Enhanced triplet state generation through radical pair intermediates in BODIPY-quantum dot complexes

Because of long lifetimes of triplet excitons,^1–3 extensive research efforts have been devoted to the efficient generation and harvesting of triplet excited states and their application in photodynamic therapy,^4–6 bioimaging,^7,8 photocatalysis,^9–11 and photon upconversion.^12–14 Traditional methods for efficiently generating triplet excited states include the intersystem crossing (ISC) by heavy atom enhanced spin-orbit coupling^15,16 and energy transfer from organometallic triplet sensitizers.^17–20 Recently, quantum dots (QDs) have been shown to be excellent triplet sensitizers,^21–27 owing to their large extinction coefficients,^28,29 tunable band structures,^30–33 and small singlet/triplet state energy gap.^34–36 Excitons can be transferred from QDs to molecular acceptors through Dexter energy transfer,^21,37,38 and significant progress has been made in understanding triplet sensitization mechanisms and improving their efficiencies.^26,39,40

More recently, QD sensitized triplet formation has also been suggested to go through charge transfer intermediate states, but the nature of the spin states in these charge separated (CS) radical pairs has not been carefully examined.^41–43 In molecular donor-acceptor or host-guest complexes, triplet excited states can be generated through a charge transfer intermediate,^44–48 by radical pair intersystem crossing (RP-ISC)^49,50 or spin-orbit coupling mechanism (SOCT-ISC).^51,52 Charge separated states between QDs and adsorbed molecules often decay by charge recombination to form a singlet ground state, implying that these radical pairs are likely in their spin singlet state.^31,33,53 In a recent study of thiol-modified bis(diarylamino)4,4′-biphenyl (TPD) attached on CdS QDs, it was demonstrated that electron transfer from TPD to CdS forms charge separated states (with an oxidized TPD and a trapped electron in CdS) that undergo charge recombination to generate both TPD ground and triplet excited states.⁵⁴ EPR studies reveal that the charge separated intermediates consist of spin-correlated singlet and triplet radical pairs, suggesting that these hybrid QD-adsorbate complexes may be a novel platform for spintronics applications.^55,56 However, because of strong spin-orbit coupling in QDs,³⁶ it remains unclear whether such a spin correlated radical pair is a general characteristic and whether the spin states can be effectively controlled and selected.

In this work, we demonstrate that in CdSe QD-modified BODIPY (shown as Compound 3 in Scheme S1) complexes, long lived triplet excited state of BODIPY can be generated by visible light excitation of BODIPY with 27% quantum efficiency. The triplet excited state is formed through a charge separated state immediate consisting of a CB electron in the QD and an oxidized BODIPY radical cation [Fig. 1(a)].

CdSe QDs with a Cd(oleate)₂ surface ligand and modified BODIPY with a carboxylic acid anchoring group were synthesized following reported procedures^23,57 (see supplementary material S1 and Figs. S1–S3 for ¹H and ¹³C NMR spectra). Figure 1(b) shows the UV-Vis absorption spectra of BODIPY, CdSe quantum dots, and their complex. The S₀-S₁ absorption peak of BODIPY lies at 656 nm, with a shoulder peak at 600 nm corresponding to the 0–1 vibrational band of the same transition.⁵⁸ Because of further conjugation induced by benzene rings attached to the main body of the BODIPY structure, the S₀-S₁ peak is red-shifted by around 150 nm compared to unmodified BODIPY molecules.⁸ Upon adsorption onto CdSe QDs, the absorption spectrum shows additional features in the range from 700 nm to 750 nm, which can be attributed to the aggregation of BODIPY molecules on the QD surface.⁵⁹ The adsorption of BODIPY does not affect the lowest energy 1S_3/2-1S_e exciton transition of the CdSe QD, centered at 584 nm.⁶⁰ Because there is no absorption of CdSe at 656 nm, selective excitation of BODIPY can be achieved in this wavelength region.

The transient absorption spectra of BODIPY at indicated delay times after 650 nm excitation are shown in Fig. 2(a). Optical excitation generates a BODIPY singlet excited state (¹BODIPY^*), which results in the ground state bleach (GSB) signal centered at 656 nm. The negative peak centered at 740 nm is attributed to the stimulated emission (SE) of ¹BODIPY^* given that its position matches that of BODIPY steady state fluorescence. The TA spectra also contain positive bands (induced absorption, IA) ranging from 430 to 600 nm and from 800 to 913 nm, which are attributed to the transition from the first (S₁) to higher singlet excited states (S₁ → S_n) of BODIPY.⁶¹ The TA spectra show the simultaneous decay of the amplitudes of ground state (GSB) and singlet excited state (SE and IA) features with clear isosbestic points and no change of spectral shape, which suggests that the only decay channel of ¹BODIPY^* is to regenerate the ground state, and there is negligible formation of the triplet excited state of BODIPY (³BODIPY^*) through the intersystem crossing (ISC). The kinetics of ¹BODIPY^* decay can be well fit by convolution with a single exponential decay function with a time constant of (4.20 ± 0.09) ns, as shown in Fig. 2(b) and Table S1.

Femtosecond transient absorption spectra of BODIPY-CdSe at indicated delay times after 650 nm excitation are shown in Fig. 3(a). Since the 650 nm pump pulse selectively excites BODIPY to its singlet excited state, the initial spectra at 1–10 ps are similar to isolated BODIPY molecules in the solution [Fig. 2(a)], showing well resolved features of GSB, SE, and IA (S₁- > S_n). Note that because of the aggregation of BODIPY on the QD surface, there is slightly more overlap between GSB and SE features at this initial time range. Interestingly, from 10 to 1000 ps, the decay of the ¹BODIPY^* signal leads to the growth of the QD 1S exciton bleach centered at 584 nm and a positive peak centered at around 800 nm. The positive peak is not resolved at <1 ns because of the saturation of the probe pulse at 800 nm but can be well observed at >1 ns using a different probe light source [see Fig. 3(b)]. Because the 1S exciton bleach, resembling those of photoexcited CdSe QDs (see Fig. S4 in supplementary material S2), has been shown to be mainly caused by the state filling of the conduction band 1S level,^62,63 its growth suggests either energy or electron transfer from the ¹BODIPY^* to CdSe.³⁸ Energy transfer can be excluded, given that there is no overlap of the fluorescence spectrum of BODIPY with the absorption spectrum of CdSe [see supplementary material S3, Fig. S5 and Fig. 1(b)].^38,64 The remaining possibility is electron transfer, which generates a charge separated (CS) state BODIPY^+·-QD^−·. This process is consistent with the energy level alignment of CdSe and BODIPY shown in Fig. S6 (see supplementary material S4 for details). The positive peak at around 800 nm from 50 to 1000 ps can thus be assigned to absorption of BODIPY cation radicals (BODIPY^+·).⁶⁵ 

Further spectral evolutions in the 1 ns–1 µs time scale are shown in Fig. 3(b). In these spectra, the positive peak of BODIPY^+· centered at 797 nm can be well resolved. The amplitude of the charge separated state signal reaches a maximum at 1–5 ns and then decays within 100 ns. The decay of the CS state leads to the formation of a new species with three positive peaks centered at 412 nm, 570 nm, and 750 nm. Its signal amplitude reaches a maximum at ∼100 ns, and, as shown in Fig. S7, its decay kinetics can be fit with a single exponential function with a time constant of (20.0 ± 0.3) µs. Because this signal is long-lived, and there are no remaining spectral features of the QD, it can only be assigned to the BODIPY triplet excited state (³BODIPY^*) formed by back electron transfer from the QD^−· to BODIPY^+·. The positive absorption peaks in the TA spectra can be attributed to the T₁- > T_n absorption bands of ³BODIPY^*. This assignment is further confirmed by the faster decay of the signal when the system was exposed to oxygen (Fig. S8).

In order to fully understand the competing processes and mechanism for ³BODIPY^* generation, we further analyze the TA spectra of BODIPY-CdSe from 10 ps to 1000 ns, which are the combination of ¹BODIPY^*, ³BODIPY^*, and CS state signals. The spectra of ¹BODIPY^* can be obtained independently from the TA spectra of free ¹BODIPY^*, and the spectra of ³BODIPY^* are obtained from the TA spectra of BODIPY-QD at a long delay time, when only ³BODIPY^* exists. The Multivariate Curve Resolution (MCR)-alternating least squares method was applied to extract the CS spectrum.⁶⁶ (see supplementary material S7 for details). The fitting result is shown in Fig. 4(a). As expected, the resulting CS spectrum shows the features of the QD exciton bleach and BODIPY cation radical absorption. The CS spectrum can also be obtained by subtraction of the ¹BODIPY^* spectrum signal from the TA spectrum of CdSe-BODIPY [shown in Fig. 3(a)] (see supplementary material S8 for details). The kinetics for each species can be obtained from linear regression and the result is shown in Fig. 4(b). Furthermore, ¹BODIPY^* kinetics can be determined independently at the near infrared spectral region where the spectral overlap is negligible (Fig. S9a), and the kinetics of each species can also be extracted independently by subtraction at isosbestic points (see supplementary material S8 for details). As shown in Figs. S9b and S10, the spectra and kinetics obtained from the subtraction and MCR analysis agree well with each other, further supporting the data analysis procedure. As shown in Fig. 4(b), from 1 to 10 ps, there is an initial fast decay of ¹BODIPY^* with little growth of the CS state, which can be attributed to nonradiative decay induced by partial aggregation of BODIPY, as mentioned above.⁵⁹ At later delay time ranges, the kinetic traces show the decay of ¹BODIPY^* within 10 ns, which leads to the corresponding growth of the CS state, and the decay of the CS state at longer delay times produces ³BODIPY^*.

The three kinetic traces were globally fit according to the kinetics model shown in Fig. 4(c), in which two configurations of BODIPY binding to the QD surface with a percentage of a_i are assumed (see supplementary material S9 for details). The global fitting result is shown in Fig. 4(b) and Table S2. From the obtained rate constants, the efficiency of each step can be calculated from the branching ratio,

in which Φ_j is the efficiency of step j and k_other is the rate constant for the competing steps. The calculated efficiencies are shown in Fig. 4(c). The average time constants for specific processes can be calculated as

The average charge separation time constant in the ¹BODIPY^*-QD state is (6.33 ± 1.13) ns, which results in (53.0 ± 5.7)% of charge separation (to form BODIPY^+·-QD^−·) and (47.0 ± 5.0)% of decay back to the ground state. The average charge recombination rates from the BODIPY^+·-QD^−· state to the BODIPY triplet excited state and singlet ground state are (158 ± 28) ns and (123 ± 21) ns, respectively, which results in (51.3 ± 4.1)% yield of ³BODIPY^* formation. From the energy alignment of CdSe and BODIPY shown in Fig. S6, the driving force for charge recombination from BODIPY^+·-QD^−· to form the ground state is large enough (ΔG ≈ −1.52 eV) to fall into the Marcus inverted region, resulting in the slow charge recombination rate.⁶⁷ Because the energy of ³BODIPY^* is unknown, we could not determine the exact driving force of charge recombination to form the triplet excited state. However, the similar rates of the two charge recombination pathways and the difference in their driving force indicate the possible different coupling strengths of these two pathways. The efficiency of the overall triplet excited state yield is determined to be Φ = (27.2 ± 3.0)%, limited by competing pathways in both the charge separation and recombination steps. The efficiency of charge separation can be improved by increasing the rate of electron transfer (ET) from ¹BODIPY^* to the QD, which according to an extensive previous study can be achieved by increasing the ET coupling strength and/or driving force.^31,68,69

Extensive previous studies in organic donor-acceptor complexes have shown that triplet excited states can be generated from charge separated state intermediates through radical pair intersystem crossing (RP-ISC) or spin-orbit charge transfer intersystem crossing (SOCT-ISC). The first mechanism involves conversion from the spin-correlated singlet radical pair¹(Acceptor^−·-Donor^+·) to the triplet radical pair³(Acceptor^−·-Donor^+·) state when the hyperfine interactions within the radical centers are larger than the exchange interaction between them.^49,54 In the second mechanism, the triplet excited state is formed if charge recombination is accompanied by a significant change in the orbital angular momentum.^51,52 A previous study of the TPD-CdS QD has shown that the TPD triplet state is generated from the charge separated state by exciting either the CdS QD or TPD.⁵⁴ Detailed EPR characterization and the magnetic field effect (MFE) of the ³TPD^* yield show that the RP-ISC mechanism with spin-correlated radical pair intermediates dominates in this system, although 20% of the triplets are formed through the SOCT-ISC. The dominance of RP-ISC can be understood by the large hyperfine interactions in nanocrystal systems, in which the spin of electrons can be readily flipped compared to organic chromophores. Furthermore, because of the relatively large size of the particles (on the order of 3 nm in diameter) compared to organic chromophores, the center-to-center distance between the spins is relatively large, which may lead to a small exchange interaction between the electrons in the QD and on the surface adsorbed molecule.^70,71

In our TA study, we cannot distinguish¹ (CdSe^−·-BODIPY^+·) from³(CdSe^−·-BODIPY^+·) radical pairs, and therefore cannot distinguish the RP-ISC or SOCT-ISC pathways for triplet formation. Because ³BODIPY^* are generated with similar time scales of charge separation (hundreds of picoseconds) and charge recombination (tens to hundreds of nanoseconds) to previously reported CdS-TPD complexes,⁵⁴ we propose that RP-ISC may also play a dominant role in CdSe-BODIPY complexes. One notable difference between our observation and the previous report is the nature of the electron in the QD in the radical pair state. In the study of CdS-TPD complexes, the optical signal of the QD was not directly observed and the g-factor measured by the EPR indicates that the electron is localized on the QD surface.^54,72 While our study does not directly measure the g-factor, direct observation of the bleach of the QD exciton band in the radical-pair state confirms that the electron is at the 1S level of the QD, delocalized throughout the whole particle. Future EPR characterization and MFE experiments are required to reveal whether the spins in the CdSe^−·-BODIPY^+· radical pair are correlated.

In summary, we have demonstrated the generation of the long-lived triplet excited state of modified BODIPY attached on CdSe QDs. TA spectroscopy confirms that the photo-excitation of a BODIPY singlet state leads to electron transfer to the QD to form a charge separated state (with an oxidized BODIPY and CB electron in the QD) with a quantum efficiency of 53.0%, competing with radiative and nonradiative decays within BODIPY. The charge separated state undergoes charge recombination to form a BODIPY singlet ground state (48.7%) and triplet exited state (51.3%). The overall quantum efficiency for BODIPY triplet excited state generation from its singlet excited state is calculated to be (27.2 ± 3.0) %. Future experiments, such as EPR, are required to elucidate the mechanisms of triplet excited state generation from the radical pairs. This report suggests that triplet excited state formation through the charge transfer intermediate may be a general approach for these QD-adsorbate hybrids. Furthermore, the observation of long-lived radical pairs suggests that these hybrid materials may be an interesting spintronics platform for further exploration.

See the supplementary material for Scheme S1, Figs. S1–S10, Tables S1 and S2, synthesis of BODIPY and CdSe QD, transient absorption spectroscopy setups, cyclic voltammetry measurement, spectra, and kinetic analysis, and fitting methods.

T.L. gratefully acknowledge the financial support from the National Science Foundation (Grant Nos. CHE-1709182 and CHE-1726536). E.A. acknowledges support from the National Science Foundation (Grant No. CHE-1821863).

The authors declare no competing financial interests.