Molecular Simulation of Enantiodifferentiating Photoisomerization of Cyclooctene by Chiral Sensitizers
Hiromitsu HASHIMOTO, Tetsuro SHIMO, Mikito ATSUCHI, Masaru MITSUSHIO and Kenichi SOMEKAWA
1. 1 Chirality control
The origin of biomolecular homochirality in the biosphere is one of the most interesting and contentious issues in the chemical evolution. The founders of stereochemistry, van't Hoff and Le Bel, had suggested the use of circularly polarized light for "absolute asymmetric synthesis (AAS)" in the 19th century. Many kinds of AAS for chiral control and for useful products were developed by use of peculiar chiral catalysts by several Nobel prizers, etc. in the last half of 20th century. The field of "asymmetric photochemistry" has however undergone an accelerated development only in the past 20 years. "Chiral Photochemistry" edited by Y. Inoue and V. Ramamurthy  provides a range of articles with various aspects of controlling the chirality of photochemical reactions so-called "photochirogenesis". It contains, (1) direct asymmetric photochemistry with circularly polarized light,
(2) magnetochiral anisotropy in asymmetric photochemistry,
(3) enantiodifferentiating photosensitized reactions,
(4) chirality in photochromism,
(5) chiral photochemistry with transition metal complexes, and
(6) enantioselective photochemical reactions in various molecular aggregates or in the solid media, etc..
Very few authors however described the chiral mechanisms by the molecular orbital (MO) method.
1. 2 Enantiodifferentiating photoisomerization
Various and interesting experimental studies by Inoue et al. [1, 2] on the enantiodifferentiating photosensitized reactions in solution, have stimulated our requisition for the mechanism and factors that control the stereochemical outcome of the enantio- and diastereodifferentiating process in the excited state. Namely, the sensitized enantiodifferentiating photoisomerizations (action by 2a in Scheme 1) by Inoue et al. [2 - 5] were especially appearing as a method of efficient chirality transfer. The chirality transfer is unusually effected by weak energies of the reaction temperature and pressure etc.. Development of "Entropy-control" reactions by the enantiodifferentiating photoisomerization is now in one of international big projects of ICORP. The biological application example of enantiomeric selectivity in the bringing of (R)-or (S)-E-cyclooctene (1E) to ethylene (hormone) receptor antagonist is shown in section 3.4.
The conformation and energy profile at the stereo-differentiating processes in the excited state have however not been promoted. Dynamic molecular orbital (MO) method may be effective for understanding of the whole process [6, 7]. We have elucidated major factor and origin for peri-, site-, regio- and stereoselectivities of many kinds of photocycloadditions by frontier MO (FMO)  and transition state (TS) analysis using improved MOPAC-PM5 and UCIS (or UB3LYP) level of calculations [9 - 12]. The TS energy change by frontier orbital interactions, ionic interactions and hydrogen-bonding interactions is effective for the product-selectivities. PM5 is thought to be effective for calculation of such singlet excited photoreactions. We now elucidate the energy and stereochemical profile of photosensitized enantiodifferentiating Z-E isomerization of cyclooctene by some benzenepolycarboxylates (2b-2d in Scheme 1) by molecular simulation using the MOPAC-PM5 level.
The photosensitized isomerization process is supposed as the solid line via 2 exciplexes, Ex1 and Ex2 in Figure 1.
Figure 1. Process of sensitized enantiodifferentiating photoisomerization
2 Experimental data and calculation method
2. 1 Photoisomerization data
Sensitized photoisomerization data of 1Z to 1E by many kinds of benzenecarboxylates and cyanobenzenes are given in the literature . 1E/1Z ratio by 2a, 2b, 2c and 2d are nearly 0.01, 0.1, 0.6, and 0.3, respectively. The enantioselectivity, (S)-1E/(R)-1E ratio by 2a is higher than the one by 2d.
2. 2 Calculation
All geometry optimization and energy minimum path at the interactions of the two substrates at ground states and excited singlet states were performed using PM5, which are available in the molecular orbital package WinMOPAC3.5 by Fujitsu Ltd. [13, 14].
1) Keyword for PM5
Ground state: EF PM5 PRECISE, Excited singlet state: EF PM5 EXCITED SINGLET OPEN(2,2) , Saddle point: SADDLE PM5, Transition state: TS PM5, Normal mode analysis: FORCE ISOTOPE PRECISE (XYZ) .
2) Two molecular interaction
Interactions between excited singlet state 2 (2*) and cyclooctene 1 were calculated by approach of the proper parts of the two molecules . The stable structures Ex1 in Figure 3, and Ex2 in Figure 5 etc., were optimized. The TS structures such as TS (C) in Figure 3, C in Figure 5, Saddle in Figure 6, and TS in Figure 8 were checked by normal mode analysis with only imaginary frequency, which is evidence of the TS point. The frequency of the C point in Figure 4 was only imaginary: -1125cm-1 (GNorm = 5.3). We need precise checking for other data (GNorm = 16-95) of "too large GNorm". They however seem to be reasonable similarly to Figure 5. We utilize them for discussion.
3) The excited state MO of 2a was not obtained. It might be caused by the many alkyl conformations.
Reasons of MOPAC-PM5 utilization and the accuracy
MOPAC program has been improved in accuracy from PM3 to PM5 and PM6 , and has also advantages in cost-performance and consecutive graphs. The accuracy of PM5 for Diels-Alder reaction TS analysis is very good as well as one of B3LYP(/6-31+G(d)) [17, 18]. PM5 method was the best among semi-empirical ones and the usage could reproduce the crystal structure as well as B3LYP/LANL2DZ . For TS analysis of photocycloaddition selectivities, PM5 method was also effective as well as UCIS (at singlet reactions) and UB3LYP (at triplet reactions) [10 - 12, 14].
3 Results and discussion
3. 1 Energy (HOF) diagram and stereochemical profile of (Z)- and (E)-cyclooctenes, 1Z and 1E
The heat of formation (HOF) of the lowest energy conformer of 1Z by PM5 method is -14.7 kcal/mol, and is 11.7 kcal/mol lower  than that of 1E (dihedral angle: F8123 = 143°). Their relative results and stereochemistry are shown in Figure 2.
Figure 2. Relative energy (HOF) versus dihedral angle of 1Z and 1E.
1Z (F8123 = -1.5°, F4567 = -107°) has the asymmetry conformer 1Zc (F8123 = 1.5°, F4567 = 107°), and the conformation barrier (4.0 kcal/mol) for equilibrium may be concerned with the chiral sensitization by chiral sensitizers as shown in section 3.4. 1Z is correlated to (R)-(-)-1E (F8123 = -143°, F4567 = 103°) by a quarter rotation.
Calculated TS point (relative HOF = 61kcal/mol) between 1Z and 1E in Figure 2 was introduced by use of keyword SADDLE, and then TS. We have not succeeded in "normal mode with only imaginary frequency" which was shown at section 2.2.
The two excited singlet states (1Z* and 1E*(R or S)) of 1Z and 1E are intersected at F8123 = 94°, and have other conformers. They explain the population data (1Z/1E 0.5) at the direct photoisomerizations by Inoue et al. .
3. 2 Sensitized photoisomerization of 1Z to 1E by 2b*
3. 2. 1 Exciplex Ex1 formation between 1Z and 2b*
Equation 1 and Figure 3 show the calculated energy diagram
Figure 3. Interaction between 1Z and 2b*.
and stereochemical changes by approach between C1 (2b*) and C1 (1Z). It supports the experimental result of preparation of the exciplex 1Z.2b* (Ex1: B), and nothing of photoadduct via a biradical (D), which is similar to biradical D in Figure 5, caused by the high barrier (C). The HOF of Ex1 is 34kcal/mol lower than the sum of (1Z+2b*).
Figure 4. Clear graphs of Ex1 and Saddle intermediates from (1z+2b) reaction in Figure 3 and Figure 6.
The clear graphs of Ex1 in Figure 3 and also "saddle" in Figure 6 are shown in Figure 4. The olefin part of 1Z in Ex1 shows small twist (F8123 = -19.8°) and faces the benzene part of 2b. The distances of 1(1z)-1(2b), and 2(1z)-2(2b) are 3.12 A and 2.82 A like as exciplexes of singlet [2+2]photoadditions [9, 14]. Four ester carboxyl groups of 2b all turn to olefin or allyl hydrogens, and the distances are 2.56, 2.66, 2.69 and 2.72 A.
The stabilization of Ex1 is inferred from the p/p* interaction, four hydrogen bondings and C-H/p interactions between the carbonyl oxygens or p components, and C(p)-H or allylic hydrogens observed in B conformation (Ex1). The dihedral angle F8123 (1Z) in B was twisted (20°). We then checked two possibilities of : (a) ground state isomerization of equation 2,
and (b) energy transfer of equation 3,
For (a) the change to 94° of the dihedral angle F8123 (1Z) in 1Z.2b* similar to Figure 2 needed high energy, 49 kcal/mol, which shows the (a) process is impossible. We next calculated the energy for (b).
3. 2. 2 Exciplex Ex2 formation between 1E* and 2b, and energy transfer
Figure 5 shows the interaction between 1E* and 2b for Ex2 formation in equation (3). The Ex2 energy (-257 kcal/mol) is 15 kcal/mol lower than the total energy of 1E* and 2b, and 3 kcal/mol higher than Ex1's in Figure 3. Ex2 is followed by quenching to 1E (and 1Z). These data may be used for explanation of E/Z ratio 0.1 at the (1Z+2b) photoreaction.
Figure 5. Interaction between 1E* and 2b.
We also simulated the transition process from Ex1 (1Z.2b*) to Ex2 (1E*.2b) by use of the PM5 keyword, SADDLE, and show the data in Figure 6. The Saddle structure in Figure 6 has following data:
Figure 6. Energy transfer from 1Z.2b* (Ex1) to 1E*.2b(Ex2).
1E: F8123 = -121.8°, F4567 = -59.2°,
r(1(1E) -1(2b)) = 2.82 A, r(2(1E)- 2(2b)) = 3.30 A,
r(2(1E) -5(2b)) = 2.94 A, r(CH-O=C)
= 2.50, 2.69, 2.75, 2.89 (A),
r(1-H(1E)- 1(2b)) = 2.39 A, r(2-H(1E) -4(2b)) = 2.46 A.
The data are near to those of Ex1 except F8123, and like as a singlet exciplex of [3+2] addition type. Since the TS energy is very low, the energy transfer equilibrium is inferred to be smooth, and to be caused from many hydrogen bondings, and CH/p interactions (the distances: 2.3 - 2.9 A) at the TS state.
3. 3 Sensitized photoisomerization of 1Z to 1E by 2c
The 1E/1Z ratio of the sensitized reaction of 1Z to 1E by 2c is relatively high (1E/1Z = 0.6) . Figure 7 shows the calculated energy diagram and stereochemistry by approach between C2 (2c*) and C1 (1Z). It suggests formation of the exciplex 1Z.2c*(Ex1) of the conformation B and HOF (-319.1 kcal/mol), which is 32 kcal/mol lower than the start (1Z + 2c*). The stabilization is brought also from H/F hydrogen bonding, placing alongside.
Figure 7. Interaction between 1Z and 2c*.
Figure 8 shows the interaction between 1E* and 2c for Ex2 (1E*.2c: -319.9 kcal/mol). In Ex1 1Z is on the 2c*, and in Ex2 1E* is alongside of 2c possessing intermolecular interactions of hydrogen bonding, CH/p, H/F etc. The energy transfer process data (-318.8 kcal/mol in Figure 9) from Ex1 (1Z.2c*) to Ex2 (1E*.2c) by use of keyword, SADDLE suggest that the isomerization is very smooth. 1E/1Z ratio by 2c is larger than that by 2b . From difference between Figure 9 and Figure 6, Ex1> Ex2 in the energy (HOF) is inferred to bring high 1E/1Z ratio.
Figure 8. Interaction between 1E* and 2c.
Figure 9. Energy transfer from 1Z.2c* (Ex1) to 1E*.2c (Ex2).
3. 4 Chiral isomerization of 1Z to 1E by chiral menthyl 3,5-bis(trifluoromethyl)benzoate (2d)
3. 4. 1 Isomerization by (R)-menthyl isomer (2dr)
Approach between excited singlet state (2dr*) of 2dr and 1Z on PM5 program showed the existence of the exciplex 1Z.2dr* (Ex1 in Figure 10) (HOF = -366.7 kcal/mol, dihedral angle F8123 (1Z) = -21.4°) which is 16.9 kcal/mol lower than the total of (1Z+2dr*). Similar approach between the excited singlet state (1E*) of 1E and 2dr showed the existence of the exciplex 1E*.2dr (Ex2 in Figure 10) (HOF = -368.9 kcal/mol), which is 4.5 kcal/mol lower than the total of (1E*+2dr), and 2.1kcal/mol lower than Ex1. The transition simulation from Ex1 (1Z.2dr*) to Ex2 (1E*.2dr) by the keyword (SADDLE) showed the TS energy (-366.6 kcal/mol) and the steric conformation in Figure 10. Since the TS energy is very small (0.1 kcal/mol), the energy transfer leading to the isomer 1E is inferred to be very easy similarly to the case of 1c. The dihedral angle F8123 (1Z) also changed from -21.4° to -34.2° and -85.2° in the transfer process, which suggests to go to (R)-1E (F8123 = -143°) by one-side rotation.
Figure 10. Ex1 and Ex2 from reaction (1Z+2dr) and the energy transfer.
3. 4. 2 Isomerization by (S)-menthyl isomer (2ds) and enantiodifferentiation
Similar approach between the unstable excited singlet state 2ds* and 1Z showed existence of the exciplex 1Z.2ds* (Ex1 in Figure 11) (HOF = -366.3 kcal/mol, F8123 = -29.3°), which is 0.4 kcal/mol unstable comparing with 1Z.2dr* in 3.4.1. Exciplex 1E*.2ds (Ex2 in Figure 11) (HOF = -368.8 kcal/mol) between 1E* and 2ds is 2.5 kcal/mol lower than Ex1.
Figure 11. Ex1 and Ex2 from reaction (1Z+2ds) and the energy transfer.
The transition simulation from Ex1 to Ex2 in Figure 11 showed low TS energy (-366.5 kcal/mol) almost the same as Ex1. By comparison between Figure 10 and Figure 11, 1Z (F8123 = -1.5°) is inferred to be introduced more effectively to Ex1 and Ex2 by 2dr than 2ds, and to be followed by (R)-1E (F8123 = -143°), which was examined experimentally by Inoue et al. . The enantiodifferentiation may come from the proper intermolecular interactions of hydrogen bondings between p-H or allyl groups in 1Z, and asymmetric ester carbonyl in 2dr. The CF3 groups in 2dr are also thought to be effective for the interactions in the stereochemistry.
Figure 12 shows the clear graphs of Saddle in Figure 10 and of Saddle in Figure 11. The structural data of the former and the latter are as follows. Saddle/(1Z + 2dr): F8123 = -33.4°, F4567 = -109.8°, r(1(1Z) -1(2dr)) = 3.32 A, r(1(1Z)- 2(C=O)) = 3.02 A, r(2(1Z) -2(O=C)) = 2.63 A, r(CH-O=C) = 2.34, 2.46 (A), r(CH/F) = 3.06 A, 1Z/iso-propyl (2dr): exo.
Saddle/(1Z + 2ds): F8123 = -28.2°, F4567 = -109.8°, r(1(1Z) -1(2ds)) = 3.32 A, r(1(1Z)- 2(C=O)) = 3.07 A, r(2(1Z) -2(O=C)) = 2.71 A, r(CH-O=C) = 2.45, 2.57 (A), r(CH/F) = 2.99 A, 1Z/iso-propyl (2ds): endo.
Figure 12. Clear graphs of Saddle from 1Z.2dr* (Figure 10) and Saddle from 1Z.2ds* (Figure 11).
The saddle/(1Z + 2dr) seems to have advantages in p/p*, and hydrogen bonding interactions, ionic interactions by ester carbonyl, and steric and conformational factors by the iso-propyl group. The smaller r(CH/F) in the latter saddle also suggests delicate balance between their interactions and steric factors. They are inferred to bring smaller advantage in F8123 of the former, which is followed by (R)-1E.
1Z (F4567 = -107°) has a conformational isomer 1Zc of opposite F signs. The isomer 1Zc was simulated to become rich (S)-1E or not by use of 2ds as shown in Figure 13. Thus the energy Ex1 (1Zc.2ds*) (-366.5 kcal/mol) was lower than that of the diastereomeric exciplex Ex1 (1Zc.2dr*) (-364.4 kcal/mol). The latter data also suggests some margin of error [10, 17]. By comparison of three steric structures of Ex1 in Figures 10, 12, 13, we infer that isopropyl group in the menthyl substituent is more of hindrance than an interaction.
Figure 13. 1Zc and the Ex1(1Zc.2ds*) from reaction (1Zc+2ds).
Pirrung et al. recently demonstrated that (R)-riched (E)-cyclooctene (1E) was more effective than the (S)-riched 1E as ethylene receptor (ETR1) antagonist. Since ethylene is used as a hormone to control physiological processes of plants, the enantiomeric selectivity may give a search for the asymmetric ETR1 protein-composed environment. The enantiomerically enriched 1E was prepared by the method of photosensitized isomerization of 1Z using (R)- and (S)-hexakis (1-methylheptyl) benzenehexacarboxylates as effective chiral sensitizers, which were developed by Inoue et al. .
Interesting experimental results on enantiodifferentiating photoisomerization of (Z)-cyclooctene (1Z) to chiral (E)-isomer (1E) sensitized by chiral polyalkyl benzenepolycarboxylates (2a-2d) were elucidated on the energy and stereochemical profile by molecular simulation by use of MOPAC-PM5 program.
Energy, stereochemistry, and equilibrium of ground states of 1Z (and conformer 1Zc), 1E, 2(2b-2d) and the excited singlet states of 1Z*, 1E*, 2* (2b*-2d*) were first elucidated. One asymmetric conformar 1Z (F4567 = -107°) was inferred to have the preferential one-side rotation to go to (R)-1E (F8123 = -143°) by photoisomerization. The photoisomerizations between 1Z and 2 were inferred to proceed via two exciplexes Ex1 (1Z.2*) and Ex2 (1E*.2). The transition state (TS) for the energy transfer process is low and it is followed by quenching to 1E.
The sensitization ratio 1E/1Z may be related to the energy difference (DHOF) between each Ex1 and Ex2 as follows. Since 2b gives stable Ex1 because of p*/p and four ester-carbonyls for the effective intermolecular interactions such as C=O/HC (p and allyl) hydrogen bondings, the 1E/1Z ratio is low (0.1) . As 2c gives stable Ex2 because of alongside interactions by the meta-CF3 and ester carbonyl for Ex2, the 1E/1Z ratio is relatively high (0.6) .
The enantiodifferentiating photoisomerization of asymmetric 1Z to chiral 1E by chiral 2 may be estimated by calculation of diastereomeric Ex1 energy and TS energy for Ex2. Enantiomeric (R)-1E via low Ex1 and small TS process from 1Z by 2dr (or 1Zc by 2ds) was speculated to be preferred. The preference depends on sum of the steric interaction and repulsion energies including hydrogen bondings between asymmetric 1Z and enantiomeric 2d*. It will be influenced by 2d substituents and environment.
Such various and weak interactions are caused from intra- and intermolecular interactions and the entropy-control factors of the reaction temperature etc., may happen chirality control of the photosensitized isomerization in solution.
We hope that our molecular simulation by MO method for enantiodifferentiating phenomena by weak intra- and intermolecular interactions suggests some ideas for development of such interesting interaction environments like enzymes.
We are grateful to Prof. Y. Inoue at Osaka University for advice and encouragement to this simulation research. We also thank Ms. T. Ooto at Kagoshima University for paper work.
[ 1] Y. Inoue, V. Ramamurthy, Chiral Photochemistry, Marcel Dekker, New York (2004).
[ 2] Y. Inoue, T. Yokoyama, N. Yamasaki, A. Tai, Nature, 341, 225 (1989).
[ 3] Y. Inoue, N. Yamasaki, T. Yokoyama, A. Tai, J. Org. Chem., 57, 1332 (1992).
[ 4] Y. Inoue, E. Matsushima, T. Wada, J. Am. Chem. Soc., 120, 10687 (1998).
[ 5] T. Mori, Y. Inoue, CRC Handbook of Organic Photochemistry and Photobiology, Second edition, edit. by W. Horspool, E. Lenci, CRC Press (2003), Chap.16-1.
[ 6] I. Fleming, Frontier Orbitals and Organic Chemistry Reactions, John Wiley & Sons, Ltd (1976).
[ 7] J. L. Broeker, E. Eksterowicz, A. Z. Belk, K. N. Houk, J. Am. Chem. Soc., 117, 1847 (1995).
[ 8] T. Suishu, T. Obata, T. Shimo, K. Somekawa, Nippon Kagaku Kaishi, 2000, 167.
[ 9] T. Shimo, K. Somekawa, CRC Handbook of Organic Photochemistry and Photobiology, Second edition, edit. by W. Horspool, E. Lenci, CRC Press (2003), Chap.82-1.
 Y. Odo, T. Shimo, K. Hori, K. Somekawa, Bull. Chem. Soc. Jpn., 77, 1209 (2004).
 H. I. Omar, T. Shimo, K. Somekawa, J. Mol. Struct.(Theochem), 763, 115 (2006).
 T. Shimo, K. Sato, W. Wang, T. Obata, T. Iwanaga, T. Shinmyozu, K. Somekawa, Bull. Chem. Soc. Jpn., 81, 894 (2008).
 J. J. P. Stewart, MOPAC2002 Manual, Fujitsu Limited (2001).
 D. Tokunaga, T. Shimo, H. Hashimoto, T. Ooto, K. Somekawa, J. Comput. Chem. Jpn., 6, 283 (2007).
 K. Hori, S. Yamasaki, Keisan-Kagaku-Jikken, Maruzen (1995), 43.
 J. J. P. Stewart, Computational Chemistry-MOPAC Home Page.
 S. Sakaki, Kagaku-Sousetsu, No.47, Gakkai-Shuppan Center (2000), 179.
 S. Tokita, K. Somekawa, Ryoshi-Kagaku no Kiso, Shokabo (2005), 147.
 H. Sakiyama, A. Kazama, S. Suzuki, Y. Nishida, J. Comput. Chem. Jpn., 7, 27 (2008).
 M. C. Pirrung, A. B. Bleecker, Y. Inoue, F. I. Roseifuwz, N. Sugawara, T. Wada, Y. Zou, B. M. Binder, Chemistry & Biology, 15, 313 (2008).