Excited-state antiaromaticity relief drives facile photoprotonation of carbons in aminobiphenyls

A combined computational and experimental study reveals that ortho-, meta- and para-aminobiphenyl isomers undergo distinctly different photochemical reactions involving proton transfer. Deuterium exchange experiments show that the ortho-isomer undergoes a facile photoprotonation at a carbon atom via excited-state intramolecular proton transfer (ESIPT). The meta-isomer undergoes water-assisted excited-state proton transfer (ESPT) and a photoredox reaction via proton-coupled electron transfer (PCET). The para-isomer undergoes a water-assisted ESPT reaction. All three reactions take place in the singlet excited-state, except for the photoredox process of the meta-isomer, which involves a triplet excited-state. Computations illustrate the important role of excited-state antiaromaticity relief in these photoreactions.


Introduction
4][5] In 2002, Wan et al. reported the rst example of direct photoprotonation of a carbon atom of an aromatic ring via excited-state intramolecular proton transfer (ESIPT). 6,7Irradiating 2-phenylphenol in D 2 O-CH 3 CN solution led to regioselective D-exchange at the 2 0 and 4 0 positions.Deuteration at the 2 0 position was believed to proceed through ESIPT involving an initial hydrogen bonding interaction between the OH group of the phenol moiety and the p-system of the adjacent phenyl moiety.Upon irradiation, the phenolic OH proton shis to the 2 0 position of the unsubstituted phenyl ring.][10][11][12][13][14] Many examples of ESIPT and ESPT involving proton transfer, either from an OH group [15][16][17] or an amine group [18][19][20] to a basic carbon atom have been reported.These ndings are at rst sight somewhat surprising, since protonation of a carbon atom of an aromatic ring typically is a slow process in the ground state. 21,22et, why do some aromatic ring carbons gain basicity in the excited-state?Some of us recently related the phenomena of excited-state proton transfer in aromatic organic compounds to a reversal of aromaticity and antiaromaticity in the lowest pp* excitedstates. 23o-Salicylic acid, 24 for example, is 4n + 2 aromatic in the ground state, but shows enhanced antiaromatic character in the rst 1 pp* state.Proton transfer from an OH group to the carbonyl site disrupts cyclic 4n + 2 conjugation in the six membered ring and this reaction alleviates excited-state antiaromaticity (Scheme 1a).The low barriers of many other excited state proton and electron transfer reactions, for example, ESIPT in phenol-benzoxazoles, 25 the photoacidity of naphthols, 26 and proton coupled electron transfer reactions in phenols 27 have been attributed to the effects of excited-state antiaromaticity relief.The concepts of excited-state aromaticity and antiaromaticity were rst explored by Dewar, 28 Zimmerman, [29][30][31] and later Dougherty 32 to explain the mechanisms of pericyclic reactions.Later, Baird proposed, based on a set of perturbation molecular orbital theory analyses, that "the rules for ground state aromaticity are reversed in the rst 3 pp* state: 4n rings display 'aromatic' character whereas 4n + 2 systems display 'antiaromaticity'. 33][39][40] Here, we relate the potential for carbon atom photobasicity in aromatic ring to a reversal of aromaticity and antiaromaticity in the lowest pp* excited states.In the ground state, protonating an aromatic ring and disrupting cyclic 4n + 2 p-conjugation is a slow process due to a loss of aromaticity.But in the lowest singlet or triplet excited state, these rings gain antiaromatic character and the ring carbon atoms display increased basicity, as protonation now affords a channel to alleviate excited-state antiaromaticity (Scheme 1b).To model the effects of excited-state antiaromaticity relief in the photoprotonation of aromatic ring carbons, here, we investigate the photophysics and photochemistry of a series of ortho-(1), meta-( 2) and para-( 3) aminobiphenyls (Scheme 1c).These chromophores can be found in many structures with important applications in supramolecular chemistry or material science. 41We now report a combined experimental and computational study, showing that the aromatic rings of aminobiphenyls can gain signicant antiaromatic character in the singlet and/or triplet states, demonstrating that antiaromaticity relief is a major driving force for the photoreactivities of ortho-(1), meta-(2) and para-(3) aminobiphenyls.

Photophysical properties
Absorption and emission spectra for the aminobiphenyls 1-3 were recorded in CH 3 CN (see Fig. S1-S3 in the ESI †).The absorption spectra of 1 and 2 show maxima at 305 nm, and that of 3 shows a maximum at 283 nm (Fig. 1, top).These absorption bands correspond to the S 1 ) S 0 transition and are characterized as pp* states as conrmed by the symmetry of the computed orbitals.The higher singlet states of 1 (z250 nm, shoulder), 2 (236 nm), and 3 (200 nm) indicate signicant charge transfer character.The emission spectra of 1 and 2 nearly overlap and show maxima at 390 nm, while that of 3 shows a maximum at 361 nm (Fig. 1, bottom).Stokes' shis of all three isomers range between 6600-7500 cm −1 (Table S1 in the ESI †).Quantum yields of uorescence were measured by use of N-acetyl-tryptophanamide in water as a reference (F f = 0.12). 42In accordance with an observed lower photoreactivity, 3 (F f = 0.55) shows higher uorescence compared to 1 (F f = 0.21) and 2 (F f = 0.26).
Time-correlated single photon counting (TC-SPC) experiments were performed to measure the decay of uorescence for 1-3 in CH 3 CN solutions (Fig. S7-S9 and Table S2 in the ESI †).
The decay times do not t to a single-exponential function.In addition to the major decay component (2.5-6.6 ns), a short decay component was detected (<100 ps); however, a precise measurement for the short decay component was not possible with the setup used.Dual uorescence for aminonaphthalenes have been observed and attributed to emission from the L a and L b states. 43Our computations agree with this interpretation, since the excitation at 280 nm used for the SPC can give rise to the adiabatic transition to the S 2 state.The other, less likely interpretation for dual emission would be that it originates from different vibrational levels of the excited states. 44ddition of a protic solvent (i.e., H 2 O) to CH 3 CN increases F f for all three compounds (Fig. S4-S6 in the ESI †).We surmise that the presence of H 2 O may hinder vibrational modes that lead to a non-radiative decay from the S 1 state, thereby enhancing uorescence.Direct comparisons of the emission spectra for 1-3 in CH 3 CN-H 2 O and in CH 3 CN-D 2 O, reveal stronger uorescence and a slower decay in D 2 O (Fig. 2, S10, S11 and Table S3 in the ESI †), especially for 1.This nding strongly indicates the involvement of H 2 O in blocking a non-radiative decay pathway from the singlet excited state; D 2 O shows slower decay kinetics due to a kinetic isotope effect.These ndings are in line with the observed ESIPT reactivity of 1, which takes place only in the presence of H 2 O as a protic solvent.

Acid-base properties
UV-vis and uorescence titrations in acidic media were performed to determine the pK a values and corresponding pK * a values in the S 1 state for 1-3 (Table 1).Fig. 3 and 4 show the spectra for 3, and those of 1 and 2 are included in the ESI (Fig. S12-S18).† Increasing the acidity of the solution resulted in a hypochromic shi of the lowest-energy absorption band for 1-3 (i.e., due to protonation of the nitrogen atom) and a quenched uorescence.The lower uorescence quantum yields of 1-3 in acidic solutions are in line with an observed efficient photoreaction in acidic media.As shown in Table 1, compounds 1-3 display decreased basicity in the S 1 state (lower pK * a values).These results suggest that upon photoexcitation, electron density on the nitrogen atom is delocalized into the aromatic ring, and the nitrogen atom becomes less basic (see computed NPA charges for N in Fig. 6).

Irradiation experiments
Irradiating 1-3 in CH 3 CN-D 2 O (2 : 1) at pD = 2 at z300 nm led to regiospecic deuteration: specically, at the C2 0 position of the phenyl ring in 1 and 3, and the C4 position of the aniline ring in 2. Additionally, irradiating 2 gave a photoredox product 4 (vide infra).The acidity of the mixture was adjusted by addition of D 2 SO 4 or a NaOD-D 3 PO 4 buffer.The content of deuteration and position of the deuterium in molecules 1-3 was determined aer aqueous workup from MS and 1 H NMR, respectively.Experimental procedures and graphs showing dependence of Dexchange with the irradiation time, the D 2 O concentration, and the pH (pD) are included in the ESI (Fig. S19-S34).† Experiments for the naphthyl derivative of 1 under similar acidic conditions (pD = 2 in the presence of D 2 SO 4 ) also showed efficient D-exchange. 20rradiating  S3. †  (2 : 1, v/v, pD = 2) was determined by use of KI/KIO 3 as an actinometer (F 254 = 0.74). 45,46This value is lower compared to the quantum yield for D-exchange in 2-phenylphenol (F 254 = 0.041 ± 0.004), 6 as expected by the lower acidity of the NH 2 versus OH group.We found that the efficiency of D-exchange increases upon increased D 2 O content (between 0.1% to 1% D 2 O) and levels off at concentrations higher than 1% (see Fig. S22 in the ESI †).Interestingly, a higher D 2 O concentration does not lead to regioselective D-exchange at the more distant C4 0 position.These results contrast with Wan's 2002 report on 2-phenylphenol, which showed both C2 0 and C4 0 deuterated photoproducts. 6It was found that, upon irradiating 2-phenylphenol, D-exchange at the C2 0 position takes place via an ESIPT mechanism, whereas at higher D 2 O concentration, D-exchange at the C4 0 position occurs via a H 2 O-mediated ESPT mechanism.Thus, deuteration of position C4 0 may also take place in 1, but it is signicantly less efficient compared to the ESIPT to position 2 0 .
Note that we detected some tri-deuterated compound 1 aer long irradiation (Fig. S19 in the ESI †), which may be due to the less efficient deuteration of the 4 0 position, not detectable by 1

H NMR (<5%).
To investigate whether D 2 O is required for D-exchange, we prepared a deuterated version of 1 (ND 2 , see the ESI † for details) and irradiated it in anhydrous CH 3 CN.No D-exchange occurred at the phenyl rings, suggesting that D 2 O is required for Dexchange to occur.To investigate whether an NH 2 group is required for D-exchange, we prepared the corresponding dimethylated derivative 1(NMe 2 ).Irradiated and non-irradiated samples of 1(NMe 2 ) in CH 3 CN-D 2 O (2 : 1) both showed comparable minor D-exchange, more than ten times less efficient than for 1 (Table S7 †   Formation of a hydroxylamine photoproduct 4 suggests that photoexcitation of 2 triggers a photoredox process.To investigate how 4 was formed, compound 2 was irradiated in CH 3 CN-H 2 O, at different pH values and in the presence of different acids (H 2 SO 4 and HCl).Product 4 was formed regardless of the pH values and the acid used, and these results suggest that an acidic media is not required for the observed photoredox process.We surmise that a proton-coupled electron transfer (PCET) reaction occurs between 2 and H 2 O, generating an OH radical and H atom (vide infra), which then forms 4 and releases H 2 .Although we did not detect the formation of H 2 , it is a probable product based on the proposed photoredox mechanism.
Irradiating  S9 in the ESI †), suggesting that D-exchange may involve some triplet excited-state species.However, some quenching can occur in the S 1 state.We expect that singlettriplet energy transfer can give rise to singlet oxygen and produce the triplet excited state of 1.Note that the energy gap between the S 1 and T 1 state of 1 (E = 27.2 kcal mol −1 ) is within a reasonable range to populate 1 O 2 (E = 22.4 kcal mol −1 ).Spin is conserved in this process.D-exchange in 2 and 3 were not signicantly quenched by O 2 , suggesting that these reactions do not involve triplet excited-states.The photoredox reaction of 2 giving hydroxylamine 4 was completely quenched by O 2 , indicating that the suggested PCET reaction for 2 involves a triplet excited state.
In summary, the three aminobiphenyl isomers 1-3 are likely undergoing three different photoreactions.The ortho-derivative, 1, undergoes ESIPT.We expect that the meta-derivative, 2, undergoes water-assisted ESPT and a photoredox reaction via PCET.We also expect that the para-derivative, 3, undergoes a water-assisted ESPT reaction.All of these reactions appear to take place in the singlet excited state, except for the photoredox process of 2, which involves a triplet excited state.

Laser ash photolysis
Laser ash photolysis experiments were carried out to probe for triplet excited states and reactive intermediates in the photoreactions of 1 and 2. All spectra and experimental details are included in the ESI (Fig. S35-S42 †).In Ar-purged CH 3 CN and CH 3 CN-H 2 O solutions, a transient signal with a maximum of absorption at z350 nm was observed for 1 and 2 (Fig. 5).The transient decayed within nanoseconds (s z 100-300 ns) and was quenched by O 2 .
We assigned the transient signals to the triplet excited states of 1 and 2 by comparison to the transient absorption spectra of hydroxybiphenyls, 47,48 and triplet excited states of aniline derivatives. 49,501-aQM was not detected, possibly due to a short lifetime and the low efficiency of its formation.It is anticipated to have a short lifetime, similar to aza-QM, detected from 2-(2aminophenyl)naphthalene, 20 or similar QM derivatives formed from phenols. 48The radical cations of 1 and 2 and solvated electron also were not detected, which preclude a photoionisation mechanism.The tentative assignments of the transients observed in the LFP of 1 and 2 is provided in the ESI.†

Computed vertical excitations, charges, and nucleusindependent chemical shis (NICS)
Computed vertical excitation energies for 1-3 at the ADC(2)/cc-pVDZ level are in good agreement with experimental UV-vis data (Tables S10-S12 †).Plots of orbitals involved in the rst three singlet excited states are included in Fig. S43 Computed Natural Population Analysis (NPA) charges at the CPCM(CH 3 CN)-PBE0-D3/def2-TZVPP level for 1-3 in the S 0 and S 1 states are shown in Fig. S52 of the ESI, † and these results explain the regioselectivity of D-exchange in 1-3 (partial charges computed at the MP2/cc-pVDZ and ADC(2)/cc-pVDZ levels are included to Fig. S49-S51 †).For 1 and 3, all carbon atoms of the phenyl ring (besides the linker carbons) become more negatively charged in the S 1 state; this is consistent with an observed deuteration at the phenyl ring (i.e., Ring-1).In the S 1 state of 1, C4 0 of the phenyl ring bears the most negative charge, and the proximity of C2 0 to the NH 2 explains regioselective D-exchange via ESIPT at this position (Fig. S52 †).In the S 1 state of 3, C2 0 and C4 0 of the phenyl ring bear the most negative charge, yet the C2 0 position is closer to the NH 2 group and thus may be more accessible for water-assisted ESPT.For 2, all carbon atoms of the aniline ring (besides C3, which is attached to the NH 2 group) become more negatively charged in the S 1 state; this is consistent with an observed deuteration at the aniline ring (i.e., Ring-2).In the S 1 state of 2, C4 of the aniline ring is the most negatively charged ring carbon atom, which supports the observed regioselective D-exchange at this position.
Computed nucleus-independent chemical shis (NICS) at the CPCM(CH 3 CN)-PBE0-D3/def2-TZVPP level, based on the S 0 and S 1 state minimum geometries of 1-3, are shown in Fig. 6.NICS(1) zz values were computed at the phenyl (Ring-1) and aniline (Ring-2) ring centers.NICS for the S 1 states were computed as triplet states based on geometries optimized in the S 1 state with TD-CPCM(CH 3 CN)-PBE0-D3/def2-TZVPP.In the ground state, all three compounds show large negative NICS(1) zz values for Ring-1 and Ring-2, indicating strong aromatic character.In the S 1 state, computed NICS for 1 and 3 show that Ring-1 is non-aromatic (i.e., NICS (1) zz values close to zero) and Ring-2 is antiaromatic (i.e., positive NICS(1) zz values).
In contrast, computed NICS(1) zz for the S 1 state of 2 reveals a localized antiaromatic character in Ring-2, while Ring-1 retains aromaticity.Differences in the (anti) aromaticity patterns of 1-3 in the S 1 state may have mechanistic consequences.We note that in 1 and 3, excited-state antiaromatic character of the aminobiphenyl compound is delocalized between Ring-1 and Ring-2.But in 2, it is localized in Ring-2.A higher photoreactivity of the aniline fragment in 2 may explain why it undergoes a photo-induced PCET reaction and uniquely deuterates at the aniline ring.In accordance, computed spin densities for the T 1 state of 2 shows more spin in the aniline ring, while those of 1 and 3 display delocalized spins across the two rings (Fig. S53 in the ESI †).

Antiaromaticity relief in the ESIPT reaction of 1
A minimum energy pathway (MEP) was computed at the ADC(2)/cc-pVDZ level to investigate the ESIPT reaction of   electron delocalization across the biphenyl linker (note shortened linker C-C bond length in B, as shown in Fig. 7a).( 3) As D forms, the S 1 and S 0 state surfaces cross and a conical intersection brings the excited-state species back to the ground state.A computed MEP for 1 with H 2 O is included to Fig. S54, Tables S17 and S18 in the ESI, † showing a similar energy prole for ESIPT as that shown in Fig. 7. Cartesian coordinates of points A-D are included to Tables S13-16.† Nucleus-independent chemical shi tensor components of the principal axis perpendicular to the ring plane (NICS(1) zz ) [51][52][53][54] were computed at 1 Å above the centroids of Ring-1 (Fig. 7c, white triangles) and Ring-2 (Fig. 7c, black triangles) for each energy point along the MEP for the ESIPT reaction of 1. 53 At A, Ring-1 (NICS(1) zz = −11.5 ppm) is aromatic and Ring-2 (+99.5 ppm) is strongly antiaromatic (see also spin density for A in Fig. S53 †).From there, Ring-1 remains relatively nonaromatic throughout the ESIPT process (B to D) ranging from −5.5 ppm to +1.3 ppm.Yet, Ring-2 alleviates antiaromaticity as A (+99.5 ppm) relaxes to B (+25.8 ppm), and re-gains aromaticity as C (+11.7 ppm) evolves to D (−21.7 ppm).Changes in the geometries from A to D also are consistent with the effects of antiaromaticity relief and aromaticity gain in Ring-2 (Fig. 7a 23,25,54 ESIPT in 1 is accompanied both by antiaromaticity relief (i.e., from relaxation of the Franck-Condon structure) and by aromaticity gain (i.e., proton transfer from the NH 2 group C2 0 in Ring-2).

Antiaromaticity relief in the PCET reaction of 2
To investigate the excited-state PCET reaction of 2, we approximated the MEP with a linearly interpolated pathway in internal coordinates computed at the ADC(2)/cc-pVDZ level.We expect this to be the rst step in the photoredox reaction of 2 leading to 4. Upon photoexcitation of 2, the aniline moiety undergoes a PCET reaction, and the resulting H atom splits water to produce an OH radical, which then forms a bond with the N atom, producing 4 (Scheme 6).A scheme of the MEP for the PCET reaction of 2, including only the S 0 , S 1 , and T 1 states, is shown in Fig. 8b.Plots of MEPs at the ADC(2) level suggest that the S 1 is populated, and an intersystem crossing from S 1 to T 2 followed by internal conversion to T 1 is involved in the PCET process (see relevant structures (A-D) along the MEP of the PCET reaction in the S 1 and T 1 states are shown in Fig. 8a and a plot of the MEP is shown in Fig. 8b).Single-point energies at the MP2/cc-pVDZ level were computed for the S 0 states along the MEP.
Structure A corresponds to the Franck-Condon point.Structure B is the lowest energy point on the MEP and corresponds to the minimum energy structure of the S 1 state.Structure C is a postcrossing point where the S 1 and T 1 state surface cross.At structure D, the T 1 and S 0 surface cross, and 2 can return to the ground state by intersystem crossing.A few important observations emerge from Fig. 8: (1) the PCET reaction of 2 is a nonadiabatic process.The reaction likely begins in the S 1 state and crosses to a triplet state (a more elaborate energy plot including  and 20, Ring-1 and Ring-2 undergo a drastic drop in NICS(1) zz magnitude, corroborating a PCET process that restores aromaticity at these points in rings 1 and 2, respectively.
Antiaromaticity relief in the water-assisted ESPT reaction of 3 Irradiating 3 in CH 3 CN-D 2 O showed selective D-exchange at the C2 0 position (Scheme 5).To investigate the origin of this Dexchange, a MEP of 3 in a micro-solvated environment of ve water molecules was computed on the S 1 with ADC(2)/cc-pVDZ (Fig. 9a and b).Note that it is only a model, and in a real system more H-bonded H 2 O molecules are involved.Furthermore, such a relay processes are entropically unfavoured, and therefore, the process takes place with a low quantum yield, as observed in the experiment.properly described by the approach applied here.Optimized Cartesian coordinates for structures A-D are included to Tables S23-S26.†

Conclusions
Excited-state antiaromaticity relief can play important roles in many light-driven proton and electron transfer reactions.Here, we investigate and contrast the photoprotonation mechanism of three aminobiphenyl isomers.In all three cases, a proton formally migrates from an NH 2 group to an aromatic ring carbon atom.All three isomers, ortho-, meta-and para-, can undergo excited-state proton transfer involving water, as documented by deuterium exchange experiments.In addition, the meta-isomer can undergo a photoredox reaction, involving proton-coupled electron transfer and water-splitting.We note that, in the meta-isomer, excited-state antiaromatic character is localized in the aniline moiety while the benzene moiety remains weakly aromatic (Fig. 6), and argue that strong antiaromaticity in the aniline ring may be responsible for driving the observed photoredox reaction.

Experimental and computational methods
Preparative irradiation experiment in the presence of D 2 O 2-Aminobiphenyl (1) (100 mg, 0. 62 mmol) was dissolved in CH 3 CN-D 2 O (9 : 1, v/v, 100 mL) and transferred to a quartz Erlenmeyer ask with a rubber septum.The solution was purged with a stream of N 2 for 30 min and irradiated in a Luzchem reactor equipped with 8 lamps (1 lamp z 8 W) with the maximum output at 300 nm over 4 days.The ask equipped with a stirring bar was centered in the middle of the reactor, separated 15 cm from the lamps on each side.Information of the lamps irradiance can be found from the Luzchem supplier. 55Aer the irradiation, the solution was transferred to a separation funnel and water (100 mL) was added.An extraction with CH 2 Cl 2 was carried out (3 × 75 mL).The extracts were dried over anhydrous MgSO 4 , ltered and the solvent was removed on a rotary evaporator.The residue was chromatographed on a column of silica gel by using CH 2 Cl 2 as an eluent to remove high-weigh material formed in the photolysis.The residue (80 mg, 80%) was analysed by 1 H NMR and MS.Since the 1 H NMR indicated exchange of only one H by D, the sample was dissolved again in CH 3 CN-D 2 O (100 mL, 9 : 1), purged with N 2 and irradiated in the Luzchem reactor over 4 days.Aer the same workup and chromatographic purication, 45 mg (45%) was obtained that was analyzed by 1 H NMR and MS.Details of all photolyses conducted under different conditions for molecules 1-3 can be found in the ESI.† 2-Amino-2,2 0 -bisdeuteriobiphenyl (1-2D) 1 H NMR (600 MHz, CD 3 OD): d/ppm 7.43 (d, 2H, J = 7.8 Hz, H-3 0 ), 7.33 (t, 1H, J = 7.8 Hz, H-4 0 ), 7.10 (dt, J = 1.5 Hz, J = 7.9 Hz, H-4), 7.04 (dd, 1H, J = 1.5 Hz, J = 7.5 Hz, H-6), 6.83 (dd, 1H, J = 0.9 Hz, J = 7.9 Hz, H-3), 6.77 (dt, J = 1.5 Hz, J = 7.5 Hz, H-5). 13  (23), 172 (100), 173 (13).

Computational methods
Density functional theory (DFT) calculations were carried out in Gaussian 16 (revision C.01). 568][59][60][61] For each point of interest NICS(1) zz points were manually created using the Molecule soware program.Bq atoms were placed 1 Å above each ring.Time-dependent density functional theory (TD-DFT) calculations were also computed in Gaussian 16 using the Tamm-Dancoff approximation.All Cartesian coordinates of structures optimized with DFT or TD-DFT are included in the ESI, Tables S28-S36.† Ground state geometries were optimized at MP2/aug-cc-pVDZ 62 employing the resolution-of-identity (RI) approximation. 63Vertical excitation energies and oscillator strengths were computed using the algebraic diagrammatic construction to second order ADC(2) method [64][65][66] as implemented in Turbomole 7.6. 67Calculations were performed with the cc-pVDZ and aug-cc-pVDZ basis sets both in the gas phase and in solution using the implicit solvation model COSMO 68,69 with default parameters for acetonitrile (ACN).Excited states have been characterized in terms of natural transition orbitals 70,71 computed by retaining only singly excited coefficients of the ADC(2) wave functions. 72eaction pathways in the S 1 state have been optimized using the double-ended reaction path optimization scheme woel-ing 73 of Turbomole.All reaction pathways start at the Franck-Condon geometry and end at the S 1 /S 0 conical intersection (CI).CI geometries were optimized using the sequential penalty constrained optimization method of Levine et al. 74 with default initial values of a = 0.025 Hartree and s = 3.5.

Fig. 3
Fig. 3 Absorption spectra of 3 (c 0 = 3.67 × 10 −5 M) in CH 3 CN-H 2 O (1 : 4) at different pH values (top), and dependence of the absorption at 274 nm on the pH (bottom).The black dots are experimental values and the red line corresponds to a model involving a one-step protonation equilibrium.
), indicating that an NH 2 group is necessary for D-exchange.To investigate whether photoinduced electrophilic attack of D 3 O + to the carbon atom site contributes to D-exchange, we irradiated 1 in CH 3 CN-D 2 O in the presence of deuterated sodium phosphate buffer at different pD values (Table S6 and Fig S23 in the ESI †).D-exchange is low at pD < 3 and 1-D is observed at pD values 3-8, suggesting that an electrophilic D 3 O + attack mechanism is unlikely (although it cannot be completely ruled out that some D-exchange may operate at very low pD values with much lower efficiency).These results obtained at pH values higher than the pK a also indicate that ESIPT involves proton transfer from an NH 2 group (and not NH 3 + ).Electrophilic attack of D 3 O + to the deuterated carbon atom site (Scheme 3) is only competitive in very acidic solutions (i.e., when the concentration of D 3 O + is high enough so that bimolecular reaction can compete with fast deactivation of 1 from S 1 : s = 2.5 or 3.6 ns; Tables S2 and S3 †).Compounds 2 and 3 cannot undergo ESIPT, since the acidic NH 2 group is too far away from the neighboring phenyl ring.But water (D 2 O)-mediated excited-state proton transfer may give rise to D-exchange at distal sites.Irradiating 2 in CH 3 CN-D 2 O gave two products, a compound 2-D resulting from regioselective D-exchange at the C4 position of the aniline ring, and a photoredox product 4 (Scheme 4).The photoproduct 4 could not be separated from the starting material.Attempts of HPLC separation provided their mixture or the starting compound 2. The structure of photoproduct 4 was therefore assigned based on NMR characterization of the mixture with 2 and MS analysis (see Fig. S24-S30 in the ESI †).The 15 N-1 H HMBC NMR correlation of the mixture clearly points to two different N-atoms (Fig. S29 †).The structure for 4

Fig. 4
Fig. 4 Fluorescence spectra (l ex = 260 nm) of 3 (c 0 = 4.51 × 10 −6 M) in CH 3 CN-H 2 O (1 : 4) at different pH values (top), and dependence of the fluorescence intensity at 382 nm on the pH (bottom).The black dots are experimental values and the red line corresponds to a model involving a one-step protonation equilibrium.
was also conrmed by chemical synthesis.We performed a reduction of 3-nitrobiphenyl with Raney nickel and obtained the same mixture as the photolysis of 2 (see Fig.S31in the ESI †).2-D only is observed in acidic solution and may occur by electrophilic attack of D 3 O + at the C4 position via water-assisted Dexchange.Computed partial charges for 2 support the observed regioselectivity, showing that in the S 1 state, C4 is the most negatively charged ring carbon atom in the aniline ring (see Fig.6, and discussion below).
in CH 3 CN-D 2 O led to regioselective D-exchange at the C2 0 position of the phenyl ring, giving 3-D (Scheme 5).Dexchange occurs only at very low pD values (see also Fig. S32-S34 in the ESI †) with a quantum yield of <10 −4 .The reaction takes place only in acidic solution and possibly proceeds through a water-assisted ESPT reaction involving D 3 O + .Computed partial charges for 3 support the observed regioselectivity, showing that in the S 1 state, the C2 0 and C4 0 positions are most negatively charged ring carbon atoms in the phenyl ring (Fig. S52 †).D-exchange at the C2 0 position may be explained by a closer proximity to the NH 2 group, making it a more accessible site via water-assisted ESPT.Note also that there are two C2 0 positions but only one C4 0 position, which may explain the observed protonation at C2 0 .Do these photoreactions happen in the singlet or triplet excited state?To investigate whether the ESPT of 1-3 and the photoredox of 2 occur from the singlet or triplet excited state, the experiments were repeated in CH 3 CN-D 2 O at pD = 2 and purged by N 2 or O 2 .O 2 strongly quenches triplet excited-states, and thus reactions that occur in the triplet state would be signicantly quenched.For 1, we observed some quenching by O 2 (Table -S45 of the ESI, † while the corresponding electron density difference plots are included in Fig. S46-S48.† All three compounds have pp* character in the S 1 state and show decreased electron density at the N atom and increased electron density in the phenyl ring in the S 1 state.These ndings are consistent with a red-shied UVvis spectra and quenched uorescence at lower pH due to protonation of the N atom.

1 .
Relevant structures (A-D) along the MEP are shown in Fig. 7a.Plots of the S 1 state MEP and S 0 state single-point energies at MP2/cc-pVDZ are shown in Fig. 7b.Structure A is the Franck-Condon point.Structure B is the lowest energy point on the MEP.Structure C is a transition state-like structure (i.e., proton is transferring-not necessarily a rst-order saddle point) and the highest energy point aer B. Structure D is a post-transition structure and the last point on the MEP considered.A few important observations emerge from Fig. 7: (1) the ESIPT reaction of 1 is a near-barrierless process in the S 1 state.(2) As 1 relaxes from the Franck-Condon point, there is signicant

Fig. 6
Fig. 6 Computed NICS(1) zz values for the S 0 and S 1 state minimum geometries of 1-3.NICS(1) zz for the S 1 state minimum geometries were computed as triplet states.
).From A to B, the linker C-C bond shortens (from 1.486 Å to 1.435 Å) and the dihedral angle of the biphenyl linker tends towards planarization (4 = −53.1°to 4 = −38.8°,Fig. 7a), indicating increased quinoidal character in both rings in B. From C to D, the incipient C-H bond forms, and the linker C-C bond lengthens (from 1.432 Å to 1.483 Å) as the dihedral angle of the biphenyl linker twists even more towards planarity (from 4 = −34.1°to 4 = −29.2°),resulting in rearomatization of Ring-2 in D. NICS(1) zz values for all S 1 state structures along the MEP were computed as triplet states at the CPCM(CH3CN)-PBE0-D3/ def2-TZVPP level.In contrast to previous examples of ESPT reactions explored by us and by others,

Fig. 7
Fig. 7 (a) Selected structures on the S 1 state MEP for ESIPT in 1: the Franck-Condon point (A), the lowest energy point on the S 1 state MEP (B), the highest energy point after (B) and (C), and a post-transition structure (D).d is the distance between the migrating H and the proton accepting carbon atom.F is the CCCC dihedral angle across the central linker CC bond.(b) A plot of the S 1 state MEP computed at ADC(2)/cc-pVDZ for the ESIPT reaction of 1. Single-point S 0 state energies were computed for each structure at MP2/cc-PVDZ.(c) NICS(1) zz values computed at CPCM(CH 3 CN)-PBE0-D3/def2-TZVPP for structures along the MEP (cf.orange curve in Fig. 7b).

Fig. 8
Fig. 8 (a) Geometric features of the Franck-Condon state (A), the minimum of the S 1 state the lowest energy point on the S 1 state MEP (B), the highest energy point on the MEP (C), and a post-surface crossing structure (D).(b) Computed MEP plots at ADC(2)/cc-pVDZ for the PCET reaction of 2; the T 1 and S 0 state surfaces are computed single-point energies based on fully optimized S 1 state geometries.d is the distance between the migrating H and the O atom on water.F is the CCCC dihedral angle across the central linker CC bond.(c) NICS(1) zz values computed at CPCM(CH 3 CN)-PBE0-D3/Def2-TZVPP for structures along the MEP for the PCET reaction of 2. (d) Dominant natural transition orbital (NTO) pairs for structure A-D.% values are the contributions of an NTO pair to the excited-state wavefunction (S 1 for A and B, T 1 for C and D).
computed energies in the S 2 , T 1 , and T 2 states is included in the ESI Fig.S55 †).(2) As 2 relaxes from the Franck-Condon point, there is signicant electron delocalization across the biphenyl linker (note shortened linker C-C bond length in B). (3) Computed Natural Transition Orbitals (NTO's) for structures A and B show pp* character, while structures C and D show ps* state character, suggesting that the PCET reaction is a nonadiabatic process involving electron transfer from the excitedstate p-system followed by N-H bond breaking.(4) At structure D, the T 1 and S 0 state surfaces cross and intersystem crossing brings the excited-state species back to the ground state.Changes in the geometry of 2 along the MEP for an excitedstate PCET reaction show the effects of antiaromaticity relief.The C-C bond linker in structure A (1.483 Å) shortens as the structure relaxes from the Franck-Condon region to B (1.436 Å), and the dihedral angle of the two phenyl rings tends towards planarization (from 4 = 43.6°to 4 = 12.5°).Planarization allows the triplet spin to delocalize which alleviates excited-state antiaromaticity in Ring-2 (cf.Fig. 8c).Already in B, the forming O/H bond shortens from 2.091 Å (in A) to 1.877 Å. Upon electron transfer from the aniline ring to the amine H atom, the geometry at C suggests the O-H bond is almost fully formed (d = 1.121Å) and the C-C bond linker length (1.474 Å) and dihedral angle (35.2°) is returned to a Franck-Condon-like geometry.Computed T 1 NICS(1) zz values for Ring-1 and Ring-2 along the MEP show that Ring-1 remains slightly aromatic during the PCET process (hovering around −12 ppm), and Ring-2 experiences a gradual decrease in antiaromaticity.At points 19 Fig. 9 (a) Geometric features of the Franck-Condon state (A), S 2 -S 1 near-crossing on MEP (B), ESPT to Ring-1 (C), and S 1 -S 0 conical intersection (D).(b) Computed MEP at the ADC(2)/cc-pVDZ for the water-assisted ESPT reaction of 3 in the S 1 state, and singlet-point energies in the S 0 and S 2 states.(c) Geometry for C showing proton transfer to the 2 0 position in the benzene ring.

Table 1
Measured pK a and pK * a values of 1-3 by UV-vis and fluorescence titration a Compound pK a (UV-vis) pK a (uorescence) pK a All titration experiments were conducted in CH 3 CN-H 2 O (1 : 4) at 25 °C.