Control of intramolecular singlet fission in a pentacene dimer by hydrostatic pressure

Singlet fission (SF), which produces two triplet excitons from a singlet exciton, has been identified as a novel nanointerface for efficient (photo)energy conversion. This study aims to control exciton formation in a pentacene dimer through intramolecular SF using hydrostatic pressure as an external stimulus. We reveal the hydrostatic-pressure-induced formation and dissociation processes of correlated triplet pairs (TT) in SF by means of pressure-dependent UV/vis and fluorescence spectrometry and fluorescence lifetime and nanosecond transient absorption measurements. The photophysical properties obtained under hydrostatic pressure suggested distinct acceleration of the SF dynamics by microenvironmental desolvation, the volumetric compaction of the TT intermediate based on solvent reorientation toward an individual triplet (T1), and pressure-induced shortening of T1 lifetimes. This study provides a new perspective on the control of SF by hydrostatic pressure as an attractive alternative to the conventional control strategy for SF-based materials.


Introduction
Singlet ssion (SF) is a reversible photophysical process in which two chromophores in the ground state (S 0 ) and an excited singlet state (S 1 ) interact to form correlated triplet pairs ( 1 (TT) and 5 (TT)). These pairs then relax, forming two individual excitons (T 1 ) with an extremely high T 1 quantum yield (F T ) of up to 2 (Fig. 1a). [1][2][3] This process is very promising and even more successful than other recently developed photophysical processes, including thermally activated delayed uorescence (TADF), 4 aggregation-induced emission (AIE), 5 and upconversion, 6 toward the construction of photo-relevant materials and optical chemosensors. [7][8][9] For example, the highly efficient generation of multiple excitons allows for a wide array of applications for SF-based materials, including the construction of solar cells, 10 photosensitizers, 11 singlet oxygen generators for photodynamic therapy, and relevant biological systems. 12 Nevertheless, the fabrication of SF-based materials requires that the important energy balance at the excited singlet and triplet levels be met (E (S 1 ) $ 2E (T 1 )), 1-3 which in turn can hamper their wide range of molecular design. One appropriate solution for adjusting such rigidity or non-tunability of SF properties is the exible/dynamic control achieved by external stimuli such as temperature, [13][14][15][16] solvent, [17][18][19] or reaction (aggregation [20][21][22][23] and supramolecular complexation 19,24,25 ). This control strategy has recently attracted signicant attention although it is yet to be demonstrated on a trial-and-error basis.
Hydrostatic pressure is a mechanical isotropic stimulus and one of the most signicant state quantities, enabling scientists to precisely control thermodynamic equilibria and kinetic rates, particularly in solutions in both ground and excited states. [26][27][28] Thus, studies on hydrostatic pressure have been widely conducted since the early 1960s. 29 In the 1980s, advances were made in the understanding of photophysical and photochemical characteristic processes, such as excimer/exciplex formation, 30,31 photoinduced electron transfer, 32 and twisted intramolecular charge transfer 33 upon hydrostatic pressurization. In this way, the effects of hydrostatic pressure on the properties of solutions have been investigated on a long-term basis, in addition to solid-state high-pressure chemistry using a diamond anvil cell (DAC). 34,35 Although such high-pressure chemistry has been extensively researched, recently, the emergence of mechanochemistry 36 and mechanobiology, 37 in which "pressure" as a mechanical force plays a key role, has brought solution-state hydrostatic pressure chemistry under the spotlight once again. We recently revealed that the TADF properties in a mechanochromic material 38 and the on-off AIE behavior of dynamic AIE-active polymers 39 can be controlled by the application of hydrostatic pressure. These results prompted us to examine how hydrostatic pressure affects or "controls" the SF processes in functional molecules.
In this study, to implement the hydrostatic pressure control concept in SF chemistry, we focused on a biphenyl-bridged pentacene dimer (Pc-BP-Pc, Fig. 1b), which possesses a more exible biphenyl linker than its corresponding reference monomer (Pc-ref, Fig. 1b). Recently, Asbury et al. investigated the solid-state pressure effects of Pc-ref using the DAC technique, revealing efficient triplet-pair separation. 40 This sophisticated case provided us with a breakthrough in the design of smart solar cells and relevant materials in the thin lm state. However, control of solution-state SF processes upon the application of hydrostatic pressure, for example, photodynamic therapy in physiological (buffer) solutions, remains a major challenge in current multidisciplinary chemistry.
As a guideline for molecular design, the choice of linkers between bichromophores 41 is signicant for achieving the stated purpose. This signicance is highlighted by the results shown in Fig. 1c: the SF kinetics (k SF ) in the m-and o-phenylene bridges in toluene were 2.1 × 10 9 s −1 and 1.2 × 10 11 s −1 , 42 respectively, indicating that adjusting the distance and angle of each chromophore is critical. These results allowed us to choose a relatively exible biphenyl linker with a k SF of 1.8 × 10 9 s −1 in tetrahydrofuran (THF) under atmospheric conditions (0.1 MPa). 17 We now report the unprecedented excited-state dynamics of Pc-BP-Pc, in which the SF kinetics are drastically facilitated by hydrostatic pressurization. This study enabled us to examine the extent to which hydrostatic pressure affects the SF dynamics. The present demonstration of hydrostatic pressure control highlights the potential for a wide variety of attractive applications using SF processes in solution systems.

Results and discussion
Investigation of pressure-induced structural relaxation First, to investigate the degree of aggregation of the pentacene dimer under hydrostatic pressure (∼320 MPa), concentrationdependent (14-228 mM) UV/vis spectra were measured at 0.1 (atmospheric pressure), 160, and 320 MPa in toluene. As shown in Fig. S2 in ESI, † the absorbance increased at a constant rate with increasing concentration without any spectral changes in each hydrostatic pressure range, with the standard absorbance curves exhibiting strong linear relationships. Therefore, we conrmed that pressure-induced aggregation and crystallization based on intermolecular interactions of Pc-BP-Pc do not occur in these concentration ranges, which enables us to treat the pentacene analogs as well-dispersed, "monomeric" states in the following experiments. Fig. 1 (a) Schematic of the mechanism of SF in a TIPS-pentacene dimer. k f and k nr : deactivation processes from S 1 ; k SF : SF from S 1 to TT; k Rec : recombination process from TT to S 0 ; k Diss : dissociation process from TT to T 1 + T 1 (2T 1 ); k T : deactivation process from 2T 1 to S 0 . Considering the reported energies of S 1 (1.9 eV) and TT (1.6 eV), the reverse triplet-triplet annihilation process (TT / S 0 + S 1 ) can be omitted because of the large exergonic trend (∼0.3 eV). 42  Next, through steady-state UV/vis and uorescence spectrometry under hydrostatic pressure, we investigated the effect of hydrostatic pressure on both ground-state absorption and excited-state uorescence properties. As a control experiment, we performed similar tests using Pc-ref, a monomeric portion of the target pentacene dimer, and compared the pressureinduced spectral changes. As shown in Fig. 2a-f, under hydrostatic pressure, the spectra of both Pc-BP-Pc and Pc-ref showed gradual increases in absorbance and stepwise bathochromic shis without signicant changes in the spectral shape. This result suggests that intramolecular p-stacking does not occur in the ground state and in the excimer species in the excited state even under high pressure. It is well known that an increase in pressure induces a considerable change in solvent polarizability, causing the absorption and uorescence peaks to decrease in energy. 29 In addition, the monotonic hyperchromic effect on absorbance is simply due to the increase in the effective concentration upon pressurization. Therefore, such pressure-induced wavelength shis can provide us with significant information about the pentacene analogs in hydrostaticpressurized solutions; the absorption and uorescence slopes in toluene were calculated to be −0.750 and −0.856 cm −1 MPa −1 for Pc-BP-Pc, and −0.848 and −0.891 cm −1 MPa −1 for Pcref, respectively, as listed in Table S1. † These values appear to be very similar to those observed in other p systems such as anthracene, pyrene, and perylene. 28 Hence, these results conrm that the intramolecular stacking behavior (vide supra) and substantial conformational relaxation around the biphenyl linker (particularly in the S 1 transition) may not be affected by hydrostatic pressure. Parallel experiments in methylcyclohexane (MCH) and THF showed similar slopes from −0.588 to −0.755 cm −1 MPa −1 , verifying the occurrence of no signicant conformational changes even in a wide variety of dipole moments (0.00-1.75 (Table S1 and Fig. S3-S10 in ESI †)).

Intramolecular SF dynamics
We estimated the effects of hydrostatic pressure on the intramolecular SF kinetics of the pentacene analogs. The uorescence lifetimes of Pc-BP-Pc and Pc-ref under hydrostatic pressure were measured in toluene ( Fig. 3a and b), with the resulting decay prole obtained for Pc-BP-Pc containing multiple components. The prole reasonably tted to the sum of two exponential functions, in contrast to that of the monoexponential function observed for Pc-ref (see Fig. S11-S16 and Tables S2-S4 in ESI †). Very short-lived decays (s 2 ) in Pc-BP-Pc are observed in the enlarged gure. The exible Pc-BP-Pc dimer adopts some conformers, 17,43 in which the s 1 species emits the uorescence as a monomer (without the SF process) and the s 2 species mainly decays as a deactivation path (involved the SF process). Indeed, the strong uorescence quenching of Pc-BP-Pc at 0.1 MPa, rather than Pc-ref, was clearly observed. 17 Certainly, other Pc dimers also showed the same decay behavior; the long-lived species (monomer conformer) and the short-lived species (TT process). 44 Thus, the rate constant of SF, k SF,app , for the generation of correlated TTs can be written as eqn (1): where k 0 represents the decay rate constant from the excited singlet state Pc-ref, which can further be divided into the sum of k f (uorescence rate constant) and k nr (radiationless deactivation rate constant). According to our previous study, 17 the rate constant in the SF process should be expressed k SF,app as an apparent rate because the reverse TTA process has not been proved in this dimer. As listed in Tables 1 and S5, † the k SF,app values increased from 1.65 × 10 9 s −1 to 1.96 × 10 9 s −1 , consistent with the elevated pressures (0.1-180 MPa) in toluene. Interestingly, a clear increasing trend of k SF,app from 1.23 × 10 9 s −1 to 1.67 × 10 9 s −1 (40-160 MPa) in the more polar THF upon hydrostatic pressurization was also observed, whereas the k SF,app values in nonpolar MCH were almost constant in the range of pressures examined. These ndings indicate that hydrostatic pressure critically affects the SF process with solvent polarity.
To further elucidate the hydrostatic-pressure-induced SF dynamics more quantitatively, we calculated the activation volume change (DV ‡ ) in the transition state according to eqn (2): The natural logarithm of k SF,app was plotted against pressure, with a linear relationship obtained in each solvent (Fig. 3c). This indicates that the excited-state reaction toward the transition state proceeds by only a single mechanism, that is, the SF process, in the range of the hydrostatic pressures examined. A distinct difference in the negative DV ‡ was observed for the different solvents: −2.5 cm 3 mol −1 in toluene and −3.5 cm 3 mol −1 in THF, compared to the almost zero DV ‡ (−0.2 cm 3 mol −1 ) in MCH (see Table 1). Considering that DV ‡ is the degree of compactness in a transition state depending on not only structural changes, such as differences in bond lengths and angles, 26 but also solvation, 28,45 in this case relating to the intramolecular SF system, the latter solvation contribution seems to be preferable. Hence, these results can be explained in terms of the intramolecular charge density change in the correlated TT. As the desolvation process proceeds, the transition-state structure becomes more compact in polarized toluene (0.38 D) or THF (1.75 D), as shown in Fig. 3d. This scenario may be supported by a very small contribution of DV ‡ in nonpolarized MCH (0.00 D). Here, as the volume of one toluene or THF molecule is 106 cm 3 mol −1 or 81 cm 3 mol −1 , the desolvation contribution to each DV ‡ is approximately 2% in one toluene molecule or 4% in one THF molecule. This indicates that intramolecular SF processes can be accelerated by critical desolvation, which allows the transition-state assembly involving the solvent cluster to be much more compact. Therefore, the solvation/desolvation contribution plays a decisive role in the precise control of intramolecular SF dynamics.

Hydrostatic pressure effects of triplets
We elucidated the hydrostatic-pressure-induced intramolecular SF processes, at which the active exciton, TT, eventually becomes T 1 . Nanosecond transient absorption (nsTA) spectrometry is an effective analytical tool for elucidating the photophysical properties of T 1 as a nal step of SF. 17 The construction of a newly designed optical system for hydrostatic pressure nsTA spectrometry is detailed in the Materials and methods section. This system was used to measure the solvent-dependent spectra of toluene, MCH, and THF solutions in the triplet absorption band (Fig. 4a). As shown in Fig. 4b and S17a-c, † spectral measurements at 0.1 (atmospheric pressure), 160, and 320 MPa also showed pressuredependent decays and slight pressure-induced bathochromic shis, similar to those observed in the steady-state UV/vis absorption spectra upon hydrostatic pressurization. In addition, the nsTA decay proles of the generated individual T 1 at each hydrostatic pressure in Fig. 4c and S17d † can be reasonably tted to a monoexponential function with triplet lifetimes (s T ) of 0.7-1.6 ms (Fig. S18-S19 and Table S6 †). So far, we have revealed that s T of Pc-BP-Pc can be prolonged from 0.36 ms in THF to 1.0 ms by mixing paraffin into THF, based on the suppression of collisional deactivation by the polar solvent. 17 We achieved precise hydrostatic pressure control of s T without changing the solvent, in which the shortened lifetimes in response to hydrostatic pressurization are responsible for T 1 deactivation under higher viscosity solvent conditions, that is, acceleration of solvent interactions with T 1 excitons (vide infra: result of the bulkier T 1 structure). As shown in Fig. S27, † the pressure-induced viscosity changes play signicant roles in the excited-state processes rather than the polarizability, density, and polarity; the latter ones almost affect to the absorption behavior. 29 Indeed, Lacour and Vauthey et al. demonstrated the SF control by solvent viscosity. 19 Finally, the S 1 , TT, and T 1 quantum yields (F S , F TT , and F T ) were calculated (see the Materials and methods section), enabling us to estimate the SF thermodynamics, that is, the reaction volume changes as DV S/TT in the equilibrium between S 1 and TT (eqn (3)) and DV TT/T in the equilibrium between TT and T 1 (eqn (4)).
As shown in Table 2 and Fig. S22, † the values of DV S/TT in toluene and MCH were negative and almost zero, respectively, indicating that the TT structure comprising the solvent core was thermodynamically more compact than the S 1 structure upon desolvation. This is consistent with the fact that the kinetically formed TT transition-state complex is also more compact owing to the desolvation-driven behavior (vide supra). In contrast, the values in Table 2 and Fig. S24 † for DV TT/T were positive. Strangely, these values in both toluene and MCH are approximately +5.8 cm 3 mol −1 ; such a solvent-independent trend suggests that a different mechanism is at play here, rather than the aforementioned solvation/desolvation. These results are likely attributable to the change in conformers in Pc-BP-Pc during the TT dissociation process (TT / 2T 1 ), as shown in Fig. 4d. In a previous study on the intramolecular SF of pentacene dimers using time-resolved electron paramagnetic resonance measurements, TT dissociation motion was observed, and the pentacene dimer exhibited a dynamic change in the dihedral angle between chromophores. 43 Similarly, Pc-BP-Pc is highly likely to undergo a conformational change associated with the dihedral angle change around the biphenyl linker during the TT dissociation process even under hydrostatic pressure. This was the case with the D-A-D triad that showed the excited-state conformational change under hydrostatic pressure. 38 Hence, the nature of DV TT/T may be attributed to the solvent reorientation/interaction occurring aer the conformational change. Because the TT dissociation process is spatially divided by spin interactions, in contrast to the TT formation (via the exciton coupling of S 0 and S 1 ), a possible interpretation of this result is that the dependence of the solvation term (toluene vs. MCH) may be extremely small. The contribution of the conformational change during dissociation is further related to the stepwise reduction of F T with increasing hydrostatic pressure (Table 2), which is responsible for the gradually enlarged solvent interactions in Pc-BP-Pc due to the increase in hydrostatic-pressure-driven solvent viscosity. This fact strongly supports the relationship among thermodynamically expanded or "bulkier" structures, positive DV TT/T ; and the graduals T shortening trend. In detail, the thermodynamically bulkier T 1 exciton by clustering a large number of solvent molecules is strongly deactivated by the solvent attack in the solvent core to shorten s T .

Conclusions
In this study, for the rst time, we realized the signicance of hydrostatic-pressure-induced intramolecular SF behavior. This was demonstrated using the biphenyl-linked pentacene dimer as a model SF material, in which hydrostatic-pressurizationbased solvent-induced property changes are key factors. Although remarkable conformational changes in the dimer were not observed, through hydrostatic pressure steady-state spectrometry, the uorescence and triplet absorption lifetime measurements enabled us to recognize that the rates of the correlated triplet pairs from the singlet were greatly accelerated by changing the hydrostatic pressure in toluene only. This revealed that the desolvation process in polar solvents plays an important role in the SF dynamics. More importantly, we have shown that the entire process of SF involving ssion (S 0 + S 1 / TT) and dissociation (TT / 2T 1 ) under hydrostatic pressure can be precisely controlled by not only kinetics in the transition state but also thermodynamics in the equilibria on the basis of microenvironmental desolvation and solvent reorientation. Finally, it should be emphasized that by using hydrostatic pressure as an external stimulus, the dynamic control concept of intramolecular SF kinetics observed in this study can be further expanded to other SF scaffolds and relevant systems that are difficult to control in both ground and excited states.

Materials
All commercial reagents and solvents were used without further purication. Sample solutions dissolved in spectroscopic grade toluene, methylcyclohexane (MCH), and tetrahydrofuran (THF) were deaerated by ve freeze-pump-thaw cycles saturated with N 2 for uorescence lifetime measurements or by Ar bubbling for nanosecond transient absorption (nsTA) measurements. The SF-based material (Pc-BP-Pc) was synthesized according to the literature. 17 Pc-ref was commercially available.

Instruments
UV/vis and uorescence spectra were recorded in a highpressure cell (path length: 2 mm) by using a JASCO V-650 or a JASCO FP-8500. Fluorescence lifetimes were measured in a high-pressure cell by a Hamamatsu Quantaurus-Tau single photon counting apparatus tted with an LED light source (l ex = 405 nm). Nanosecond transient absorption (nsTA) measurements were performed by using a Unisoku TSP-2000 ash spectrometer-pump pulse source: Surelite-I Nd:YAG (Continuum, 4-6 ns fwhm) laser with the second harmonic at 532 nm, monitor light source: xenon lamp (150 W), and detector: photomultiplier tube. 1 H NMR spectrum of Pc-BP-Pc was recorded on an ECS-400 spectrometer.

Hydrostatic pressure spectroscopy
Steady-state UV/vis absorption/excitation uorescence spectroscopy and uorescence lifetime decay measurement were conducted using a custom-built high-pressure apparatus. 28,45 As the process has previously reported in detail in a previous study, here, we briey describe this. A quartz inner cell was lled with the sample solution, and then the cell was set into the outer cell, where sapphire or quartz windows were tted. A tightly closed outer cell, which was hydrostatically pressurized by water, was placed in the spectroscopic apparatus (Fig. S1a-e †).

Hydrostatic pressure nsTA
The outer cell was placed on the nsTA optical system, which allowed us to apply hydrostatic pressure in an appropriate manner (Fig. S1f †). The T 1 quantum yield (F T ) of a toluene solution of Pc-BP-Pc measured in this optical system under an atmospheric pressure (0.1 MPa) was 176%, which is identical to that (F T = 176%) 17 of a THF solution observed in a regular cuvette at 0.1 MPa. This proves the validity of the nsTA measurements under hydrostatic pressure using the new optical system prepared in this study.

Determination of S 1 quantum yield (F S ) of Pc-BP-Pc
Absolute S 1 quantum yields F P0 (F S at 0.1 MPa; 0.0140 in toluene, 0.0514 in MCH) were evaluated using a spectrouorometer (FP-8500) tted with an integrating sphere at 0.1 MPa, and then relative S 1 quantum yields can be determined using the following eqn (5): 46 where A, I, n, and D are the absorbance at the excitation wavelength, intensity of the excitation light, refractive index, and area ratio in the uorescence spectrum, respectively (Tables S7  and S8 †). The data for A were extracted from Fig. 2a and S5a. † The data for I P /I P 0 were 1 because the measurements were performed at the same excitation wavelength ( Fig. 2c and S6a †). The refractive indices used under atmospheric pressure (n P 0 ) were identical to the values reported in a previous study, 47 and the refractive indices under pressure (n P ) were calculated using the Eykman equation (eqn (6)) from the change in the density of each solvent with respect to pressure: 48,49 n 2 À 1 n þ 0:4 Constant C was calculated using a known value. 47 The data for D were obtained by tting the area of the uorescence spectra in toluene (Fig. 2c) and MCH (Fig. S6a †) using the MATLAB soware with a three-component Gaussian function (eqn (7)) (see Fig. S20 and S21 and Tables S9 and S10 †): Determination of the TT quantum yield (F TT ) of Pc-BP-Pc The F TT values were calculated using eqn (8) (see Table S11 †): k SF;app k 0 þ k SF;app ¼ F TT (8) where the data for k 0 and k SF,app were extracted from Tables S2,  S3 and S5, † respectively.

Determination of T 1 quantum yield (F T ) of Pc-BP-Pc
The F T,Pc-BP-Pc value was calculated as the relative F T quantum yield (see Tables S12 and S13 †) in relation to that of zinc tetraphenylporphyrin (ZnTPP) or zinc tetratolylporphyrin (ZnTTP), using the following eqn (9) where F T , 3 T , DA and Abs are the T 1 quantum yield, excitation coefficient, delta absorbance in nsTA, and absorbance at 532 nm in the steady-state absorption measurements, respectively. F T,ZnTTP and 3 T,ZnTTP were acquired from a previous study 50 (l ex = 532 nm, l obs = 470 nm in toluene; F T,ZnTTP = 0.88, 3 T,ZnTTP = 69 000 M −1 cm −1 ), and 3 T,Pc-BP-Pc from another paper 17 (l ex = 532 nm, l obs = 516 nm in THF; 3 T,ZnTPP = 54 900 M −1 cm −1 ). DA ZnTPP was observed at 470 nm and DA Pc-BP-Pc was observed at 521 nm for toluene and 516 nm for MCH, taking into account the solvent-induced shi of the excited-state absorption T 1 -T n band (Fig. 4a). Data of Abs (532 nm,ZnTPP) and Abs (532 nm,Pc-BP-Pc) are shown in Fig. S23b, 2a, and S5a, † respectively.

Data availability
The data supporting this article have been uploaded as part of the ESI. †