak band, radical cation) and at 1579 cm-1 (strong band, radical anion).5859

We have previously characterized the fsIR radical anion C=C peak for a similar xanthene-

bridged PDI dimer,59 but find that the present excimer spectra do not match the intensity progression and peak positions of the PDI radical anion. Given that excimer states possess CT



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 31

character, further work is needed to understand how this influences the fsIR spectra of PDI excimers.

Structural Dynamics of Excimer Relaxation. Overall, for excitation at either 492 or 530 nm, the spectra for *Exc and Exc are quite similar. Notably, no monomer-like spectra emerged as a precursor for the later species-associated spectra, even when an early-time 0.5 ps component was fixed in the kinetic model to capture the coherence artifact. This observation supports a rapid (~200 fs) excimer formation time, and highlights how the spectral congestion in the visible transient absorption experiments can obscure the nature of the initially detected states. Coherent oscillations on the order of fs between vibrational states have been both predicted and observed in molecular aggregates.12,

29-30, 60-61

In a theoretical treatment of exciton trapping in PDI

aggregates, Engel and Engels put forth a mechanism in which an initial exciton spanning multiple chromophores localizes to dimers, before localizing to the lower-lying Frenkel state.60 Recent work has extended this mechanism to a more controlled solution aggregate of a PDI derivative, where excimer formation was seen for the smallest dimeric noncovalent aggregate, in contrast to larger aggregates for which localized exciton migration occurred.9, 12 Indeed, as π-π overlap is strong in the ground state structure of these dimers, and excitons in larger aggregates have been known to span multiple chromophores,62-63 it is very likely that the initial (unresolved) photogenerated excited state is delocalized across both chromophores immediately, followed by rapid (

Characterization of Excimer Relaxation via Femtosecond Shortwave

Recommend Documents