The control and manipulation of dynamic spin processes in multispin molecular excited states is critical to developing large spin polarization effects [1][2][3][4] for molecular spintronics [5][6][7][8] and for engineering entangled spin systems for use in molecular quantum information processing technologies. [9][10][11][12] A key goal toward understanding electronic structure contributions that govern dynamic excited state spin processes focuses on the nature of the photoexcited states involved and the interactions between multiple spin centers that are defined by pairwise magnetic exchange interactions (J i ). Magnetostructural correlations, [13][14][15][16] where the exchange interaction is directly related to a structural parameter, have greatly impacted our understanding of electronic structure contributions to J i [17][18][19][20][21][22] in the ground state and guided fields ranging from bioinorganic chemistry 23,24 to molecular magnetism. 25,26 In marked contrast, there exists a dearth of excited state magnetostructural correlations, particularly in cases where multiple exchange interactions are present. This has prompted us to design new molecular systems where photoexcitation generates excited states with multiple spin centers that are exchange coupled to one another. This allows for correlations to be developed between excited state J values and the underpinning geometric and electronic structure of the molecule, as well as the nature of excited state spin polarizations, dynamics, and lifetimes.
One strategy for designing such a tractable system is to covalently attach a spin center (e.g., an organic radical) to a closed shell donor-acceptor chromophore. Upon photoexcitation, the open shell singlet nature of the chromophore results in the emergence of excited state J i between the localized chromophore spins, which constitute the chromophore's excited singlet (S) and triplet (T) states, and the spin localized on the organic radical. Thus, the radical spin center can function to admix components of the chromophore S and T states via the excited state exchange couplings. This has the potential to control entanglement of spins, and facilitate large spin polarization effects without the need for intersystem crossing. 1,2,4,11,12 Importantly, the new J i that result from appending an additional spin to a chromophore play a critical role in modulating excited state spin dynamics, resulting in a strategy for exerting wave function control over excited state lifetimes. 1,2,4,[27][28][29] Although molecular electron and nuclear spin manipulations may be controlled using pulsed NMR and EPR techniques, 30,31 our focus here is on spin manipulation via excited state magnetic exchange interactions.
Herein, we present a series of molecules (Figure 1) that conclusively demonstrate the manipulation of excited state wave functions and spin polarizations by variation of the pairwise magnetic exchange interaction between a stable radical and one of the spins that derive from the open shell nature of the excited state donor-acceptor dyad.
Electronic Absorption Spectra and Molecular Orbital Description of the Chromophores. The electronic absorption spectra for (t-Bu 2 bpy)Pt(Cat-R), 1-t-Bu, and three radical-elaborated complexes of the structural motif (t-Bu 2 bpy)Pt(Cat-R), where R = NN (1-NN), 2,5-thiophene-NN (1-Th-NN), para-phenyl-NN (1-Ph-NN) and Cat = 3-tertbutyl-ortho-catecholate and NN = nitronylnitroxide radical (S = 1/2), are displayed in Figure 2. A broad, featureless ligand-to-ligand charge transfer [32][33][34] (Cat → bpy LL′CT) band is observed at ∼16 000 cm -1 in the low energy region of these electronic absorption spectra. This relatively intense Cat → bpy LL′CT band obscures a localized NN radical based transition (viz. NN(SOMO) → Ph-NN(LUMO)-Ph-NN(HOMO) → NN(SOMO)), which possesses a low oscillator strength in the 16 000-18 000 cm -1 region of the spectra of 1-NN, 1-Th-NN, and 1-Ph-NN. 35 Neither the pendant NN radical spin nor the nature of the bridge (Ph or Th) in these (t-Bu 2 bpy)Pt(Cat-R) chromophores result in dramatic changes to the LL′CT band relative to the nonradical elaborated 1-t-Bu. The 1-t-Bu parent molecule is only slightly red-shifted (∼1000 cm -1 ) relative to the radical-elaborated species, and this principally derives from the electron withdrawing nature of the NN moiety.
Figure 3A,B shows the frontier orbitals that contribute to the observed LL′CT excitations in 1-t-Bu and the radicalelaborated complexes, respectively. The NN SOMO is absent in the molecular orbital scheme for the 1-t-Bu parent complex, and this leads to a diamagnetic |S 0 ⟩ singlet ground state configuration for 1-t-Bu. Thus, a one-electron promotion from the Cat HOMO to the bpy LUMO in 1-t-Bu results in both singlet, |S 1 ⟩, and triplet, |T 1 ⟩, charge-separated, biradical excited states with the hole localized primarily on the Cat donor (viz. a SQ radical) and the promoted electron localized predominantly on the bpy acceptor. 32 The Cat-bpy chromophore components of these configurations are depicted within the color-coded boxes of Figure 3C. The nature of this LL′CT transition has been evaluated using time-dependent density functional theory (TD-DFT), and the analysis indicates that the LL′CT transition is ∼99% HOMO → LUMO (CAT → bpy) in nature with only a minor contribution from the Pt center. 32 The S 0 → S 1 transition in 1-t-Bu is spin-allowed, while the S 0 → T 1 transition is spin-forbidden. A similar manifold of excited states is present at higher energy. These states are generated from one-electron promotions originating on the Cat HOMO-1 to the bpy LUMO, and these |S 2 ⟩ and |T 2 ⟩ excited states are polarized along the short, in-plane axis, perpendicular to the xpolarized S 0 → S 1 and S 0 → T 1 transitions. 32 Accurate assignments of the S 0 → S 2 and S 0 → T 2 transitions are complicated by the spin-forbidden nature of the latter, and the fact that the S 0 → S 2 transition possesses a weaker oscillator strength since it is polarized along the short axis of the molecule, perpendicular to the charge transfer direction.
As mentioned above, appending a persistent NN radical to 1t-Bu results in the appearance of the localized NN radical SOMO in the molecular orbital scheme, Figure 3B. This does not change the gross orbital parentage of the one-electron promotions that contribute to the LL′CT transitions observed in 1-NN, 1-Th-NN, and 1-Ph-NN. However, in contrast to 1-t-Bu, when the NN-elaborated complexes are optically excited to their low-energy Cat → bpy LL′CT excited states the near quantitative charge transfer 32
There are now two dominant pairwise exchange interactions in the 3-spin LL′CT excited states for 1-NN, 1-Th-NN, and 1-Ph-NN, and the nature of these excited state exchange interactions are discussed below.
Excited State Exchange Interactions. When the NN radical spin is appended to the 1-t-Bu chromophore, the ground state changes from S T = 0 to an S T = 1/2 spin doublet, | S 0 ,1/2⟩. The S 0 in the |S 0 ,1/2⟩ function for (t-Bu 2 bpy)Pt(Cat-R) indicates the ground state spin singlet nature of the (t-Bu 2 bpy)Pt(Cat) chromophore and the "1/2" indicates the total spin of the system (Figure 3C). This results in a more complex excited state spin manifold relative to the |S i ⟩ and |T i ⟩ states of 1-t-Bu. The new excited state manifolds are now described in terms of spin-quartet |T i ,3/2⟩, and spin-doublet |T i ,1/2⟩ and | S i ,1/2⟩ excited states due to the exchange interaction between the S = 1/2 NN radical spin center and the S i and T i excited states of the chromophore (Figure 3C), respectively. Thus, we may now take advantage of the spin-Hamiltonian formalism to determine the energy levels of the |T i ,3/2⟩, |T i ,1/2⟩, and |S i ,1/ 2⟩ excited states in terms of the magnetic exchange interactions between the individual SQ•, bpy•, and NN• spin centers. These energy splittings are defined by the pairwise exchange Hamiltonian for a linear triad in eq 1. This approach will also allow for the determination of excited state wave functions, spin populations, and spin polarization effects.
Here, the J i are the excited state pairwise exchange interactions between the three spin centers, SQ, bpy•, and NN (the "•" on SQ and NN and t-Bu 2 on bpy have been omitted for brevity), in the (bpy)Pt(Cat-R) complexes, which can be evaluated using experimental data as described below. The paramagnetic 1-NN, 1-Th-NN, and 1-Ph-NN complexes are all linear spin triads in the LL′CT excited state with no direct NN-bpy connectivity. Therefore, the far weaker J NN-bpy exchange interaction can be ingored in the analysis of the data. 13,[36][37][38] Since the LL′CT excited state of each complex is characterized by a unit charge transfer, and J varies linearly with spin densities, J SQ-bpy will be constant in this series. 18,[39][40][41][42] This is reflected in the similarity of the MCD spectra for the series, where the interdoublet splitting is dominated by the magnitude of J SQ-bpy . Thus, the J SQ-NN -dependent |S 1 ,1/2⟩, |T 1 ,1/2⟩, and |T 1 ,1/2⟩ energies are plotted in Figure 4 using a constant J SQ-bpy = 1400 cm -1 (vide inf ra). The exchangedependent mixing of the |S 1 ,1/2⟩ and |T 1 ,1/2⟩ doublet states are an interesting consequence of a linear triad of spins with two unequal, pairwise exchange interactions 43 as described by eq 1. 26 In this general case, the |T i ,1/2⟩ and |S i ,1/2⟩ states with the same S T = 1/2 total spin can mix with one another via an off-diagonal matrix element in the exchange Hamiltonian. 26,44 This is important, since it means that the magnetic exchange interaction (J i ) between an electron spin and a photogenerated electron-hole pair can facilitate a localized intersystem crossing (|S i ,1/2⟩ → |T i ,1/2⟩) between the chromophore LL′CT S i and T i excited states. 1,2,29,45 The magnitude of the |S 1 ,1/2⟩ and |T 1 ,1/2⟩ excited state wave function mixing can be parametrized by λ, 44,46 and this quantifies the amount of |T 1 ,1/2⟩ (|S 1 ,1/2⟩) character admixed into the |S 1 ,1/2⟩ (|T 1 ,1/2⟩) wave function by the J SQ-NN exchange interaction according to
(2)
where the primed functions indicate exchange-admixed states (see also Figure 4). Thus, knowledge of the individual J SQ-NN and J SQ-bpy pairwise exchange interactions (vide infra) allows us to determine the value of λ, and the mixing coefficient, sin λ, for each of our radical-elaborated chromophores according to eq 4:
yielding the excited state admixed wave functions (|S 1 ,1/2⟩′ and |T 1 ,1/2⟩′) directly from experimental observables. The J SQ-NN exchange values are known with high accuracy from ground state magnetic susceptibility measurements, 19 and the J SQ-bpy value is determined using magnetic circular dichroism (MCD) spectroscopy.
Magnetic Circular Dichroism Spectroscopy. Both the S 1 and T 1 LLC′T excited states of 1-t-Bu possess A 1 symmetry at the local C 2v symmetry of the Pt site (Figure 3A). As such, there is no spin orbit matrix element that connects these two states and ⟨T 1 |L i |S 1 ⟩ = 0 (the L i transform as a 2 , b 1 , and b 2 in C 2v ). Thus, conversion of the S 1 state to the T 1 state by intersystem crossing (ISC) is forbidden, and fast nonradiative relaxation back to the S 0 ground state is observed. 32 In contrast, spin orbit mixing of the low energy S 1 and T 1 ( 1 A 1 and 3 A 1 ) states with the higher energy T 2 and S 2 ( 3 B 1 and 1 B 1 ) states is symmetry allowed. 32 In fact, the presence of |S 1 ,1/2⟩ -|T 1 ,1/2⟩ wave function mixing, the paramagnetic nature of the ground states, and Pt-induced excited state spin orbit (SOC) "switches on" an exchange-dependent magnetooptical activity in these radicalelaborated complexes, which can be probed exclusively and at high resolution by MCD spectroscopy. A pictorial mechanism for the observed dependence of MCD activity on λ is summarized in Figure 5. 47,48 No temperature-dependent C-term MCD is observed for diamagnetic chromophores like the 1-t-Bu parent complex since a paramagnetic ground state is required. 47,48 Furthermore, C-term MCD is not expected for organic radicals like the catechol-bridge-NN ligands due to the small SOC constants of the constituent atomic centers. 47,48 In contrast, C-term MCD signals for radical elaborated 1-NN, 1-Th-NN, and 1-Ph-NN are observed and the corresponding spectra are presented in Figure 6. Thus, the incorporation of a pendant, persistent radical to the (t-Bu 2 bpy)Pt(Cat) chromophore imparts magnetooptical activity in 1-NN, 1-Th-NN, and 1-Ph-NN with a magnitude that is both temperature-and bridgedependent. Although only one LL′CT transition (|S 0 ⟩ → |T 1 ) is observed in the electronic absorption spectrum of 1-t-Bu, two MCD bands are observed in this LL′CT region for radical- 1) between the |T 1 ,1/2⟩ and |S 1 ,1/2⟩ configurations, as well as between the two higher energy |T 2 ,1/2⟩ and |S 2 ,1/2⟩ configurations. This is important as it relaxes the spin selection rule and allows for the onset of C-term MCD activity. elaborated 1-NN, 1-Th-NN, and 1-Ph-NN. We assign the higher-energy of these two MCD bands as the spin-allowed |S 0 ,1/2⟩ → |S 1 ,1/2⟩ transition and the lower energy MCD band as the |S 0 ,1/2⟩ → |T 1 ,1/2⟩ transition within the lowest energy LL′CT manifold (Figures 3C and5). The assignment of these two, low-energy MCD bands as arising from excited states with the same orbital parentage is supported by their relative energies with respect to the LL′CT band in the electronic absorption spectra, and the fact that the vibronic progressions observed for the |S 0 ,1/2⟩ → |T 1 ,1/2⟩ and |S 0 ,1/2⟩ → |S 1 ,1/2⟩ transitions (0-0 to 0-1 = 1276 cm -1 ) are identical and are thus built upon the same progression forming mode.
The magnitude of the J SQ-bpy exchange interaction can be directly addressed from the low-temperature MCD spectra presented in Figure 6, where both the |S 0 ,1/2⟩ → |S 1 ,1/2⟩, and |S 0 ,1/2⟩ → |T 1 ,1/2⟩ transitions that possess Cat → bpy LL′CT character are observed. The pure electronic 0-0 transitions at 18 300 cm -1 (|S 1 ,1/2⟩) and 15 800 cm -1 (|T 1 ,1/2⟩) are clearly present in the 1-NN spectra and less so in the 1-Th-NN spectra. This allows J SQ-bpy to be estimated from the energy difference between the |S 0 ,1/2⟩ → |S 1 ,1/2⟩, and |S 0 ,1/2⟩ → |T 1 ,1/2⟩ 0-0 transitions (2J SQ-bpy ≈ 2500 cm -1 ). As mentioned previously and depicted in Figure 1C, the excited state J SQ-NN values detailed in eq 1 can be approximated by the corresponding ground-state biradical J SQ-NN values. These J SQ-NN values (Table 1) have been determined from magnetic susceptibility measurements on Tp Cum,Me Zn(SQ-bridge-NN) complexes that possess the same charge/spin distributions and the same bridge fragments as is found in the lowest energy LL′CT excited of 1-NN, 1-Th-NN, and 1-Ph-NN: Tp Cum,Me Zn(SQ-NN) (J SQ-NN = +550 cm -1 ), Tp Cum,Me Zn(SQ-Th-NN) (J SQ-NN = +200 cm -1 ), 19 and Tp Cum,Me Zn(SQ-Ph-NN) (J SQ-NN = +100 cm -1 ). 19,40 The J SQ-NN -dependent |S 1 ,1/2⟩, |T 1 ,1/2⟩, and |T 1 ,3/2⟩ state energies that were given in Figure 4 utilize the experimentally observed energy gap between the 0-0 transitions associated with |S 0 ,1/2⟩ → |S 1 ,1/2⟩ and |S 0 ,1/2⟩ → |T 1 ,1/2⟩ transitions in 1-NN as a starting point to iteratively determine the value of J SQ-bpy (1400 cm -1 , Table 1) as J SQ-NN → 0 cm -1 . Thus, a 2J SQ-bpy splitting of 2800 cm -1 is determined to be the intrinsic excited state |S 1 ⟩ -|T 1 ⟩ singlet-triplet gap for the parent complex, 1-t-Bu. Furthermore, using the experimentally determined J SQ-NN and J SQ-bpy values in the context of eq 4 allows us to determine λ which varies from 11.5 for 1-NN to 1.83 for 1-Ph-NN (see Table 1). In the present case, J SQ-bpy is effectively a constant, while J SQ-NN is an exchange-modified variable (Figure 4 and Table 1). Our experimentally determined values for J SQ-bpy and J SQ-NN , when evaluated in the context of eq 1, allow us to demonstrate the effects of J SQ-NN modulation on the |S 1 ,1/2⟩′ and |T 1 ,1/2⟩ spin eigenfunctions, as well as the degree of excited state spin polarization in these radical-elaborated complexes. The details of these exchange modulated excited state spin polarizations are discussed in detail below.
Implication for Spin Polarization Effects as a Function of Radical-Chromophore Exchange. Figure 7 shows the excited doublet net spin populations on the NN, SQ, and bpy centers (green, blue, and red, respectively, in Figure 7) that are calculated by applying a Heitler-London approach 26,49 using experimentally determined pairwise exchange parameters within the Heisenberg-Dirac-Van Vleck (HDVV) formalism as detailed by Kahn. 26 Importantly, the spin populations localized on the NN, SQ, and bpy centers are determined by the magnitude of the interdoublet state mixing parameter, λ. 44 In the absence of a J SQ-NN exchange interaction, the spin density of the SQ-bpy LL′CT dyad singlet state is zero and the MCD intensity is also zero. The magnitude of J SQ-NN modulates the excited state doublet wave function mixing, altering the SQ-bpy spin density distribution, and the MCD intensity is found to vary linearly with λ, Figure 6B. This plot of MCD intensity vs λ clearly shows that I MCD → 0 as λ → 0, indicating that there will be no spin population on the chromophoric SQ or bpy sites in the |S 1 ,1/2⟩ wave function in the absence of exchange-mediated mixing with the |T 1 ,1/2⟩ state. 46 Optical excitation from the |S 0 ,1/2⟩ ground state to |S 1 ,1/2⟩′ in 1-NN, 1-Th-NN, and 1-Ph-NN results in an exchangedependent net spin population transfer from NN (100% in the electronic ground state) to SQ and bpy in the |S 1 ,1/2⟩′ excited state. The J SQ-NN excited state exchange interaction that allows the |S 1 ,1/2⟩ and |T 1 ,1/2⟩ states to mix with one another also provides a novel mechanism for a highly efficient |S 1 ,1/2⟩′ → |T 1 ,1/2⟩′ internal conversion (IC) process. Population of the We note that both the magnitude and sign of these spin populations can be further controlled by the nature of the J SQ-NN interaction through its effect on λ. Most importantly, the |T 1 ,1/2⟩′ → |T 1 ,3/2⟩ ISC is both spin-and symmetryforbidden in these systems resulting in no net population of |T 1 ,3/2⟩ and no observed |T 1 ,3/2⟩ → |S 0 ,1/2⟩ phosphorescence. Thus, a nonradiative |T 1 ,1/2⟩′ → |S 0 ,1/2⟩ IC process dominates repopulation of the |S 0 ,1/2⟩ ground state. This |T 1 ,1/2⟩′ → |S 0 ,1/2⟩ relaxation causes the NN spin to flip again, with a concomitant temporal redistribution of the |S 0 ,+1/ 2⟩ and |S 0 ,-1/2⟩ ground state populations as Boltzmann equilibrium is reached. This also provides a mechanism for the ultrafast hyperpolarization of nuclear spins located on, or attached to, the bpy acceptor. The bpy acceptor nitrogen atoms are located ∼8 Å from the NN bridgehead carbon, and are therefore remote from the spatial confinement of the ground state electron spin localized on the NN radical. Future experiments will probe hyperpolarization of the NN spin and critical |T 1 ,1/2⟩′ → |S 0 ,1/2⟩ relaxation processes, including the potential for magnetic exchange control of |T 1 ,1/2⟩′ → |S 0 ,1/2⟩ IC rates to repopulate the |S 0 ,1/2⟩ ground state.
We have developed a chromophoric spin system where each spin-1/2 subsystem in the tripartite excited doublet states is entangled with the remaining two spins, and we have probed the magnetooptical properties of these systems using variabletemperature MCD spectroscopy. The open-shell singlet nature of the LL′CT charge separated state provides a convenient platform for facilitating strong pairwise excited state exchange coupling between the three spin centers (NN, SQ, and bpy•) in (t-Bu 2 bpy)Pt(Cat-Bridge-NN). A critical magnetic exchange mixing is observed between the excited |S 1 ,1/2⟩ and |T 1 ,1/2⟩ wave functions, providing a means of accessing the "dark" T 1 triplet configuration of the bpy-Cat chromophore that is otherwise spectroscopically inaccessible in the nonradical elaborated 1-t-Bu parent complex. Moreover, the |T 1 ,1/2⟩′ excited state wave function is shown to possess dramatically different NN, SQ, and bpy• spin populations relative to the ground (|S 0 ,1/2⟩) and excited (|S 1 ,1/2⟩′) doublet states, and these spin populations are controlled by the pairwise exchange interactions (J SQ-bpy and J SQ-NN ) and the interdoublet mixing parameter, λ. Remarkably, the MCD intensity is found to be a linear function of λ, correlating magnetooptical activity with spin polarization effects. Thus, our results show that magnetic "exchange fields" generated by radical attachment to molecular chromophores can control the admixture of chromophore singlet and triplet states, modulate entanglement of remote spin centers, facilitate spin polarization transfer processes with high efficiency, and yield exchange dependent magnetooptical activity. Furthermore, this tripartite system can be generated on an ultrafast time scale via the spin-allowed |S 0 ,1/2⟩ → |S 1 ,1/2⟩′ photoexcitation process (Figures 1B and3C) and can be modulated by well-defined synthetically accessible individual pairwise exchange interactions, which determine interstate mixing between exchange-coupled |S 1 ,1/2⟩ and |T 1 ,1/2⟩ states through λ. 32,50 ■ EXPERIMENTAL SECTION Synthesis. All synthetic procedures were carried out under a N 2 atmosphere and used reagents purified by standard literature procedures. NN-CatH 2 , NN-Ph-CatH 2 , NN-Th-CatH 2 , and (di-tertbutyl-bipyridine)PtCl 2 were synthesized by literature methods. 2-Methyltetrahydrofuran was purchased from Alfa Aesar and purified by passage down a column of basic alumina immediately before use. (Ditert-butyl-bipyridine)Pt(Cat) compounds were synthesized by reaction of (di-tert-butyl-bipyridine)PtCl 2 , protonated catechol ligand, and tert-BuOK in MeOH. Complexes 1-t-Bu, 1-NN, 1-Ph-NN, and 1-Th-NN were prepared using literature methods (see the Supporting Information, SI for synthetic details). The solution, room-temperature EPR spectra (see Figure S1 for spectra) of these NN-elaborated complexes are characteristic of an organic NN radical and display hyperfine splitting due to electron spin coupling to two equivalent I = 1 14 N nuclei. The isotropic EPR spectra possess spin-only g values (g = 2.001) with no evidence of hyperfine coupling to the 195 Pt nucleus (I = 1/2). These spectral characteristics derive from the fact that the NN singly occupied molecular orbital (SOMO) possesses a node at the carbon atom that is bonded to the Cat ring. 18,41,51 This effectively limits both NN spin-and electron delocalization, and results in no Pt character being present in the NN SOMO and no measurable spin delocalization onto the Pt ion. Additional synthetic details and characterization are available in the SI.
MCD Spectroscopy. MCD spectra were collected on an Oxford SM4000T magnetooptical cryostat interfaced with a Jasco J-810 spectropolarimeter. Samples were made up as 2-methyltetrahydrofuran solutions and injected into a custom brass sample holder containing spectrosil quartz windows, a butyl rubber spacer, and rubber o-rings. The samples were frozen in liquid N 2 and loaded into the MCD cryostat. Depolarization was checked by use of a nickel tartrate sample placed before and after the cryostat; samples that displayed less than 5% depolarization were deemed suitable. Baseline correction was performed by subtraction of a 0 T spectrum. Where necessary, B-term spectral contributions were eliminated by subtraction of a hightemperature (50 K), high-field (7 T) spectrum from the lowtemperature spectra. Samples with very weak S/N were further enhanced by the collection of spectra in positive and negative applied magnetic fields. Spectral subtraction (positive -negative) then yields MCD spectra that are free of zero-field contributions.
EPR Spectroscopy. Fluid solution room temperature EPR spectra were collected on a Bruker EMX EPR spectrometer. Sample concentrations were ∼0.1 mM in 2-methyltetrahydrofuran.
Electronic Absorption Spectroscopy. Electronic absorption spectra were taken on a Hitachi U-4100 double beam spectrophotometer. Room temperature samples were made up as 2-methyltetrahydrofuran solutions and loaded into a microvolume quartz cuvette. Low temperature absorption spectra were taken with a custom Janis liquid He flow cryostat mounted in a Hitachi U-3100 spectrophotometer. Samples were loaded into a brass sample holder identical to that used for MCD measurements, and baseline corrected with a frozen solvent sample. Temperature was monitored with a silicon diode and a Lakeshore temperature controller and was adjusted by changing the flow of liquid He through the cryostat and with a nichrome wire heater located above the sample.
Computational Methods. All calculations were performed with the ORCA 3.0.1 or 3.0.2 program suite. 52 Density functional theory (DFT) calculations were performed using the def2-TZVP basis 53 and the PBE GGA functional. 54 CASSCF/NEVPT2 55-58 calculations used quasi-restricted orbitals (QROs) from the DFT calculations as the initial guess orbitals. Minimal active space calculations (CAS (3,3) or CAS(2,2)) for radical elaborated or nonelaborated compounds, respectively) were first performed, and the molecular orbitals obtained were used for subsequent calculations using larger active spaces. All CASSCF calculations utilized the RIJCOSX approximation. 59 ■ ASSOCIATED CONTENT
J. Am. Chem. Soc. 2018, 140, 2221-2228