Although electronic excited states have been studied by a myriad of spectroscopic techniques that have been complemented by theoretical investigations, the ability to control and manipulate excited state processes and lifetimes remains an important challenge. [1][2][3] Electronic relaxation between singlet and triplet excited states occurs via spin-orbit mediated intersystem crossing (ISC), and this allows access to long-lived excited states that promote photoredox reactions, 4 large spin-polarizations, [5][6] and photo-/electroluminescence. 3,7 Recently, we developed new molecular donor-acceptor chromophores that possess a covalently-attached stable organic radical in order to generate multiple unpaired electron spins in photoexcited states. 8 These spin centers exchange couple to generate excited state spin polarizations that are a function of both the sign and magnitude of the pairwise exchange interactions and allow for ground state magnetooptical activity to arise. The same magnetic exchange interactions that determine variable excited state spin polarizations also promote magnetic exchange dependent excited state wavefunction mixing, 8 and this provides a novel way to control and manipulate excited state lifetimes via the radical-chromophore exchange interaction. Understanding these exchange interactions is important, as they figure prominently in the exciton-polaron interaction present in trion quasiparticles. [9][10][11] The dynamics of these trions have recently been studied in carefully charge-doped single-walled carbon nanotubes following photoexcitation and exciton formation, 12 and excited state exchange interactions are expected to affect the lifetimes of trion and higher-order multipartite quasiparticles.
While several studies have demonstrated how electron spin affects photophysical properties, 11,[13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28] the evaluation of multiple pairwise exchange interactions and their effect on photoexcited states remain relatively unexplored. Our prior work has focused on using a combination of spectroscopy and magnetometry to determine the nature of exchangedependent wavefunction mixing between the spin doublet donor-acceptor charge transfer excited states of these radical-elaborated molecules. 8 To continue our study of magnetic exchange effects on the excited states of donor-acceptor chromophores, we desire specific chromophores that produce excited states with a high degree of biradical character. 3 An excellent chromophoric scaffold for our studies are square planar ligand-to-ligand charge transfer (LL'CT) complexes of the general formula (BPY)Pt II (dichalcogenolene) (dichalcogenolene = benzene-1,2-dithiolate, catecholate, etc.; BPY = 2,2'-bipyridine), which are characterized by LL'CT transitions to excited states that are either short-lived and nonemissive (t ≤ 1 ns; e.g. catecholate (CAT) complexes), 1,29 or long-lived and emissive (t ≥ 10 ns; e.g., benzene-1,2-dithiolate complexes). 1,[30][31] The short-lived, non-emissive excited states of (BPY)Pt II (CAT-R) complexes (Fig. 1A,B) allow us to investigate the effects of excited state radical-chromophore exchange interactions on non-radiative excited state decay rates. Our choice of this chromophore is based not only on its short-lived, non-emissive LL'CT excited state, but also on the charge-and spin distribution of the LL'CT excited state 1,3 (Fig. 1B) and the resemblance of the excited state SQ-NN interaction to those in our fully-characterized ground-state LZnSQ-B-NN complexes (Fig. 1C). 8,[32][33][34][35][36][37][38] Ground state magnetic exchange couplings between nitronylnitroxide radical (NN) and semiquinone (SQ) in LZnSQ-B-NN complexes [33][34][35][36]38 very closely approximate the corresponding exchange couplings in the LL'CT excited states of the NN-elaborated platinum complexes since a full unit of charge is transferred in the LL'CT excited states (Fig. 1B,C). 3 Moreover, we recently demonstrated how these excited state wavefunctions can be determined from an a priori knowledge of the experimentally-determined exchange parameters. 8 In this report, we show that radicalelaborated donor-acceptor (D-A) dyads reveal a remarkable relationship between excited state lifetimes and wavefunction mixing, which derives from the pairwise excited state magnetic exchange couplings between R, D, and A spins.
CAT ➝ BPY Charge Transfer, 3-Spin LLCT Excited States, and Excited State Doublet Wavefunction Mixing. The synthesis of radical-elaborated 1-NN, 1-Th-NN, and 1-Ph-NN has been reported previously, 8 and the new compounds, 1-m-Ph-NN and 1-Ph2-NN, that have been prepared for this study and are characterized as described in the SI (Fig. 1A). The bridge fragments, B, for the (BPY)Pt II (CAT-B-NN) complexes were selected because they correspond to those in our donor-acceptor biradical complexes, LZnSQ-B-NN, Fig. 1, and thus provide a range in both the magnitude and sign of JSQ-NN values. [33][34][35][36][37][38][39][40] This choice results in the greatest range of wavefunction mixing allowed by this particular spin system (vide infra).
These complexes feature a broad, solvatochromic CAT → BPY LL'CT transition spanning the 450 -800 nm region (Fig. 2). The unit charge transfer results in the LL'CT excited state of the NN radical-containing complexes possessing excited state triradical character (BPY•)Pt II (SQ•-B-NN•) (B = bridge; Fig. 1B;"•" on NN and on SQ omitted hereafter for brevity). Two pairwise magnetic exchange interactions play a crucial role in defining the (BPY•)Pt II (SQ-B-NN) LL'CT triradical excited state energies and photophysics: that between the BPY• and SQ radicals (JSQ-BPY = 1400 cm -1 and constant within the series), 8 and that between the SQ and NN radicals (JSQ-B-NN = variable -depending on the bridge, B: -32 to +550 cm -1 , Fig. 1C). [32][33][34][35][36]38 The JBPY-NN exchange interaction is expected to be markedly smaller than the adjacent JSQ-B-NN and JSQ-BPY couplings due to the long NN-BYP distance and poor BPY-NN orbital overlap. We therefore ignore the effects of JBPY-NN in our analysis. [41][42][43][44][45] The JSQ-BPY and JSQ-B-NN couplings give rise to two doublets and one quartet state in the excited state LL'CT manifold: Dsing, Dtrip, and Q, respectively, Fig. 3. Importantly, Dsing and Dtrip are admixed by the JSQ-BPY and JSQ-B-NN exchange interactions, 8 and the degree of this admixture is controlled by the variable JSQ-B-NN exchange interaction as per Eqs. (1-3). 8 It is important to note that the JSQ-B-NN exchange mediated by the meta-phenylene bridge in 1-m-Ph-NN is antiferromagnetic, and this results in the dark Q state lying energetically between the Dsing and the Dtrip states. The quartet state, Q, does not mix with these doublets and is therefore unaffected by the degree of Dsing -Dtrip exchange mixing. 8 The cosλ and sinλ coefficients in Eqs. ( 1) and ( 2) define the degree of pure Dsing and Dtrip doublet (|S1,1/2> and |T1,1/2>, respectively) character that is admixed to form the magnetic exchange perturbed wavefunctions. In these equations, S1 and T1 represent the singlet and triplet excited states of the (BPY)Pt II (CAT) core chromophore. The pairwise excited state exchange coupling constants, JSQ-B-NN 33,35,46 and JSQ-BPY, 8 are related to this wavefunction mixing through λ as defined in Eq. ( 3), and the wavefunction mixing coefficients (sin λ, cos λ) are listed in Table 1. Although the Q state is included in Fig. 2, there is no evidence of intersystem crossing (ISC) to Q, and this is consistent with the lack of doublet-quartet exchange mixing and the absence of ISC to the excited 3 LL'CT in the parent complex, 1-t-Bu (vide supra). 1 With respect to wavefunction control of the excited state lifetimes, our focus is solely on the excited LL'CT doublets, Dsing and Dtrip, and their bridge-modulated admixture.
An important consequence of Dsing -Dtrip mixing in (BPY•)Pt II (SQ-B-NN) LL'CT excited states results in a spin-and dipole-allowed D0 → Dsing transition 8 that relaxes by rapid internal conversion (IC) to the Dtrip state (Fig. 3). With respect to the (BPY)Pt II (CAT) core chromophore (the "red" and "blue" spins in Fig. 3), this IC represents a localized S1 → T1 spin conversion within the chromophore that is inaccessible to 1-t-Bu and its non-radical elaborated catecholate-containing derivatives. 8 Thus, the core chromophore triplet character present in the Dtrip excited state predicts that the non-radiative Dtrip → D0 lifetime will be dependent on the magnitude of sinl, and longer Dtrip → D0 lifetimes are expected as this wavefunction mixing decreases due to the increase in chromophore spin triplet character now present in Dtrip.
Summarizing, the magnitude of the bridge-dependent JSQ-B-NN exchange interaction will determine the degree of Dsing character admixed into Dtrip, resulting in the NN-chromophore spin exchange functioning as a molecular rheostat to modulate excited state lifetimes. 4A, these spectra exhibit bleaching of both the ground state UV band ~380 nm and the LL'CT band ~575 nm, as well as transient absorptions in the visible region of the spectrum (~550 and 750 nm). Transient absorption spectral features are in good agreement with the optical transitions observed in the ground state absorption spectra of the LZnSQ-B-NN complexes. 33- 36, 38, 40 The transient absorption spectra at early delay times (~ 1-5 ps) in Fig. 4A display spectral features that span the same wavelengths as those at longer delay times (44~650 ps), but they possess distinctly different spectral band shapes. This observation suggests that more than two electronic states contribute to the transient absorption spectral envelope, and we assign these states to D0, Dsing, and Dtrip (Fig. 3). The Dsing and Dtrip LL'CT excited states possess the same electronic configuration 8 and are therefore expected to exhibit nearly identical excited state distortions relative to the electronic ground state. This results in highly nested Dsing and Dtrip excited state potential energy surfaces and very similar transient absorption features.
The kinetic traces of the spectral transients for each radical-substituted complex are shown in Figs. 4B-F. These data may be fit with as many as four kinetic decay components on the order of hundreds of femtoseconds (τSR; solvent reorganization), ~1 ps (τVR; vibrational relaxation), tens of ps (τIC; internal conversion), and hundreds of ps (τCR; charge recombination), (Fig. 5; Global analysis kinetic fits give charge recombination lifetimes that agree with those from single-wavelength kinetics, see SI for a complete fit analysis of the kinetic data). The short-lived components τSR and τVR can be assigned to solvent reorganization, and vibrational relaxation and redistribution, as has been reported previously for 1-t-Bu. 29 The τIC is tentatively assigned to Dsing → Dtrip internal conversion. Finally, the longest lifetime component is easily assignable as the Dtrip→D0 charge recombination step (tCR). Since the measured charge recombination time constant, tCR, is markedly longer than any of the other time constants, it is not convoluted with the short-lived components included in our analysis of the kinetic data. A Jablonski diagram that shows only doublet states and summarizes the photophysical events with their characteristic time constants is shown in Fig. 5. Importantly, the magnitude of tCR is a function of the excited state exchange-mediated Dsing -Dtrip mixing and therefore the ensuing discussion will focus on this important exchangedependent relaxation process. ps) clearly shows that there is a marked increase in lifetime as the excited state exchange coupling between the SQ and NN radical centers decreases (JSQ-B-NN, Table 1 and Fig. 6). The observed increase in the Dtrip→D0 charge recombination lifetime is counter to what is predicted by the energy gap law, 49 using LL'CT energies observed in the electronic absorption spectra and the results of our electrochemical studies (see SI). Additionally, the observed lifetime increase is also inconsistent with a vibrational "loose bolt" effect 49 The analysis presented above points to exchange-mediated Dtrip→D0 relaxation as the dominant mechanism that accounts for the observed differences in ground state recovery rates between 1-Ph2-NN, 1-m-Ph-NN, 1-Ph-NN, 1-Th-NN, and 1-NN (Fig. 6). This excited state exchange mechanism reveals itself in the form of a dramatic power-law dependence of the charge recombination lifetimes, tCR, as a function of (sinλ) 2 , the latter of which is the amount of Dsing character admixed into the Dtrip wavefunction by the JSQ-B-NN exchange interaction (Eq. 1, Fig. 3 and Table 1). We observe that tCR asymptotically approaches large values as the Dtrip wavefunction tends toward the "pure" Dtrip = |T1,½> that has zero Dsing character. Thus, as (sinl) 2 → 0, the Dtrip → D0 charge recombination event takes on a larger degree of spin forbiddeness due to the increased chromophore triplet character (e.g., 1-Ph2-NN). This leads to dramatic exchange-modulated increases in the charge recombination lifetime. Conversely, an increase in Dsing character admixed into the Dtrip function (e.g., 1-NN), reduces the spin forbiddeness and the Dtrip lifetime is observed to decrease. Importantly, the covalently attached NN radical provides a mechanism for these complexes to attain lifetimes characteristic of the chromophoric triplet state, but without spin-orbit mediated ISC. The new excited state exchange interactions that result from covalent attachment of a radical spin to a chromophore have previously been shown to be important in changing excited state spin dynamics, 16,[21][22][50][51][52] and it was suggested 8 that this could be used as a strategy to exert wave function control over excited state lifetimes, including the charge recombination process observed here. In our prior work, we used MCD spectroscopy and magnetic susceptibility measurements to determine excited state magnetic exchange couplings that between the electron spins in the excited states of these molecules, and that there was no direct spin-orbit coupling matrix element that connects the Dsing and Dtrip functions. Mixing between Dsing and Dtrip was shown to occur via the exchange interaction. Our results therefore provide a unique example of creating entangled triad of electron spins with concomitant spin polarization using visible light. These results are of interest to efforts in molecule-based quantum information science since this excited state spin polarization may be transmitted to the recovered ground state. Dtrip → D0 charge recombination is supported by the functional form of the Dtrip lifetime vs. (sinλ) 2 plot and the fact that that Dtrip lies below the Q state in 1-m-Ph-NN. Thus, the effect of excited state exchange on lifetime can be described as a non-radiative decay counterpart to the exchange mechanism that activates spin forbidden electronic transitions, originally proposed by Tanabe, 53 indicating that spin-orbit coupling using heavy metals is not required to access the core chromophore triplet state when multiple exchange interactions are operative.
We have presented a time-resolved kinetic and spectroscopic study of NN radicalsubstituted (BPY)Pt II (CAT-B-NN) complexes with charge recombination lifetimes that are inversely proportional to the magnitude of the excited state Dsing -Dtrip mixing. This wavefunction mixing and lifetime modulation is a direct consequence of the radicalchromophore magnetic exchange coupling. We show that tCR represents an exchangemediated T1→S0 ISC process that is not present in the (BPY)Pt II (CAT) core chromophore. Our study provides an important mechanism for precise control of charge-separated excited state lifetimes, including IC ground state recovery rates without the stringent requirement of spinorbit coupling. Exchange-mediated wavefunction mixing 8 and its effects on excited state lifetimes may play an important role in charge-doped materials including conjugated organic polymers and carbon nanotubes. More specifically, the implications for further manipulation of these exchange coupled excited states is tremendous, and the concepts detailed here will translate to furthering our understanding of spin-coupled trion states, such as those found in hole-doped carbon nanotubes. 12 This is impactful, for it suggests that exchange coupling of a singlet exciton with a hole polaron could result in the formation of a triplet exciton without the need for SOC. We note that the JSQ-NN values (|JSQ-NN| ~30-550 cm -1 ) are small compared to most excited state singlet triplet gaps (>1000 cm -1 depending on the chromophore), and small compared to heavy transition metal spin orbit coupling constants (~1000 cm -1 ).
Remarkably, the magnitude of the excited state exchange interaction will not have to be large to affect the Dsing -Dtrip spin interconversion and have a dramatic and predictable effect on lifetimes. We have shown that JSQ-B-NN exchange interactions as small as 20 cm -1 in 1-Ph2-NN can dramatically affect charge recombination lifetimes relative to 1-NN. The data presented in Fig. 6 clearly show that even smaller 1-B-NN exchange interactions will lead to markedly longer charge recombination lifetimes. Importantly, the JSQ-B-NN exchange interaction must be large enough to enable conversion from Dtrip to Dsing, but small enough to maximize the chromophore triplet character in Dtrip.