A new approach exploiting thermally activated delayed fluorescence molecules to optimize solar thermal energy storage

We propose a new concept exploiting thermally activated delayed fluorescence (TADF) molecules as photosensitizers, storage units and signal transducers to harness solar thermal energy. Molecular composites based on the TADF core phenoxazine–triphenyltriazine (PXZ-TRZ) anchored with norbornadiene (NBD) were synthesized, yielding compounds PZDN and PZTN with two and four NBD units, respectively. Upon visible-light excitation, energy transfer to the triplet state of NBD occurred, followed by NBD → quadricyclane (QC) conversion, which can be monitored by changes in steady-state or time-resolved spectra. The small S1-T1 energy gap was found to be advantageous in optimizing the solar excitation wavelength. Upon tuning the molecule’s triplet state energy lower than that of NBD (61 kcal/mol), as achieved by another composite PZQN, the efficiency of the NBD → QC conversion decreased drastically. Upon catalysis, the reverse QC → NBD reaction occurred at room temperature, converting the stored chemical energy back to heat with excellent reversibility.


Synthesis of Compound 2 and 3.
We started from phenoxazine that reacted with 4-iodo-bromobenzene, potassium carbonate, copper(I) iodide to obtain compound 2. Next, a solution of 2 (2.0 g, 5.91 mmol) in 40 mL of anhydrous THF was cooled to -78 °C under a nitrogen atmosphere and was treated dropwise with 2.5 M n-BuLi in hexane (2.84 mL, 7.1 mmol). The resulting mixture was stirred at -78 ℃. After 1 h, the mixture was quickly added to a solution of cyanuric chloride (5.45 g, 29.6 mmol) in 40 mL of THF at 0 ℃. The resulting mixture was stirred at room temperature overnight.
The reaction was quenched with water and extracted with CH2Cl2. The combined organic layers were dried over MgSO4, filtered and evaporated. The crude product was purified by silica gel column chromatography with Hexane/CH2Cl2 mixture as eluent to afford 3 (0.98 g, 40%) as deep purple solids.

Synthesis of Compound PZDN.
A solution of bicyclo [2.2.1]hepta-2,5-diene-2-methanol (300 mg, 2.5 mmol) in 10 mL of anhydrous THF was added to 60% NaH (94 mg, 2.4 mmol) in 10 mL anhydrous THF under nitrogen atmosphere. The resulting mixture was stirred at RT. After 1 h, the mixture was added dropwise to a solution of 3 (250 mg, 0.61 mmol) in 10 mL of THF. The resulting mixture was stirred at room temperature overnight. The reaction was quenched with water and extracted with CH2Cl2. The combined organic layers were dried over MgSO4, filtered and evaporated. The crude product was purified by silica gel column chromatography with Hexane/CH2Cl2 mixture as eluent to afford PZDN (245 mg, 70%) as bright yellow solids.

Synthesis of Compound PZQN.
A solution of 5 (600 mg, 1.84 mmol) in 10 mL of anhydrous THF was added to 60% NaH (60 mg, 1.5 mmol) in 10 mL anhydrous THF under nitrogen atmosphere at 0 ℃. After 1 h, the mixture was added with a solution of 3 (250 mg, 0.61 mmol) in 10 mL THF dropwise. The resulting mixture was stirred at room temperature overnight. The reaction was quenched with water and extracted with CH2Cl2. The combined organic layers were dried over MgSO4, filtered and evaporated. The crude product was purified by silica gel column chromatography with Hexane/CH2Cl2 = 1/2 as eluent to afford PZQN (413 mg, 69%) as orange oil.

S14
The excited state-dynamics is based on thermally activated delayed fluorescence coupled with energy transfer as given in Supplementary Fig. 18. Kinetic rates regarding radiative plus non-radiative decay for the lowest singlet and triplet state, energy transfer of triplets from PXZ-TRZ to NBD moiety, intersystem crossing (S1 to T1), and reverse intersystem crossing (T1 back to S1) are labeled with kF, kP, kET, kISC, and kRISC, respectively. Numerical analysis on five rate constants was performed with Wolfram Mathematica 10.1 1 by solving the ordinary differential equation followed by optimized nonlinear fitting. The generated bi-exponential decay curves accord to the deconvolution fit from TCSPC measurements.

Supplementary Figure 18. The energy transfer was proposed to couple the TADF dynamics in titled molecules.
Vertical radiative and non-radiative transition rates were summed into one term, namely, kF and kP for convenience.
The rate constants were simulated based on the following equations 2,3 . We start from accessing the total decay of S1 state, which a valid initial guess is given by Supplementary equation (1) -(3). Additionally, kP is fixed as 10 3 s -1 throughout the process. where k S1* and k T1* refer to the total decay pathway of S1 and T1. With equation (S1.3) as the initial condition, we performed iterative, non-linear fitting on kF, kISC and kRISC through Supplementary equations (4) -(6).
where kprompt(kdelay) and Aprompt(Adelay) are the rate constant and pre-exponential factor associated to the fast(slow) population decay obtained from the fitting result of TCSPC. Note that equation (S2.3) is added here to optimize the simulated turning point from the prompt to decay region. S15 Supplementary Figure 19. Nonlinear fitting curves (blue, solid line) to the TCSPC data (red dot) generated from five optimized rate constants. The decay lifetimes and pre-exponential factors of the simulated decay curves coincide well with the deconvolution fitting of TCSPC results (confer Supplementary Fig.15). The prompt decays from data points are seemingly longer than simulated ones due to the convolution of excitation light and exponential decay for TCSPC measurements.  Supplementary Fig. 19.

Photochemistry and Isomerization
We were preparing a photoisomerization measurement of PZDN. First, the PZDN was dissolved with cyclohexane-d12, then degassed and transferred into a valved NMR tube to prevent influences from O2. After that, the NMR tube was irradiated in different minutes by a metal halide lamp.

S20
The reverse reaction for PZDQC and PZTQC with cobalt tetraphenylporphyrin (CoTPP) as catalyst: Dissolved PZDQC or PZTQC in toluene (5 mL), then CoTPP (1 mg) was added and the mixture was stirred at room temperature for 4 h. After that, the solvent was removed in vacuo and the residue was chromatographed on silica gel to afford corresponding PZDN or PZTN in quantitative yield. NMR spectra were identical as shown in Supplementary Fig. 27, 28. As shown in Supplementary Fig. 31, we observe a red-shifted excitation spectrum upon monitoring at 600 nm emission, compared to that when monitoring the emission at 450 nm. The emission wavelength dependent excitation spectra render evidence of intermolecular interactions and thus possible aggregation. This aggregation effect is more significant in high concentration polymer films and pure PZDN powder. The molecular aggregation results in lowering the triplet state energy, which is unfortunately lower than that of the NBD's triplet state, prohibiting the Dexter-type triplet-triplet energy transfer efficiency. As a result, we are unable to observe the changes in population lifetime (monitored at emission peak) upon irradiating 405 nm-light in pure solid PZDN film, PZDN doped PE film or PMMA film, where the solubility of PZDN is low and aggregation becomes dominant. As for doping PZDN in PVC and PS films, the presence of aggregation seems to be suppressed and there is still a small portion of photo-conversion for PZDN monomer, as implied by the increase of population lifetime during the photolysis. The corresponding kinetic traces and emission spectra as a function of irradiation time are shown in Supplementary Fig. 32 and 33. Clearly, in both PVC and PS hosts, the emission spectrum gradually blue shifts due to the PZDN→PZDQC conversion of monomer. Additionally, PZDN@PVC has a better conversion efficiency than in PS, indicating more homogeneous dispersion of PZDN in PVC film. Furthermore, we managed to access the reversibility between PZDN and PZDQC in both PVC and PS through irradiating the 405 nm light (3 W LED lamp, 15 minutes to ensure full conversion of PZDN→PZDQC, marked as red dot and red line in Supplementary Fig. 34, 35) and heating at 95°C in a clean oven for an hour (back-conversion of PZDQC→PZDN, marked as blue dot and blue line in Supplementary Fig. 34, 35). The results of photo-thermal conversion cycles are shown in Supplementary Fig. 34 and 35 for PZDN@PVC and PZDN@PS, respectively.  Plot (a) of Supplementary Fig. 34 and 35 shows the intensity monitored at emission peak, while (b) and (c) refer to the corresponding emission spectra and decay profile in the photo-thermal conversion cycles 0, 1 and 2. The results show that PZDN in PVC solid films, albeit good photo-conversion efficiency at the first cycle, suffers from severe aggregation and polymerization 4 during the next light irradiation and heating, resulting in a dramatic decrease in emission intensity and shortening in the decay kinetic trace. As for PZDN@PS, the trend in emission intensity holds for 3 cycles (increases after irradiation, decreases after heating), which again verifies PS to be a more suitable host for NBD/QC system 5 . Note that the anomalous increase in intensity (2 nd red dot to 3 rd blue dot) is caused by the inhomogeneous thickness of the film, where the steady state measurements for each cycle deviates from the focused excitation source.