Excited-State Planarization in DonorâBridge Dye Sensitizers

Subscriber access provided by READING UNIV

Article

Excited state planarization in donor-bridge dye sensitizers: phenylene vs. thiophene bridges Zachary Piontkowski, and David W. McCamant J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.8b05463 • Publication Date (Web): 09 Aug 2018 Downloaded from http://pubs.acs.org on August 10, 2018

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 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 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.

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 36 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

Journal of the American Chemical Society

Excited state planarization in donor-bridge dye sensitizers: phenylene vs. thiophene bridges Zachary Piontkowski, David W. McCamant* Department of Chemistry, University of Rochester, Rochester, New York 14627, United States.

Abstract Donor-π-acceptor complexes for solar energy conversion are commonly composed of a chomophore donor and a semiconductor nanoparticle acceptor separated by a π bridge. The electronic coupling between donor and acceptor is known to be large when the π systems of the donor and bridge are coplanar. However, the accessibility of highly coplanar geometries in the excited state is not well understood. In this work, we clarify the relationship between bridge structure and excited state donor-bridge coplanarization by comparing rhodamine sensitizers with either phenylene (O-Ph) or thiophene (O-Th) bridge units. Using a variety of optical spectroscopic and computational techniques, we model the S1 excited state potential surfaces of O-Ph and O-Th along the dihedral coordinate of donor-bridge coplanarization, τ. We find that O-Th accesses a nearly coplanar (τ = 8°) global minimum geometry in S1 where significant intramolecular charge transfer (ICT) character is developed. The S1 coplanar geometry is populated in < 10 ps and is stable for ca. 1 ns. Importantly, O-Ph is sterically hindered from rotation along τ and therefore remains at its initial S1 equilibrium geometry (τ = 56°) far from coplanarity. Our results demonstrate that donor-bridge dye-sensitizers utilizing thiophene bridges should facilitate strong donor-acceptor coupling by an ultrafast and stabilizing coplanarization mechanism in S1. The coplanarization will result in strong donor-acceptor coupling, potentially increasing electron transfer efficiency . These findings provide further explanation for the success of thiophene as a bridge unit and can be used to guide the informed design of new molecular sensitizers.

ACS Paragon Plus Environment

1

Journal of the American Chemical Society 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 36

1. Introduction The resurgence of interest in rhodamines stems in part from their recent application as effective sensitizers in the photo-generation of hydrogen1-4 and dye-sensitized solar cells (DSSCs).5-8 Due to their large molar absorptivities (ε ≈ 100,000 M-1 cm-1) and photostability, rhodamines have the potential to serve as robust and efficient light harvesters in solar energy applications. The rapidly increasing demand for clean burning, renewable fuels requires that systems utilizing molecular sensitizers for solar light harvesting be optimized at a rapid pace to stay competitive with other contemporary alternative energy schemes. Herein we present new physical insights regarding rhodamine photochemistry, particularly regarding the aryl bridge moiety, which will enable informed design of new donor-bridge dye-sensitizers. Rhodamines O-Ph and O-Th are composed of a xanthilium core chromophore substituted at the C9 position with either a phenylene or thiophene aryl group respectively (Figure 1). In hydrogen production and DSSC schemes, the molecule is attached to a semiconductor acceptor such as TiO2 by an anchoring group on the aryl moiety.1-11 The aryl group acts as a bridge which mediates the electron transfer process, affecting forward and backward electron transfer rates. For short bridges and molecular accepting groups, electron transfer has been shown to occur largely by an adiabatic tunneling mechanism, termed “super-exchange”, where there is strong coupling between donor and acceptor moieties.12 In this case, electron transfer occurs in the presence of fixed nuclei and the role of the bridge is to provide virtual orbitals which electronically couple the donor and acceptor. The rate of electron transfer depends on the donor-acceptor electronic coupling strength, HDA, which is modulated by the thermally fluctuating geometry of the bridge.13 Super-exchange has also been shown to be the dominant mechanism by which locally excited donors reduce semiconductor nanoparticle acceptors.14 Sensitizers such as O-Ph and O-Th are expected to follow a similar mechanism due to their bright, visible excitation being localized on the core chromophore and separated from the acceptor by the aryl groups. It has previously been shown that HDA is largest when the π systems of the donor and bridge are coplanar.15-18 It has also been shown that thiophene bridges have a smaller distance attenuation parameter β compared to phenylene which suggests that thiophene bridges will be more effective conduits for charge transport.19 This is consistent with the superior efficiencies observed for thiophenes in DSSC work, allegedly due to excellent charge transport properties.10 High performance has routinely been observed for thiophenes and related thieno(3,2-b)-thiophene bridges in photovoltaic and electroluminescent devices.20-29 ACS Paragon Plus Environment

2

Page 3 of 36 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


It is understood then that coplanar thiophene bridges should initiate the strongest coupling between donor and acceptor, facilitating enhanced charge separation. However, there is a lack of understanding regarding the accessibility of such highly coplanar geometries in the excited state. In a recent study from our group, some of the photophysical effects of substituting the rhodamine aryl group with phenylene or thiophene have been discerned and explained with regard to flexibility along the dihedral angle coordinate τ, defined by atoms C13, C9, C14, and C15 (see Figure 1).30 Steady state fluorescence experiments showed that OTh only has only 25-40% of the fluorescence quantum yield of O-Ph. Further, ultrafast transient absorption measurements found O-Th to have a unique sub-10-ps time constant associated with large amplitude decay of stimulated emission (SE) and excited state absorption (ESA) that was not observed in O-Ph. The ground state potential surfaces along the coordinate τ were calculated to compare torsional flexibility of phenylene and thiophene groups with respect to the core. It was found that, in S0, O-Th has a more coplanar equilibrium geometry and a softer potential barrier to more coplanar geometries than O-Ph. It was thus proposed that, due to the smaller energetic barrier to coplanarity, O-Th forms an equilibrium on S1 with a secondary excited-state S1’ that is accessed by rotation on the torsional coordinate τ. Access of S1’ was proposed to open unique non-radiative decay pathways for O-Th that are not accessible to O-Ph due to its rigidity along τ, thereby explaining the observed higher fluorescence quantum yield and lack of the sub-10-ps time constant for O-Ph. However, questions about the nature of the S1 ⇌ S1’ process remain. Specifically, the structure of S1’ is not known, and it is not clear whether the molecule relaxes to a stable equilibrium structure on the S1’ surface during a significant structural reorganization or instead rapidly relaxes from S1’ to S0 through a curve crossing. Since reentry of S0 would deactivate electron transfer, it is important to determine whether highly coplanar geometries, despite their high HDA coupling, deactivate the excited state via an ultrafast non-radiative pathway such as a conical intersection.


3


Page 4 of 36

Figure 1. O-Ph and O-Th. The dihedral angle τ (red) is defined by the planes of the core and aryl bridge substituents (C10, C9, C14, and C15). The dihedral angle δ (green) is defined by the plane of the dimethylamine group and the core. Note that two resonance structures exist for these compounds, making the left and right dimethylamines equivalent in solution.

In this work, we clarify the photophysics of O-Ph and O-Th by providing an explicit characterization of the S1 potential surfaces. Direct comparison is made between the O-Ph and O-Th S1 surfaces along τ to demonstrate how the S1 dynamics depend on bridge structure. This is achieved by using optical lineshape modelling to extract the S1 equilibrium geometry in the Franck-Condon (FC) region from experimental resonance Raman (RR) excitation profiles, absorption, and emission spectra. We assume the independent mode displaced harmonic oscillator (IMDHO) model so that RR intensities can be directly related to photoinduced structural changes.31-38 The spectra are then calculated by DFT/TD-DFT at the same level of approximation33-34 and compared to experiment to determine the accuracy of different functionals in calculating S1 geometries. Importantly, this comparison of the experimentally modelled spectra to the TD-DFT results allows us to validate the shape of the excited-state surface predicted by TD-DFT with much more accuracy than is traditionally available. To reveal the energetic landscape beyond the FC region, potential energy scans along the torsional coordinate τ are carried out in S1.


4

Page 5 of 36 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 main results to be discussed in the following sections are (1) the sub-10 ps decay of excited state signal for O-Th is due to room temperature access of a coplanar global minimum on the S1 potential surface with low oscillator strength and red shifted emission energy, (2) the S1 global minimum is far away from degeneracy with S0 explaining the long ca. 1 ns S1 lifetime of O-Th despite its highly distorted geometry, (3) at coplanarity O-Th forms an ICT state with HOMO intensity localized on the core and LUMO intensity more localized on the thiophene moiety, and (4) O-Ph is sterically hindered from rotation to coplanar geometries in S1, inhibiting its potential for strong coupling to an acceptor. The torsional flexibility of thiophene compared to phenylene is a general result that can be used to explain extended conjugation lengths common in thiophene containing donor-π-acceptor systems and guide the design of new sensitizers. This point is elaborated in the final section where the observed photodynamics are interpreted in the context of light harvesting by dyesensitization of semiconductor nanoparticles. 2. Results and Discussion

2.1 Experimental and modelled cross sections. Femtosecond stimulated Raman spectroscopy (FSRS) has been established as a technique which enables collection of ground and excited state RR spectra free of strong fluorescence background signals.39-40 FSRS spectra of the ground electronic state were collected at multiple excitation wavelengths within the visible absorption bands of O-Ph and O-Th in order to fully map the RR excitation profiles. Absolute RR scattering cross-sections were obtained using the expression 41

=

!" #$ ! & " #%

(1)

with,

'∥ ) *

= +∥ ) , =

∥ + ,

∥ + ,

(2)

where only the parallel Raman intensities, - ‖ , were collected. The intensity was obtained by fitting the baseline corrected, solvent subtracted FSRS spectrum to a sum of Gaussians or Lorentzians and then integrating the area under each peak. The differential scattering cross sections of the 918 cm-1 acetonitrile band at each of the Raman pump wavelengths were extrapolated from previous literature results (1.15, 0.939, 0.775, 0.646, 0.581 x 10-30


5


Page 6 of 36

Å2/(molecule sr) for 510, 535, 560, 585, and 600 nm respectively).42 All excitation energies are assumed to be resonant with only one electronic transition so that depolarization ratios / are set equal to 0.33 for each solute peak.28 The known / of 0.01 was used for the MeCN

peak.41-42 Time-domain methods for computing absorption and RR lineshapes have been thoroughly explored.32, 37-38, 43 The model used to fit the experimental spectra utilizes a Brownian oscillator model of solvation44-45 and explicitly accounts for temperature effects.37 The adjustable parameters include the FC displacements ∆, homogeneous linewidth Γ, static

inhomogeneous linewidth Θ, transition length 2, and 0-0 vibronic energy gap 34. A detailed description can be found in the SI. The FC displacements, ∆, represent the structural changes induced along vibrational normal coordinates upon photoexcitation. The ∆’s for O-Ph and O-Th were first estimated from the normalized relative scattering cross-sections. ∆’s, as well as electronic and solvent model parameters, were adjusted until the best fit to the RR excitation profiles, absorption spectrum, and the Stokes shift between absorption and emission maxima was achieved. Figure 2 shows the experimental absorption and emission spectra and RR excitation profiles (blue) as well as the resulting fits (red) for O-Ph and O-Th. The set of fit parameters for OPh and O-Th are summarized in Table 1.


6

Page 7 of 36 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. RR spectrum, RR excitation profiles, and absorption/emission spectrum of O-Ph and O-Th. (a) 600 nm (16,667 cm-1) Raman pump FSRS spectrum (blue) and the fit used to compute absolute cross sections (grey fill) of O-Ph. The asterisk marks a residual feature from solvent subtraction. (b)-(d) O-Ph experimental RR cross sections of the most intense vibrational modes (labeled in (a)) at each Raman pump wavelength (blue circles) with the corresponding RR excitation profile calculated using parameters from Table 1 (red lines). RR profiles are offset for comparison. (e) The experimental (blue solid) and calculated (red dashed) absorption and emission spectra. Emission spectra are calculated stimulated emission spectra multiplied by 5 for comparison to spontaneous emission and scaled to match absorption intensity (see text). (f)-(j) are the corresponding plots for OTh.

Table 1. O-Ph and O-Th experimental Raman frequencies and displacements used to model the experimental spectra. O-Ph

O-Th

Freq. (cm-1)

∆(a)

Freq. (cm-1)

∆(b)

242

0.8

251

0.8

290

0.65

290

0.65

342

0.52

347

0.6

424

0.15

450

0.15

507

0.45

539

0.19

506

0.34

580

0.12

574

0.15

635

0.4

631

0.39

668

0.22

742

0.21

802

0.081

745

0.21

850

0.13

838

0.11

932

0.095

912

0.1

965

0.095

966

0.15

1125

0.17

1142

0.1

1213

0.29

1214

0.37

1287

0.19

1275

0.15

1320

0.17

1360

0.27

1364

0.22

1414

0.15

1410

0.12

1449

0.1

1450

0.13

1496

0.15

1535

0.31

1519

0.34


7


1596

0.15

1594

0.15

1647

0.36

1646

0.33

Page 8 of 36

(a) O-Ph ∆’s were established after fitting the RR excitation profiles, absorption, and emission spectra with the following parameters: 34 = 17,750 cm-1, 2 = 1.7 Å, Θ < 50 cm-1, 6 = 730 cm-1 (7-1 = 32 fs, 8 = 315 cm-1, 9 solv = 240 cm-1), 9int = 677 cm-1, 9 tot = 917 cm-1, 9stokes = 802 cm-1. (b) O-Th: 34 = 17,250 cm-1, 2 = 1.83 Å, Θ < 50 cm-1, 6 = 630 cm-1 (7-1 = 27 fs, 8 = 272 cm-1, 9 solv = 179 cm-1), 9 int = 761 cm-1, 9tot = 940 cm-1, 9stokes = 762 cm1 . For both molecules, the refractive index : was set to the value for MeCN of 1.33. ; = 0.1, and < = 298 K. All fit parameters are assumed to have error within 10% of their listed values.

For O-Ph and O-Th, good agreement is obtained regarding the shape and absolute intensity of the absorption spectrum, fluorescence Stokes shift, and RR excitation profiles. Compared to O-Ph, O-Th has a more intense FC shoulder in the absorption and emission spectra (Figure 2 (e) and (j)) suggesting a larger overall internal reorganization energy, λint, as observed (Table 1). The source of the difference in λint between the two molecules is made clear by their respective spectral distributions of ∆. Table 1 shows that O-Th has many more observed features than O-Ph. In O-Th, some strong features that were also observed in O-Ph appear to have been split apart allowing for a clear distinction. For instance, the peak at 635 cm-1 in O-Ph consists of one strong band. In O-Th however, at 631 cm-1, the peak intensity is slightly lowered but broadened with a strong side band at 668 cm-1. This redistribution of intensity occurs throughout the mid-frequency region between 240 and 1000 cm-1. To determine if this region is the source of the difference in internal reorganization energy, or simply a redistribution of the same amount of reorganization energy, we have calculated λint associated with the 240-1000 cm-1 and 1000-1700 cm-1 regions separately. The results of this calculation, shown in Table 2, are that the 77 cm-1 difference in internal reorganization energy between O-Ph and O-Th is almost entirely due to enhanced RR activity in the < 1000 cm-1 region. Since the two molecules only differ in their aryl components, we can infer from this analysis that the different λint in the < 1000 cm-1 region reflects differences in structural distortions along normal modes which contain displacements on both core and aryl moieties.


8

Page 9 of 36 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


Table 2. Internal reorganization energies λint for different spectral regions of the O-Ph and O-Th vibrational spectrum. λint (cm-1)

240-1000 cm

-1

1000-1700 cm Total

-1

O-Ph

O-Th

297

374

380

387

677

761

O-Th was fit with a lower solvent reorganization energy (9 = 179 cm-1, 6 = 630

cm-1) than O-Ph (9 = 240 cm-1, 6 = 730 cm-1) to balance the increase in linewidth resulting from a larger internal reorganization energy. The differences in solvent

reorganization energy observed between these molecules reflect differences in the solvent response to the electronic transition. The difference is however very small, hardly affecting the solvent relaxation timescale 7-1, suggesting that the solvent relaxation processes of O-Ph and O-Th are very similar. The static inhomogeneous broadening parameter, =, represents fluctuations in the local environment on time-scales slower than the RR process and, for similar molecules, it should mostly change as a function of the solvent. Previously, rhodamine 6G was fit using = = 300 cm-1 in a methanol/water mixture.35 When we included temperature in our simulations, we found that O-Ph and O-Th did not require any additional static inhomogeneous broadening (i.e. Θ < 50 cm-1), suggesting that the previously determined Θ for rhodamine 6G was actually capturing the effects of temperature broadening due to vibrational excitation in S0. The Stokes shift, which results from both internal and solvent reorganization, reproduces the experimental Stokes shift quite well for both O-Ph and O-Th. The FC shoulders in the experimental emission spectra of O-Ph and O-Th show lower intensity than absorption. This is due to the lower density of states at lower emission energies which reduces emission amplitude in this region. To account for this in the simulation, the calculated emission cross sections have been scaled by a factor of 5 so that they are representative of the spontaneous emission spectrum.46-48 Upon scaling, the calculated emission spectra are in good agreement with the experimental spontaneous emission spectra.


9


Page 10 of 36

A comparison of the experimental emission bands of O-Ph and O-Th is made in Figure S1 which shows that although O-Ph and O-Th have similar absorption linewidths, the emission linewidth of O-Th is broadened on the red side with a long tail extending out to the detector limit at 800 nm (< -2000 cm-1 in Figure S1). The significance of this result will be discussed in later sections. Overall, the lineshape modelling results provide an experimentally determined S1 equilibrium structure which can be used to verify DFT/TD-DFT models of the excited state surface.

2.2 Raman intensities by DFT/TD-DFT. The first five electronic transitions were calculated in vacuo using functionals B3LYP, BHLYP, PBE0 and BP86. The orbital contribution, oscillator strength, and energy for each transition can be found in Tables S1 and S2. The highest oscillator strength transition is found to be the lowest energy transition and is dominated by a single HOMO → LUMO transition for each molecule and functional except BP86. BP86 predicts the lowest energy transition to have much lower oscillator strength. The second lowest energy transition predicted by BP86 is the high oscillator strength, HOMO → LUMO transition for both O-Th and O-Ph. The calculated HOMO → LUMO transition is found to have a substantially higher transition energy than the observed values, regardless of functional, consistent with previous calculations performed with polarizable continuum solvation.30, 49 Therefore, in comparing the calculated absorption and emission spectra and RR excitation profiles to experiment, we have set all 34 values equal to the value determined from the modelling of the experiment. In this way, we are assessing the agreement between the DFT absorption and RR lineshapes rather than just the E0 value. Similarly, the broadening parameters and transition lengths for each molecule are kept the same as the experimental when calculating the lineshapes for the DFT spectra. Agreement between the DFT calculated and experimental dimensionless displacements can be assessed by comparing the internal reorganization energies, the shape of the absorption spectra, the Stokes shifts, and the shape of the Raman spectra taken at the same excitation wavelength within the RR excitation profile. Table S3 compares the internal reorganization energies and Stokes shifts between experiment and DFT for each functional for both O-Ph and O-Th. Figure 3 shows the experimental and DFT absorption and RR spectra for O-Ph and O-Th.


10

Page 11 of 36 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 3. Comparison between calculated and experimental spectra: The experimental and DFT absorption and emission spectra (a) and RR spectra (b) of O-Ph, as well as those for O-Th, (c) and (d). The experimental absorption spectrum is shown in blue with the modeled spectra superimposed in red. Above the experimental and modelled spectra are the DFT spectra calculated using the labeled functional. In (b) and (d), the corresponding RR spectra are shown for a 560 nm Raman pump wavelength for each functional and the experiment; functionals appear in the same order as in (a). The normal mode frequencies for each functional have been scaled according to literature scaling factors: B3LYP 0.9613, BHLYP 0.9244, BP86 0.9914, PBE0 0.9512.50

For O-Ph and O-Th, the agreement in the shape of the absorption spectrum with the observed spectrum is generally good for each functional (Figure 3 (a) and (c)). The differences are most apparent in the FC shoulder suggesting that each functional calculates a different distribution of ∆’s for the high- and low-frequency modes. To explore this, the RR spectrum taken with a 560 nm excitation wavelength is also shown in Figure 3 (b) and (d) along with the DFT results. B3LYP, BP86, and PBE0 are all similar in that they exhibit high intensity in the high- and low-frequency regions of the RR spectrum and low intensity in the


11


Page 12 of 36

mid-frequency range. The high intensity in the high- and low-frequency regions agrees well with experiment and contributes to the large λint and λStokes. The lack of intensity in the midfrequency is inconsistent with experimental results and represents a shortcoming of these functionals’ ability to accurately represent the FC force in this spectral region. The BHLYP functional shows weak high-frequency intensity which explains its lack of absorption intensity in the FC shoulder in the absorption spectrum and its comparatively low λint and λStokes (Table S3). The mid- and low-frequency regions however demonstrate excellent agreement with experiment both in the relative intensities of each peak as well as the absolute intensity of all the peaks. BHLYP is the only functional found to reproduce the lowfrequency triplet at ~300 cm-1 as well as the correct relative intensities of the series of midfrequency peaks spanning from 500 cm-1 to 1000 cm-1. Based on the analysis of Table 2, the low- and mid-frequency regions contain the normal modes which most strongly differ between O-Ph and O-Th. These modes are shown in Figure S2 to carry the largest contribution to displacement along τ as well as the C9-C14 bond, bC9-C14, making their accurate characterization critical to describing S1 coplanarization. Due to near quantitative agreement in this spectral region, we consider BHLYP to be the most suitable functional for accurate calculations of the S1 potential surfaces along τ.

2.3 Impulsive vibrational spectra. Impulsive vibrational spectroscopy (IVS) is a technique for collecting resonant and off-resonant Raman spectra in the time-domain by impulsively exciting coherent vibrational wavepackets on ground or excited state surfaces.51-53 DFT predicts very low-frequency normal modes, below 200 cm-1, with significant RR intensity. This low-frequency region is inaccessible by FSRS due to scattering of the Raman pump. To verify whether these lowfrequency peaks predicted by DFT are experimentally observed, we collected IVS spectra with a pre-resonant pump pulse, centered at 620 nm. The pre-resonance of the pump pulse ensures that we observe resonance enhancement of the ground state while avoiding contamination with transient absorption and oscillatory signals from the excited state. Figure S3 (a) shows the wavepacket motion apparent in the time domain as oscillations at wavelengths around the ground state absorption band of O-Ph and O-Th. For every probe wavelength, a Fourier transform is carried out along the time axis resulting in a probe-dependent Fourier transform spectrum (Figure S3 (b) and (e)) with bands of intensity


12

Page 13 of 36 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


occurring at vibrational frequencies of the dyes. To compare the IVS spectra to FSRS and DFT RR spectra, we have summed the Fourier transform magnitude along the probe wavelength axis within the absorption band range of O-Ph from 520 to 580 nm and from 540 to 600 nm for O-Th. Figure 4 compares the IVS, FSRS, and BHLYP DFT spectra of O-Ph and O-Th.

Figure 4. Comparison of O-Ph (a)-(c) and O-Th (d)-(e) IVS (620 nm), FSRS (600 nm), and BHLYP DFT (620 nm). O-Ph: (a) IVS Fourier transform magnitude summed from 520 to 580 nm (black) compared to FSRS (red). (b) IVS spectrum as in (a) compared to the BHLYP DFT Raman spectrum simulated with a 620 nm Raman wavelength (blue). (c) Same as (b) zoomed into the low frequency region. In (a) and (b), the spectra have been scaled to normalize the intensity of the 635 cm1 peak. O-Th: (d) Same as (a) for O-Th with the sum going from 540 to 600 nm. The spectra have been normalized to the 631 cm-1 peak. (e) and (f) are the same as (b) and (c) but for O-Th.

Figure 4 (a) and (d) demonstrates the excellent frequency alignment between the peaks in IVS and FSRS spectra of O-Ph and O-Th respectively. There is also very good agreement in the relative intensities, especially in the mid-frequency range between 450 cm-1 and 1000 cm-1. Figure 4 (b) and (e) compares O-Ph and O-Th IVS to BHLYP DFT and shows similar agreement in the mid- and high-frequency regions. Figure 4 (c) and (f) makes the same comparison but in the very low-frequency region. Although the relative intensities between IVS and BHLYP DFT Raman are not in agreement here (a scaling factor of 25 was required to see the IVS features around 100 cm-1), IVS display similar low-frequency features to those predicted by BHLYP DFT. Since the magnitude of these features is on the order of the noise, assignment becomes complicated. The low-frequency features can however be unambiguously differentiated from surrounding noise by considering the wavelength dependence of the Fourier transform signal in the frequency–wavelength contour plot


13


Page 14 of 36

(Figures S4 and S5 for O-Ph and O-Th respectively). Pre-resonant wavepacket motion should produce Fourier transform signal centered around the ground state absorption band of the molecule, typically in the form of lobes of intensity centered around a node. Figure S4 clearly shows that features at 52 and 150 cm-1 are peaks of O-Ph, since they result from a band of intensity centered at the molecule’s absorption. These peaks are labeled in Figure 4(c). The features around 100 and 120 cm-1 have a much less clearly defined structure in the contour plot and therefore we refrain from assigning these as peaks. In Figure 4 (f), many features below 190 cm-1 can also be assigned as peaks of O-Th, since they are centered at the absorption wavelengths and several of them also exhibit a nodal structure in the IVS 2D Fourier transform contour plot (Figure S5). These features are labeled in Figure 4 (f). Peaks around 150 cm-1 do not exhibit a clear wavelength dependence and are therefore not assigned as peaks.

2.4 Structural changes from Resonance Raman. The experimental and DFT calculated FC displacements ∆ represent the photoinduced structural changes along each vibrational normal coordinate. These changes are more conveniently analyzed after transformation to internal coordinates, which can be carried out by applying Eq. 20 of the SI. Further, analysis of the frontier molecular orbitals involved in the transition and their resultant FC energy gradient can be used to rationalize the internal coordinate displacements. Figure S6 shows the HOMO and LUMO molecular orbitals, as well as the FC gradient vectors at the S0 equilibrium geometry for O-Ph and O-Th. At the FC geometry, the HOMO→LUMO transitions of both molecules are found to be localized on the xanthilium core chromophore, with a small amount of LUMO amplitude appearing on the aryl groups. The gradients shown in Figure S6 show that the S0→S1 transition induces rotation along τ in addition to contraction of bC9-C14. To determine which peaks in the RR spectrum are sensitive to changes along these internal coordinates, we have converted the dimensionless normal mode displacements computed by BHLYP DFT to internal coordinate displacements along τ, δ, bC9- C14 and bC9-C10. . The values of these coordinate changes predicted by DFT/TD-DFT, determined by extrapolating the FC slope to predict the equilibrium geometry in the IMDHO model, along with comparison to values from a direct geometry optimization on the S1 surface are shown in Table 3. The S1 geometry optimizations using the BHLYP functional show rotation around τ towards coplanarity as


14

Page 15 of 36 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


well as contraction of bC9-C14 and expansion of bC9-C10 compared to S0 for both O-Ph and OTh. The δ coordinate is nearly unaffected by the transition, remaining in the plane of the core near δ = 0°. Table 3. Internal coordinates positions of the O-Ph and O-Th S0 X-ray, DFT S0 and S1 minimum and S1 IMDHO geometries τ (°) δ (°) bC9-C14 (Å) bC9-C10 (Å)

S0 X-ray

(a)

DFT S0 (b) TD-DFT S1 TD-DFT S1 IMDHO.(d)

(c)

O-Ph

O-Th

O-Ph

O-Th

O-Ph

O-Th

O-Ph

O-Th

59.8(2)

49.3(4)

0.7(9)

2.9(3)

1.483(4)

1.472(4)

1.4108(13)

1.405(4)

66.4

61.1

0.5

0.7

1.4797

1.4658

1.4007

1.4030

56.3

52.1

0.9

0.9

1.4675

1.4471

1.423

1.4210

30.0

27.6

1.4

0.2

1.4437

1.3668

1.5451

1.4814

(a) X-ray crystallography structure taken from ref. 30 (b) S0 equilibrium geometry using the BHLYP functional (c) S1 equilibrium geometry using the BHLYP functional. (d) Calculated by converting the dimensionless displacements determined by the FC gradient in BHLYP into the displacement along the specified internal coordinate from the BHLYP ground state using Eq. 20 of the SI.

We find that DFT/TD-DFT in the IMDHO model overestimates the displacement along the angle τ by about 30° for both O-Ph and O-Th. This is to be expected given the strongly anharmonic nature of torsional degrees of freedom and likely reflects a breakdown in the assumption of harmonic potential surfaces along every normal mode. The IMDHO model also slightly overestimates the contraction of bC9-C14 and expansion of bC9-C10, likely due to similar shortcomings of the harmonic approximation. However, the errors are small (< 6%) since the high-frequency backbone stretching modes that contribute to the bC9-C14 and bC9-C10 displacements are expected to be much closer to harmonic compared to the normal modes contributing to aryl rotations.

2.5 Potential energy surfaces along τ The S1 geometry optimization results for BHLYP show rearrangement about τ towards a more coplanar structure from the S0 minimum geometry, in agreement with our previous hypotheses about motion in S1.30 Although this change places O-Th at a more coplanar geometry than O-Ph, it is unclear what the subsequent mechanism is that results in such drastically different excited state dynamics. To explore the effect of motion along τ on the excited state dynamics, we carried out relaxed potential scans along this coordinate in S0


15


Page 16 of 36

and S1 using the functionals described above. Since the DFT BHLYP spectra are in nearquantitative agreement with features between 150 and 800 cm-1, we choose to emphasize analysis of these results. See the SI Section 5 for S0 and S1 surfaces of other functionals and a discussion about how they differ from BHLYP. The S0 and S1 potential scans for O-Ph and O-Th are shown in Figure 5.

Figure 5. S0 and S1 relaxed potential scans along τ. BHLYP S0 and S1 relaxed potential energy scans along the coordinate τ of (a) O-Ph and (b) O-Th with their Boltzmann populations at 298K in colorscale and oscillator strength at the S1 geometry (green dashes). Black arrows show the vertical excitation and relaxed emission. Red circles correspond to the energy in the FC region, that is the energy in S1 at the S0 minimum geometry. Structures at the FC and coplanar minima for O-Ph and O-Th are provided in Figure S13. Coordinate files of these structures as well as the S0 minimum structures are also provided in the SI to enable visualization from multiple perspectives.

In Figure 5 the difference between the FC geometry and the S1 relaxed geometry at the FC τ reflects reorganization along non-τ coordinates. Energetic differences between these points are 0.039 eV and 0.043 eV for O-Ph and O-Th respectively. Once relaxation along τ is allowed, there is an additional relaxation to a local minimum at a slightly more coplanar geometry. The energy difference between the FC point and the local S1 minimum of O-Ph and O-Th are 0.0475 eV (383 cm-1) and 0.0484 (390 cm-1) eV respectively, in close agreement with the internal reorganization energies calculated by the IMDHO model using the BHLYP functional. Figure 5(a) shows the S0 and S1 τ scans for O-Ph. The S0 potential scan is in agreement with previous ground state scans of O-Ph.30 Namely, there are two minima at 66° and 114° (-66°), symmetric around a small 28 meV barrier at 90°. The barrier to rotation becomes steeper as τ approaches 14° before relaxing into a shallow, high-energy minimum at


16

Page 17 of 36 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


0°. In S1, the minimum τ is closer to coplanarity at 56° with a 32 meV barrier at 90°. A large energetic barrier (220 meV) peaked at 31° must be overcome to access a second minimum in the potential surface at 4.5°. The magnitude of this energetic barrier is well above kBT at room temperature which suggests that the initial rearrangement in the FC region will lead to a stable excited state with τ near 56° rather than accessing any highly-coplanar geometries near τ = 0°. For O-Th (Figure 5(b)), we find a similar S0 potential surface with two minima at 61° and 119° (-61°) symmetric around a small 19.5 meV barrier at 90°. In agreement with previous results for O-Th30, we find a softer barrier as τ approaches a high-energy minimum at 0°. Like O-Ph, the S1 equilibrium τ is displaced to a more coplanar geometry at τ = 52°, and has a higher potential barrier of 31 meV at 90° than in S0. The key difference between OPh and O-Th is that as O-Th approaches τ = 0° in S1, it must overcome a much smaller barrier of 10 meV at 35°. Beyond this barrier is a second minimum at 8° which is lower in energy than the local minimum near the FC point. Since the barrier at 35° is well below kBT at room temperature, and the τ = 8° minimum is more energetically stable than the FC minimum, at thermal equilibrium most of the population will be at the τ = 8° geometry. This is represented by the Boltzmann distribution color map in Figure 5 where, for O-Th, the ratio of population at the initial FC minimum to the τ = 8° minimum is 0.05, while all O-Ph population remains near the FC minimum. Previously, our group carried out transient absorption and steady state spectroscopy on O-Ph and O-Th in CH2Cl2 and MeCN.30 Given our new results, a few key results from our prior transient absorption study30 can now be interpreted to establish how the excited states of O-Ph and O-Th evolve: (1) In MeCN, the O-Ph excited state decays with a long singlet lifetime of 2111 ps, while O-Th exhibits a long singlet lifetime of 1044 ps after an initial 6.25 ps reorganization event. With the 6.25 ps time constant, the ESA decays to half of its amplitude and the SE immediately to the red of the ground state bleach (GSB) loses amplitude. (2) The short time constant slows to ~40 ps in a 1:1 (v:v) mixture of MeOH and glycerol. (3) Both dyes have longer lived excited states and higher quantum yields in the less polar CH2Cl2 solvent compared to MeCN. This is attributed to formation of a charge separated state between the dimethylamines and the core which will be stabilized by a polar solvent thus opening alternative decay channels from S1. However, the O-Th quantum yields and lifetimes are much lower than those of O-Ph in either solvent suggesting that O-Th has an additional decay pathway.


17


Page 18 of 36

Sabatini et. al. have suggested that the key to the differences in the excited state dynamics between O-Ph and O-Th lie in the appearance of the short time constant only observed for O-Th.30 It was proposed that the rotational flexibility of the thiophene group around τ allows for access of a new excited state with a delocalized π system and different radiative and non-radiative couplings to the ground state. From the new excited state, new non-radiative decay channels are accessed thus reducing the overall quantum yield. O-Ph due to its rigidity around τ does not access a similar coplanar geometry and thus does not exhibit the same excited state depopulation dynamics. Our more extensive S1 potential energy scans allow for a more precise understanding of the nature of the new excited state formed in O-Th proposed by Sabatini et. al.. O-Th remains on the S1 surface, but thermally accesses a global minimum at coplanarity after overcoming a small energetic barrier. At coplanarity, there is an increased non-radiative coupling to S0, by the energy gap law, resulting in the faster recovery of GSB compared to OPh. The energy of the emission is also a function of rotation about τ. The BHLYP calculated energy gap of the local minimum near the FC region of O-Th is 3.17 eV, but at the S1 global minimum the energy gap is only 2.4 eV. When corrected for BHLYP’s overestimation of the energy gap compared to experiment, this would correspond to a new stimulated emission band appearing in the near-IR region with very low intensity (f ≈ 0.5), consistent with the loss of the visible SE band and appearance of weak SE in the NIR region observed in prior studies.30 This is also in support of the red-side broadening of the spontaneous emission spectrum observed for O-Th (Figure S1). Furthermore, the 2.4 eV energy gap indicates that the newly formed coplanar O-Th in S1 will not rapidly access S0 through a curve crossing, but will instead remain stable, consistent with the observed 1 ns lifetime. The loss of the fluorescent population of S1 near the FC geometry to the lower energy coplanar minimum will tend to reduce the observed fluorescence quantum yield measured via the visible fluorescence. For O-Th at thermal equilibrium in S1, the Boltzmann ratio of population in the initial geometry to that of the coplanar geometry is 0.05. The reduced population of the initial S1 geometry results in the observed low fluorescence quantum yield (ΦFL = 0.113) but the amount of population at the initial geometry does seem to be too small to account for the observed quantum yield. It is known that TD-DFT tends to stabilize charge-transfer (CT) states54-55 and, as will be shown below, O-Th develops CT character at the coplanar geometry due to delocalization of electron density onto the thiophene group in the LUMO relative to a localized HOMO. Therefore, the calculated energy at the coplanar


18

Page 19 of 36 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


minimum may contain slight errors that tend to place the coplanar minimum energy too low relative to the minimum near the FC geometry. Nevertheless, the qualitative picture provides insight into the relaxation pathway along τ. Furthermore, our results are consistent with previous configuration interaction with double excitation (CID) studies on BODIPY analogs with varying degrees of steric hindrance between the chromophore and aryl moieties which found that only the most flexible aryl groups exhibited a thermally accessible coplanar potential minimum with potential surfaces similar in topography to those calculated for O-Th and O-Ph.56 In ref. 56 it was also determined that the lowest lying transition was dominated by a simple HOMO to LUMO single excitation at both perpendicular and coplanar (BODIPY-core)-(BODIPY-aryl) geometries. This mirrors our results where the HOMOLUMO transition accounts for 93% and 96% of the lowest lying transition at 90° and 0° respectively. This finding supports our use of TD-DFT to calculate the potential surface of the S1 state since there should be no build-up of strong double excitation character at coplanarity which could result in failure to describe a conical intersection.55 Figure 5 also shows the oscillator strength of O-Ph and O-Th as a function of τ (green dashes). For both molecules, the oscillator strength at the S0 minimum τ is around f = 1.3 while at the S1 minimum τ it decreases to f = 1.2. Rotation towards τ = 0° results in a further decrease in oscillator strength as the molecule accesses an optically “dim” geometry where the oscillator strength is < 0.6. The oscillator strength for O-Th at the coplanar minimum τ = 8° is only 0.53. The dependence of oscillator strength on τ was found to follow almost the exact same dependence as the bC9-C14 length and τ. This is shown in Figure S14. This can be understood by considering the delocalization of the LUMO at coplanar geometries, shown in Figure 6. At coplanar τ, the LUMO extends onto the aryl group with large amplitude occurring at the C9 position. Relative to the HOMO, this results in much more double bond character at and contraction of bC9-C14. The oscillator strength follows the bC9-C14 and τ dependence because as the π system becomes delocalized, the oscillator strength becomes redistributed over a larger chromophore. At coplanarity, O-Th puckers to accommodate the steric hindrance between the aryl hydrogens and the core hydrogens (Figure S13). The LUMO is significantly delocalized onto the aryl group at coplanarity while the HOMO is entirely localized on the core. This suggests that an intramolecular charge transfer (ICT) state is formed as τ rotates into coplanarity. This is further illustrated by the S1 difference density plot shown in Figure S15(e). Note that our calculations are performed in vacuo, but inclusion of a solvent will further stabilize the ICT


19


Page 20 of 36

state relative to the FC minimum. The unique formation of an ICT state at coplanarity by OTh will be discussed in the following section in the context of light harvesting by dye sensitization.

Figure 6. HOMO and LUMO molecular orbitals of O-Th at the coplanar S1 minimum. (a) The LUMO is delocalized onto the aryl group while in (b) the HOMO remains localized on the core, even at coplanarity. The LUMO → HOMO emission at coplanarity is thus an ICT transition. See Figure S15 for a comparison to the FC geometry as well as corresponding difference density plots.

3. Implications for Light Harvesting In past studies, O-Ph, O-Th and related dyes have been used as sensitizers for DSSC and solar hydrogen production systems in which the dye is bound to a TiO2 nanocrystal via an acidic functional group on the aryl bridge.1-2, 5, 7-8 The initial photogeneration of excited state population in these systems is followed by the crucial process of electron transfer to the semiconductor nanoparticle. Electron transfer must occur on timescales shorter than the excited state lifetime of the sensitizer which requires that the dyes be very strongly coupled to ACS Paragon Plus Environment

20

Page 21 of 36 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 conduction band acceptor of the semiconductor. This must also however be balanced by mitigation of back electron transfer where the hole on the sensitizer is recombined with the electron on the acceptor. To achieve an optimal balance between forward and backward electron transfer rates, a bridge spacer is introduced between the donor and acceptor. As stated previously, the strongest coupling between donor and acceptor is achieved when the π systems of donor and bridge are coplanar (in the case of molecular acceptors coplanarity with the acceptor further enhances electronic coupling).17, 19 We have shown that rhodamine sensitizers with thiophene bridge units exhibit unique coplanarization dynamics in S1 which, in less than 10 ps, result in the formation of an ICT state where LUMO amplitude is delocalized onto the bridge. For dyes bound to a semiconductor surface, this delocalization would place electron density in close proximity to the acceptor, enhancing the forward electron transfer process. We have shown that geometries with strong donor-bridge electronic coupling can be rapidly accessed and are in fact more stable than non-coplanar geometries when steric hinderance is low, as for O-Th. Strong coupling for forward electron transfer however also suggests strong coupling for backward electron transfer, or electron-hole recombination. To fully understand how the forward and backward electron transfer rates from a coplanar donor-bridge sensitizer compare requires an understanding of how the sensitizer geometry changes upon electron transfer. Charge separation between O-Th and a semiconductor acceptor will place O-Th in the coplanar, oxidized ground state. However, because the coplanarization of O-Th is caused by population of the LUMO, once the LUMO is depopulated by electron transfer, the molecule is expected to relax back towards a τ angle closer to the S0 equilibrium. Calculating the minimum geometry of the oxidized ground state, 84> , starting from the S1 coplanar geometry, we find that the 84> equilibrium τ is 60°

(Figure S16), close to the S0 τ of 61°. Hence, this reorganization from the hot, coplanar 84> would effectively turn off the strong coupling between the thiophene and the xanthylium core, isolate the hole away from the semiconductor surface, and therefore reduce the rate of back electron transfer. This mechanism where back-electron transfer is mitigated by relaxation away from strongly coupled coplanar geometries could explain the relative success of thiophene units as bridges for donor-π-acceptor charge separation. However, to fully understand the forward and backward electron transfer processes requires explicit studies of dye sensitized semiconductor nanoparticles.


21


Page 22 of 36

Although other sensitizers utilizing thiophene bridge units may not exhibit exactly the same excited-state dynamics, the low steric hindrance and charge transport properties of thiophenes lends them generally to facilitating increased conjugation length within donor-πacceptor systems. Our explicit maps of the ground and excited state potential surfaces of OPh and O-Th exemplify these notions thus providing groundwork for the design of new dyesensitizers.

Conclusion

In order to establish the mechanism by which torsional coordinates of bridge moieties can influence photoinduced electron transfer rates in molecular sensitizers, RR excitation profiles of O-Ph and O-Th were collected using FSRS and modelled using time-dependent wavepacket theory. IVS spectra were collected to assign normal modes in the very-lowfrequency region below 200 cm-1. The results were compared with spectra calculated using DFT/TD-DFT under an IMDHO model using a variety of functionals. The functional found to best reproduce experiment in the low- and mid- frequency regions, BHLYP, was then used to calculate S1 potential surfaces of O-Ph and O-Th with respect to the dihedral coordinate τ. We found that the less sterically hindered O-Th rapidly accesses a second minimum on the S1 potential surface, and that as this minimum is accessed the oscillator strength of the transition is reduced and ICT character develops due to delocalization of the LUMO. The rapid coplanarization of O-Th in the excited state suggests that thiophene moieties are ideal as bridges in dye sensitization as they allow for access of geometries with strong electronic/exchange coupling between donor and acceptor. O-Ph is restricted from rotating towards coplanarity and is therefore unable to initiate strong donor-acceptor coupling. These results support the previous successes of thiophene as a bridge unit and encourage further development of torsionally flexible donor-bridge dye sensitizers.

Experimental Femtosecond Stimulated Raman Spectroscopy. Raman pump pulses (510, 535, 560, and 585 nm) were generated by a homebuilt two-stage noncollinear optical parametric amplifier (NOPA) pumped by the frequency doubled output of a regeneratively amplified Ti:sapphire laser (Spectra-Physics, 800 nm, 100 fs, 0.3 mJ/pulse, 1 kHz) The NOPA output was tunable in the range 500-650 nm with a 10 nm bandwidth, a much narrower bandwidth than typical, which was


22

Page 23 of 36 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


achieved by adding 1-cm of water to the white-light seed path to additionally chirp it prior to amplification by a ~120 fs 400 nm pulse. The output bandwidth of the NOPA was then dispersed by an 1800 gr/mm holographic grating and focused by a 30 cm focal length spherical lens through a 90%) by one singly-excited configuration (HOMO → LUMO) at all angles of τ. The singly-excited character, which cannot be confirmed by TD-DFT, is supported by previous CID torsional scans of BODIPY analogs which found the S0→S1 transition to be dominated by a single HOMO → LUMO transition at all arylcore torsional angles.

56

Acknowledgements The authors would like to thank Dr. Mark Kryman, Jacqueline Hill and Michelle Linder in the laboratory of Prof. Michael R. Detty at the University at Buffalo for synthesizing compounds O-Ph and O-Th. This work was supported by the U.S. National Science Foundation (NSF) via grant number CHE-1566080 and a Graduate Assistance in Areas of National Need (GAANN) fellowship from the U.S Department of Education (P200A150052). Supplementary Information Theoretical description of the model used to simulate optical lineshapes; TD-DFT transition properties, reorganization energies and normal mode decomposition; 2D IVS spectra; extended discussion on structural changes predicted by RR; S1 scans using other DFT functionals; relaxed bC9C14

dependence on τ scans; FSRS spectra of each compound at each wavelength with fits; Files

containing XYZ coordinates for structure visualization. Movies of O-Ph and O-Th evolution in S1 along τ. The code used to compute optical lineshapes can be downloaded at https://github.com/McCamant-group. *Corresponding Author: [email protected] References 1. Sabatini, R. P.; Eckenhoff, W. T.; Orchard, A.; Liwosz, K. R.; Detty, M. R.; Watson, D. F.; McCamant, D. W.; Eisenberg, R., From Seconds to Femtoseconds: Solar Hydrogen Production and Transient Absorption of Chalcogenorhodamine Dyes. J. Am. Chem. Soc. 2014, 136 (21), 7740-7750. 2. He, J.; Wang, J.; Chen, Y.; Zhang, J.; Duan, D.; Wang, Y.; Yan, Z., A dye-sensitized Pt@UiO66(Zr) metal-organic framework for visible-light photocatalytic hydrogen production. Chemical Communications 2014, 50 (53), 7063-7066. 3. McCormick, T. M.; Calitree, B. D.; Orchard, A.; Kraut, N. D.; Bright, F. V.; Detty, M. R.; Eisenberg, R., Reductive Side of Water Splitting in Artificial Photosynthesis: New Homogeneous Photosystems of Great Activity and Mechanistic Insight. J. Am. Chem. Soc. 2010, 132 (44), 1548015483. 4. Li, G.; Mark, M. F.; Lv, H.; McCamant, D. W.; Eisenberg, R., Rhodamine-Platinum Diimine Dithiolate Complex Dyads as Efficient and Robust Photosensitizers for Light-Driven Aqueous Proton Reduction to Hydrogen. J. Am. Chem. Soc. 2018, 140 (7), 2575-2586. 5. Kryman, M. W.; Nasca, J. N.; Watson, D. F.; Detty, M. R., Selenorhodamine Dye-Sensitized Solar Cells: Influence of Structure and Surface-Anchoring Mode on Aggregation, Persistence, and Photoelectrochemical Performance. Langmuir 2016, 32 (6), 1521-1532. 6. Mann, J. R.; Gannon, M. K.; Fitzgibbons, T. C.; Detty, M. R.; Watson, D. F., Optimizing the Photocurrent Efficiency of Dye-Sensitized Solar Cells through the Controlled Aggregation of


25


Page 26 of 36

Chalcogenoxanthylium Dyes on Nanocrystalline Titania Films. J. Phys. Chem. C 2008, 112 (34), 1305713061. 7. Mulhern, K. R.; Orchard, A.; Watson, D. F.; Detty, M. R., Influence of Surface-Attachment Functionality on the Aggregation, Persistence, and Electron-Transfer Reactivity of Chalcogenorhodamine Dyes on TiO2. Langmuir 2012, 28 (17), 7071-7082. 8. Mulhern, K. R.; Detty, M. R.; Watson, D. F., Aggregation-Induced Increase of the Quantum Yield of Electron Injection from Chalcogenorhodamine Dyes to TiO2. J. Phys. Chem. C 2011, 115 (13), 6010-6018. 9. Gong, J.; Liang, J.; Sumathy, K., Review on dye-sensitized solar cells (DSSCs): Fundamental concepts and novel materials. Renewable and Sustainable Energy Reviews 2012, 16 (8), 5848-5860. 10. Mishra, A.; Fischer, M. K. R.; Bäuerle, P., Metal-Free Organic Dyes for Dye-Sensitized Solar Cells: From Structure: Property Relationships to Design Rules. Angewandte Chemie International Edition 2009, 48 (14), 2474-2499. 11. Zhou, N.; Prabakaran, K.; Lee, B.; Chang, S. H.; Harutyunyan, B.; Guo, P.; Butler, M. R.; Timalsina, A.; Bedzyk, M. J.; Ratner, M. A.; Vegiraju, S.; Yau, S.; Wu, C.-G.; Chang, R. P. H.; Facchetti, A.; Chen, M.-C.; Marks, T. J., Metal-Free Tetrathienoacene Sensitizers for High-Performance DyeSensitized Solar Cells. Journal of the American Chemical Society 2015, 137 (13), 4414-4423. 12. Weiss, E. A.; Ahrens, M. J.; Sinks, L. E.; Gusev, A. V.; Ratner, M. A.; Wasielewski, M. R., Making a Molecular Wire: Charge and Spin Transport through para-Phenylene Oligomers. Journal of the American Chemical Society 2004, 126 (17), 5577-5584. 13. Zhao, Y.; Liang, W., Charge transfer in organic molecules for solar cells: theoretical perspective. Chemical Society Reviews 2012, 41 (3), 1075-1087. 14. Jin, S.; Snoeberger, R. C.; Issac, A.; Stockwell, D.; Batista, V. S.; Lian, T., Single-Molecule Interfacial Electron Transfer in Donor-Bridge-Nanoparticle Acceptor Complexes. The Journal of Physical Chemistry B 2010, 114 (45), 14309-14319. 15. Davis, W. B.; Ratner, M. A.; Wasielewski, M. R., Conformational Gating of Long Distance Electron Transfer through Wire-like Bridges in Donor−Bridge−Acceptor Molecules. Journal of the American Chemical Society 2001, 123 (32), 7877-7886. 16. Benniston, A. C.; Harriman, A.; Li, P.; Patel, P. V.; Sams, C. A., The effect of torsion angle on the rate of intramolecular triplet energy transfer. Physical Chemistry Chemical Physics 2005, 7 (21), 3677-3679. 17. Stasiw, D. E.; Zhang, J.; Wang, G.; Dangi, R.; Stein, B. W.; Shultz, D. A.; Kirk, M. L.; Wojtas, L.; Sommer, R. D., Determining the Conformational Landscape of σ and π Coupling Using paraPhenylene and “Aviram–Ratner” Bridges. Journal of the American Chemical Society 2015, 137 (29), 9222-9225. 18. Hanss, D.; Wenger, O. S., Conformational Effects on Long‐Range Electron Transfer: Comparison of Oligo‐p‐phenylene and Oligo‐p‐xylene Bridges. European Journal of Inorganic Chemistry 2009, 2009 (25), 3778-3790. 19. Kirk, M. L.; Shultz, D. A.; Stasiw, D. E.; Lewis, G. F.; Wang, G.; Brannen, C. L.; Sommer, R. D.; Boyle, P. D., Superexchange Contributions to Distance Dependence of Electron Transfer/Transport: Exchange and Electronic Coupling in Oligo(para-Phenylene)- and Oligo(2,5-Thiophene)-Bridged Donor–Bridge–Acceptor Biradical Complexes. Journal of the American Chemical Society 2013, 135 (45), 17144-17154. 20. H., D. J.; Yanrong, S.; Jun‐Ho, Y.; Tissa, S.; Elisa, D. O.; Stephen, B.; K., N. M.; R., M. S.; Michael, G., The Role of π Bridges in High‐Efficiency DSCs Based on Unsymmetrical Squaraines. Chemistry – A European Journal 2013, 19 (5), 1819-1827. 21. Lee, M.-W.; Kim, J.-Y.; Lee, D.-H.; Ko, M. J., Novel D-π-A Organic Dyes with Thieno[3,2b]thiophene-3,4-ethylenedioxythiophene Unit as a π-Bridge for Highly Efficient Dye-Sensitized Solar Cells with Long-Term Stability. ACS Applied Materials & Interfaces 2014, 6 (6), 4102-4108.


26

Page 27 of 36 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


22. Li, W.; Wu, Z.; Wang, J.; Zhang, W.; Wu, M.; Zhu, W., Organic sensitizers with different thiophene units as conjugated bridges: molecular engineering and photovoltaics. Science China Chemistry 2017, 60 (2), 231-236. 23. Ren, X.; Jiang, S.; Cha, M.; Zhou, G.; Wang, Z.-S., Thiophene-Bridged Double D-π-A Dye for Efficient Dye-Sensitized Solar Cell. Chemistry of Materials 2012, 24 (17), 3493-3499. 24. Shen, P.; Liu, X.; Jiang, S.; Wang, L.; Yi, L.; Ye, D.; Zhao, B.; Tan, S., Synthesis of new N, Ndiphenylhydrazone dyes for solar cells: Effects of thiophene-derived π-conjugated bridge. Dyes and Pigments 2012, 92 (3), 1042-1051. 25. Shen, P.; Liu, Y.; Huang, X.; Zhao, B.; Xiang, N.; Fei, J.; Liu, L.; Wang, X.; Huang, H.; Tan, S., Efficient triphenylamine dyes for solar cells: Effects of alkyl-substituents and π-conjugated thiophene unit. Dyes and Pigments 2009, 83 (2), 187-197. 26. Wang, X.; Sun, Y.; Chen, S.; Guo, X.; Zhang, M.; Li, X.; Li, Y.; Wang, H., Effects of π-Conjugated Bridges on Photovoltaic Properties of Donor-π-Acceptor Conjugated Copolymers. Macromolecules 2012, 45 (3), 1208-1216. 27. Wu, W.; Zhang, J.; Yang, H.; Jin, B.; Hu, Y.; Hua, J.; Jing, C.; Long, Y.; Tian, H., Narrowing band gap of platinum acetylide dye-sensitized solar cell sensitizers with thiophene [small pi]-bridges. Journal of Materials Chemistry 2012, 22 (12), 5382-5389. 28. Yuan, Y.; Chen, J.-X.; Chen, W.-C.; Ni, S.-F.; Wei, H.-X.; Ye, J.; Wong, F.-L.; Zhou, Z.-W.; Tong, Q.-X.; Lee, C.-S., The locally twisted thiophene bridged phenanthroimidazole derivatives as dualfunctional emitters for efficient non-doped electroluminescent devices. Organic Electronics 2015, 18, 61-69. 29. Li, Y., Molecular Design of Photovoltaic Materials for Polymer Solar Cells: Toward Suitable Electronic Energy Levels and Broad Absorption. Accounts of Chemical Research 2012, 45 (5), 723-733. 30. Sabatini, R. P.; Mark, M. F.; Mark, D. J.; Kryman, M. W.; Hill, J. E.; Brennessel, W. W.; Detty, M. R.; Eisenberg, R.; McCamant, D. W., A comparative study of the photophysics of phenyl, thienyl, and chalcogen substituted rhodamine dyes. Photochemical & Photobiological Sciences 2016, 15 (11), 1417-1432. 31. Myers, A. B.; Mathies, R. A., Resonance Raman intensities: a probe of excited-state structure and dynamics. In Biological applications of Raman spectrometry - Volume 2: Resonance Raman spectra of polyenes and aromatics, Spiro, T. G., Ed. John Wiley & Sons, Inc.: New York, 1987. 32. Myers, A. B.; Mathies, R. A.; Tannor, D. J.; Heller, E. J., Excited state geometry changes from preresonance Raman intensities: Isoprene and hexatriene. The Journal of Chemical Physics 1982, 77 (8), 3857-3866. 33. Petrenko, T.; Neese, F., Analysis and prediction of absorption band shapes, fluorescence band shapes, resonance Raman intensities, and excitation profiles using the time-dependent theory of electronic spectroscopy. The Journal of Chemical Physics 2007, 127 (16), 164319. 34. Petrenko, T.; Neese, F., Efficient and automatic calculation of optical band shapes and resonance Raman spectra for larger molecules within the independent mode displaced harmonic oscillator model. The Journal of Chemical Physics 2012, 137 (23), 234107. 35. Shim, S.; Stuart, C. M.; Mathies, R. A., Resonance Raman Cross-Sections and Vibronic Analysis of Rhodamine 6G from Broadband Stimulated Raman Spectroscopy. Chemphyschem 2008, 9 (5), 697-699. 36. Shoute, L. C. T.; Loppnow, G. R., Excited-state dynamics of alizarin-sensitized TiO2 nanoparticles from resonance Raman spectroscopy. The Journal of Chemical Physics 2002, 117 (2), 842-850. 37. Shreve, A. P.; Mathies, R. A., Thermal Effects in Resonance Raman Scattering: Analysis of the Raman Intensities of Rhodopsin and of the Time-Resolved Raman Scattering of Bacteriorhodopsin. The Journal of Physical Chemistry 1995, 99 (19), 7285-7299. 38. Tannor, D. J.; Heller, E. J., Polyatomic Raman scattering for general harmonic potentials. The Journal of Chemical Physics 1982, 77 (1), 202-218.


27


Page 28 of 36

39. McCamant, D. W.; Kukura, P.; Yoon, S.; Mathies, R. A., Femtosecond broadband stimulated Raman spectroscopy: Apparatus and methods. The Review of scientific instruments 2004, 75 (11), 4971-4980. 40. Philipp Kukura, D. W. M., Richard A. Mathies, Femtosecond Stimulated Raman Spectroscopy. Annual Review of Physical Chemistry 2007, 58, 461-488. 41. Myers, A. B., Laser techniques in chemistry. John Wiley & Sons, Inc.: New York, New York, 1995; Vol. 23. 42. Li, B.; Myers, A. B., Absolute Raman cross sections for cyclohexane, acetonitrile, and water in the far-ultraviolet region. The Journal of Physical Chemistry 1990, 94 (10), 4051-4054. 43. Page, J. B.; Tonks, D. L., On the separation of resonance Raman scattering into orders in the time correlator theory. The Journal of Chemical Physics 1981, 75 (12), 5694-5708. 44. Li, B.; Johnson, A. E.; Mukamel, S.; Myers, A. B., The Brownian oscillator model for solvation effects in spontaneous light emission and their relationship to electron transfer. Journal of the American Chemical Society 1994, 116 (24), 11039-11047. 45. Mukamel, S., Principles of Nonlinear Optical Spectroscopy. Oxford University Press: New York, 1995. 46. Kelley, A. M., Condensed-Phase Molecular Spectroscopy and Photophysics. John Wiley & Sons, Inc.: Hoboken, 2013. 47. Svelto, O., Principles of Lasers. 5 ed.; Springer: New York, 2010. 48. Druzhinin, S. I.; Ernsting, N. P.; Kovalenko, S. A.; Lustres, L. P.; Senyushkina, T. A.; Zachariasse, K. A., Dynamics of Ultrafast Intramolecular Charge Transfer with 4-(Dimethylamino)benzonitrile in Acetonitrile. J. Phys. Chem. A 2006, 110 (9), 2955-2969. 49. Savarese, M.; Aliberti, A.; De Santo, I.; Battista, E.; Causa, F.; Netti, P. A.; Rega, N., Fluorescence Lifetimes and Quantum Yields of Rhodamine Derivatives: New Insights from Theory and Experiment. J. Phys. Chem. A 2012, 116 (28), 7491-7497. 50. Merrick, J. P.; Moran, D.; Radom, L., An Evaluation of Harmonic Vibrational Frequency Scale Factors. J. Phys. Chem. A 2007, 111 (45), 11683-11700. 51. Liebel, M.; Schnedermann, C.; Wende, T.; Kukura, P., Principles and Applications of Broadband Impulsive Vibrational Spectroscopy. The Journal of Physical Chemistry A 2015, 119 (36), 9506-9517. 52. Johnson, A. E.; Myers, A. B., A comparison of time‐ and frequency‐domain resonance Raman spectroscopy in triiodide. The Journal of Chemical Physics 1996, 104 (7), 2497-2507. 53. Kuramochi, H.; Takeuchi, S.; Tahara, T., Femtosecond time-resolved impulsive stimulated Raman spectroscopy using sub-7-fs pulses: Apparatus and applications. Review of Scientific Instruments 2016, 87 (4), 043107. 54. Dreuw, A.; Head-Gordon, M., Failure of Time-Dependent Density Functional Theory for LongRange Charge-Transfer Excited States: The Zincbacteriochlorin−Bacteriochlorin and Bacteriochlorophyll−Spheroidene Complexes. Journal of the American Chemical Society 2004, 126 (12), 4007-4016. 55. Levine, B. G.; Ko, C.; Quenneville, J.; MartÍnez, T. J., Conical intersections and double excitations in time-dependent density functional theory. Molecular Physics 2006, 104 (5-7), 10391051. 56. Kee, H. L.; Kirmaier, C.; Yu, L.; Thamyongkit, P.; Youngblood, W. J.; Calder, M. E.; Ramos, L.; Noll, B. C.; Bocian, D. F.; Scheidt, W. R.; Birge, R. R.; Lindsey, J. S.; Holten, D., Structural Control of the Photodynamics of Boron−Dipyrrin Complexes. The Journal of Physical Chemistry B 2005, 109 (43), 20433-20443. 57. Lee, J.; Challa, J. R.; McCamant, D. W., Pump power dependence in resonance femtosecond stimulated Raman spectroscopy. J. Raman Spectrosc. 2013, 44 (9), 1263-1272. 58. Lee, J.; Challa, J. R.; McCamant, D. W., Ultraviolet Light Makes dGMP Floppy: Femtosecond Stimulated Raman Spectroscopy of 2′-Deoxyguanosine 5′-Monophosphate. J. Phys. Chem. B 2017, 121 (18), 4722-4732.


28

Page 29 of 36 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


59. Challa, J. R.; Du, Y.; McCamant, D. W., Femtosecond Stimulated Raman Spectroscopy Using a Scanning Multichannel Technique. Applied Spectroscopy 2012, 66 (2), 227-232. 60. Akturk, S.; Gu, X.; Kimmel, M.; Trebino, R., Extremely simple single-prism ultrashort-pulse compressor. Opt. Express 2006, 14 (21), 10101-10108. 61. Neese, F., The ORCA program system. Wiley Interdisciplinary Reviews: Computational Molecular Science 2012, 2 (1), 73-78.

Table of Contents Graphic


29


Figure 1. O-Ph and O-Th. The dihedral angle τ (red) is defined by the planes of the core and aryl bridge substituents (C10, C9, C14, and C15). The dihedral angle δ (green) is defined by the plane of the dimethylamine group and the core. Note that two resonance structures exist for these compounds, making the left and right dimethylamines equivalent in solution. 93x124mm (300 x 300 DPI)


Page 30 of 36

Page 31 of 36 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. RR spectrum, RR excitation profiles, and absorption/emission spectrum of O-Ph and O-Th. (a) 600 nm (16,667 cm-1) Raman pump FSRS spectrum (blue) and the fit used to compute absolute cross sections (grey fill) of O-Ph. The asterisk marks a residual feature from solvent subtraction. (b)-(d) O-Ph experimental RR cross sections of the most intense vibrational modes (labeled in (a)) at each Raman pump wavelength (blue circles) with the corresponding RR excitation profile calculated using parameters from Table 1 (red lines). RR profiles are offset for comparison. (e) The experimental (blue solid) and calculated (red dashed) absorption and emission spectra. Emission spectra are scaled to match absorption intensity and are calculated stimulated emission spectra are multiplied by ω^3 for comparison to spontaneous emission and scaled to match absorption intensity (see text). (f)-(j) are the corresponding plots for O-Th. 228x139mm (96 x 96 DPI)



Figure 3. Comparison between calculated and experimental spectra: The experimental and DFT absorption and emission spectra (a) and RR spectra (b) of O-Ph, as well as those for O-Th, (c) and (d). The experimental absorption spectrum is shown in blue with the modeled spectra superimposed in red. Above the experimental and modelled spectra are the DFT spectra calculated using the labeled functional. In (b) and (d), the corresponding RR spectra are shown for a 560 nm Raman pump wavelength for each functional and the experiment; functionals appear in the same order as in (a). The normal mode frequencies for each functional have been scaled according to literature scaling factors: B3LYP 0.9613, BHLYP 0.9244, BP86 0.9914, PBE0 0.9512.50 293x255mm (96 x 96 DPI)


Page 32 of 36

Page 33 of 36 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 4. Comparison of O-Ph (a)-(c) and O-Th (d)-(e) IVS (620 nm), FSRS (600 nm), and BHLYP DFT (620 nm). O-Ph: (a) IVS Fourier transform magnitude summed from 520 to 580 nm (black) compared to FSRS (red). (b) IVS spectrum as in (a) compared to the BHLYP DFT Raman spectrum simulated with a 620 nm Raman wavelength (blue). (c) Same as (b) zoomed into the low frequency region. In (a) and (b), the spectra have been scaled to normalize the intensity of the 635 cm-1 peak. O-Th: (d) Same as (a) for O-Th with the sum going from 540 to 600 nm. The spectra have been normalized to the 631 cm-1 peak. (e) and (f) are the same as (b) and (c) but for O-Th. 216x101mm (96 x 96 DPI)



Figure 5. S0 and S1 relaxed potential scans along τ. BHLYP S0 and S1 relaxed potential energy scans along the coordinate τ of (a) O-Ph and (b) O-Th with their Boltzmann populations at 298K in colorscale and oscillator strength at the S1 geometry (green dashes). Black arrows show the vertical excitation and relaxed emission. Red circles correspond to the energy in the FC region, that is the energy in S1 at the S0 minimum geometry. Structures at the FC and coplanar minima for O-Ph and O-Th are provided in Figure S13. Coordinate files of these structures as well as the S0 minimum structures are also provided in the SI to enable visualization from multiple perspectives. 79x37mm (600 x 600 DPI)


Page 34 of 36

Page 35 of 36 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 6. HOMO and LUMO molecular orbitals of O-Th at the coplanar S1 minimum. (a) The LUMO is delocalized onto the aryl group while in (b) the HOMO remains localized on the core, even at coplanarity. The LUMO → HOMO emission at coplanarity is thus an ICT transition. See Figure S15 for a comparison to the FC geometry as well as corresponding difference density plots. 132x216mm (96 x 96 DPI)



269x129mm (96 x 96 DPI)


Page 36 of 36

Excited-State Planarization in DonorâBridge Dye Sensitizers

Recommend Documents

Excited-State Planarization in DonorâBridge Dye Sensitizers