© 2005 Nature Publishing Group

Catalytic enantioselective reactions driven byphotoinduced electron transferAndreas Bauer1, Felix Westkamper1, Stefan Grimme2 & Thorsten Bach1

Photoinduced electron transfer is an essential step in the conver-sion of solar energy into chemical energy in photosystems I and II(ref. 1), and is also frequently used by chemists to build complexmolecules from simple precursors2. During this process, lightabsorption generates molecules in excited electronic states thatare susceptible to accepting or donating electrons. But althoughthe excited states are straightforward to generate, their shortlifetimes makes it challenging to control electron transfer andsubsequent product formation — particularly if enantiopureproducts are desired. Control strategies developed so far usehydrogen bonding, to embed photochemical substrates inchiral environments3 and to render photochemical reactionsenantioselective through the use of rigid chiral complexingagents4. To go beyond such stoichiometric chiral informationtransmission, catalytic turnover is required5. Here we present acatalytic photoinduced electron transfer reaction that proceedswith considerable turnover and high enantioselectivity. By usingan electron accepting chiral organocatalyst that enforces a chiralenvironment on the substrate through hydrogen bonding, weobtain the product in significant enantiomeric excess (up to70%) and in yields reaching 64%. This performance suggeststhat photochemical routes to chiral compounds may find use ingeneral asymmetric synthesis.In a catalytic photochemical reaction, the catalyst acts as an

antenna collecting the light and transferring it to the substrate viasensitization. Sensitization can occur by energy or electron transfer.The most successful enantioselective sensitizers rely on chiralitytransfer in a conformationally restricted exciplex6. The best resultsreported for a unimolecular reaction are 77% enantiomeric excess(77% e.e.; 20mol% catalyst) on an analytical scale7 and for abimolecular reaction 58% e.e. (15mol% catalyst) at a productyield of 1% (ref. 8).

The reaction we have devised for studying an enantioselectivephotoinduced electron transfer (PET) sensitization is depicted inFig. 1. It is based on previously reported PET catalysed conjugateadditions of a-amino alkyl radicals to enones that had beenperformed non-enantioselectively9,10. It proceeds from substrate 1(see Supplementary Scheme a and Supplementary Information pageSI 4) to a chiral spirocyclic pyrrolizidine, which in an achiralenvironment is obtained as a mixture of 2 and its enantiomer,ent-2. The simple diastereoselectivity of the reaction is perfect. Onlyone diastereoisomer is formed, the configuration of which wasproved by 1H-NMR nuclear Overhauser effect (NOE) experiments.In the presence of a catalyst, ultraviolet irradiation induces a PET

from the amine to the photoexcited catalyst. Subsequent protonloss from the intermediate cation radical presumably leads to ana-aminoalkyl radical (see also Fig. 3), which adds intramolecularly tocarbon atom C-4 of the quinolone. After the radical additionreaction, back electron transfer from the catalyst generates anenolate, which is eventually protonated to yield the products. Asmentioned above, similar intramolecular9 and intermolecular10

addition reactions to enones are known. A suitable PET catalyst forthese reactions was 4,4

0-dimethoxybenzophenone (3, Fig. 2). By

employing ketone 3 as a catalyst (10mol%), the desired reaction1 ! 2/ent-2 proceeded in good yield (Table 1, entry 1) but of coursewithout any enantioselectivity. The chiral catalyst 4 that we employedfor enantioselective reactions has two key elements. First, it allows forbinding of the substrate 1 by two hydrogen bonds at the bridgeheadlactam. Second, it contains the catalytic benzophenone unit, which isbound to the 1,5,7-trimethyl-3-azabicyclo[3.3.1]nonan-2-one via arigid oxazole. It therefore serves not only as a PET catalyst, but also asa stereocontrolling device inducing the desired enantiofacial differ-entiation in the cyclization step. Indeed, more flexible catalysts werefar less successful in the attempted enantioselective reaction. Thesynthesis of compound 4 and its enantiomer ent-4 was straight-forward, based on our previous work (see Supplementary Scheme band Supplementary Information pages SI 6–8). Experiments wereconducted varying the catalyst loading and the substrate concen-tration. The results are summarized in Table 1.Even with only 5mol% catalyst, a reasonable product yield of 61%

was obtained (Table 1, entry 2). This gives a calculated turnovernumber (12.2) that is unprecedented when considering that theproduct was formed with significant enantiomeric excess (20%).Upon raising the catalyst concentration, the reaction time decreasedand the enantioselectivity increased (entries 3, 5, 6). The enantio-meric excess reached 70% for a catalyst loading of 30mol%, whichgave a calculated turnover number of 2.1 (entry 6). Note that theseturnover numbers have not been corrected for any uncatalysedprocesses, even though a racemic background reaction evidentlyoccurs and causes the decrease of e.e. upon decreasing the amount ofcatalyst (entries 6, 5, 3, 2). In fact, when quinolone 1 was irradiated in

LETTERS

Figure 1 | PET-catalysed cyclization of the prochiral substrate 1 to the chiralpyrrolizidine 2 and its enantiomer ent-2.

1Lehrstuhl fur Organische Chemie I, Technische Universitat Munchen, Lichtenbergstr. 4, D-85747 Garching, Germany. 2Universitat Munster, Organisch-Chemisches Institut,Corrensstr. 40, D-48149 Munster, Germany.

Vol 436|25 August 2005|doi:10.1038/nature03955

1139

© 2005 Nature Publishing Group

the absence of a PET catalyst, the compound decomposed and25% of racemic pyrrolizidine 2/ent-2 was isolated after 5 h. Theuncatalysed reaction probably involves radical chain reactions asproposed in similar cyclizations10, and is increasingly suppressed inthe presence of increasing amounts of photocatalyst 4. (In theunlikely event that the background reactions occurred to the samedegree in the presence of 4, the corrected turnover number for entry 2would be 7.2.)If the other antipode of the catalyst was employed, the absolute

configuration of the product was expectedly the opposite, that is,enantiomer ent-2 prevailed (entry 4). The product configuration wastentatively assigned on the basis of known radical processes thatoccur in the presence of chiral complexing agent 5 (refs 11, 12). Forcomparison. a reaction was conducted in the presence of stoichio-metric amounts of complexing agent 5 (entry 7), which has the sameabsolute configuration as catalyst ent-4. The product enantiomerent-2 was formed preferentially. Proof for the assignment wasobtained by comparing the calculated specific rotation ([a]D) andcircular dichroism (CD) spectra with experimental values13,14. Anenantiomerically pure compound with the given structure 2 wascalculated to show a negative specific rotation with negative Cottoneffects at 270, 235 and 210 nm and a positive Cotton effect at 255 nm(see Supplementary Information pages SI 10–13). Experimentally,enantiomer 2 obtained from the reactions with catalyst 4 waslaevorotatory and exhibited the predicted Cotton effects in the CDspectra. The facial differentiation is explained in Fig. 3, assumingradical 6 as intermediate.Mechanistic details of the PET catalysed reaction have not yet been

elucidated.The single steps of the catalytic cycle will be studied infurther experiments. The hydrogen atom transfer, for example, mayin fact occur directly from the suggested ketyl radical and not via theproposed back electron transfer. The concept of chirality multipli-cation via a hydrogen-bonded catalyst like 4 should be successfullyapplicable to other photochemical reactions. Work along these linesis currently in progress in our laboratory.

Received 5 April; accepted 27 June 2005.

1. Nelson, N. & Ben-Shem, A. The complex architecture of oxygenicphotosynthesis. Nature Rev. Mol. Cell Biol. 5, 971–-982 (2004).

2. Schmoldt, P., Rinderhagen, H. & Mattay, J. in Molecular and SupramolecularPhotochemistry Vol. 9 (eds Ramamurthy, V. & Schanze, K. S.) 185–-225(M. Dekker, New York, 2003).

3. Grosch, B. & Bach, T. in CRC Handbook of Organic Photochemistry and Photobiology(eds Horspool, W. & Lenci, F.) 61/1–-61/14 (CRC Press, Boca Raton, 2004).

4. Bach, T., Bergmann, H., Grosch, B. & Harms, K. Highly enantioselective intra-and intermolecular [2þ2] photocycloaddition reactions of 2-quinolonesmediated by a chiral lactam host: Host-guest interactions, productconfiguration, and the origin of the stereoselectivity in solution. J. Am. Chem.Soc. 124, 7982–-7990 (2002).

5. Noyori, R. Asymmetric Catalysis in Organic Synthesis Ch. 1 (Wiley, New York, 1994).6. Inoue, Y. in Molecular and Supramolecular Photochemistry Vol. 11 (eds Inoue, Y.

& Ramamurthy, V.) 129–-177 (M. Dekker, New York, 2004).7. Hoffmann, R. & Inoue, Y. Trapped optically active (E)-cycloheptene generated by

enantiodifferentiating Z-E photoisomerization of cycloheptene sensitized bychiral aromatic esters. J. Am. Chem. Soc. 121, 10702–-10710 (1999).

8. Asaoka, S., Wada, T. & Inoue, Y. Microenvironmental polarity control ofelectron-transfer photochirogenesis. Enantiodifferentiating polar addition of1,1-diphenyl-1-alkenes photosensitized by saccharide naphthalenecarboxylates.J. Am. Chem. Soc. 125, 3008–-3027 (2003).

9. Jeon, Y. T., Lee, C.-P. & Mariano, P. S. Radical cyclization reactions of a-silylamine a,b-unsaturated ketone and ester systems promoted by single electrontransfer photosensitization. J. Am. Chem. Soc. 113, 8847–-8863 (1991).

10. Bertrand, S., Hoffmann, N. & Pete, J.-P. Highly efficient and stereoselectiveradical addition of tertiary amines to electron-deficient alkenes—Application tothe enantioselective synthesis of necine bases. Eur. J. Org. Chem., 2227–-2238(2000).

11. Bach, T., Aechtner, T. & Neumuller, B. Enantioselective Norrish-Yangcyclization reactions of N-(q-oxoalkyl)substituted imidazolidinones in solutionand in the solid state. Chem. Eur. J. 8, 2464–-2475 (2002).

12. Aechtner, T., Dressel, T. & Bach, T. Hydrogen bond mediatedenantioselectivity of radical reactions. Angew. Chem. Int. Edn Engl. 43,5849–-5851 (2004).

13. Polavarapu, P. L. Optical rotation: Recent advances in determining the absoluteconfiguration. Chirality 14, 768–-781 (2002).

14. Diedrich, C. & Grimme, S. Systematic investigation of modern quantumchemical methods to predict electronic circular dichroism spectra. J. Phys.Chem. A 107, 2524–-2539 (2003).

15. Bergmann, H., Grosch, B., Sitterberg, S. & Bach, T. An enantiomerically pure1,5,7-trimethyl-3-azabicyclo[3.3.1]nonan-2-one as 1H-NMR shift reagent for theee determination of chiral lactams, quinolones, and oxazolidinones. J. Org.Chem. 69, 970–-973 (2004).

Supplementary Information is linked to the online version of the paper atwww.nature.com/nature.

Acknowledgements This work was supported by the DeutscheForschungsgemeinschaft and by the Fonds der Chemischen Industrie. We thankO. Ackermann for technical assistance and T. Straßner for discussions.

Author Contributions A.B. and F.W. contributed equally to the experimentalpart of the study. S.G. performed the DFT calculations.

Author Information Reprints and permissions information is available atnpg.nature.com/reprintsandpermissions. The authors declare no competingfinancial interests. Correspondence and requests for materials should beaddressed to T.B. (thorsten.bach@ch.tum.de).

Table 1 | Enantioselective catalytic PET reactions of substrate 1 (seeFigs 1 and 2)

Entry no Catalyst Equiv.* Time (h) Product e.r.† e.e.‡ (%) Yield§ (%)

1 3 0.1 3.5 2/ent-2 50/50 — 712 4 0.05 5 2 60/40 20 613 4 0.1 2.5 2 69/31 38 554 ent-4 0.1 3 ent-2 31/69 38 525 4 0.2 2 2 77/23 54 576 4 0.3 1 2 85/15 70 647 3/5k 0.1/1.2 2 ent-2 14/86 72 39

*The reactions were carried out in deaerated toluene as the solvent at 260 8C (irradiationsource: Orginal Hanau TQ 150) and with a substrate concentration of 4 mM (seeSupplementary Information page SI 5).†The enantiomeric ratio (e.r.) was determined by 1H-NMR shift experiments (seeSupplementary Information page SI 9)15.‡The enantiomeric excess (e.e.) was calculated from the e.r. based on the uncertainty of the1H-NMR integration, the variance of e.e. data are estimated as ^2%.§Yield of isolated product.kA stoichiometric amount (1.2 equiv.) of the chiral complexing agent 5 was added to thereaction mixture.

Figure 3 | Explanation for the facial differentiation in thePET-catalysed cyclization of the prochiral substrate 1 via radicalintermediate 6.

Figure 2 | Structures of the achiral PET-catalyst 3, of the chiral enantio-meric PET-catalysts 4 and ent-4, and of the chiral complexing agent 5.

LETTERS NATURE|Vol 436|25 August 2005

1140