Barrier-Lowering Effects of Baird Antiaromaticity in Photoinduced Proton-Coupled Electron Transfer (PCET) Reactions

Abstract

Many popular organic chromophores that catalyze photoinduced proton-coupled electron transfer (PCET) reactions are aromatic in the ground state but become excited-state antiaromatic in the lowest ππ* state. We show that excited-state antiaromaticity makes electron transfer easier. Two representative photoinduced electron transfer processes are investigated: (1) the photolysis of phenol and (2) solar water splitting of a pyridine–water complex. In the selected reactions, the directions of electron transfer are opposite, but the net result is proton transfer following the direction of electron transfer. Nucleus-independent chemical shifts (NICS), ionization energies, electron affinities, and PCET energy profiles of selected [4n] and [4n + 2] π-systems are presented, and important mechanistic implications are discussed.

Photoinduced proton-coupled electron transfer (PCET) reactions are the critical steps to a myriad of energy conversion processes in organic and bio-organic photochemistry.^1,2 In the case where an electron moves first, these reactions also can be called electron-driven proton transfer—an aromatic chromophore absorbs light, triggering the migration of an electron, and a proton follows. These reactions typically have low barriers, and here, we explain the origin of such low barriers, making the connection between facile electron transfer and the concepts of ground- and excited-state (anti)aromaticity. We note that the most common organic chromophores (e.g., phenol,³ pyridine,⁴ and indoles^5,6) are Hückel [4n + 2] aromatic in the ground state,⁷ but become Baird [4n + 2] antiaromatic in the lowest ππ* state.⁸ Excited-state antiaromaticity makes charge separation easier, and this relationship has important mechanistic consequences for the photochemical reactions of aromatic compounds.

A representative example is the photolytic O–H bond fission of phenol. Flash photolysis and transient absorption studies of phenol in the vapor phase and in aqueous solution indicate the formation of neutral phenoxy radicals.^9,10 Because photolysis produces H^• and PhO^•, it is tempting for the trained organic chemist to illustrate bond dissociation by two sets of single-headed arrows showing homolytic cleavage of the OH σ-bond (Figure 1a, top). But the reaction cannot happen this way. The >95 kcal/mol homolytic O–H bond dissociation energy is much too high. According to ab initio studies and kinetic experiments, UV irradiation first generates an optically active ππ* state, then O–H σ-bond fission resolves in a dark πσ* state, reached by nonadiabatic interaction of the two surfaces.^5,11–13 Photolytic O–H and N–H bond fission processes for other aromatic chromophores proceed through similar dissociative charge transfer.^3,6,14 Crossing of the ππ* and πσ* states implies an alternative mechanism for OH photolysis that has never been recognized explicitly. We suggest that photolysis of phenol occurs through heterolytic cleavage of the polar O–H bond. Upon photoexcitation, an electron is transferred from the π-ring to H, and the polar O–H bond breaks heterolytically (Figure 1a, bottom).

Figure 1.

(a) Two different mechanistic representations for the photolysis of phenol. (b) Potential energy profiles along the O–H stretching coordinate, in the S₀ (blue squares), ¹ππ* (orange rhomboids), and ¹πσ*(yellow circles) states of phenol, and (c) NICS(1)_zz values computed along the photoinduced PCET pathway (see markers with black border in 1b).

Figure 1b shows the potential energy profiles of the S₀, ¹ππ* (locally excited, LE), and ¹πσ* (charge transfer, CT) states of phenol along the O–H stretching coordinate (at the S₀− and LE-state minima, and at points between O–H = 0.9 to 2.0 Å, at 0.1 Å intervals), computed at CASPT//CASSCF/6–311+G(d,p) (see full methods in the Supporting Information, SI). Homolytic cleavage of the O–H bond requires >95 kcal/mol in both the S₀ and the ¹ππ* states; see parallel displaced S₀ and ¹ππ* curves in Figure 1b. Heterolytic cleavage of the polar O–H bond requires much less energy and occurs readily upon crossing of the ¹ππ* and ¹πσ* states. Estimated barriers to electron transfer (ΔE_ET) were derived by the energy at the crossing of the interpolated ππ* and πσ* curves minus the energy of the ππ*-state minimum. Note that the intersecting point of the yellow and orange curves in Figure 1 is not a minimal energy crossing point (MECP) on the potential energy surface, since the geometries for each state were optimized separately. Past the conical intersection, an electron transfers from the photoexcited [4n + 2] π-ring (“antiaromatic”) to the phenolic H through a low estimated barrier (ΔE_ET ≈ 20 kcal/mol, at O–H = 1.21 Å), and the ¹πσ* state is stabilized as the proton follows the electron, producing H^• and PhO^•. A heterolytic bond O–H bond dissociation pathway is further supported by charge analysis data included in Figure S9. We note that closely related reactions such as the photo-Fries rearrangement of phenyl esters also are commonly illustrated in Google searches and textbooks as homolytic O–R bond cleavage processes. We suggest that the arrow pushing mechanisms of these reactions also ought to be redrawn.^15,16

But why should the barrier to electron transfer be so low? Note in Figure 1b, that a low ΔE_ET barrier is in part due to a high ¹ππ*-state energy. This may be explained by Baird antiaromaticity of the [4n + 2] π-ring. Phenol is Hückel [4n + 2] aromatic in the S₀ state but becomes Baird [4n + 2] antiaromatic in the lowest ¹ππ* state,^17–20 and reversal of the Hückel rule drives electron transfer. Computed nucleus-independent chemical shifts, NICS(1)_zz, at geometries along the photoinduced PCET pathway (Figure 1c) confirm that phenol is antiaromatic in the ¹ππ* state (large positive NICS(1)_zz values) and show that paratropicity of the π-ring drops immediately past the ¹ππ*- and ¹πσ*-state intersection (see also Figure S5 for T₁-state results). NICS(1)_zz data were computed at 1 Å above the ring centers including only the out-of-plane (zz) shielding tensor with an inverted sign.^21–23 NICS(1)_zz data were computed at CASSCF/6–311+G(d,p) for 1, 1′, phenol, and the (Py)–water complex and at CASSCF/6–31G(d,p) for 2 and 2′. Negative NICS(1)_zz values indicate aromaticity, and positive NICS(1)_zz values indicate antiaromaticity.

Of course, photoexcitation promotes the energy of any molecule, increasing both the ability to donate and to accept an electron. But this effect is especially pronounced for compounds with antiaromatic character in the lowest ππ* states. We show this connection in Figure 2, by comparing the computed ionization energies (IE) and electron affinities (EA) of a set of [4n + 2] and [4n] annulenes to their nonaromatic isomers, in the S₀ and T₁ states.²⁴ Zhu and Schleyer have shown that the isomerization energies of the T₁ states of such pairs provide reliable energetic measures for triplet state (anti)aromaticity²⁵ Toluene is Hückel [4n + 2] aromatic in the S₀ state but Baird antiaromatic in the T₁ state and shows significantly reduced IE (ΔIE = 203.0–122.9 = 80.1 kcal/mol) and EA (ΔEA = 48.6 + 45.9 = 94.5 kcal/mol) values in the excited state (Figure 2a, left), compared to those of the nonaromatic methylenecyclohexadiene (ΔIE = 36.9 kcal/mol, ΔEA = 39.5 kcal/mol) (Figure 2a, right). Figure 2b shows the opposite example. Planar methyl-cyclooctatetraene (COT) is Hückel [4n] antiaromatic in the S₀ state but Baird aromatic in the T₁ state and exhibits only modestly lowered IE (ΔIE = 8.0 kcal/mol) and EA (ΔEA = 8.5 kcal/mol) values in the excited state (Figure 2b, left), compared to those of the nonaromatic isomer (ΔIE = 36.1 kcal/mol, ΔEA = 36.3 kcal/mol, Figure 2b, right). A significant drop in IE and EA values, because of excited-state antiaromaticity, makes charge transfer easier.

Figure 2.

Computed ionization energies (IE), electron affinities (EA), and NICS(1)_zz values for (a) toluene vs methylenecyclohexadiene and (b) planar methyl-COT vs planar methylenecyclooctatriene.

Figure 3 compares the T₁-state PCET profiles of 4-methylphenol (1, [4n + 2]) and 1-hydroxy-5-methyl-COT (2, [4n]) with those of their nonaromatic isomers (1′ and 2′). Potential energy profiles along the O–H stretching coordinate were computed at 0.1 Å intervals in the ³ππ* and ³πσ* states (see Figure S4 for the S₁-state results). As the O–H bond stretches, the ³ππ* and ³πσ* curves intersect, and crossing of the two functions marks the point at which an electron transfers from the π-ring to the phenolic H. 1 (Baird antiaromatic) displays a high T₁-state energy (80.7 kcal/mol), and the ³ππ* to ³πσ* intersection occurs “early” through a relatively low barrier (ΔE_ET ≈ 27 kcal/mol, crossing at O–H = 1.26 Å) (Figure 3a, left). In contrast, 1′ (nonaromatic) exhibits a lower T₁-state energy (38.3 kcal/mol), and the conical intersection occurs “late” through a nearly doubled barrier (ΔE_ET ≈ 57 kcal/mol, crossing at O–H = 1.47 Å) (Figure 3b, left). Computed NICS(1)_zz values for 1, at geometries along the photoinduced PCET pathway, show an abrupt drop in paratropicity past the ³ππ* to ³πσ* intersection (Figure 3a, right, note the sign change of NICS(1)_zz values from positive to negative), while those of 1′ remain constant for both the ³ππ* and ³πσ* states, having values close to zero (Figure 3b, right). The high T₁-state energy of 1 is a result of Baird antiaromaticity, and facile electron transfer is the escape from it.

Figure 3.

Potential energy profiles along the O–H stretching coordinate in the ³ππ* (orange rhomboids) and ³πσ* (yellow circles) states for (a) 1, (b) 1′, (c) 2, and (d) 2′. NICS(1)_zz values were computed along the photoinduced PCET pathway as indicated by points with a black marker border on the potential energy curve; the vertical blue dotted lines indicate crossing of the ³ππ* and ³πσ* states.

Potential energy profiles for the T₁ states of 2 vs 2′ show the opposite. 2 (Baird aromatic) exhibits a low T₁-state energy (5.9 kcal/mol), and crossing from the ³ππ* state to the ³πσ* state involves a high barrier (ΔE_ET ≈ 70 kcal/mol, at O–H = 1.56 Å) due to a largely stabilized T₁ state (Figure 3c, left). In comparison, 2′ (nonaromatic) has a higher T₁-state energy (31.5 kcal/mol) and a lower barrier to electron transfer (ΔE_ET ≈ 56 kcal/mol, at O–H = 1.47 Å) (Figure 3d, left, note the similar T₁ energies and ΔE_ET values compared to 1′, cf. Figure 2b). Computed NICS(1)_zz values at geometries along the photoinduced PCET pathway of 2 increases in paratropicity past the ³ππ* to ³πσ* intersection (Figure 3c, right, note sign change of NICS(1)_zz values), while those of 2′ remain relatively constant for both the ³ππ* and ³πσ* states (Figure 3d, right). Computed gauge-including magnetically induced current (GIMIC)²⁶ plots, ¹H chemical shifts, and the harmonic oscillator model of aromaticity (rHOMA)²⁷ agree with NICS; see the data in the SI.

We further considered the phototriggered water-splitting reaction of a model pyridine (Py)–water complex. In this example, an electron transfers from water to the aromatic core. Photodeactivation through PCET was suggested as a reason for the absence of fluorescence of pyridine in water.^4,28–31 Water splitting through this route requires two photons. In the first step, pyridine absorbs light (ππ*, LE state), and an electron moves from water to the photoexcited π-ring followed by proton transfer (ππ* or nπ*, CT state), generating a PyH^• and OH^• radical pair. The triplet CT state is degenerate with the singlet state and can be reached through efficient intersystem crossing. Photoreactions on the excited singlet and triplet state surfaces were shown to be quite similar. In the next step, a second photon detaches H^• from PyH^• and regenerates the catalytic pyridine. Without the chromophore, homolytic bond cleavage of the water O–H bond (in the gas phase) is is 5.1 eV (117.6 kcal/mol).³² Here, we examine the first step of the reaction, showing that in the lowest ππ* state the catalytic pyridine ring is Baird antiaromatic and that moving an electron from water to the pyridine ring alleviates excited-state antiaromaticity.

Figure 4b shows the computed energy profiles of the (Py)–water complex at 0.1 Å intervals along the N–H bond forming coordinate (i.e., H moving from water to the pyridinyl N) in the S₀ state, LE state (¹ππ*), and CT state (¹nπ*, i.e., electron transfer from the hybridized lone pair of O). It was shown that crossing from the ¹ππ* LE state to the ¹ππ* CT state (i.e., electron transfer from the unhybridized lone pair of O) produced a similar energetic profile.⁴ As the N–H bond forms, the ¹ππ* and ¹nπ* curves intersect, and crossing of the two functions marks the point at which an electron transfers from water to the π-ring, first forming a zwitterionic complex, followed by proton transfer leading to the homolytic products PyH^• and OH^• (Figure 4a, bottom). Without electron transfer from water to the pyridinyl ring, the water O–H bond breaks heterolytically to give the products, PyH⁺ and OH⁻ (Figure 4a, top). Computed NICS(1)_zz values at geometries along the reaction pathway show a drop in paratropicity past the LE- and CT-state intersection (Figure 4c, see Figure S6 for T₁ results). A heterolytic bond O–H bond dissociation pathway leading to homolytic products is further supported by charge analyses data (Figure S10).

Figure 4.

(a) Two different mechanistic representations for O–H bond breaking in the water-splitting reaction of (Py)–water. (b) Potential energy profiles along the N–H bond forming coordinate, in the S₀ (blue squares), ¹ππ* (orange rhomboids), and ¹nπ* (yellow circles) states for the (Py)–water complex, and (c) NICS(1)_zz values computed along the photoinduced PCET pathway.

Traditionally, the driving force for photoinduced electron transfer has been explained by the Rehm–Weller model,³³ where Gibbs free energy for charge separation is estimated by oxidation/reduction potentials of the electron donor/acceptor. Here, we show that a more complete picture emerges when the effects of excited-state (anti)aromaticity^34–43 are considered. Even though all photoexcited molecules can be thought to be “unstable,” Baird antiaromaticity is a useful effect for tuning the energy levels of ππ* states and the many reactions that proceed through these bright states.