Organopolymer with dual chromophores and fast charge-transfer properties for sustainable photocatalysis

Photocatalytic polymers offer an alternative to prevailing organometallics and nanomaterials, and they may benefit from polymer-mediated catalytic and material enhancements. MPC-1, a polymer photoredox catalyst reported herein, exhibits enhanced catalytic activity arising from charge transfer states (CTSs) between its two chromophores. Oligomeric and polymeric MPC-1 preparations both promote efficient hydrodehalogenation of α-halocarbonyl compounds while exhibiting different solubility properties. The polymer is readily recovered by filtration. MPC-1-coated vessels enable batch and flow photocatalysis, even with opaque reaction mixtures, via “backside irradiation.” Ultrafast transient absorption spectroscopy indicates a fast charge-transfer process within 20 ps of photoexcitation. Time-resolved photoluminescence measurements reveal an approximate 10 ns lifetime for bright valence states. Ultrafast measurements suggest a long CTS lifetime. Empirical catalytic activities of small-molecule models of MPC-1 subunits support the CTS hypothesis. Density functional theory (DFT) and time-dependent DFT calculations are in good agreement with experimental spectra, spectral peak assignment, and proposed underlying energetics.


Supplementary
Setup for hydrodehalogenation reactions. a Photoreactor setup for parallel batch reactions with reaction vessels suspended above the stir plates by an elastic cord. The reactor was equipped with a thermometer which confirmed that the temperature with the heat from the lights and stir plates was consistently 37 °C. b Overlapping MPC-1-0 absorption and blue LED emission spectra indicating the suitability of the selected irradiation source. Steady-state absorption was measured on a Cary Bio 50 UV-Vis spectrometer (Agilent Technologies) with a quartz cuvette in chloroform using the chloroform-soluble component of MPC-1-0 which passed through a cotton plug. Blue    The ideality of the different MPC-1 preparations was first assessed using NMR and GPC. A comparison of 1 H NMR data is provided in Supplementary Figure 12 and analysis of the MPC-1-2 spectrum is presented in Fig. 2a. GPC results are summarized

S98
To elucidate the possibility of structural defects in MPC-1 chains, a series of DFT frequency calculations were conducted in conjugation with FTIR measurements, with the focus being nitrile stretching modes present in various environments due to possible polymer chain defects. Geometry optimizations and frequency calculation were carried out at B3LYP/6-31g(d) level of theory.
With the measured peaks from CHR1 and CHR3, Gaussian fits were carried out to locate the maximum of each peak. Using these maxima and the predicted vibrational frequencies of the isolated chromophore models, a scaling factor was deduced to adjust the transition frequencies predicted by DFT for all investigated defect models. The overall scaling factor for all transitions was chosen to be 0.9503, while the precompiled correction factor for B3LYP/6-31g(d) is reported as 0.960 1 .
A depiction for the predicted modes for all investigated structures is shown in Supplementary Figure 17a Figure 17n, gray sticks) occur at a slightly lower frequency (~2231-2237 cm -1 ) than the observed peak center. These transitions are not far enough away from the peak center to conclusively assess the presence or absence of this type of defect. However, an argument can be made that the presence type-2 defects in abundance would shift the overall nitrile peak center to a lower frequency and skew the peak to a less symmetrical shape.

Supplementary Note 2 Use of the Frenkel-Davydov exciton model
In general, electronic states of a homodimer (X-X) or a heterodimer (X-Y) with similar chromophores can be well described by the Frenkel-Davydov exciton model [2][3][4][5][6] . In this model, the ground state electronic wavefunction, Ψ '' is approximated by the product of the two chromophores' ground-state wavefunctions, |)⟩ and |+⟩: Ψ '' = |)⟩. |+⟩. The lowest two valence states (VSs) are associated with the promotion of an electron from HOMO to LUMO of each chromophore. HOMO and LUMO of the two chromophores within the investigated heterodimer DXY are depicted in Figure 5b. Wavefunctions of the two first excited-state diabats of the dimer can therefore be approximated as Ψ /' = |) * ⟩. |+⟩ and Ψ '/ = |)⟩. |+ * 1, where |) * ⟩ and |+ * ⟩ are electronic wavefunctions of the excited chromophores and "g" and "e" denote "ground" and "excited", respectively. Energy separation between the two electronically excited eigenstates of the dimer is determined by the unperturbed energies of the diabats and their coupling strength, they can be approximated using second-order perturbation theory. At the limit of degenerate states, wavefunctions of the eigenstates are linear combinations of the diabatic wavefunctions with same weight, Ψ 23 where the "±" sign indicates the symmetry of the overall wavefunction.
In addition to neutral excitons (vide supra), charge-transfer (CT) excitons are also incorporated into the heterodimer model.
Wavefunctions of CT diabats may be phenomenologically expressed as Ψ :; = |) < ⟩. |+ = ⟩ and Ψ ;: = |) = ⟩. |+ < ⟩, where "a" and "c" denote "anion" and "cation", respectively. Similar to the valence states, vibronic and other coupling mechanisms including spin-orbit interactions mix the two CT diabats to form adiabatic CT states (CTSs), wavefunctions of which at the degenerate limit may be written as Ψ >? Transitions from the ground electronic state (gg) to the VSs, "ge" and "eg", are expected to dominate the steady-state optical absorption spectrum, while transition dipole moments between "gg" and CTSs are expected to be small because they involve promotion of an electron from HOMO of X to LUMO of Y or vice versa. Indeed, calculated oscillator strengths for the electronic transitions confirm these predictions (Supplementary Table 10).

Supplementary Note 3 Ground-state geometries
Cartesian coordinate information of the ground state optimized structures of X, Y, and DXY are listed below.
X Y Z    Figure 5b), an electron transfers from HOMO to HOMO-1 (Supplementary Figure 5a), or LUMO+1 to LUMO (Supplementary Figure 5b) to create the CTS1-equivelent, at which a partial positive charge is localized on Y part of the DXY dimer, and a partial negative charge is localized on X part of the DXY dimer.
Finally, Supplementary Figure 6 shows a comparison between the maps of electrostatic potentials (MESP) of GS, VS1, and CTS1 geometries, the charge distribution is consistent with our current model of eventual charge transfer from Y-chromophore unit to Xchromophore unit within the DXY dimer, either through VS1 state, or VS2 state.

Supplementary Note 5 Steady-state absorption/photoluminescence acquisition and decomposition analysis
Spectral shape and position were highly similar for all excitation energies explored for both MPC-1-1 and MPC-1-2. To better understand the influence of chromophores X and Y on the energetics of the absorption (Abs) and photoluminescence (PL) spectra ( Fig. 5), a decomposition routine was applied to both MPC-1-1 and MPC-1-2 spectra. However, intensity of the PL signal is determined not only by the transition dipole moment (µ), which the TD-DFT calculations predict, but also by the transition energy (ν). Explicitly, the PL signal is proportional to ν 3 . Experimentally obtained PL spectra were therefore scaled by 1/ν 3 (eV -3 ) then normalized to the maximum, while Abs spectra were normalized to the maximum, prior to the decomposition analysis. Normalized

Supplementary Note 6 Transient absorption kinetics analysis
Assuming all processes follow a first order reaction, and since charge transfer rates are much faster than photoluminescence rates, the reaction mechanism was approximated to be of a consecutive nature. The four kinetic traces extracted from the TA spectra were fitted simultaneously with an IRF-convoluted sum of exponential growths and decays, such that: Initial guesses of all parameters were approximated through a trial-and-error process while simulating and visually assessing the fit lines. Both samples were fitted with the same initial guess to avoid any bias, and the fit was repeated multiple times to ensure

Supplementary Note 10 Triplet state investigation
Triplet state involvement in MPC-1 photophysics was considered. Initially, a series of DFT calculations at the B3LYP/6-31G(d) level of theory in chloroform was employed. Following geometry optimization of heterodimer DXY to its singlet ground-state equilibrium geometry, triplet state vertical excitation calculations located the lowest triplet state at 2.08 eV (595 nm). However, attempting to optimize the DXY geometry to target the lowest triplet state failed to converge.
Experimentally, an attempt was made to detect any phosphorescence emission from MPC-1-2 dissolved in chloroform under ambient conditions, as well as after degassing by purging the solution with N2 gas for >20 minutes. These measurements utilized a phosphorescence detection scheme available in the spectrofluorometer (LS 55 PerkinElmer). During an acquisition cycle in this scheme, a single monochromatic excitation shot with a full width at half maximum of ~20 µs is sent to the sample, followed by monochromatic detection with a photomultiplier tube detector. The detection scheme is controlled by two variables: delay (waiting time between the release of the excitation shot and the start of the detector signal accumulation) and gate (the time window during which the detector accumulates the signal). Both variables can be adjusted in 10 µs steps. The resultant spectra from these experiments are summarized in Supplementary Figure 74.
Comparison between ambient and degassed conditions should reveal phosphorescence emission through increase in intensity at the wavelength range corresponding to the triplet state energy. However, change was not observed (Supplementary Figure 74a-b).
Also, delay times larger than 0 µs should better isolate the singlet-related emission from triplet-related emission due to the significant disparity between triplet state lifetimes (typically > 1 µs) and singlet state lifetimes (typically < 100 ns). However, no change in intensity or spectral shape was observed when changing the delay time (Supplementary Figure 74c-d). It is worth noting here that any acquisition with a delay > 30 µs did not result in any meaningful intensities. Similarly, a longer acquisition window, i.e., gate width, should accumulate higher intensity corresponding to delayed triplet-related emission, yet no change in intensity or spectral shape was observed (Supplementary Figure 74e-f).
The spectra presented in Supplementary Figure 74 show no indication of triplet-state population; however, our instrument's temporal resolution is relatively low (namely, 20-30 µs). By comparison, ladder-type polymers with a higher conjugation degree than MPC-1 were reported to have phosphorescence lifetimes on the order of 10s to 100s of microseconds, and their oligomer subunits to have a lifetime on the order of 100s of milliseconds [26][27][28] .
Another argument can be made that the emission decay we measured with TRPL from the VS1 state had a single exponential decay of ~10 ns without any longer components detected, which was consistent with photoluminescence lifetimes deduced from timecorrelated single photon counting measurements for a single-chromophore-type polymer similar to MPC-1-2 29 . Also, the < 20 ps charge transfer rates between vibronically coupled states VS1 and CTS1 leaves little chance for the slow, spin-forbidden, intersystem crossing process (typical > 10s ns) to occur from VS1 to any available triple state. Therefore, we concluded that triple states involvement in the photophysical and catalytic processes of MPC-1 was minimal.
An alternative synthetic procedure using environmentally benign aqueous PS-750-M surfactant was also employed 30 .
The vessel was shaken briefly to more evenly incorporate material on the walls of the vessel, and then the vessel was placed in a sand bath heated to 90 °C to stir at 800 rpm. The mixture was monitored every 5 minutes by briefly removing the vessel and shaking so as to evenly incorporate material on the walls of the vessel; changes in the coating left on the vessel walls and the solution appearance and viscosity were observed over time. At 50 minutes the liquid had become too viscous to effectively coat the vessel walls when shaking. At 60 minutes the stir bar was observed to intermittently slow, struggling against the increased viscosity. At 70 minutes the viscosity had become sufficiently high that the stir bar was whipping tiny bubbles into the reaction mixture, and at this point the vessel was removed from heating and allowed to cool to room temperature.
After cooling, the vessel contents solidified. The vessel was then unsealed and 4 mL deionized water was added, which caused the solidified polymer to break from coating the walls. The vessel was resealed and the contents were subjected to stirring at 1700 rpm while heating at 110 °C in a sand bath for 10 minutes, with the stirring being intermittently turned on and off to help break the stir bar free from the polymer. The suspension was then stirred vigorously at room temperature for 5 minutes. The supernatant liquid and some of the suspension was removed by syringe and transferred to a test tube. The reaction vessel was then thrice rinsed by stirring with 2 mL portions of deionized water which were also transferred into the test tube. The test tube was centrifuged, and its supernatant liquid was removed. The test tube contents were then twice admixed with 2 mL deionized water, centrifuged, and likewise separated from the supernatant liquid. The test tube contents were dissolved in chloroform and transferred back into the reaction vessel. Additional chloroform was added to bring the total volume added to 10 mL, and then 10 mL deionized water was S124 also added. The mixture was stirred, allowed to settle, and the chloroform layer was syringe-transferred to a test tube. The aqueous layer was then extracted with 2 x 1 mL chloroform to complete the transfer. The combined chloroform layers were then mixed with 1 mL deionized water, and the aqueous layer was transferred to the reaction vessel.
Subsequently, the combined chloroform layers were combined with 10 mL methanol, which caused the majority of the polymer to precipitate. The test tube was subjected to centrifugation and the supernatant liquid was decanted away. The centrifuge pellet was dissolved in chloroform, transferred to a glass storage vial, subjected to rotary evaporation under reduced pressure to remove the chloroform, and then placed in a vacuum oven. The oven was evacuated to 28 in Hg and then heated to 130 °C and allowed to sit for 14 h. The vacuum was then re-established to 28 in Hg (from 15 in Hg) and allowed to sit for 2 h, and which point the vacuum was briefly re-established, and then the oven was vented, and the sample were allowed to cool. Obtained 369 mg of translucent A stir-bar-equipped 20-mL vial was charged with 3,3,3',3'-tetramethyl-1,1'-spirobiindane-5,5',6,6'-tetraol (2 equiv., 12.0 mmol) and potassium carbonate (4 equiv., 24.0 mmol). The vessel was fitted with a rubber septum and thrice evacuated/argon-backfilled before adding 18 mL dry DMF by syringe. 1,2-Dibromoethane (1 equiv., 6.0 mmol) was added by syringe and then the septum punctures were covered with electrical tape and the septum was wrapped with PTFE tape. The mixture was allowed to stir at room temperature for 5 min before it was placed to stir in a sand bath pre-heated to 100 °C. After 12 h, the vessel was allowed to cool to room temperature. The reaction mixture was combined with 20 mL ice-cold deionized water and extracted with ethyl acetate (3x25 mL  A stir-bar-equipped 4-mL vial was charged with 2,3,5,6-tetrafluoroterephthalonitrile (1 equiv., 0.019 mmol), S3 (2.2 equiv., 0.042 mmol), and flame-dried potassium carbonate (4.5 equiv., 0.086 mmol). The vessel was fitted with a rubber septum and thrice evacuated/argon-backfilled before adding 200 µL dry DMF by syringe. The septum puncture was sealed with electrical tape and the septum was wrapped with PTFE tape. The vessel was placed to stir in a reaction block pre-heated to 90 °C for 2.5 h. The vessel was allowed to cool to room temperature. The reaction mixture was diluted with 0.5 mL ice-cold water and extracted (3x0.5 mL ethyl acetate). Combined organic layers were dried over sodium sulfate. Solvent was removed under reduced pressure. Crude material was purified by flash chromatography (hexanes/ethyl acetate). Obtained product as 33 mg (98%) yellow solid, Rf 0.55 to 0.36 (7:3, hexanes/ethyl acetate). 1 H NMR (400 MHz, CDCl3) δ 6.88-6.71 (m, 6H), 6.71-6.56 (m, 2H), 6.55-6.35 (m, 6H), 6.34-  A spin-vane-equipped 3-mL test tube was charged with 2,3,5,6-tetrafluoro-4-((1,3,5-trimethyl-1H-pyrazol-4yl)sulfonyl)benzonitrile (1 equiv., 0.030 mmol), S3 (2.2 equiv., 0.066 mmol), and potassium carbonate (4.5 equiv., 0.135 mmol). The vessel was fitted with a rubber septum and thrice evacuated/argon-backfilled before adding 300 µL dry THF by syringe. The septum puncture was sealed with electrical tape and the septum was wrapped with PTFE tape. The vessel was placed to stir in a sand bath pre-heated to 65 °C for 10 h. After TLC indicated that the reaction was still not complete, the solvent was removed under reduced pressure, the vessel was thrice evacuated/argon-backfilled, and 300 µL dry DMF was added by syringe. The vessel was placed to stir in a sand bath pre-heated to 90 °C for 2 h. The vessel was allowed to cool to room temperature. The reaction mixture was diluted with 300 µL chloroform and admixed with ca. 0.5 g crushed ice. The organic layer was set aside and the aqueous layer extracted (3x300 µL chloroform). Combined organic layers were washed (1x1 mL ice-cold water, 1x1 mL brine) and the dried over sodium sulfate. Solvent was removed under reduced pressure. Crude material was purified by flash chromatography (hexanes/DCM). Obtained product as 30 mg (52%) yellow solid, Rf 0.21 to 0.03 (7:3, hexanes/ethyl acetate