Spectroscopic features of the low-lying singlet states of some N-alkyl retinylnitrone model systems and their involvement in oxaziridine formation

Abstract

The photo-excitation process and non-radiative decay channels of the model compounds of some N-alkyl retinylnitrones are studied at the CASSCF/6-31G*, CASMP2/6-31G* and PM3/CI level of theories. A relaxed planar geometry of the first excited singlet state is reached after the initial photo-excitation, which is followed by non-radiative decay processes through conical intersection (S₀/S₁) channels. Their first-excited singlet states (S₁) have mixed biradical–ionic nature, and are found to be dominated by configurations arising from HOMO² → LUMO², HOMO − 1 → LUMO and HOMO → LUMO excitations. Conical intersection geometries originating from the one-bond-flip and Hula-twist motions in the central part of these molecules are found at higher energies in comparison to their terminally twisted counter parts. In the N-methyl nitrone system, the lowest-energy intersection point arises due to a twist in the terminal part with an out-of-plane CNO-kink (R_C–O = 2.12 Å, R_N–O = 1.38 Å) or oxygen-bridge structure. Following the directions of its gradient difference vectors, the probable oxaziridine ground-state geometry (R_C–O = 1.38 Å, R_N–O = 1.44 Å, <OCN = 62.3^°, <ONC = 57.6^°) has been located as a saddle point, which is the only experimentally reported photoproduct of N-methyl retinylnitrone compound under room light. The radiative transition studies on the allowed S₀ → S₁ transitions at the ground state equilibrium geometry have given transition moment values between 4.5 and 5.0 Debye.

---

1. Introduction

The photo-irradiation of nitrones and subsequent formation of oxaziridines along with other photochemical products has been the subject of interest over the last few decades. Analysis of the nitrone–oxaziridine conversion^1,2 and cis–trans isomerization^3,4 of nitrones was carried out long back. The study of the photoirradation products of several nitrones reported by Splitter et al.^1,2 is still considered to be one of the major experimental works done on these systems. The N-alkyl-α-arylnitrones were reported to give the more stable oxaziridines on photo-irradiation, in comparison to the N,α-diarylnitrones. The stability of the oxaziridine ring and the nature of its cleavage were found to depend primarily on the types of substituents present on nitrogen and carbon. In fact, stable and experimentally isolated oxaziridines were found to contain alkyl groups on either or both nitrogen and carbon atoms; on the other hand, the presence of aryl groups on both these atoms decreases their stabilities significantly, and their existence can be seen only in solution. The photochemical and thermal isomerization studies done^3,4 on trans and cis-α-cyano-α-phenyl-N-phenylntirones are also worthy of a special mention in the isomerization studies of nitrones. This cis–trans isomerization reaction was found to occur thermally or in the presence of photosensitizers, such as uranine, eosine, iodine etc. The latter process was reported to involve the triplet excited states, while the normal photo-excitation of the ground state nitrone species was found to give oxaziridine and some rearrangement products through the involvement of only singlet states. Photo-chemical studies on α-(2-naphthyl)-N-methylnitrone and its N-tert-butyl derivative were reported⁵ to give stable oxaziridines, while its N-(p-tolyl) derivative was found to be highly unstable. The stereospecificity of the nitrone–oxaziridine conversion was initially doubted due to the formation of both cis and trans oxaziridines from the trans isomer of methyl nitrones,⁶ however, it was soon established by Splitter and Calvin² that this is due to the thermal ring opening of the oxaziridine in two possible conrotatory modes which results in a mixture of cis and trans isomers of the nitrone.

Structural and spectral aspects of similar nitrones have been reported by several other groups, too;^7–18 this includes quantum mechanical^14–17 and molecular dynamics¹⁸ studies, as well. However, the number of studies reported on the conjugated long-chain nitrones is less in comparison to the studies done on the above-mentioned types of nitrones. Almost thirty years back, some important conjugated, long-chain nitrones were synthesized and analyzed by Balogh Nair et al.¹⁹ A new class of retinoids with a polar nitrone group was their subject of study, and these nitrones were found to be chemopreventive. It should be mentioned here that retinoids at that time were known to be effective in the chemoprevention of cancer. Their role in the reduction of neoplasia caused due to the deficiency of vitamin A and participation in controlling epithelial cell differentiation were already established. The retinyl nitrones (Fig. 1a) studied by Balogh-Nair et al.¹⁹ were reported to be effective in reversing keratinisation in hamster tracheal organ cultures. Their ability to control epithelial cell differentiation, in vitro, was observed. These nitrones were prepared by the treatment of all-trans retinal with the appropriate N-alkylhydroxylamine. The N-methylnitrone compound was found to be of highest activity, almost matching the activity of all-trans retinoic acid. This compound is stable in the dark at room temperature. Exposure to room light in methanol solution converts it slowly to the corresponding oxaziridine, which has a tendency to be partially converted back to the nitrone in the dark. In fact, almost all N-substituted nitrones studied by them were found to be light sensitive and more or less of similar nature. It is quite obvious that their low-lying electronic states are responsible for these light-sensitized properties, however, spectroscopic features of the excited states of such conjugated nitrone systems are yet to be explored, so far.

Fig. 1 (a) Structure of N-alkyl retinyl nitrone. (b) Studied model systems of methyl and isopropyl-substituted retinyl nitrones.

This present work aims to study the mechanism of the photo-excitation process of some conjugated long-chain nitrone systems by revealing the nature of their excited electronic states. The chosen conjugated nitrone systems (Fig. 1b) have methyl (I) and isopropyl (II) groups on the nitrogen, and may act as model compounds of their corresponding N-alkyl retinylnitrone systems. Unlike many other nitrones, these compounds and their oxaziridines were reported to be of better stability. The N-methyl retinyl nitrone compound was found to give¹⁹ isolable oxaziridine. A major goal of this present study is to explore the nitrone–oxaziridine conversion mechanism of the above-mentioned model compounds. However, the expected outcome of this work may have far reaching consequences in terms of revealing the actual photo-excitation mechanism of several other nitrones, as well. The first singlet excited state (S₁) of nitrone was proposed to have more biradical nature¹⁴ in comparison to the ionic form, and the oxaziridine was predicted to be produced from this S₁ state.^1–4 The complicated nature of this excited state due to mixed ionic–biradical contributions has made it very difficult to analyze them, and probably the reason behind the lack of any established mechanism for the conversion of this S₁ state to the oxaziridine. This work has attempted to put forward a proper theoretical background of this process through high-level investigations on the important points on the excited state surface. Computational studies on the low-lying electronic states of the above-mentioned nitrone molecules are based on the CASSCF, CASMP2 and PM3/CI level of theories. A separate level of study on the radiative transition properties have been carried out through the calculations of transition moment, oscillator strength and Einstein's coefficient (A₂₁) values for the low-lying vertical transitions at the ground state equilibrium geometries. These various levels of quantum mechanical studies are expected to give a significant amount of information as the outcome of this work, which is likely to establish a certain mechanism for the photo-excitation process of the conjugated long-chain nitrones.

2. Computational methods

The complete active space self-consistent field (CASSCF) method with 6-31G* basis sets in Gaussian 09 program²⁰ and configuration interaction calculations with the semiempirical-based PM3 Hamiltonian in MOPAC have been used for the ground state and the first excited singlet state analysis. The CASSCF method has been employed for locating the minimum energy geometries, transition states and conical intersection points on the potential energy surfaces (PES). Both (6,6) and (4,4) varieties of CASSCF/6-31G* methods^21–26 have been tested for the ground and excited state equilibrium geometry optimizations; however, for the conical intersection and transition state optimizations, a (4,4) active space has been chosen. Justification of this choice is given in the next section. The dynamic correlation effect of the MP2 level has been included through single point calculations on top of the CASSCF/6-31G* optimized points. For locating the transition states both QSTN-based QST2 methodology^27,28 and the normal TS technique²⁹ based on the Berny-algorithm have been employed. The minimum energy path from the transition state has been followed by the intrinsic reaction coordinate (IRC) method.^30–32 In the semiempirical CI treatment, the NDDO-based PM3^33,34 Hamiltonian has been used in the multielectron configuration interaction technique (MECI). Considering the fact that the first excited singlet state (S₁) might have both ionic and biradical contributions, the second CI root has been analyzed using both with and without the biradical option as the keyword in the input file of PM3/CI calculations. In addition to these studies, the GUGA-based CISD technique has also been used for some important calculations through the GAMESS^35–39 suite of programs. Radiative transition^40–42 calculations have been carried out between the two CI wavefunctions at the ground state equilibrium geometry, based on this GUGA CI code. A rough estimation of the radiative lifetime of the first excited singlet state has been obtained from the Einstein's coefficient, A₂₁. It is well-known that Einstein's coefficient (A₂₁) is defined in terms of the total rate of spontaneous emission (W₂₁) from an upper level (2) to a lower level (1) and the number of atoms (n₂) in the upper level.

W₂₁ = A₂₁n₂

Its value is the reciprocal of the spontaneous radiative lifetime (t) of level 2, if level 2 decays through radiative emission to level 1.

A₂₁ = 1/t

Einstein's coefficient (A₂₁) and oscillator strength values are obtained from the output of the transition moment calculations. Electrostatic potential-based atomic charges are calculated for the ground and excited state species using the Merz–Kollman^43,44 scheme in Gaussian 09. Rough estimations of these charges are also done at the semiempirical level in MOPAC. For visualization of the output files, Chemcraft software⁴⁵ has been employed throughout this work. It should be mentioned here that the semiempirical CI calculations in this work has been utilized in the beginning for a rough understanding of the whole process. It is quite obvious that this level of calculation is far from the reality; however, sufficient hint has been given by these studies to make an initial guess of the photo-excitation processes of these unexplored systems, and helped us to finally clarify the results obtained from the CASSCF and CASMP2 level of theories.

3. Results and discussion

3.1. Optimized ground states and relaxed geometry of first singlet excited states

The ionic form of nitrone is composed of three possible structures, (i), (ii) and (iii) (Fig. 2). Structure (i) dominates the ground state (S₀) in the nitrone systems which is evidenced by the optimized ground state geometry (Table 1) of our currently studied molecules from both ab initio and semiempirical level of studies. This state is found to be dominated by the HOMO² configuration (nature of the molecular orbitals are shown in Fig. 3).

Fig. 2 Possible ionic and non-ionic canonical forms of nitrone.

Table 1 Structural parameters of the optimized ground state geometries at various level of calculations

System	Method	C3–C4	C4–C5	C5–N	N–O	N–C6
a (4,4).b (6,6) CASSCF level of theories.
I	CASSCF/6-31G(d)^a	1.330	1.451	1.276	1.265	1.459
	CASSCF/6-31G(d)^b	1.330	1.451	1.309	1.253	1.461
	RHF/6-311G(d,p)	1.328	1.451	1.276	1.271	1.459
	PM3/CI	1.347	1.436	1.351	1.245	1.501
	CASSCF/6-31G(d)^a	1.331	1.452	1.307	1.299	1.479
II	CASSCF/6-31G(d)^b	1.349	1.453	1.301	1.271	1.482
	RHF/6-311G(d,p)	1.328	1.450	1.275	1.271	1.503
	PM3/CI	1.348	1.434	1.351	1.242	1.534

---

Fig. 3 HOMO and LUMO of the studied nitrone systems at their ground state geometries at the RHF/6-311G(d,p) level.

However, the ground state zwitterionic species certainly has a minor contribution in the planar first excited singlet state (A) which is confirmed by a substantial increase in the C5–N bond length of this state (Fig. 4). This increase of length may not result in pure single bond formation, as seen from the CASSCF results. Analysis of the second CI root (S₁ state) has clearly indicated that this planar excited state is a mixture of configurations arising primarily due to the HOMO − 1 → LUMO and HOMO² → LUMO² excitations with a significantly leading contribution from the latter one (Fig. 4). Some minor contributions of HOMO → LUMO and other configurations are also found in this state. It must be mentioned here that in our earlier work³⁹ on the protonated Schiff base (iminium ion) systems it was found that their ionic S₁ state (second CI root) is dominated by the HOMO → LUMO configuration and their biradical S₂ state (third CI root) is mainly contributed to by the HOMO − 1 →LUMO and HOMO² → LUMO² configurations.

Fig. 4 Optimized excited state (A) geometries at the CASSCF/6-31G* ^a(4,4) and ^b(6,6) level of theories with dominant configurations at the respective geometries (on the right).

Coming back to the discussion of the possible ionic forms of nitrone, structure (ii) in Fig. 2 is probably non-existent,^15,16,46 while the other ionic form (iii) has been reported to be another resonating form of (i) which contributes to the allyl anionic 1,3-dipolar form of nitrones with negligible contribution. Between the two possible non-ionic forms (Fig. 2), the biradical form (v) with odd electrons on carbon and oxygen seems to be structurally quite interesting, as a closer approach between them may result in a C–N–O triangle formation; though no experimental evidence has been found in favor of its existence, so far. Minor contributions of the two non-ionic forms in the ground state were predicted by Komaromi et al.¹⁵ at the CASSCF level of study. However, the first excited singlet state structure has not been analyzed much, except for few theoretical studies.¹⁴ Our detailed analysis of the excited state topography in the subsequent portions has clearly shown the possible importance of the biradical form in the oxaziridine formation process.

3.2 Important points on the excited state PES

3.2.1 Stable non-planar geometries from PM3/CI. The possible existence of S₀/S₁ conical intersection points has been explored in the next step of our work. Before going to this optimization run through CASSCF method some prior guesswork was done using the PM3/CI results on the excited state surface. The semiempirical-based CI method employed is of 2 × 2 (2 pi electrons in 2 orbitals) type and understandably predicts geometry of poor quality; however, they have given some interesting results.

Two non-planar excited state geometries of ionic (B₁) and biradical (B₂) nature have been obtained (Fig. 5). These states are found to be more stable than the planar excited states. A lowering of the negative charge on oxygen has been noticed as we move from the ground state to the planar excited state (Table S1†), which continues in the stable non-planar forms. In case of the biradical (B₂) form, the ESP-derived charges indicate the presence of a lone pair on nitrogen, an odd electron on C5 and a reduced electronic cloud on oxygen.

Fig. 5 Non-planar excited state geometries obtained at the PM3/CI level of calculations.

A closer look at the structures of B₁ and B₂ clearly shows that their deviations from planarity have some resemblance to the one-bond-flip (OBF) and the Hula-twist (HT) motions, respectively. The turn in B₁ is similar to the ionic excited state rotation (OBF) of the protonated Schiff base (PSB) iminium systems, while the nature of the geometry of the B₂ species resembles the out-of-plane H-bridge-type (here oxygen-bridge) geometry normally seen in the biradical excited state of the neutral conjugated polyene systems. The movement of the N-terminal part of the C–N bond in B₁ stops at the geometry where the N–O bond is shifted 90^° from the initial plane without any significant movement of the C-terminal part and looks like a TICT-CI geometry, while the twist in B₂ is similar to half-way HT motion. It is interesting to see the involvement of the terminal portions (involving N–O bond) only in these motions, which is certainly due to the use of a 2x2 method and the electrons have not experienced much delocalization; therefore the whole process is happening at the terminal part where the charge separations have occurred exclusively. The participation of nitrogen and oxygen is also quite expected, as there is an initial electron transfer from oxygen to nitrogen, thus reducing its electronegative behaviour. This rough estimation of the photo-excitation process through this crude method has helped us enormously to progress further, and some interesting results have been predicted subsequently in this work by the ab initio-based CASSCF level of studies.

3.2.2 Conical intersection geometries at higher energies. First of all, the choice of the active space is required to be clarified at this point. The proper choice of the relevant molecular orbitals for setting up the CAS space requires some amount of prior knowledge of the system under investigation. It is always challenging to identify the relevant MOs during the course of optimization or investigation along the reaction coordinate. Calculations reported in Table 3 of our present work have been done with 4 active electrons in 4 active orbitals (4,4). This choice of active space in our present work has been done based on chemical intuition looking at the experimental photochemical study done on the retinyl nitrone systems.¹⁹ These studies have clearly revealed that the conjugated nitrone system experiences a twist at the terminal C–N–O moiety during the passage towards oxaziridine formation without any significant change in the conjugated chain part. We have also experienced similar twist of the CNO moiety, especially in the biradical B₂ geometry, during the PM3/CI level of calculations excluding the electrons in the C–C conjugated part. Based on the experimental and PM3/CI results, a possible reaction path of this photochemical process is shown in Fig. 6. This clearly indicates the initial involvement of a C–N π bond (breaking of π bond) and p_z orbital (holding the negative charge on oxygen which is partly shifted towards N afterwards) which after photo-excitation results in a C–N σ bond, and a possible close approach of carbon and oxygen (probably through a transient biradical species) results in an oxaziridine system. Based on this intuition we have chosen a possible appropriate active space consisting of four active electrons in four orbitals (Fig. S1†). The HOMO (MO no. 30) consists of π symmetry on the CNO moiety while the LUMO (MO no. 31) is having the corresponding π* symmetry. The HOMO − 1 is of σ symmetry on the same moiety and the LUMO + 1 has σ* symmetry. This has been done using the visualization software package, Chemcraft. The latter part of the conjugated chain has very little role to play in the whole process. So basically our process is biased towards the possible reaction path and we have tried to make an accurate minimal choice of the active space leaving out the possible less important extended conjugated side. Quite obviously for some other process of these conjugated nitrones, this chosen active space may not be accurate. However, for this oxaziridine conversion process, this choice of active space is found to be quite satisfactory. Higher active spaces were also tested for the whole process; they were unable to produce the oxaziridine type twist at the terminal CNO part. It is a well-known fact that there is no single correct active space in a molecule; the active space choice depends on the particular process being carried out.

Fig. 6 A possible scheme of nitrone to oxaziridine conversion.

Considering the probability of finding both OBF-type CI and Hula-twist-type CI in these systems, we have taken the geometries of B₁ and B₂ as the starting guess geometries for conical intersection^47–49 optimization. As expected, in each case we have obtained conical intersection points between S₀ and S₁ at the CASSCF (4,4)/6-31G* level of theory (Fig. 7a and b). Moreover, the optimized geometries (CI₁ and CI₂) at these points represent their respective characteristics. The reason behind the presence of these both types of CIs can be understood if we have a closer look at the ESP-charges (Table 2) of the optimized excited state geometry (A) at the CASSCF/6-31G* level and compare it with the ground state. There is a clear indication of the biradical mode of charge separation in the excited states; however, some ionic contribution is also present. Mixed contributions of the HOMO² → LUMO², HOMO − 1 → LUMO and HOMO → LUMO configurations (Fig. 4) in this state further support this fact. This dual nature of the first excited singlet state may drive the system for both OBF-type and HT-type conical intersections. These systems can be considered to be roughly in between the nonpolar polyene and polar PSB systems. The ground state is a zwitterionic (dipolar) species and hence electrically neutral; this may indicate its resemblance to the neutral polyene systems. However, a positive charge on nitrogen may bring some similarity with the polar iminium ions. Hence a mixed contribution of biradical (similar to S₁ of polyene) and ionic (similar to S₁ of PSB) forms can be expected in the S₁ state of these nitrones, which is actually found to be happening in these systems. The photo-excitation processes of the nonpolar 1,3,5-hexatriene system and 2,4-pentadien-1-iminium cation can give some interesting insights into this matter (Fig. 8a and b). Both these systems are known to form covalent ground states (A_g); however, their S₁ (and S₂) states are exactly opposite in nature.^39,50 The first excited state has biradical characteristic in the conjugated triene system (A_g symmetry), while in the PSB system, an ionic nature has been found for the same (B_u symmetry). In the former system S₀ → S₁ is forbidden; the initial photo-excitation takes it to the S₂ (B_u, ionic) state. This is followed by an ultrafast relaxation to the S₁ state^51–55 which subsequently undergoes a conical intersection with the ground state through a triangular (kinked) structure.^56–58 On the other hand, in the protonated Schiff base, the allowed S₀ → S₁ transition is followed by an ultrafast non-radiative decay due to a barrierless process of one-bond flip (OBF)^50,59–63 around a single bond (C–C) in the S₁ state which was originally a double bond (C=C) in the ground state. During this rotation, the S₁ state becomes stabilized, whereas the ground state energy increases. At a certain torsion angle their energies will match and a conical intersection (TICT CI) between these two surfaces takes place; consequently, a radiationless transition occurs to the ground state. The features of the first excited singlet states of the studied nitrones are somewhat in between (Fig. 8c) the S₁ states of the two above mentioned extreme possibilities. As a consequence, the excited state is likely to undergo two different types of motions. In the one-bond-flip process (CI₁), the negative charge on C5 has been delocalized on the conjugated part and hence the flip has taken place with respect to the C3–C4 single bond (Fig. 7a and b). In other words, the C2–C3–C4–C5 torsion angle has changed, which is unlike the terminal rotation seen in the PM3/CI calculations (Fig. 5). However, the CI₂ geometry is not significantly different from the one obtained at the semiempirical CI level, as the turn here is also operating in the N-terminal part. The reason is probably again related to the charge distribution in this species. An electron cloud seems to be residing on the C5 atom (Table 2) which is not fully delocalized on the rest of the conjugated chain. Due to the lack of charge distribution, the turn is restricted to the terminal part only. This structure, somewhat looks like a C–N–O kink, roughly comparable to the H₁-bridging (here oxygen-bridging) conical intersection⁵⁶ operating in the terminal part of the non-polar conjugated polyene systems. The possible involvement of the oxygen atom is quite unique; however, it must be kept in mind that the photo-excitation process in these systems is probably kicked off by the initial electronic transfer from the oxygen, as evidenced by the ESP-charges of all the excited state structures. This seems to have some similarity with the well-established single electron transfer (SET) process observed in nitrones;⁶⁴ the only difference is that here it seems to be intramolecular in nature. The C5–O bond distance in this CI₂ species is roughly 2.28 Å, which is shorter than this distance in the CI₁ geometry (2.32 Å). A comparison of the energies of the excited relaxed structure (A) and the two above mentioned conical intersection points has revealed that the two CI geometries are having slightly higher energies than the A form at the CASSCF (4,4) level of calculation (Table 3).

Fig. 7 (a) Optimized conical intersection geometries (CI₁, CI₂ and CI₃) for three different types of motions at the CASSCF/6-31G* level of system I. The gradient difference and derivative coupling vectors are shown on the right. [d_{(3–4–5–6)} indicates <C3–C4–C5–N torsion angle, d_{(4–5–6–7)} indicates <C4–C5–N–C6 torsion angle], (b) optimized conical intersection geometries (CI₁, CI₂ and CI₃) for three different types of motions at the CASSCF/6-31G* level of system II. The gradient difference and derivative coupling vectors are shown on the right. [d_{(3–4–5–6)} indicates <C3–C4–C5–N torsion angle, d_{(4–5–6–7)} indicates <C4–C5–N–C6 torsion angle].

Table 2 Atomic charges determined from electrostatic potential (using Merz–Kollman scheme) at the CASSCF/6-31G* level

System	Atom	Ground state	Planar excited state (A)	Conical intersection CI1	Conical intersection CI2	Conical intersection CI3	Conical intersection CI4	Conical intersection CI5
I	C(1)	−0.3688	−0.3662	−0.5139	−0.2958	−0.4664	−0.4783	−0.4522
	C(2)	−0.0727	−0.0601	−0.0536	−0.1773	−0.1039	−0.0455	0.0349
	C(3)	−0.1406	−0.2888	−0.1862	−0.0965	−0.0657	−0.0246	−0.3501
	C(4)	−0.1790	−0.0874	−0.0925	−0.1622	−0.2255	−0.3986	0.0285
	C(5)	−0.1348	0.0347	−0.2037	−0.3303	−0.0967	0.3155	−0.0344
	N	0.5057	0.1880	0.4260	0.4537	0.2871	−0.1696	−0.1606
	C(7)	−0.4480	0.0770	−0.4471	−0.2812	−0.3348	−0.2795	−0.1923
	O	−0.6097	−0.3050	−0.3899	−0.4294	−0.4096	−0.2833	−0.1497
II	C(1)	−0.3606	−0.3425	−0.4626	−0.3058	−0.4619	—	—
	C(2)	−0.0830	−0. 0496	−0.1076	−0.1698	−0.1035	—	—
	C(3)	−0.1558	−0.3240	−0.0940	−0.0572	−0.1008	—	—
	C(4)	−0.1207	0.0329	−0.1925	−0.2387	−0.2157	—	—
	C(5)	−0.1964	−0.0948	−0.0695	−0.2761	−0.0611	—	—
	N	0.4204	0.0845	0.1721	0.3247	0.1453	—	—
	C(7)	0.2588	0.2873	0.3934	0.3432	0.3295	—	—
	O	−0.5905	−0.2989	−0.4176	−0.4195	−0.3970	—	—

---

Fig. 8 Schematic representation of the difference in the ground and excited singlet state properties of (a) non-polar conjugated hexatriene (b) polar 2,4-pentadien-1-iminium cation (c) zwitterionic conjugated nitrone systems.

Table 3 Relative energy values (with respect to the relaxed planar excited states) at various important geometries on the potential energy surfaces

Molecular geometry	System I				System II
	CASSCF		CASMP2		CASSCF		CASMP2
	Energy in hartree	Relative energy in kcal mol^−1	Energy in hartree	Relative energy in kcal mol^−1	Energy in hartree	Relative energy in kcal mol^−1	Energy in hartree	Relative energy in kcal mol^−1
Ground state	−361.664593	−78.08	−362.760885	−97.28	−439.737309	−76.72	−441.11865	−101.39
Excited state (A)	−361.540167	0	−362.605840	0	−439.615044	0	−440.95707	0
CI1	−361.532279	4.95	−362.601268	2.88	−439.605950	5.71	−440.95213	3.10
CI2	−361.534738	3.40	−362.608382	−1.59	−439.607217	4.91	−440.95858	−0.95
CI3	−361.533757	4.02	−362.606357	−0.32	−439.608575	4.06	−440.95739	−0.21
CI4	−361.565300	−15.81	−362.629875	−15.08	—	—	—	—
CI5	−361.569177	−18.21	−362.652302	−29.15	—	—	—	—
Ox1	−361.644040	−65.17	−362.725954	−75.36	—	—	—	—
Ox2	−361.640834	−63.16	−362.734703	−80.36	—	—	—	—

---

A conical intersection optimization run on the optimized excited state (A) is found to result in a different type of conical intersection geometry (CI₃). At CASSCF level, this is situated at roughly 4 kcal mol⁻¹ above the excited state geometry (A) in both methyl and isopropyl systems. This geometry looks quite comparable to the conventional kinked-CI geometry found in the neutral polyenes arising from the Hula-twist motion in the central part of the molecule. The gradient-difference and the derivative coupling vectors corresponding to these conical intersection geometries are shown in Fig. 7a and b. Inclusion of the dynamic correlation effect through MP2 is found to lower the energies of these CI geometries as shown in Table 3.

3.2.3 Conical intersection geometries at lower energies. In the above portion we have discussed three important types of conical intersection geometries; two of them are related to the twist around the central part of the molecule (CI₁ and CI₃) and the third one is the terminal out-of-plane bridging-type (CI₂). For more rigorous analysis of the photo-excitation process of these systems, we have taken the N-methyl substituted nitrone (I) as the representative system. A transition state has been obtained from the QST2 optimization using the optimized excited state (A) and the CI₂ conical intersection geometry (Fig. S2a†). Two more transition states have been detected on the excited state surface (Fig. S2b and 2c†) using the normal TS optimization method. A conical intersection optimization run on one of these latter transition states results in a lower-energy CI geometry (CI₄) which is (Fig. 9a) situated around 15 kcal mol⁻¹ below the A state. However, the lowest-energy conical intersection point (CI₅) has been obtained starting from a slightly different guess geometry with an elongated N–O bond. The C4–C5–N–O torsion angle in this conical intersection geometry is twisted away (Fig. 9b) from the plane of the molecule by a substantial amount (∼100^°); the N–O bond is highly stretched (1.38 Å), and a very close approach of the carbon (C5) and oxygen atom (2.12 Å) is seen. This geometry is found to be situated at 60 kcal mol⁻¹ above the ground state and 18 kcal mol⁻¹ below the relaxed excited state at the CASSCF (4,4) level. The CASMP2 calculations have shown the existence of this CI point at a further 11 kcal mol⁻¹ lower energy level. Few structural similarities of this intersection geometry and the CI₂ geometry can be noticed as both have a comparable C–N–O kink; however, as seen, the C5–O bond distance is much shorter at this lowest-energy intersection point with a substantial difference in the atomic charges (Table 2). Looking at the gradient difference vector (Fig. 9b), it seems that this lowest-energy conical intersection point is heading towards oxaziridine geometry with an out-of-plane C–N–O triangle. On the other hand, the derivative coupling vector indicates some other relaxation channel on the ground state. It must be added here that this lowest-energy intersection geometry is having a predominantly biradicaloid character (Table 2) unlike the relaxed excited state which has a mixed ionic–biradical nature. The possible presence of a lone pair on nitrogen can be also noticed from the ESP-derived charges. In search of the oxaziridine ground state geometry, we have chosen a guess structure following the arrow directions of the gradient difference vector (Fig. 9b). However, instead of minima, we have obtained two saddle points as the probable trans (Ox₁) and cis (Ox₂) oxaziridine ground state species. Both of them have two small imaginary frequencies and almost the same geometrical parameters (R_C–O = 1.38 Å, R_N–O = 1.44 Å, R_C–N = 1.41 Å, <OCN = 62.3^°, <ONC = 57.6^°) with an out-of-plane C–N–O kink or oxygen-bridge (Fig. 10a and b). Normally the oxaziridine geometry is reported⁶⁵ to have a C–N–O triangle having C–O and N–O bond lengths of 1.40 Å and 1.50 Å, respectively, while the C–N bond length is around 1.44 Å; the bond angles <OCN = 63.7^° and <ONC = 56.8^° are also similar to our obtained results. So clearly, the Ox₁ and Ox₂ geometries resemble the oxaziridine; however, the trans variety (Ox₁) is found to be more stable than the cis one (Ox₂) by roughly 2 kcal mol⁻¹ at the CASSCF (4,4) level. Inclusion of the dynamic correlation effect is found to make the latter geometry more stable than the Ox₁ geometry by 5 kcal mol⁻¹. A bond formation between C5 and oxygen is clearly visible from the overlap of orbitals on these respective atoms in the HOMO of the probable oxaziridine species (Fig. 10a).

Fig. 9 Conical intersection geometries (a) CI₄ (b) CI₅ of system I at lower energies with their gradient difference and derivative coupling vectors. [d_{(3–4–5–6)} indicates <C3–C4–C5–N torsion angle, d_{(4–5–6–7)} indicates <C4–C5–N–C6 torsion angle].

Fig. 10 (a) Probable trans oxaziridine (Ox₁) ground state geometry and the HOMO at this geometry of system I (b) probable cis oxaziridine (Ox₂) ground state geometry of system I.

A possible scheme of the whole process is been shown in Fig. 11. The nitrone–oxaziridine conversion process through the biradicaloid lowest-energy conical intersection geometry seems to have some similarity with the benzene–prefulvene photo-isomerization process, commonly known as the channel 3 decay.^66–68 The CI₅ geometry is quite similar to the prefulvenic conical intersection having an out-of-plane –(CH)₃– kink with odd electrons on C1 and C5, which on coupling gives prefulvene intermediate, and ultimately leads to fulvene and other products. More interestingly, similar to our obtained Ox₁ and Ox₂ geometries, the prefulvene ground state was also reported to be a saddle point instead of a minimum, which was found to connect two prebenzvalene structures. However, the kinked-CI (CI₃) structure with a conventional –(CH)₃– kink in the middle-part of the chain, is probably less important in these nitrone systems. This point has been discussed in Section 3.4.

Fig. 11 A schematic representation of the non-radiative decay processes involving different conical intersections in system I. (Terms inside the parentheses indicate the relative energies in kcal mol⁻¹ with respect to the ground state energy at CASSCF^a and CASMP2^b level).

3.3 Radiative transition properties and vertical excitation energies

Radiative transition calculations have been carried out in the present work using the GUGA-CI code in GAMESS. Oscillator strength, transition moment and Einstein's coefficient values corresponding to the lowest energy transitions (S₀ → S₁) have been predicted (Table 4) at the ground state equilibrium geometry. The transition moment value of this transition in system I is found to be higher than that of system II. It should be mentioned here that in the neutral conjugated polyene systems, the S₀ → S₁ transition is not allowed, while in the 2,4-pentadieniminium cation and in its N-substituted analogues this is highly allowed. However, the allowed transition moment values of the S₀–S₁ transitions in these nitrones (4.75 D) are much lower than that of the PSB systems (7.50 D). The calculated transition moment value of the S₀ → S₂ transition is found to be very low which again resembles the nature of the iminium ion systems. Radiative lifetime values of the first excited states (S₁) arising due to the vertical excitations from the ground states for the two studied systems are around 450–525 ps. A comparison of their radiative lifetime with that of the PSB systems studied previously by us³⁹ shows that the latter iminium ions have much lower radiative lifetime values (∼250 ps). It has been already discussed that in all the cases, quick non-radiative decay channels (probably of femtosecond order) are operating after the initial photo-excitation. It must be added in this discussion that no transition probability has been studied here near to the conical intersection region or along the pathways connecting the Franck–Condon and the conical intersection points. Our studies in this section are related to the Franck–Condon excited state, reached vertically from the equilibrium ground state geometry and situated far away from the intersection points or strong interaction regions of the two adiabatic states. Involvement of non-adiabatic coupling terms^69–71 is obvious in the latter-mentioned regions which lead to non-radiative decay paths. However, in this present study, transition probabilities in these regions have not been attempted.

Table 4 Radiative transition properties corresponding to the vertical transition (S₀ → S₁) at the ground state equilibrium geometry and vertical excitation energies

System	VEE in eV	Transition moment (in Debye)^d				Oscillator strength^d		Einstein's coefficient	Radiative lifetime (τ) of the upper state in s
		μ	μx	μy	μz	fL	fV
a VEE values reported at PM3/CI level.b VEE values reported at ^1CASSCF/6-31G* and ^2CASMP2/6-31G* level with (4,4) active space.c VEE values reported at ^1CASSCF/6-31G* and ^2CASMP2/6-31G* level with (6,6) active space.d Radiative transition properties calculated between two CI wavefunctions using GUGA CI code with 6-311G* basis sets.
I	3.99^a, 4.13^b^1, 5.59^b^2, 3.99^c^1, 5.65^c2	4.86	−4.684	−1.296	0.002	0.706	0.195	2.1472(+9)	525(−12)
II	3.99^a, 3.93^b^1, 5.75^b^2, 4.01^c^1, 5.63^c^2	4.69	4.457	−1.466	0.270	0.702	0.175	4.6046(+9)	466(−12)

---

It is well-known that there are three types of possibilities of conical intersections; sloped intersection and two varieties of peaked intersections. In the peaked intersections, the excited molecule is funnelled either from the Franck–Condon geometry or from the excited relaxed state geometry towards the point of intersection. In the latter case the excited state passes through a transition state and hence a barrier is required to be crossed, which is probably happening in these systems before the oxaziridine geometry is reached through the CI₅ intersection point. A larger barrier here will increase radiative lifetime of the excited state and may cause fluorescence, as seen in several conjugated polyene systems.

The vertical excitation energy values of these transitions are studied at different levels of calculation. The PM3/CI and CASSCF values are found to be quite close (∼4 eV). On the other hand, the CASMP2 values deviate from the other two levels of calculation. Comparisons are shown at different levels of chosen active spaces. It can be noticed that the change in active space has negligible effect on these S₀–S₁ energy gap values obtained at the CASMP2 level.

3.4 Computed results and experimental findings

The presence of different CI geometries associated with different types of motions in a single molecule reported in this study is not completely uncommon. The Hula-twist-type CI has been reported for several iminium ion systems, though the photo-isomerization in these systems are primarily dominated by the OBF-mechanism. Especially, the approach of counter ions near the positive nitrogen in these systems was predicted to have involvement of both types of conical intersections.⁶⁰ However, a system with a dominance of photoproducts arising out of the lowest-energy conical intersection geometry occurring due to a twist in the terminal part is probably not so common. It has already been mentioned that the benzene-prefulvene conversion mechanism of the biradical-mode driven out-of-plane bridge-type CI seems to be almost comparable to our proposed nitrone–oxaziridine mechanism. The studied nitrone systems can be considered to be the model compounds of their corresponding retinyl nitrone systems. Methyl and isopropyl retinylnitrones were studied earlier under photo-excitation¹⁹ and in particular the N-methyl retinylnitrone compound was found to give isolable oxaziridine under room light, which was converted back to the parent nitrone in the dark. In fact, it has been reported previously that the presence of alkyl groups on nitrogen in nitrones is found to generate stable oxaziridines, and may probably delay the formation of amides and other photoproducts. Our predicted results match with this experimental observation. The lowest-energy intersection clearly indicates a route for probable oxaziridine through a C5–O bond formation, which is exactly the fact reported by experimental results. Any chance of isomerization about the terminal C–N bond also gets reduced due to the formation of the C–N–O triangle in oxaziridine which contains a shortened C–O bond and an elongated N–O bond.

Inspite of some similarities, a major difference probably exists between the oxaziridine-type conical intersection and the prefulvenic conical intersection geometries, as the latter are actually tetraradicaloid in nature. It has been discussed earlier that in the CI₅ geometry the nitrogen holds a lone pair of electron (Table 2), while odd electrons probably reside on C5 and oxygen. This statement can be further justified by analyzing the products obtained from prefulvene, where benzvalene and fulvene are the two major possibilities. Formation of fulvene can be compared with the possibility of amide as a photoproduct in our studied systems. Though no amide has been reported from the N-methyl retinylnitrone system under room light, it seems quite reasonable if we think that a possible route of amide formation is from oxaziridine through the breaking of the N–O bond and formation of the C5–O bond, followed by a [1,2] hydrogen shift from C5 to nitrogen. This is to some extent comparable to the prefulvene → fulvene conversion. However, unless two more odd electrons are there, no benzvalene-type product can be obtained. In fact no such analogous product has been reported on photo-excitation of any kind of nitrones, so far. This supports the possible existence of a lone pair on nitrogen in the oxaziridine system, which has also been reported several times in the past.

As discussed, it seems quite obvious that the oxaziridine and thereafter amide are probably the two expected photoproducts along with other numerous possibilities arising from the different relaxation paths. The two above-mentioned products are coming out from the out-of-plane bridging CI mechanism in the terminal part, but what about the OBF-mode or Hula-twist-mode operating in the central part? The cis-isomer with respect to the C3–C4 bond rotation is likely to be the photoproduct in these cases, though experimental studies on the alkyl-retinylnitrones have not reported this product. The present theoretical results have clearly indicated that the conical intersection geometries related to the turn or twists in the central part of the molecule are situated at higher energies in comparison to the terminally twisted CI₅ geometry, which is found to form the route towards oxaziridine formation. At the same time it is required to be mentioned that ruling out the possibilities of the above-mentioned cis-products in these model compounds through the CI₁ and CI₃ geometries can be debated, as these conical intersections may be directly accessed from the vertically excited Franck–Condon geometry. It should be stated here that the volume conserving nature of the twist happening in the terminal part must be also taken into account. Our reported CI₅ geometry in these nitrones seems to be quite matched with the H₁-bridging (here oxygen-bridging) structure in HT1-mode described by Norton and Houk,⁵⁶ which was reported to be an efficient volume conserving process. Experimentally obtained photochemical products (oxaziridine) of these nitrones indicate that this volume conserving mechanism of the terminal-twisted mode dominates the photo-relaxation process. In fact, looking at the geometries related to the central C–C bond rotation and the terminal C–N bond rotation (Fig. 7a and b) clearly suggests that the latter one requires less amount of space than the other one, and therefore may be a more favorable process.

4. Conclusions

This study has proposed the probable mechanism of the photo-excitation process of the N-alkyl-substituted conjugated long-chain nitrones. The predicted photochemical products of the studied nitrone systems match with the reported experimentally obtained results on the corresponding N-alkyl retinyl nitrone molecules. The photo-excited first singlet state is found to have a mixed contribution from the biradical and ionic structures. Rotations around the central C–C bond and the terminal C–N bond are characterized by their distinct conical intersection geometries, and make the excited state topography highly complicated. Their comparative study with the structurally similar non-polar hexatriene and polar 2,4-pentadien-iminium ion has given some valuable information, and indicates the fact that their excited state properties are somewhat in between these two conjugated systems. Slightly less efficient non-radiative decay channels can be expected in these nitrones in comparison to the PSB systems. The non-radiative decay path leading to the oxaziridine geometry is likely to pass through the lowest-energy intersection point. The nature of this conical intersection and its subsequent photoproduct resembles the benzene → prefulvene conversion pathway through the prefulvenic intersection. The reported results in this work may stimulate rigorous experimental studies on conjugated nitrone systems in the future. Overall, the importance of this present study is not only restricted to the establishment of the probable mechanism of the photo-excitation process of conjugated long-chain nitrone systems, but also it can have far reaching consequences in terms of the excited state properties of several other nitrone systems, too. The proposed mechanism of the little-known excited state photochemistry of these nitrones can reveal spectroscopic features of various structurally similar nitrone systems, and tuning of photoproducts of nitrones can be done by proper choice of substituents on nitrogen.

Acknowledgements

We gratefully acknowledge the financial support received from the Council of Scientific and Industrial Research (CSIR), Government of India, under the Scheme no. 01(2681)/12/EMR-II, for the present work.

References

1. J. S. Splitter and M. Calvin, J. Am. Chem. Soc., 1965, 30, 3427.
2. J. S. Splitter, T.-M. Su, H. Ono and M. Calvin, J. Am. Chem. Soc., 1971, 93, 4075.
3. K. Shinzawa and I. Tanaka, J. Phys. Chem., 1964, 68, 1205.
4. K. Koyano and I. Tanaka, J. Phys. Chem., 1965, 69, 2545.
5. E. Lipczynska-Kochany and J. Kochany, J. Photochem. Photobiol., A, 1988, 45, 65.
6. D. R. Boyd, W. B. Jennings, R. Spratt and D. M. Jerina, J. Chem. Soc. D, 1970, 745.
7. G. G. Spence, E. C. Taylor and O. Buchard, Chem. Rev., 1970, 70, 231.
8. M. Nastasi and J. Streith, Rearrangements in Ground and Excited States, ed. P. de Mayo, Academic Press, New York, 1980.
9. D. R. Boyd, K. M. McCombe, T. A. Hamor, W. B. Jennings and V. E. Wilson, J. Chem. Soc., Perkin Trans. 2, 1984, 95.
10. D. R. Boyd, P. B. Coulter, N. D. Sharma, V. B. Jennings and V. E. Wilson, Tetrahedron Lett., 1985, 26, 1673.
11. S. Sivasubramanian, P. Manisankar, P. Jeyaram and N. Arumugam, J. Polym. Sci., Part A: Polym. Chem., 1985, 59, 367.
12. M. Iwamura, M. Katoh and H. Iwamura, Org. Magn. Reson., 1980, 14, 392.
13. J. R. Harbour, V. Chow and J. Bolton, Can. J. Chem., 1974, 52, 3552.
14. B. Bigot, D. Roux, A. Sevin and A. Devaquet, J. Am. Chem. Soc., 1979, 101, 2560.
15. I. Komaromi, M. Jean and J. Tronchet, J. Mol. Struct.: THEOCHEM, 1996, 366, 147.
16. S. D. Kahn, W. J. Hehre and J. A. Pople, J. Am. Chem. Soc., 1987, 109, 1871.
17. K. N. Houk, J. Sims, R. E. Duke, R. W. Strozier and J. K. George, J. Am. Chem. Soc., 1973, 95, 7287.
18. R. M. Aminova and E. Ermakova, Chem. Phys. Lett., 2002, 359, 184.
19. V. Balogh-Nair and K. Nakanishi, Pharm. Res., 1984, 1, 93.
20. M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery Jr, J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, A. D. Daniels, Ö. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski and D. J. Fox, Gaussian 09, Revision B.01, Gaussian, Inc., Wallingford CT, 2009.
21. D. Hegarty and M. A. Robb, Mol. Phys., 1979, 38, 1795.
22. R. H. A. Eade and M. A. Robb, Chem. Phys. Lett., 1981, 83, 362.
23. H. B. Schlegel and M. A. Robb, Chem. Phys. Lett., 1982, 93, 43.
24. F. Bernardi, A. Bottoni, J. J. W. McDougall, M. A. Robb and H. B. Schlegel, Faraday Symp. Chem. Soc., 1984, 19, 137.
25. M. J. Frisch, I. N. Ragazos, M. A. Robb and H. B. Schlegel, Chem. Phys. Lett., 1992, 189, 524.
26. N. Yamamoto, T. Vreven, M. A. Robb, M. J. Frisch and H. B. Schlegel, Chem. Phys. Lett., 1996, 250, 373.
27. C. Peng, P. Y. Ayala, H. B. Schlegel and M. J. Frisch, J. Comput. Chem., 1996, 17, 49.
28. C. Peng and H. B. Schlegel, Isr. J. Chem., 1993, 33, 449.
29. X. Li and M. J. Frisch, J. Chem. Theory Comput., 2006, 2, 835.
30. H. P. Hratchian and H. B. Schlegel, Theory and Applications of Computational Chemistry: The First 40 Years, ed. C. E. Dykstra, G. Frenking, K. S. Kim and G. Scuseria, Elsevier, Amsterdam, 2005.
31. H. P. Hratchian and H. B. Schlegel, J. Chem. Theory Comput., 2005, 1, 61.
32. H. P. Hratchian and H. B. Schlegel, J. Chem. Phys., 2004, 120, 9918.
33. J. J. P. Stewart, J. Comput. Chem., 1989, 10, 209.
34. J. J. P. Stewart, J. Comput. Chem., 1989, 10, 221.
35. M. W. Schmidt, K. K. Baldridge, J. A. Boatz, S. T. Elbert, M. S. Gordon, J. J. Jensen, S. Koseki, N. Matsunaga, K. A. Nguyen, S. Su, T. L. Windus, M. Dupuis and J. A. Montgomery, J. Comput. Chem., 1993, 14, 1347.
36. B. Brooks and H. F. Schaefer, J. Chem. Phys., 1979, 70, 5092.
37. B. Brooks, W. Laidig, P. Saxe, N. Handy and H. F. Schaefer, Phys. Scr., 1980, 21, 312.
38. A. Chattopadhyay, J. Phys. B: At., Mol. Opt. Phys., 2011, 44, 165101.
39. A. Chattopadhyay, J. Chem. Sci., 2012, 124, 985.
40. F. Weinhold, J. Chem. Phys., 1970, 54, 1874.
41. C. W. Bauschlicher and S. R. Langhoff, Theor. Chim. Acta, 1991, 79, 93.
42. S. Koseki and M. S. Gordon, J. Mol. Spectrosc., 1987, 123, 392.
43. U. C. Singh and P. A. Kollman, J. Comput. Chem., 1984, 5, 129.
44. B. H. Besler, K. M. Merz Jr and P. A. Kollman, J. Comput. Chem., 1990, 11, 431.
45. Website of Chemcraft software: http://www.chemcraftprog.com.
46. I. A. Grigor'ev, Nitrile Oxides, Nitrones, and Nitronates in Organic Synthesis, Novel Strategies in Synthesis, ed. H. Feuer, John Wiley & Sons, Inc., Hoboken, New Jersey, 2008.
47. N. Ragazos, M. A. Robb, F. Bernardi and M. Olivucci, Chem. Phys. Lett., 1992, 197, 217.
48. M. J. Bearpark, M. A. Robb and H. B. Schlegel, Chem. Phys. Lett., 1994, 223, 269.
49. F. Bernardi, M. Olivucci and M. A. Robb, Chem. Soc. Rev., 1996, 25, 321.
50. M. Garavelli, P. Celani, F. Bernardi, M. A. Robb and M. Olivucci, J. Am. Chem. Soc., 1997, 119, 6891.
51. R. A. Mathies, C. H. B. Cruz, W. T. Pollard and C. V. Shank, Science, 1988, 240, 777.
52. J. Dobler, W. Zinth, W. Kaiser and D. Oesterhelt, Chem. Phys. Lett., 1988, 144, 215.
53. W. Zinth and D. Oesterhelt, Photobiology, ed. E. Ricklis, Plenum Press, New York, 1991.
54. T. Arlt, S. Schmidt, W. Zinth, U. Haupts and D. Oesterhelt, Chem. Phys. Lett., 1995, 241, 559.
55. D. Xu, C. Martin and K. Schulten, J. Biophys., 1996, 70, 453.
56. J. E. Norton and K. N. Houk, Mol. Phys., 2006, 104, 993.
57. M. Olivucci, F. Bernardi, P. Celani, I. Ragazos and M. A. Robb, J. Am. Chem. Soc., 1994, 116, 1077.
58. P. Celani, M. Garavelli, S. Ottani, F. Bernardi, M. A. Robb and M. Olivucci, J. Am. Chem. Soc., 1995, 117, 11584.
59. A. Cembran, F. Bernardi, M. Olivucci and M. Garavelli, Proc. Natl. Acad. Sci. U. S. A., 2005, 102, 6255.
60. D. S. Ruiz, A. Cembran, M. Garavelli, M. Olivucci and W. Fu, Photochem. Photobiol., 2002, 76, 622.
61. J. A. Dobado and M. Nonella, J. Phys. Chem., 1996, 100, 18282.
62. I. Conti, F. Bernardi, G. Orlandi and M. Garavelli, Mol. Phys., 2006, 104, 915.
63. I. Conti and M. Garavelli, J. Photochem. Photobiol., A, 2007, 190, 258.
64. M. N. Khan, S. Bhattacharjee, H. Chandra and M. C. R Simons, Spectrochim. Acta, Part A, 1998, 54, 779.
65. S. S. Murphree, Modern Heterocyclic Chemistry, ed. J. Alvarez-Builla, J. J. Vaquero and J. Barluenga, Wiley-VCH Verlag & Co. KGaA, Weinhelm, Germany, 2011.
66. A. Migani and M. Olivucci, Conical Intersections: Electronic Structure, Dynamics & Spectroscopy Advanced series in Physical Chemistry, ed. W. Domcke, D. R. Yarkony and H. Koppel, World Scientific Publishing Co. (P). Ltd, Singapore, 2004, vol. 15.
67. I. J. Palmer, I. N. Ragazos, F. Bernardi, M. Olivucci and M. A. Robb, J. Am. Chem. Soc., 1993, 115, 673.
68. J. Dreyer and M. Klessinger, Chem.–Eur. J., 1996, 2, 335.
69. M. Baer, Beyond Born Oppenheimer, Electronic non-Adiabatic coupling Terms and Conical Intersections, John Wiley & Sons Inc., Hoboken, New Jersey, 2006.
70. M. Baer, Chem. Phys. Lett., 1975, 35, 112.
71. M. Baer, Phys. Rep., 2002, 358, 75.

---

Footnote

† Electronic supplementary information (ESI) available: Non-planar geometries from PM3/CI; Table S1; MOs involved in the active space; Fig. S1; transition states; Fig. S2. See DOI: 10.1039/c3ra47186a

---