摘要:

New methodologies for the synthesis of compound libraries Sh�u Kobayashi Graduate School of Pharmaceutical Sciences The University of Tokyo Hongo Bunkyo-ku Tokyo 113-0033,† and Department of Applied Chemistry Faculty of Science Science University of Tokyo (SUT) Kagurazaka Shinjuku-ku Tokyo 162 Received (in Cambridge) 15th April 1998 New methodologies for the synthesis of compound libraries are discussed. They are based on new carbon–carbon bondforming reactions on the solid-phase (using polymersupported silyl enol ethers) library preparation using polymer catalysts and multi-component reactions in liquid phase. Development of new linker resins for efficient reactions on the solid-phase and a new method for the synthesis of monosaccharide libraries are also described.1 Introduction In modern organic synthesis research efforts have been made to pursue ‘efficiency’ as a key goal. There are many kinds of ‘efficiency’ and each research field has its own unique efficiency. For example ‘efficiency’ in natural product synthesis of complex molecules is a total ‘efficiency’ throughout all steps in order to prepare target molecules. In the development of new synthetic methodologies development of new reactions with high yields and high selectivities is believed to lead to ‘efficiency’ and research efforts have been made to achieve these. On the other hand combinatorial chemistry is now of interest mainly in the field of drug discovery. There are many reports that ‘remarkably large numbers of compounds can be prepared by using the mix & split method’ or ‘drug discovery processes have been dramatically shortened’ which seem to have little relationship to academic basic research.Are these works only † Present address. Sh�u Kobayashi was born in 1959 in Tokyo Japan. He studied chemistry at the University of Tokyo and received his PhD in 1988 (Professor T. Mukaiyama). After spending eleven years at Science University of Tokyo (SUT) he moved to the Graduate School of Pharmaceutical Sciences the University of Tokyo in 1998. His research interests include development of new synthetic methods development of novel catalysts (especially chiral catalysts) organic synthesis in water solid-phase organic synthesis total synthesis of biologically interesting compounds and organometallic chemistry.He received the first Springer Award in Organometallic Chemistry in 1997. 59 54 D improvements of technology? The author’s answer is no. The author thinks that an important task which should be solved by organic chemists is how large numbers of structurally distinct molecules can be prepared. Synthesis of large numbers of compounds is needed not only in drug discovery but also in the development of new materials functionalized compounds catalysts ligands etc. In these fields large numbers of compounds are synthesized first and among them compounds which are appropriate for a particular purpose are selected. Is it possible to prepare large numbers of structurally distinct compounds by traditional technologies in organic synthesis? The answer is of course ‘yes’ and such syntheses have been done.However a problem is whether they are truly efficient or not. Recent advances in the field of organic synthesis provide many reactions with high yields and high selectivities as described above and therefore it is now possible to synthesize a great number of compounds by the combination of ‘efficient’ reactions using man-power and time. However why can you call such synthesis efficient? The author thinks new methodologies are needed to prepare compound libraries just as new methodologies for natural product synthesis were required forty years ago (Scheme 1). In this paper recent results of our group on the development of new methodologies for the synthesis of compound libraries based on the above idea especially from a standpoint of synthetic organic chemistry are described.Natural Product Synthesis Methodology for Natural Product Synthesis Library Synthesis Methodology for Library Synthesis Scheme 1 Methodology for library synthesis. 2 Development of carbon–carbon bond-forming reactions in solid-phase Reactions on the solid-phase provide an important method for the synthesis of large numbers of compounds because procedures are very simple in most cases and application to automation systems such as solid-phase peptide synthesis is easy.1 On the other hand while solid-phase peptide and nucleic acid syntheses which basically include condensation protection and deprotection have reached a certain level of completion general synthetic organic reactions on the solidphase which deal with various types of reactions have difficulties such as low reactivities of polymer-supported reagents.We think that development of basic carbon–carbon bond forming reactions on the solid-phase is needed in order to Chem. Soc. Rev. 1999 28 1–15 1 CH3COSK DMF LiBH4 Et2O SH Cl R1R2CHCOSK DMF R1R2CHCOCl Et3N CH2Cl2 O R2 S R1 2.2 Imino aldol reactions 4 R1 = Me 5 R1 = OBn Scheme 2 Synthesis of polymer-supported silyl enol ethers. utilize organic reactions on the solid-phase for library construction. We first plan to immobilize silyl enol ethers on a polymer which are isolable enolates and versatile reagents for carbon– carbon bond formation.2.1 Preparation of polymer-supported silyl enol ethers Silyl enol ethers are versatile reagents in organic synthesis. They are used as isolable enolate equivalents and many kinds of useful reactions using silyl enol ethers have been developed. Polymer-supported silyl enol ethers (thioketene silyl acetals) were prepared according to Scheme 2.2 Chloromethyl copoly- (styrene-1%-divinylbenzene) resin (1.15 mmol g21) was treated with potassium thioacetate in DMF. Formation of thioester 1 was indicated by IR spectra showing strong carbonyl stretching vibration at 1693 cm21. A chlorine titration showed that 1 was obtained in a 95% yield. Thioester 1 thus obtained was then combined with TMSOTf and triethylamine in dichloromethane to afford silyl enol ether 2.Similarly silyl enol ethers 6 and 7 were prepared from 4 and 5 respectively. They were alternatively synthesized from thiol 3 according to Scheme 2. Thiol 3 was prepared by reducing 1 using LiBH4 in Et2O at room temperature. According to this scheme various types of polymer-supported silyl enol ethers can be prepared. Silyl enol ethers thus prepared were tested in the reaction with imines. It was reported that silyl enol ethers reacted with imines in the presence of a Lewis acid to afford b-amino ketones esters and thioesters.3 Since reduction of the adducts gives the corresponding amino alcohols these reactions using the polymer-supported silyl enol ethers provide a new method for the preparation of an amino alcohol library.The reaction of 2 (0.88 mmol g21) with N-benzylideneaniline was chosen as a model and several Lewis acids were tested. Although typical Lewis acids such as TiCl4 SnCl4 and BF3·OEt2 gave poor results catalytic amounts of new types of Lewis acids such as Sc(OTf) Several imines and polymer-supported silyl enol ethers were then screened using Sc(OTf)3 as a catalyst and the results are shown in Scheme 3. Although the reaction conditions have not 3 Hf(OTf)4 gave better results. Chem. Soc. Rev. 1999 28 1–15 2 O OSiMe3 Me3SiOTf S S Et3N CH2Cl2 1 2 3 OSiMe3 R2 Me3SiOTf S Et3N CH2Cl2 OSiMe3 S R1 3 6 R1 = Me 7 R1 = OBn R3 N R1 Sc(OTf) Sc(OTf) (10 mol%) (10 mol%) H R2 O LiBH LiBH CH CH2Cl2 rt 20 h 2Cl2 rt 20 h NHR3 4 NHR3 R2 R1 Et Et2O rt rt HO R2 R1 Yield (%) Yield (%) R1 S 6H11 11 6H11 11 R2 Ph Ph 2-furyl 2-furyl c-C c-C p-ClPh 2-furyl 2-furyl Ph Ph 2-furyl 2-furyl c-C c-C p-ClPh Ph Ph 2-furyl 2-furyl 2-thiophene 2-thiophene H H H H H Me Me Me Me Me Me Me Me BnO BnO BnO BnO BnO BnO BnO BnO 65 65 65 65 68 68 52 52 67 67 78 78 68 68 77 77 8 67 67 77 77 71 71 79 79 c-C c-C6H11 11 R3 Ph Ph Ph Ph Ph Ph p-MeOPh p-MeOPh Ph Ph Ph Ph Ph Ph p-MeOPh Ph Ph Ph Ph Ph Ph Ph Ph Scheme 3 The reactions of polymer-supported silyl enol ethers with imines.yet been optimized the desired amino alcohols were obtained after reduction in good yields.2.3 Mannich-type three-component reactions Multiple-component reactions which have been focused on by us4 and other groups,5–7 provide one of the most efficient methods for the synthesis of libraries. One of the characteristic points of a method utilizing multiple-component reactions is that high yields are expected by designing the reactions while yields are lower in linear synthetic strategies with multi-step syntheses. This is especially the case in solid phase synthesis because yields are often lower and characterization of the products in each step is generally difficult. Mannich-type three-component reactions on solid phase were successfully carried out using Sc(OTf)3 as a catalyst.8 An example is the reaction of benzaldehyde anisidine and polymer-supported silyl enol ether 6 (Scheme 4).The reaction OSiMe3 3 S 6 2 + O NHPMP Ph LiBH4 Sc(OTf) (10 mol%) NHPMP Ph PhCHO + PMPNH S HO 8 PMP = p-MeOPh Scheme 4 Three-component reactions on solid phase. proceeded smoothly in the presence of a catalytic amount of Sc(OTf)3 to afford after reductive cleavage from the support an amino alcohol in 87% yield. It should be noted that the yield was improved by employing the three-component reaction (the reaction of N-benzylideneaniline with 6 gave the amino alcohol in a 64% yield under the same reaction conditions).2 It was also found that adduct 8 was converted to b-amino acid and b-lactam respectively according to Scheme 5.O NHPMP Ph HO 2. H2O 8 O Hg(OCOCF 1. NaOMe–MeOH–THF 3)2 81% PMP N Ph 74% CH2Cl2–acetone 2.4 ‘Field synthesis’8 PMP = p-MeOPh Scheme 5 Conversion to b-amino acid and b-lactam. The amino alcohol library synthesis was carried out using ‘Field Synthesis’ which was based on the above three-component reaction. Four aldehyde ‘fields’ were set and in each field three amines and four polymer-supported silyl enol ethers (PSSEEs) were employed. The reactions were performed as follows in the presence of 10 mol% of Sc(OTf)3 and Drierite (80 mg) an aldehyde (0.24 mmol) and an amine (0.24 mmol) were stirred for 1 h at rt and then a PSSEE (0.20 mmol) was added and the mixture was stirred for 20 h.After saturated aq. NaHCO3 was added to quench the reaction the polymer was filtered washed with water water–dioxane (1:1) dioxane and ether successively and dried. The resulting polymer was combined with LiBH4 (5 equiv.) in THF (4 ml) and the mixture was stirred for 12 h at rt. After the usual work up the crude product was chromatographed on silica gel to afford a pure amino alcohol. The results are shown in Schemes 6 and 7. In every aldehyde field (Fields 1–4) combination of PSSEEs 2 6 7 9 and amines R1 Sc(OTf)3 (10 mol%) R3CHO + R2NH2 + NHR2 LiBH4 HO R3 R1 OSiMe3 R1 S R2NH2 R3CHO Amine PSSEE Aldehyde I 2 R1 = H R2 = PhNH2 A R3 = Ph S 2 II R2 = p-ClPhNH 6 R1 = Me R3 = B O III R2 = p-MeOPhNH2 7 R1 = OBn R3 = 9 R1 = (CH2)6CH3 C S 14 III 58 84 87 79 85 88 OSiMe3 a a 14 70 85 78 81 86 82 II II D R3 = c-C6H11 Amine Amine III 23 92 82 65 72 81 91 16 I I 9 7 2 9 6 2 7 PSSEE 6 PSSEE 74 84 87 55 80 76 Field 1.PhCHO Field a a Field 2. 2-Furaldehyde Field a a 72 78 69 56 49 78 II II Amine III Amine III 65 83 83 68 45 16 10 I I 47 b 9 7 2 9 7 2 6 PSSEE 6 PSSEE Field 4. c-C6H11CHO Field The reaction was not performed. Field 3. Thiophene-2-carbaldehyde Field a b 67 59 50 64 Amine The yield was improved to 72% when two equivalents of the aldehyde and the amine were used. Scheme 6 ‘Field synthesis’. In each ‘field’ one of three components is fixed.The number in each column shows the yield (%) of the threecomponent reaction. III 75 52 OSi tBuMe2 S II 51 56 54 84 98 71 I PSSEE 10 D B A C Aldehyde Field 5. PSSEE 10 Field Scheme 7 PSSEE is fixed. Chem. Soc. Rev. 1999 28 1–15 3 S 6 b a d OH a PhCHO d OSiMe3 R2 Ph e b C6H13CHO c f CHO 2.5 Aldol reactions c HO 82% 66% 56% Chem. Soc. Rev. 1999 28 1–15 gave the corresponding amino alcohols in satisfactory yields while lower yields were observed in the reactions using PSSEE 2. Several reaction conditions were then examined in the model combination of benzaldehyde aniline and PSSEE 2. It was finally found that the desired amino alcohol was obtained by using tert-butyldimethylsilyl enol ether 10 instead of 2.A PSSEE 10 field was next set (Field 5) and four aldehydes and three amines were employed. The results are shown in Scheme 7. This time the desired amino alcohols were obtained in high yields in all combinations and a 48 amino alcohol library was successfully prepared. The aldol reaction of silyl enol ethers with aldehydes (Mukaiyama aldol reaction)9 is known as one of the most important and fundamental carbon–carbon bond-forming reactions. Although the original report requires a stoichiometric amount of TiCl4 to promote the reaction we found that a catalytic amount of Sc(OTf)3 accelerates this reaction efficiently.10 In the aldol reaction of polymer-supported silyl enol ethers with aldehydes it was also revealed that Sc(OTf)3 effectively catalyzed the reaction.Silyl enol ether 6 reacted with benzaldehyde in dichloromethane at 278 °C to afford the corresponding aldol adduct which was reduced with LiBH4 to give the 1,3-diol. The yield was determined to be 82% based on the loading level of the silyl enol ether. As polymer-supported silyl enol ethers are prepared from thiol 3 preparation of a 1,3-diol library is performed starting from 3. An example is shown in Scheme 8 using two acid chlorides and six typical aldehydes including aromatic aliphatic a,b-unsaturated and heterocyclic aldehydes. In all cases the reactions proceeded smoothly to afford the corresponding 1,3-diols in good yields.11 O Scheme 8 1,3-Diol library based on aldol reactions.4 1. R1COCl; 2. Me3SiOTf–Et3N 1. R2CHO Sc(OTf) 2. LiBH4 f e CHO 55% 79% CHO S 66% CHO SH 3 OSiMe3 OBn S 7 3 (20 mol%) b f a d OH c HO R2 OBn e CHO a PhCHO d 77% 56% Ph e b 70% 80% CHO S C6H13CHO CHO c f CHO O While 1,3-diols are successfully cleaved from the support by treatment with LiBH4 it is also possible to produce b-hydroxy aldehydes or b-hydroxy carboxylic acids directly (Scheme 9). 72% OH O DIBAL-H Ph H CH2Cl2 -78 °C 19 h 73% (based on 3) OH 68% O S Ph OH O 1M NaOH–Dioxane (1:4) Ph HO 100 °C 6 h 59% (based on 3) Scheme 9 Conversion to b-hydroxy aldehyde or b-hydroxy carboxylic acid. When a-substituted silyl enol ethers were used the desired aldol reactions proceeded smoothly.On the other hand a lower yield was observed when an a-unsubstituted silyl enol ether was used in the reaction with benzaldehyde. In order to improve the yield several reaction conditions were examined and finally it was found that the yield was improved when tert-butyldimethylsilyl enol ether 10 was used (Scheme 10). Thus the aldol reaction of polymer-supported silyl enol ethers with aldehydes a basic carbon–carbon bond-forming reaction on solid phase has been successfully carried out using Sc(OTf)3 as a catalyst. It is noted that on solid phase the Lewis acid-catalyzed aldol reaction is superior to the aldol reaction of zinc enolates with aldehydes under basic conditions. Thus polymer-supported silyl enol ether 11 which was prepared from OTBS RCHO S CH Sc(OTf)3 (20 mol%) 2Cl2 -78 °C 10 O OTBS OTBS LiBH4 S R Et2O rt R HO TBS = tBuMe2Si RCHO Yield (%) PhCHO 69 a CHO 91 b Ph 95 c CHO S Containing 1,3-diols; a 9%; b 13%; c 14%.Scheme 10 Aldol reactions of 10 with aldehydes. 3 and hydrocinnamoyl chloride reacted with p-anisaldehyde followed by reduction with LiBH4 to afford 1,3-diol 12 in a 72% yield based on the starting chloromethyl resin (76% yield based on 11). An aldol reaction using a zinc enolate on solid phase was recently reported to give the same diol (12) in a 26% yield (Scheme 11).12 SH 3 O Ph Cl 2. Me3SiOTf–Et3N 1. Et3N S 11 OH HO 1. p-MeOPhCHO Sc(OTf)3 (20 mol%) 2.LiBH4 OSiMe3 Ph OMe Ph 76% 2.6 Michael reactions 12 Scheme 11 Effective synthesis of 12. Michael reactions of silyl enol ethers with a,b-unsaturated ketones are one of the most useful carbon–carbon bond-forming reactions in organic synthesis. While a stoichiometric amount of TiCl4 was used in the original liquid-phase reactions,13 it was found that a catalytic amount of Sc(OTf)3 10 was effective for the solid-phase Michael reactions of PSSEEs with a,b-unsaturated ketones.14 While the 1,5-dicarbonyl compound was obtained in a 38% yield in the model reaction of PSSEE 10 with chalcone using a stoichiometric amount of TiCl4 the yield was improved to 93% using 20 mol% of Sc(OTf)3 as a catalyst in the same reaction. In addition to the improvement of the yield it should 4- be noted that Sc(OTf)3 was easily removed from the product resins by filtration after the reaction because Sc(OTf)3 is soluble in water while the insoluble titanium residue which appeared after quenching the reaction by adding water in the TiCl mediated reaction was often difficult to remove and would contaminate the product resins.Several examples of the Michael reactions on solid-phase are shown in Scheme 12. Not only acyclic but also cyclic a,bunsaturated ketones reacted with PSSEEs smoothly to afford the corresponding adducts in high yields. 2.7 Aldol-type reactions Aldol-type reactions of PSSEEs with acetals have been successfully carried out using Yb(OTf)3 as a catalyst (Scheme 13).14 The reactions were performed at rt and the adducts were cleaved from polymer supports using LiBH4 to give 1,3-diol monoethers.The SR-MAS NMR technique was also quite effective in developing the reactions. 2.8 5-(4A-Chloromethylphenyl)pentylpolystyrene resin (CMPP resin). A new linker resin for solid-phase organic synthesis under Lewis acidic conditions The proper choice of supports and linker groups are among the most important factors in the success of organic synthesis on solid supports. Although several linkers have already been developed these are mostly optimized for biopolymer synthesis such as peptides and oligonucleotides and unsatisfactory results are sometimes obtained in the reaction sequences possible on supports. For example almost all linkers developed contain oxygen and/or nitrogen atoms including ether ester and amide functional groups which coordinate Lewis acids to be decomposed or deactivated.15 Hence these linkers cannot be used in Lewis acid-promoted reactions which provide numerous useful transformations in liquid-phase organic synthesis.In the course of our program on the development of Lewis acidcatalyzed reactions on solid-phase we were confronted with the above problem. Carbon atoms instead of nitrogen or oxygen atoms were chosen in the new linker and methylene groups were used as a spacer. The synthetic scheme of the new linker 5-(4Achloromethylphenyl) pentylpolystyrene resin (CMPP resin 16) is shown in Scheme 14.16 Friedel–Crafts acylation of copoly- (styrene-1%-divinylbenzene) resin with 5-phenylvaleryl chloride was carried out using aluminum chloride in carbon disulfide.The resulting acylated resin 17 was reduced using AlCl3–LAH in ether to afford 5-phenylpentyl resin 18. Finally 18 was chloromethylated under standard conditions to afford CMPP resin (16). The evaluation of the new linker resin was carried out by the imino aldol reactions of polymer-supported silyl enol ethers with imines.3 The polymer-supported silyl enol ethers of CMPP resin were prepared according to Scheme 15. The results of the imino aldol reactions of the silyl enol ethers derived from the polymer-supported thioacetate (19) with N-benzylideneaniline are summarized in Scheme 16. CMPP resin gave higher yields than Merrifield and Wang resins.The loading levels of the enolate moieties in CMPP resin were also examined and it was found that the best results were obtained by using CMPP resin having a 0.96 mmol g21 loading level which was prepared by using 2.0 mmol g21 of the acylating reagent in the Friedel– Crafts acylation (Scheme 15). Several examples of the Sc(OTf)3-catalyzed imino aldol reactions using CMPP resin were tested and the results are shown in Scheme 17. In all cases the yields of the desired adducts (amino alcohols) using CMPP resin were much higher (ca. 10–30%) than those using Merrifield resin. These results indicate the effectiveness of the new linker resin. 5 Chem. Soc. Rev. 1999 28 1–15 OSiR3 O R1 Sc(OTf)3 (20 mol%) S + R3 R2 CH2Cl2 -78 °C 20 h O R2 O S R1 Entry PSSEE S 1 10 a 2 10 3 10 10 4 5 OSiMe OSi tBuMe2 3 10 S 7 b 6 7 OBn 3 7 S 6 c 8 OSiMe OSiMe 3 S 9 2 d a0.94 mmol g–1.b0.88 mmol g–1. c0.96 mmol g–1. d0.76 mmol g–1. Scheme 12 Michael reactions of PSSEE with a,b-unsaturated ketones. Reagents and conditions i 1) NaOMe (10 equiv.) THF MeOH (4 1); 2) IRC-76; 3) Me3SiCl MeOH. 3 Use of the swollen-resin magic angle spinning (SR-MAS) NMR technique to determine the structure of polymer-supported reagents and catalysts directly During the investigations to develop Michael reactions on solidphase (2.6) it was found that the magic angle spinning (MAS) NMR technique was very useful to determine the structure of resins containing products directly.In solid-phase organic synthesis characterization of products is often difficult and Chem. Soc. Rev. 1999 28 1–15 6 O R2 O i H2O R3 R3 MeO R1 Yield (%) 13 91 O 87 Ph O Ph 80 14 a,b-Unsaturated Ketone O Ph Ph O 83 O 81 60 13 75 14 64 13 13 48 typical NMR techniques used in liquid-phase organic synthesis cannot be applied in many cases. Consequently characterization is carried out at the stage of resins containing products using IR or special mass spectrometers or after cleavage of products from solid supports using standard analytical methods in liquid-phase organic synthesis (NMR IR MASS etc.). It was found that characterization of our polystyrene-supported resins containing products can be successfully carried out using the MAS NMR technique using swollen resins.14 The 13C swollen-resin MAS (SR-MAS) NMR spectrum of the resins containing the Michael adduct is shown Fig.1. NMR spectra OSiMe3 R1 OMe S + OMe R2 OMe Yb(OTf)3 (20 mol%) LiBH4 HO R2 THF rt CH2Cl2 rt R1 Acetal PSSEE OMe 7 a 15 Ph OMe OMe 7 a Ph OMe OMe 7 b Ph OMe OMe 7 b OMe 6 c 15 2 d 15 Entry 1 2 3 4 5 6 a0.88 mmol g–1. b0.66 mmol g–1. c0.96 mmol g–1. d0.76 mmol g–1. Scheme 13 Aldol-type reactions of PSSEE with acetals. O O Ph Cl AlCl3 CS2 17 AlCl3 LiAlH4 Et2O 18 SnCl4 MOMCl Yield (%) 81 86 72 63 76 38 Cl 16 MOM = H3COCH2 Scheme 14 Preparation of new linker resin 16.were recorded on a JEOL JNM-LA400 (CP/MAS System) spectrometer using a special sample tube. It is concluded from the NMR spectrum that no starting materials remain and several peaks which correspond to the adducts are observed and that the desired reaction has proceeded successfully. Actually the Michael adduct was isolated after cleavage from the support in a 93% yield. The 1H SR-MAS NMR technique was also used for the preparation of PSSEEs. In our initial work for the preparation of PSSEEs IR spectra were used for characterization of the products and determination of yields of the products. It was found that the 1H SR-MAS NMR technique was much more effective in these characterizations. In addition it was also found that the SR-MAS NMR technique provided an effective and powerful tool for determining the structure of some polymer-supported catalysts.Recent research from our laboratories has revealed remarkable solvent effects in SR-MAS NMR. The effect of typical seventeen deuterized solvents is summarized in Table 1. 4 A new method for the synthesis of monosaccharide libraries (efficient synthesis of diverse monosaccharide derivatives in the solid-phase) While there are many biologically-important compounds containing sugars monosaccharides are the smallest sugar unit and are known to play important roles in their biological activities.17 In order to obtain compounds having unique activity as well as to improve lower bioactivities it is desirable to optimize the structure of monosaccharides and therefore development of new methods for the synthesis of diverse monosaccharide derivatives is in great demand.While monosaccharides have rather simple structures they may contain four five six or seven (higher sugars) asymmetric centers and the combination of various substituents at each chiral center provides a great number of structurally-different monosaccharide derivatives. Three major methods have been reported so far for the synthesis of monosaccharides. The first method is the traditional one; that is the synthesis of rare sugars from the common sugars such as glucose mannose galactose etc.18 One drawback of this method is that it requires tedious long transformations and protection and deprotection of the hydroxy groups of the monosaccharides.The second method is to utilize stereoselective reactions of three-carbon or fourcarbon alkoxyaldehydes (glyceraldehyde or threose derivatives prepared from mannitol or tartaric acids) with carbon nucleophiles such as allylmetals or enolates19 or hetero Diels–Alder reactions.20 Finally efficient methods for the synthesis of monosaccharides from simple achiral compounds using asymmetric synthesis have been reported recently.21 While these syntheses provide useful methods for the preparation of specific rare sugars much time and man-power are required for the synthesis of a diverse monosaccharide library according to these methods. A new method for the synthesis of monsaccharide libraries was intended to be developed on the basis of aldol and imino aldol reactions on the solid-phase.2,8,11 An example the synthesis of a rare sugar 6-O-benzyl-2-deoxy-L-gulose is shown in Scheme 18.The starting material chloromethylated resin or thiol resin 3 was converted to thioester resin 1. Treatment of 1 with tert-butyldimethylsilyl triflate (TBSOTf) and triethylamine provided polymer-supported silyl enol ether (PSSEE) 10. A key reaction is the aldol condensation of 10 with a chiral aldehyde (26) using a Lewis acid promoter which proceeded smoothly to afford the desired adduct with perfect stereoselectivity. Deprotection of the TBS group of the aldol adduct (27) using tetrabutylammonium fluoride (TBAF)–acetic acid induced spontaneous lactone formation and hence cleavage from the polymer support.The yield was determined to be 61% from chloromethylated resin (4 steps) at this stage. Finally reduction of 28 with diisobutylaluminum hydride (DIBAL-H) gave 6-O-benzyl-2-deoxy-L-gulose (29) ( > 80%).22 Similarly four monosaccharide derivatives (lactones) were prepared using the combination of chiral alkoxyaldehydes and PSSEEs in the solid-phase (Scheme 19). The 2-deoxy series and 2-benzyloxy series were prepared from PSSEE 10 and 7 respectively. It is noted that the stereochemistry of the stereogenic center at the C-3 position can be controlled by choice of the protecting groups of the alkoxyaldehydes. The diversity of the monosaccharide derivatives obtained according to this protocol depends on the numbers and kinds of Chem.Soc. Rev. 1999 28 1–15 7 16 Polymer S Polymer-Supported Silyl Enol Ether a b 5.1 mmol g–1 of 5-phenylvaleryl chloride was used in Scheme 10. 2.0 mmol g–1 of 5-phenylvaleryl chloride was used in Scheme 10. alkoxyaldehydes and PSSEEs. A preparation method for PSSEEs has already been shown and various types of PSSEEs can be prepared (cf. Section 2.1). On the other hand many Chem. Soc. Rev. 1999 28 1–15 Scheme 16 Effect of linker resins. 8 TBSOTf S Et3N CH2Cl2 19 O LiBH4 Et2O CH3COSK DMF RCH2COSK DMF Me3SiOTf S Scheme 15 Synthesis of polymer-supported silyl enol ethers using novel linker resin 16. 24 R = Me 25 R = OBn OTBS 22 R = Me 23 R = OBn Ph N + Ph H Sc(OTf)3 (10 mol%) LiBH RCH2COCl Et3N CH2Cl2 R HN THF rt CH2Cl2 rt Et3N CH2Cl2 Ph Ph O 4 HO OTBS S (Merrifield) OTBS S O (Wang) S OTBS (CMPP) TBS = tBuMe2Si Yield (%) 53 57 62 67 74 71 Loading Level /mmol g–1 1.15 0.71 1.82 a 1.15 a 0.96 b 0.83 b S 20 OTBS SH 21 S 3 R1 + H R2 TBS = tBuMe2Si R3 S OSiR'3 R3 R OSiMe N HN Sc(OTf)3 (10 mol%) LiBH4 HO R2 CH2Cl2 rt R1 R1 R3 R2 SiR'3 Yield (%) (Merrifield) Yield (%) (CMPP) (53) Ph H Ph Si tBuMe 74 THF rt 2 p-ClPh (60) H Ph 72 SiMe Si tBuMe2 3 (78) Me Ph Ph 90 p-MeOPh 96 (64) Me Ph SiMe (68) 2-furyl SiMe3 3 Me Ph quant.SiMe (67) OBn Ph 3 Ph 96 p-MeOPh (66) OBn Ph 93 SiMe3 (77) 2-furyl OBn Ph 90 SiMe3 Scheme 17 Imino aldol reactions using CMPP resin. alkoxyaldehydes have been synthesized from natural sources such as mannitol and tartaric acid.19,23 Alternatively we have developed an efficient method for the synthesis of alkoxyaldehydes using asymmetric aldol reactions. Recently we have developed tin(II)-mediated highly diastereo- and enantioselective aldol reactions of silyl enol ethers with carbonyl compounds.24,25 Various types of b-hydroxy thioesters have been prepared starting from simple achiral compounds using these asymmetric reactions. After protection of the b-hydroxy groups of the aldol adducts treatment of these protected adducts with DIBAL-H gave the desired alkoxyaldehydes in excellent Fig.1 13C Swollen-resin magic angle spinning (SR-MAS) NMR.a Table 1 Effect of solvents for thioacetyl resin (1) in MAS NMR Solvent Entry 2 8 6 6 7 6 8 6 4 6 5 5 3 10 Chloroform-d Dichloromethane-d THF-d Toluene-d Benzene-d DMF-d DMSO-d Dioxane-d Acetone-d Methanol-d Ethanol-d Pyridine-d 1,2-dichloroethane-d Nitrobenzene-d Acetonitrile-d Diethyl ether-d Deuterium oxide 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 a 4.06 8.06 4.38 3.30 3.83 —b —b 2.17 —b —c —c 4.40 9.20 5.79 —c —b —c Methylene peak of thioacetyl resin. b Peaks of resin overlapped with those of solvents.c Resin’s peaks did not appear. yields with excellent diastereo- and enantioselectivities (Scheme 20). 4.1 Amino sugars 3-Amino sugars were then synthesized in the solid-phase. In nature there are many biologically-interesting compounds containing 3-amino sugars such as daunosamine acosamine ristosamine etc. An example of our solid-phase synthesis of 3-amino sugars is shown in Scheme 21. Thiol resin 3 was converted to thioester resin 5 which was silylated to give Cl CH3COSK Bu4NI S/N ratio DMF O Methyl carbon Carbonyl carbon Half-height widtha/Hz S HS CH3COCl Et3N 3 1 TBSO CHO BnO OTBS TBSO TBSOTf Et3N 26 S BF3•OEt2 CH2Cl2 4 10 O 2.31 2.25 2.82 2.39 2.82 2.11 —b 1.55 2.53 —c —c 2.62 2.81 3.45 —c 1.38 —c 55.2 60.4 67.6 71.6 74.0 82.0 87.2 106.8 116.8 —c —c 63.6 57.4 73.8 —c 129.2 —c OH CH2Cl2 S CH2Cl2 TBSO BnO TBSO 27 THF OH BnO BnO OH OH OH DIBAL-H O O O TBAF AcOH OH 28 (61% from 3) 29 TBS = tBuMe2Si Scheme 18 Solid-phase synthesis of 6-O-benzyl-2-deoxy-l-gulose (29).PSSEE 7. The key three-component reaction of an aldehyde an amine and 7 proceeded smoothly in the presence of a catalytic 9 Chem. Soc. Rev. 1999 28 1–15 O OBn OBn OBn 26 OH O OH OH CHO O O + O + O 10 OH OBn OBn 7 OTBS CHO 30 OH Ph 71% (50:50) Ph O + O OH + O 10 7 OH 61% (95:5) O OBn 3 48% (50:50) Scheme 19 Synthesis of monosaccharide derivatives (lactones) in the solidphase.OSiMe + RCHO TBSO SEt + Sn(OTf)2 + Bu2Sn(OAc)2 N N Me 63% (80:20) TBS = tBuMe2Si O OH SEt R TBSO DIBAL-H TBSCl imidazole R = ( E)-CH3CH=CH 82% syn/anti = >99/1 98% ee ( syn) R = CH3 60% syn/anti = >99/1 97% ee ( syn) O H R TBSO R = ( E)-CH3CH=CH 95%; quant R = CH3 94%; quant R = 2-furyl 86% syn/anti = >99/1 99% ee ( syn) TBSO 10 R = 2-furyl 90%; quant 4.1 Alditol derivatives Chem. Soc. Rev. 1999 28 1–15 TBS = tBuMe2Si Scheme 20 Preparation of chiral aldehydes. amount of Sc(OTf)3 to afford the corresponding adduct (30) with perfect stereoselectivity.Also in this case deprotection of the TBS group induced a spontaneous cyclization to give lactone 31 which was reduced with DIBAL-H to produce 3-amino sugar derivative 32 in an 82% yield. Similarly a 2-benzyloxy series 2-deoxy series and 2-deoxy-2-methyl series of 3-amino sugars were successfully prepared in the solid-phase (Scheme 22). The present method was successfully applied to the synthesis of alditol derivatives (the reduced forms of monosaccharides). Reductive cleavage from the support instead of deprotection of the TBS groups gave alditol derivatives in good yields (Scheme 23). Since we have already demonstrated that basic cleavage (NaOMe) from the same support affords carboxylic acids uronic acid (the oxidized forms of monosaccharides) formation would be possible simply by changing the cleavage method.HS 3 Me3SiOTf Et3N CH2Cl2 BnO BnO PMP TBSO TBSO BnO OH NHPMP OBn 31 (52% from 3) Scheme 21 Solid-phase synthesis of 32. PMP = p-MeOPh TBS = tBuMe2Si 5 Library synthesis using polymer catalysts 5.1 Quinoline derivatives Although polymer-supported substrates (reagents) have often been employed for library construction there are some disadvantages to this method. First the reactions of polymersupported reagents are sometimes slow and differences in reactivity between the substrates lead to lack of diversity of the library in some cases. Secondly the loading level of polymersupported substrates is generally low ( < 0.8 mmol g21) and large-scale preparation is difficult.To overcome these problems a new methodology for library synthesis has been developed. The new method is not using polymer-supported reagents but using polymer-supported catalysts in multicomponent reactions. An example is shown in the synthesis of a large number of quinoline derivatives. We have recently developed threecomponent coupling reactions using a lanthanide triflate as a catalyst.6 Many combinations of aldehydes amines and olefins are used in this reaction and a large quinoline library could be prepared based on these combinations. Although liquid phase combinatorial synthesis may be possible our attention was focused on reactions using a polymer catalyst. Use of polymer-supported catalysts26 offers several advantages such as simplification of product work up separation isolation reuse of the catalyst etc.and may be useful for parallel library construction. On the other hand one of the drawbacks of polymer-supported catalysts is their low reactivity. Bearing in mind that the low reactivity may be ascribed to insolubility of the catalysts we searched for a new polymersupported catalyst which is partially soluble in an appropriate solvent and is precipitated after completion of the reaction and recovered quantitatively by filtration. After several trials a new scandium catalyst polyallylscandium trifylamide ditriflate (PA-Sc-TAD) has finally been developed. The synthetic route of PA-Sc-TAD is shown in Scheme 24.Polyacrylonitrile was treated with BH3·SMe in dyglyme for 36 h at 150 °C. The resulting amine (33) was reacted with Tf2O in the presence of Et3N in 1,2-dichloroethane for 10 h at 60 °C to afford sulfonamide 34. After 34 and KH were combined Sc(OTf)3 was O BnOCH2COCl Et3N BnO S CH2Cl2 5 OSiMe3 26 + PMPNH2 Sc(OTf)3 S 7 NH O S OBn 30 BnO DIBAL-H NHPMP O O Drierite CH2Cl2 TBAF THF OH OBn O OH 32 Benzyloxy Series OH OH O O O O H3C OBn PMPHN OBn PMPHN 62% (100/0) 52% (100/0) OH OH O O O BnO O O OBn PMPHN OBn PMPHN 46% (100/0) 52% (100/0) 2-Deoxy Series OH OH O O O O H3C PMPHN PMPHN 83% (73/27) 75% (89/11) OH OH O O BnO O O O PMPHN PMPHN 70% (67/33) 64% (65/35) 2-Deoxy-2-Methyl Series OH OH O O O O H3C PMPHN PMPHN 82% (82/18) 68% (88/12) OH OH O O BnO O O O PMPHN PMPHN 52% (100/0) 73% (73/27) 1All yields are from 3 2PMP = p-MeO-Ph 3Diastereomer ratios (major/minors) are shown in parentheses.Scheme 22 Synthesis of 3-amino sugar derivatives (lactones) in the solidphase. added and the mixture was stirred in THF for 48 h at rt to give 35. PA-Sc-TAD (35) is gummy but is dispersed and partially soluble in a CH2Cl2–CH3CN mixed solvent. The dispersed catalyst assembles again when hexane is added. PA-Sc-TAD (35) is especially useful for the synthesis of a quinoline library (Scheme 25).27 The procedure is very simple; just mixing the catalyst (PA-Sc-TAD) an aldehyde an aromatic amine and an alkene (alkyne) in CH2Cl2–CH3CN (2:1) at 60 °C for 12 h.Hexane was then added and the catalyst was filtered. The filtrates are concentrated to give almost pure quinoline derivatives in most cases. It is noted that PA-Sc-TAD is watertolerant28 and that substrates having water of crystallization can be used directly. PA-Sc-TAD can be easily recovered and O H O 1. NCbz OH OH O 3 7 NCbz OBn Me2AlCl CH2Cl2 3 2. LiBH4 OSiMe S 6 2. LiBH4 NH 2Cl2 PMP TBSO OH TBSO 70% from 3 CH 1. 26 + PMPNH2 Sc(OTf)3 Drierite CH BnO CH 2 CN n 33 CH CH2NH2 2 1. KH CH CH2 2. Sc(OTf) PMP = p-MeOPh TBS = tBuMe2Si Scheme 23 Solid-phase synthesis of alditol derivatives.2 CH BH3•SMe CH 3 n n CH2NH Tf 64% from 3 (65 35) Tf2O–Et3N n CH CH2NTf Sc(OTf)2 34 PA-Sc-TAD (35) Scheme 24 Preparation of PA-Sc-TAD (35). continuous use is possible without any loss of activity (Scheme 26). A characteristic feature of the present method compared to conventional combinatorial synthetic technology using polymer-supported reagents is that more than hundred milligrams scale syntheses with a large array of diverse molecular entries have been achieved with high purities (high yields and high selectivities). The number of commercially available aromatic aldehydes aliphatic aldehydes heterocyclic aldehydes and glyoxals and glyoxylates is more than 200 and more than 200 aromatic amines and 50 alkenes (and alkynes) are commercially available.Therefore a quinoline library of more than a million compounds with high quantity and quality could be prepared by using an automation system based on this method. Moreover the tetrahydroquinoline derivatives thus obtained were easily oxidized to dihydroquinoline or quinoline derivatives which could double the size of the library. 5.2 Amino ketone amino ester and amino nitrile derivatives This new polymer catalyst method will be useful for construction of other compound libraries based on multi-component Chem. Soc. Rev. 1999 28 1–15 11 NH2 R5 R3 + R6 R4 R3 R4 R2 NH O PA-Sc-TAD SPh + Cl R1CHO Cl Ph R2 Cl Ph NH NH Cl Cl 65% (nd) HN OMe HN 84%a 83% (62/38) HN Ph Cl Cl 70% (100/0) HN 71% (100/0) HN O 78% (100/0) 80% (100/0) a quant.(100/0) NH2 + PA-Sc-TAD CH2Cl2–CH3CN (2:1) Butyl ethynyl sulfide was used as a dienophile. Scheme 25 Use of PA-Sc-TAD in the synthesis of a quinoline library. Diastereomer ratios are shown in parentheses. Relative stereochemical assignment was made by 1H NMR analysis. COPh NH smoothly at room temperature to afford the corresponding bamino ketone in a 91% yield (after column chromatography on silica gel). In this reaction the molar ratio of aldehyde amine 36 was 1 1 1.1 and no side reaction was observed. After the reaction was completed the catalyst was filtered and the filtrate was concentrated in vacuo to afford the almost pure b-amino ketone.Heterocyclic and aliphatic aldehydes and a glyoxal also worked well with various amines and 36 to give b-amino ketone derivatives in high yields (Scheme 27). The reactions using the ketene silyl acetal of methyl isobutylate (37) as a silylated nucleophile were then examined. It was expected that a b-amino ester could be produced by the reaction of cyclohexanecarbaldehyde p-chloroaniline and ketene silyl acetal 37 under standard conditions. However only a trace amount of the product was obtained after 19 h at room temperature. It was assumed that water was produced in the formation of the imine from the aldehyde and the amine and that the ketene silyl acetal was decomposed by this water leading to the low yield.Magnesium sulfate (MgSO4) was then added as a dehydrating agent and the yield was dramatically improved to afford the desired adduct in a 74% yield. Under PhCOCHO•H2O + 12 1st use 90%; 2nd use 91%; 3rd use 93% yield Scheme 26 reactions. Another example is three-component reactions between aldehydes amines and silylated nucleophiles leading to amino ketone amino ester and amino nitrile derivatives.29 When benzaldehyde aniline and the silyl enol ether of propiophenone (36) were combined in the presence of polyallylscandium trifylamide ditriflate (PA-Sc-TAD) it was found that the reaction proceeded a little slower compared to that using Ln(OTf)3 as a catalyst but that the clean reaction proceeded Chem.Soc. Rev. 1999 28 1–15 Ph Cl R5 Ph Ph NH NH R6 R1 SBu H,H Ph HN NH MeO 80% (86/14) HN Ph Cl Cl Ph 92% (95/5) HN 75% (90/10) Cl 85% (99/1) HN S 91% (100/0) Cl 96% (100/0) HN COPh 99% (100/0) OSiMe3 Ph R2 36 NH R1 MeO R2 R1 R1CHO + R2NH2 + Scheme 27 + + R1CHO R2NH2 MgSO Scheme 28 Me3SiCN PA-Sc-TAD O Ph OMe 77-95% these reaction conditions several b-amino ester derivatives were obtained in high yields (Scheme 28). OSiMe3 37 PA-Sc-TAD 4 NH O 73-89% Finally cyanotrimethylsilane (TMSCN) was used as a silylated nucleophile. The three component reactions between aldehydes amines and TMSCN proceeded smoothly in the presence of PA-Sc-TAD to afford various a-amino nitrile derivatives (Scheme 29).Three-component reactions between PA-Sc-TAD Scheme 29 R1CHO + R2NH2 + aldehydes amines and silylated nucleophiles have been successfully carried out by using a polymer scandium catalyst to afford b-amino ketones b-amino esters and a-amino nitriles in high yields. The reactions are very clean and the procedure is very easy; simply mixing the catalyst (PA-Sc-TAD) and almost equimolar amounts of an aldehyde an amine and a silylated nucleophile. After filtration the filtrates are concentrated to give almost pure products in most cases. It is noted that PA-Sc- TAD can be easily recovered and that continuous use is possible without any loss of activity. These reactions provide a useful route to large numbers of structurally distinct amino groupcontaining compounds of high quality and quantity.6 Library synthesis in liquid-phase Because of the simple experimental procedure solid-phase synthesis is now popular but some problems such as low R2 NH CN R1 83-99% reactivity of polymer-supported reagents leading to low yields low loading levels of polymer reagents which prevent largescale synthesis difficulties of characterization of polymersupported compounds etc. have been identified as mentioned in the previous section. Although these disadvantages would be overcome by liquid-phase reactions rather tedious procedures required in the work up and purification processes make their application to library construction difficult.In the course of our investigations to develop new methodologies for library construction our attention has been focused on the development of efficient multiple-component reactions in liquid-phase as well as simplification of the work-up and purification processes in these reactions. The methods reported here are based on three-component reactions of aldehydes amines and silyl enolates or alkenes leading to b-amino ester3 or quinoline derivatives.3,4 One of the necessary conditions for using liquid-phase synthesis as a method for library construction is to develop truly efficient reactions. For example in multiple-component reactions equimolar amounts of each component should react smoothly to afford the corresponding adducts in nearly quantitative yields.The Ln(OTf)3-catalyzed three-component reactions of aldehydes amines and silyl enolates or alkenes meet the criteria. The reactions proceed smoothly by using almost equimolar amounts of each component. The next step is to separate the products easily from the catalyst. The lanthanide reagent is water-tolerant and soluble in water rather than in organic solvents,28 therefore the products can be separated by simple extraction. However a simpler procedure such as filtration was required. One of the important factors necessary to achieve this goal is that the catalyst is stable during the work up process and that a process of deactivation of the catalyst for example adding water is not necessary. Ln(OTf)3 is very suitable in regard to these points while many Lewis acid catalysts such as AlCl3 TiCl4 SnCl4 etc.are decomposed during the work-up process by air or water which complicates the purification process. First reprecipitation of Ln(OTf)3 was tried. After the reaction was completed the solvent was removed under reduced pressure and hexane was added. A new precipitate was observed and was separated by filtration and then the filtrate was concentrated under reduced pressure. Unfortunately it was found that the filtrate was contaminated with a small amount of the catalyst. Several similar methods along this line were then examined. After several trials it was finally found that the catalyst was separated by directly charging onto a short column without concentration.After elution by an appropriate solvent the eluent was concentrated in vacuo to afford almost pure product. Several examples of the present method for the synthesis of b-amino esters from aldehydes amines and silyl enolates were tested and in all cases the desired b-amino ester derivatives were obtained in high yields with high purities (Scheme 30).30 OSiMe3 R3 R1CHO + R5 R4 R2 NH O R5 R1 R3 R4 80-95% (purity >90%) Scheme 30 R2NH2 + 1. cat. Yb(OTf)3 MS 4 Å 2. filtration 3. concentration The procedure is very simple. An aldehyde an amine a silyl enolate ytterbium triflate (Yb(OTf)3 10 mol%) and 4 Å molecular sieves were combined in dichloromethane. The mixture was stirred for 20 h at room temperature and then the whole reaction mixture was passed through a short column 13 Chem.Soc. Rev. 1999 28 1–15 packed with silica gel. The eluent was hexane–AcOEt (6:1 ca. 80 ml) and use of medium pressure enhanced the quick separation. The eluent was concentrated under reduced pressure to afford the products directly. Similarly tetrahydroquinoline derivatives were prepared from aldehydes amines and alkenes in high yields and high purities (Scheme 31). R5 NH2 R3 + R1CHO + R6 R4 R2 1. cat. Yb(OTf)3 MS 4 Å 2. filtration 3. concentration Scheme 31 The present method is useful for the synthesis of large numbers of b-amino esters which are versatile intermediates for the synthesis of b-amino alcohols b-amino acids b-lactams etc.and quinoline derivatives. The very simple work-up and purification procedure compared to conventional methods (Scheme 32) would make it possible to apply this method to automation systems. Although small amounts of products are generally obtained in syntheses using polymer-supported reagents large-scale preparation is possible by this method. In addition the liquid-phase reactions can solve the low reactivity low yield and characterization problems often observed on the solid-phase synthesis. The method is based on Ln(OTf) catalyzed three-component reactions and the key is the efficient reactions and the simple purification process. Because many Ln(OTf)3-catalyzed reactions have been developed,31 the present method would be useful for construction of other compound libraries.extraction 83-99% (purity >85%) 3- Reaction Mixture filtration (short column) 7 Conclusions and outlook Several methods for the synthesis of compound libraries developed by our group have been overviewed. According to these methods large quantities of single compounds (not mixtures) with high purities are prepared in most cases. In drug discovery large numbers of compounds are synthesized and among them optimized compounds are selected. This method is Conventional Methods Reaction Mixture quench (adding water) The Present Method 14 concentration Chem. Soc. Rev. 1999 28 1–15 Scheme 32 Work-up and purification processes. R3 R4 R5 R6 R1 NH R2 Organic Phase Aqueous phase 8 Acknowledgements Our work was partially supported by CREST Japan Science and Technology Corporation (JST) a Grant-in-Aid for Scientific Research from the Ministry of Education Science Sports and Culture Japan and a SUT Special Grant for Research Promotion.Ryo Akiyama and Takayuki Furuta are acknowledged for the preparation of Table 1. concentration purification filtration Pure product common in drug discovery as well as in the development of new materials functionalized compounds catalysts ligands etc. and in these cases large quantities of pure compounds are needed. Solid-phase synthesis has often been used for the preparation of compound libraries. This procedure has obvious advantages over liquid-phase synthesis in its simple experimental procedures and its effectiveness in intramolecular cyclization reactions etc.In the application to automated systems the advantages of solid-phase synthesis will be unshakeable. On the other hand there are reactions which proceed smoothly in liquid-phase but do not proceed well on the solid-phase. Most resins spacers and linkers used now on the solid-phase organic synthesis are those used on the solid-phase peptide synthesis. While condensation protection and deprotection reactions are mainly carried out in peptide synthesis modern organic synthesis requires various types of reactions using organometallics Lewis acids etc. Development of new resins spacers or linkers which are appropriate for such organic reactions is needed.Another problem is stereoselectivity. Although many highly stereoselective reactions have been developed there are still very few reactions which have perfect stereoselectivities. More than 90% diastereomeric excesses or more than 90% enantiomeric excesses are believed to be enough and this is partially true from a practical point of view because separation or purification (recrystallization or column chromatography) is easy after such highly stereoselective reactions. However separation or purification is usually difficult on the solid-phase synthesis. If a reaction proceeds in a diastereomer ratio of 95:5 the 5% impurity cannot be removed. Therefore perfect reactions in both chemical yields and diastereoselectivities are required on the solid-phase synthesis.More basic and fundamental research works on the solid-phase organic synthesis are strongly demanded. The significance of library synthesis will be increasing in the next century and the role played by synthetic organic chemists will also become more and more important. Pure product dry 9 References 1 L. A. Thompson and J. A. Ellman Chem. Rev. 1996 96 555 and 2 S. Kobayashi I. Hachiya S. Suzuki and M. Moriwaki Tetrahedron 3 S. Kobayashi M. Araki and M. Yasuda Tetrahedron Lett. 1995 36 references cited therein. Lett. 1996 37 2809. 5773 and references cited therein. 4 S. Kobayashi H. Ishitani and S. Nagayama Synthesis 1995 1195. 5 I. Ugi A. Dömling and W. Hörl Endeavour 1994 18 115. 6 R. W. Armstrong A. P. Combs P.A. Tempest S. D. Brown and T. A. 7 P. A. Tempest S. D. Brown and R. W. Armstrong Angew. Chem. Int. 8 S. Kobayashi M. Moriwaki R. Akiyama S. Suzuki and I. Hachiya Keating Acc. Chem. Res. 1996 29 123. Ed. Engl. 1996 35 640. Tetrahedron Lett. 1996 37 7783. 9 T. Mukaiyama K. Banno and K. Narasaka Chem. Lett. 1973 1101. 10 S. Kobayashi I. Hachiya H. Ishitani and M. Araki Synlett 1993 472. 11 S. Kobayashi I. Hachiya and M. Yasuda Tetrahedron Lett. 1996 37 5569. 12 M. J. Kurth L. A. Ahlberg Randall C. Chen C. Melander R. B. Miller K. McAlister G. Reitz R. Kang T. Nakatsu and C. Green J. Org. Chem. 1994 59 5862. 13 K. Narasaka K. Soai and T. Mukaiyama Chem. Lett. 1974 1223. 14 S. Kobayashi R. Akiyama T. Furuta and M. Moriwaki Molecules Online 1998 2 35.15 (a) L. F. Tietze and A. Steinmetz Angew. Chem. Int. Ed. Engl. 1996 35 651; (b) S. M. Hutchins and K. T. Chapman Tetrahedron Lett. 1996 37 4869. 16 S. Kobayashi and M. Moriwaki Tetrahedron Lett. 1997 38 4251. 17 For leading references see (a) Trends in Synthetic Carbohydrate Chemistry eds. D. Horton L. D. Hawkins and G. McGarvey ACS Symposium Series 386 Amrican Chemical Society Washington DC 1989; (b) T. Kawano J. Cui Y. Koezuka I. Toura Y. Kaneko K. Motoki H. Ueno R. Nakagawa H. Sato E. Kondo H. Koseki and M. Taniguchi Science 1997 278 1626 and references cited therein. 18 For example (a) N. Prentice L. J. Cundet and F. Smith J. Am. Chem. Soc. 1956 78 4439; (b) R. Kuhn and G. Baschang Liebigs Ann. Chem. 1960 636 164. 19 For example (a) G. J. McGarvey M. Kimura and T. Oh J. Carbohydr. Chem. 1984 3 125; (b) T. Mukaiyama K. Suzuki T. Yamada and F. Tabusa Tetrahedron 1990 46 265. 20 For example S. J. Danishefsky and M. P. DeNinno in Trends in Synthetic Carbohydrate Chemistry eds. D. Horton L. D. Hawkins and G. McGarvey ACS Symposium Series 386 American Chemical Society Washington DC 1989 pp. 176–181. 21 (a) S. Y. Ko W. M. Lee S. Masamune L. A. Reed III K. B. Sharpless and F. J. Walker Tetrahedron 1990 46 245; (b) S. Kobayashi and T. Kawasuji Synlett 1993 911. 22 S. Kobayashi T. Wakabayashi and M. Yasuda J. Org. Chem. 1998 63 4868. 23 T. Mukaiyama T. Yamada and K. Suzuki Chem. Lett. 1983 5. 24 S. Kobayashi Y. Fujishita and T. Mukaiyama Chem. Lett. 1990 1455. 25 S. Kobayashi and M. Horibe Chem. Eur. J. 1997 3 1472 and references cited therein. 26 S. Kobayashi and S. Nagayama J. Org. Chem. 1996 61 2256 and references cited therein. 27 S. Kobayashi and S. Nagayama J. Am. Chem. Soc. 1996 118 8977. 28 We have found that lanthanide triflates (including scandium triflate) are water-tolerant Lewis acids and efficient catalysts in several synthetic reactions in aqueous media. (a) S. Kobayashi Synlett 1994 689; (b) S. Kobayashi in Organic Reactions in Water ed. P. Grieco Chapman & Hall London 1997 and references cited therein. 29 S. Kobayashi S. Nagayama and T. Busujima Tetrahedron Lett. 1996 37 9221. 30 S. Kobayashi S. Komiyama and H. Ishitani Biotechol. Bioeng. 1998 1 23. 31 For recent examples (a) S. Kobayashi T. Busujima and S. Nagayama Chem. Commun. 1998 19; (b) H. Oyamada and S. Kobayashi Synlett 1998 249 and references cited therein. Review 7/07429H 15 Chem. Soc. Rev. 1999 28 1–15

ISSN:0306-0012    

DOI:10.1039/a707429h

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Travelling the organometallic road a Wittig student’s journey from lithium to magnesium and beyond Friedrich Bickelhaupt Scheikundig Laboratorium Vrije Universiteit NL-1081 HV Amsterdam The Netherlands Received (in Cambridge) 3rd June 1998 Both organomagnesium compounds (Grignard reagents) and organolithium compounds are of utmost importance in synthesis. New developments in the area of difunctional organometallic compounds will be described against a historical background in which the author’s interest gradually drifted from organolithium chemistry in particular the deep red ortho-dilithiobenzene to its magnesium analogue the tetrameric ortho-phenylenemagnesium. The latter was applied in the synthesis of novel 1,2-disubstituted benzene derivatives such as 9,10-dimetallatriptycenes.1 Introduction The hundredth anniversary of Georg Wittig (1897–1987) was commemorated last year by a range of academic ceremonies. On 16 June 1997 precisely the centenary of Wittig’s birthday a special colloquium was held at the University of Marburg his alma mater in order to commemorate some aspects of his seminal contributions to organic chemistry in general and to organometallic and organoelement chemistry in particular culminating in his discovery of the phosphorus ylides; generally known as Wittig reagents and indispensable in preparative organic chemistry they earned him the Nobel Prize in 1979. At such an occasion recollections have to be by necessity selective. It happened to be the author’s duty and pleasure to make a choice and he did so by concentrating on certain developments in organometallic chemistry in which he himself had the good fortune to participate first as a student in Wittig’s Friedrich Bickelhaupt obtained his PhD in organic chemistry at the University of T�ubingen with G.Wittig in 1957. After a postdoctoral period with K. Wiesner (University of New Brunswick at Fredericton Canada alkaloid chemistry) and R. B. Woodward (Harvard University chlorophyll synthesis) he joined in 1960 the research laboratories of Boehringer Mannheim (Germany psychopharmaca). In 1964 he was appointed Professor of Organic Chemistry at the Vrije Universiteit in Amsterdam; since 1997 he has been Professor Emeritus. His research interests are in the fields of organometallic chemistry low coordination chemistry in groups 14 and 15 and valence isomers of benzene and small cyclophanes.Awards JSPS fellowship (1985); Max Planck Research Prize (with Professor Dr G. Erker 1993). Ph Ge Si C laboratory in Tübingen and Heidelberg and later in the course of his own academic career; this talk formed the basis for the present review. It will become apparent that Wittig’s spirit is permeating through much of the chemistry to be addressed; this was not always conscious to those involved but is in retroperspective clearly discernible. (1) COOMe COOMe 2 Phenyllithium Wittig’s divining rod Organolithium compounds had been known for some time but they were curiosities of interest to the specialist mainly.1 This changed in 1930 when Karl Ziegler Wittig’s colleague from the early Marburg days and his good friend (and sometimes rival) developed an easy preparative access to these highly reactive carbanionoid reagents through ‘direct synthesis’ from organic halides and lithium metal [eqn.(1)].2 2 Li + RHal —? RLi + LiHal In the same period Wittig was interested in the chemistry of triphenylmethyl type diradicals such as 1 which he hoped to obtain from the diol 2 via known pathways.3 The problem however was the synthesis of 2. An obvious approach by the Grignard route starting from dimethyl phthalate (3) and phenylmagnesium bromide failed because the organomagnesium reagent turned out to be insufficiently reactive (Scheme 1).Only one year after Ziegler’s discovery and possibly inspired by his close relationship with Ziegler Wittig made use of the superior reactivity of phenyllithium to enforce the transformation 3?2,3 and he also investigated the difference in reactivity between the organolithium and the organomagnesium reagent in other applications.4 PhMgBr OH CPh2 • CPh2 PhLi !! CPh2 CPh2 • OH 1 2 3 Scheme 1 Ever since phenyllithium has been Wittig’s divining rod (‘Wünschelrute’ as he called it) and it was to be a superb choice indeed. It led him to important discoveries in radical and carbanion chemistry amongst which are the ylides including the Nobel Prize winning Wittig reagents the Wittig rearrangement of ethers dehydrobenzene (or benzyne C6H4) ate complexes pentaphenylphosphorus the controlled aldol condensation — too many to name them all in the present context; Chem.Soc. Rev. 1999 28 17–23 17 however they have been adequately reviewed recently by his student Tochtermann.5 A particular category of ate complexes is that of the alkali metals such as ‘phenyllina’ ([Ph2Li]2Na+ 4) obtained by the interaction of phenyllithium with phenylsodium [eqn. (2)].6 Wittig was fascinated not only by the analogy with ‘his’ borate complexes such as kalignost (sodium tetraphenylborate NaBPh4) but also because the extremely reactive phenylsodium which normally destroys diethyl ether instantaneously was stabilised by one equivalent of phenyllithium in ether without losing much of its high reactivity.(2) PhLi + PhNa —? [Ph2Li]2Na+ So in 1955 he coaxed the present author (who at that time would have slightly preferred to jump on the bandwagon of the newly discovered Wittig reagent) into investigating this phenomenon more closely. It turned out to be a worthwhile effort phenyllithium proved to stabilise not only one equivalent of phenylsodium but even a tenfold excess as in 5 [eqn. (3)] making this strongly carbanionic reagent available for synthetic applications in diethyl ether an attractive polar medium. The stabilisation was further extended to combinations with the still more reactive phenylpotassium (6a) and phenylcaesium (6b) or less effectively to the corresponding methyl ate complex 7 [eqns. (4) and (5)].7 (3) PhLi + 11 PhNa —? PhLi·(PhNa)11 [in Et2O !] 5 (4) PhLi + PhM —? [Ph2Li]2M+ 6a :M = K 6b :M = Cs (5) MeLi + MeNa —? [Me2Li]2Na+ 7 A problem which at that time remained unresolved was the structural identity of the stabilised ate complexes.In analogy to NaBPh4 Wittig considered 8 to be a reasonable structure for 4 [eqn. (6)],7 although the occurrence of for example 5 indicated (6) RLi + RNa —? [R·�Li/R] Na+ ?? 8 [Ph4Li]32 [Na+(TMEDA)]3 9 a more complex mode of aggregation. Thirty years later Weiss proved by X-ray crystallography that another compound with a high PhNa/PhLi ratio having the composition PhLi·3PhNa·3TMEDA has the structure 9,8 which is reasonably close to 8 certainly in showing the capability of lithium to function as the central atom of an ate complex.3 Difunctional organometallic compounds Though pleased with these results Wittig did not want to pursue the chemistry of lithium ate complexes any further. Instead his insatiable curiosity was turning to a new topic for the second part of the author’s PhD thesis the preparation and investigation of o-dilithiobenzene (10). Undoubtedly one of the reasons why he considered the synthesis of 10 a challenge was that it is not accessible by the normal approach towards organolithium compounds (seemingly) obvious reactions of o-dibromobenzene (11) with either lithium metal according to Ziegler or bromine–metal exchange with n-butyllithium proceed stepwise and thus necessarily pass through the stage of 12 which instead of being converted to 10 in a second metallation step immediately eliminates lithium bromide to form dehydrobenzene (13) incidentally another favourite of Wittig’s (Scheme 2).5 Fortunately Vecchiotti had synthesised the mercury analogue o-phenylenemercury (14)9 from 11 and sodium amalgam; Chem.Soc. Rev. 1999 28 17–23 18 Li +i –LiBr Li 10 Li Br Br Br 12 11 + BuLi –LiBr –BuBr 13 Scheme 2 he believed it to be a dimer (with a dihydroanthracene type structure) but eventually it was shown to be a trimer with the usual linear C–Hg–C angles (Scheme 3; see also Scheme 7).10 By the classical Schlenk procedure i.e. shaking the mercury compound with metallic lithium 14 could be converted to 10 in about 80% yield.11 Compound 10 opened up a new route to (conventional) chemistry by reaction with organic electrophiles and (less conventional) transformations with main group and transition metals.However the most profound impression on the young chemist’s mind came from the intensely red colour of the compound. As very pure phenyllithium is practically colourless a deep red colour is not what one would expect from inspection of the structural formula. Undoubtedly 10 has a higher aggregation state but unfortunately a satisfactory explanation of this phenomenon is presently not available. Hg Li Br Li Na–Hg deep red ! Et2O Et2O n Li Br 10 14 (n = 3) 11 Scheme 3 After obtaining his PhD in Tübingen and after several years of travel and apprenticeship the author finally continued his research in Amsterdam where he was pleased amongst other things to realize that his own inclination towards organometallic compounds could smoothly be combined with certain investigations in the field of organomagnesium chemistry initiated by his predecessor Jan Coops including highly sophisticated high vacuum techniques.12 In the years to follow many aspects of organomagnesium chemistry have been studied in our group such as the (unexpectedly complex) formation reaction of Grignard reagents RMgX their structure and interaction with Lewis bases and finally their application in particular in organometallic synthesis.13,14 In the course of these investigations and for various reasons specific attention was paid to divalent organomagnesium compounds 15.Apart from having structural interest of their own,13 they are valuable synthons for applications in organic and especially in organometallic synthesis. Thus they open a convenient and quite general route to organometallic heterocycles 16 (Scheme 4). MgBr (CH2)n + Cl2MLn MLn (CH2)n –2 MgBrCl MgBr 16 15 a n = 1 b n = 2 c n = 3 d n � 4 Scheme 4 This strategy had already been successfully applied since the early days of Grignard chemistry to the synthesis of 16d starting from di-Grignard reagents 15d with 4 or more carbon atoms between the two organometallic functions. In contrast it failed completely for the smaller members 15a–c with 1 2 or 3 carbon atoms between the metal functions respectively for the simple reason that they could not or not in a satisfactory fashion be obtained by direct synthesis from the corresponding dihalides.13 This synthetic challenge together with the prospect that the reagents if available would offer an easy and general access to highly interesting small metallacycles such as metallacyclobutanes and analogues led us to develop several routes to small representatives of 15; this permitted us to prepare a range of four-membered metallacycles as shown in Scheme 5.13 14 In Cl2MLn 2 CH2(MgBr)2 LnM MgBr Me MgBr Me Mg n Mg 4 M M' M" = main group and/or transition metals Scheme 5 contrast to 15a and 15c and several of their derivatives which could be prepared and applied in a satisfactory way great problems were encountered in the preparation of 1,2-bisbromomagnesioethane (BrMgCH2CH2MgBr 15b).Although it could be finally synthesised15 in collaboration with G. W. Klumpp another Wittig student the low yield (10%) and the instability of the compound made preparative applications unattractive. 4 1,2-Dimetallated benzenes 4.1 Derivatives of Groups 2 and 12 Being nevertheless strongly interested in the missing link of 1,2-difunctional organometallics we were looking for alternatives and it was at this stage that we returned to ‘Wittig chemistry’ to try the synthesis of o-phenylenemagnesium (17) the magnesium analogue of 10. Performing the synthesis was easier than its conception shaking 14 with an excess of magnesium gave 17 in 65% isolated yield after crystallisation from THF (Scheme 6).16 The formation of 17 was slower (2 weeks at room temperature and 10 hours at 70 °C) than that of 10 (4 days at room temperature) but 17 had two advantages over 10 it was stable (in the absence of light) and like organomagnesium compounds in general it promised to give rise to less unwanted side reactions such as reduction when used for metathesis with transition metal salts.A big surprise was the structure of 17 while nobody had expected it to have the monomeric structure of a magnesacyclopropabenzene it turned out to be a tetramer both in solution (molecular weight determination in THF) and in the crystalline Cl2MLn MLn LnM MgBr Cl2M'Ln M'Ln LnM MgBr M"Cl4 M" MLn LnM Me Cl2MLn MLn Me Cl2MLn MLn MLn Cl2MLn Hg 3 14 THF Mg THF Mg Mg THF Mg 4 Mg Mg THF THF Scheme 6 º 17 (only one of the four phenylene units shown) state (X-ray structure determination).The crystal structure showed a slightly distorted tetrahedron of 4 magnesium atoms with each triangle of the tetrahedron capped by a phenylene unit in such a fashion that one carbon is s-bonded to one magnesium and the second carbon is m-bridging between two other magnesiums (Scheme 6; only one phenylene unit shown). Similar tetrameric structures were found for 1,2-diphenylvinylenemagnesium (18) and the 1,3-dimagnesium derivative 1,8-naphthalenediylmagnesium (19). These tetrameric structures are reminiscent of those of many organolithium compounds.We have explained the analogy by the correspondence between complexes of two monovalent ions of (R2Li+)4 and those of two divalent ions (R22Mg2+)4; apparently in both series electrostatic interactions are largely responsible for a tetrahedral arrangement.16 As the reductive potential towards transition metal salts decreases steadily in the series RLi ?R2Mg ?R2Zn we were also interested in the zinc analogue 20. It could be smoothly obtained from 14 and zinc and again had two surprises. Firstly while zinc is less electropositive than magnesium and therefore usually less reactive the transformation 14 ? 20 went much faster than the formation of the magnesium or even the lithium analogue; it took only 6 hours at room temperature for completion (Scheme 7).17 The second surprise came again from the structure in THF solution 20 is strictly and concentration independently trimeric (20a) presumably as a slight variation of the structure of 14 with somewhat smaller bond angles at zinc; however in the crystalline state the compound occurs in the dimeric form of a 9,10-dihydro-9,10-dizincaanthracene (20b).4.2 Derivatives of Group 13 So far we have applied 17 and 20 mainly in the synthesis of main group metal derivatives. With organoaluminium dichlorides RAlCl2 the results were not clear-cut. According to NMR spectroscopy in [D8]THF single species were obtained for R = Me or Et and mixtures of species for R = t-Bu Ph and of organyl groups capable of intramolecular coordination (Scheme 8); though crystals formed they easily crumbled so that X-ray structures could not be obtained.18 Very probably some of these species are of the dihydroanthracene type observed for 20b (vide supra) and for the heavier metals of Group 13 in the case of gallium (21) and indium (22); in the crystal both 2118 and 2219 have an essentially planar central ring (Scheme 9).4.3 Derivatives of Group 14 While 9,10-dihydro-9,10-disilanthracenes such as 23a were known for some time the germanium and tin analogues 23b and Chem. Soc. Rev. 1999 28 17–23 19 THF THF Zn Zn Zn THF Zn THF THF 20a Hg Hg Hg THF crystal 14 THF THF Zn Zn THF THF 20b Scheme 7 R Mg Al RAlCl2 1/n 1/4 • n THF THF n 4 17 RAlCl2 = MeAlCl2 EtAlCl2 single species ( n = 2 ?) NMe RAlCl2 = t-BuAlCl2 PhAlCl2 OMe Al Cl Al Cl 2 2 mixtures of sevel species Scheme 8 pyr Me Ga 1) 2 MeGaCl2 2) pyridine Ga ryp Me Mg 21 2/4 4 17 NMe Me2N 2 Cl In 2 Me2N Me2N In 2 In Me2N Me2N 22 Scheme 9 23c respectively were not.All three heterocycles were conveniently prepared from 17 and the dihalides R2MCl2 though the yields were mediocre (Scheme 10). It was therefore a little bit of a surprise that in the more ambitious attempt to obtain the 9,10-dimetallatriptycenes 24 from 17 and the trihalides MeMCl3 (Scheme 11) the yields of 24b and 24c were rather good especially if one considers that in this one-pot reaction 5 formal monomeric units have to Chem.Soc. Rev. 1999 28 17–23 20 Mg 2 Me2MCl2 2/4 THF THF 4 17 THF Scheme 10 combine to form 6 new bonds; an overall yield of 68% for 24b then means an average yield of 94% in each step. Furthermore it is nontrivial that the reaction partners apparently make little use of ample opportunities to crosslink and polymerise! Mg 2 MeMCl3 3/4 –3 MgCl2 4 17 Scheme 11 2 The first hint as to what was going on came from the attempt to prepare 25 the phenyl analogues of the dimethyltriptycenes 24 because the yield of 25 was zero! A closer investigation revealed that the situation was not quite as hopeless as it seemed because after deuterolysis and regardless of whether the stoichiometrically required 2 equivalents of PhMCl3 were applied or only one equivalent we did obtain the trideuterated tetraphenylmetal compounds 26 in practically quantitative yield (Scheme 12).Mg 1) 2 PhMCl3 2) D2O 3/4 4 17 Scheme 12 This result suggested that with remarkable specificity one equivalent of PhGeCl3 and three (monomer) units of 17 had combined to form the tri-Grignard reagent 27b in quantitative yield (Scheme 13). This assumption was supported not only by deuterolysis yielding 26b but also by the addition of 1 molar equivalent of MeGeCl3 to the intermediate reaction mixture the digermatriptycene 28 bearing two different substituents at the bridgehead position was obtained in rather high yield (83% Scheme 13).Two aspects of these results deserve further comment. In the first place the high specificity of the formation of 27b is particularly unexpected against the background that Group 14 Me Me M M Me Me 23a 23b 23c M = Si 30% M = Ge 32% M = Sn 27% Me M M Me 24a 24b 24c M = Si 2% M = Ge 68% M = Sn 42% Ph M M Ph 25 Ph D M 3 26b M = Ge 26c M = Sn MgCl Ph Mg Ge PhGeCl3 3/4 3 4 17 27b MeGeCl3 D2O Ph D Ph Ge Ge Ge 3 Me 26b 28 Scheme 13 trihalides are well known to react in a nonselective manner with Grignard reagents; thus 3 molar equivalents of PhMgBr and RMCl3 will give a mixture of 29 30 and 31 (Scheme 14). The R M + 3 PhMgBr 2 + RMPh2Cl + RMPh3 Cl Cl 30 RMPhCl 29 31 Cl R 1 M Cl Cl R Mg M 2 Mg Cl Cl Mg Cl Cl 3 32 4 17 R Mg 3 + "C6H4Mg" MgCl 27 Scheme 14 question arises why is 17 so specific in substituting all three chorines of one molecule of RMCl3 before attacking the next one especially in view of the general rule that the reactivity of an organometallic polyhalide decreases with decreasing number of halogens? We feel that the answer must come from the unusual tetrameric structure of 17.As shown in Scheme 14 the first encounter between 17 and RMCl3 presumably leads to a complex 32 in which one Mg–Cl bond has been replaced by Mg–C(1); the replaced Cl may reside on a magnesium at position 2 of the phenylene ring (or an equivalent position).Whereas normally the following step is the attack by a second external organometallic reagent in 32 all the ingredients for further reactions are closely assembled in one agglomeration due to the tetrameric structure of the starting material 17 which gives the entropic advantage of an intramolecular process. Thus rather than attacking a second molecule of RMCl3 the two remaining chlorines of the MCl2 group in 32 are arylated first with formation of 27; one equivalent of C6H4Mg is left over and may engage anew in aggregation and arylation. The formation of the dimetallatriptycenes thus proceeds in two stages first formation of a tri-Grignard reagent such as 27 followed by a triple ring closure to yield the triptycene.This hypothesis also explains the second surprising observation concerning this reaction i.e. the fact that the second stage is of the ‘Go/NO GO’ type as illustrated in Scheme 15 the R MgCl R M M R'M'Cl3 M' –3 MgCl2 3 R' 27 R' Yield M' M R HH Me 20–30% 73% 2% Si Si Si Si Si Si H Me Me Me Me Ph 68% 83% 0% Ge Ge Ge Ge Ge Ge Me Ph Ph H Me 76% 0% Si Si Ge Ge Ph Ph Me Ph 42% 0% Sn Sn Sn Sn Me Ph Scheme 15 triptycene is formed either in good to reasonable yield or not at all. While the size of the Group 14 metal and its substituent have no influence in the first stage i.e. the formation of the tri- Grignard reagent 27 (as monitored by quantitative yield of the trideuterated product on deuterolysis) steric factors are apparently decisive in the second stage to an extent far beyond our intuition.Thus the small silicon tolerates only one methyl group whereas the larger germanium allows ring closure with up to one methyl and one phenyl group but not with two phenyl groups; tin behaves similarly. This leads to the seemingly absurd situation that in the reaction of 3 equivalents of 17 with 2 molar equivalents of PhMCl3 (Scheme 12) 27 is formed selectively and sits next to the second equivalent of PhMCl3 in solution without reacting!18 That the second phenyl group in 25 cannot be introduced by this approach is not a consequence of these compounds being incapable of existence.This was proven by the synthesis of 33 from 25a and of 35 from 34 (Scheme 16).20 H Ph Si Si 2 PhLi Si Si Ph H 33 25a H Ph Si Si PhLi Ge Ge Ph Ph 35 34 Scheme 16 4.4 Other bridging atoms The reaction of 17 with Group 15 trihalides proceeds easily. In particular we investigated the reaction with arsenic trichloride which gives 36 (Scheme 17),20 the oldest known triptycene and 21 Chem. Soc. Rev. 1999 28 17–23 Mg 2 AsCl3 3/4 4 17 AsCl3 MgCl 37 MgCl Ph Ge ECl3 3 27b Scheme 17 prepared as early as 1927.21 Its tendency of formation is so high that in this case it is difficult to stop the reaction at the intermediate tri-Grignard stage of 37 (which is analogous to the readily formed and stable 27 see Scheme 13).Probably 36 is so readily formed because it is practically strain-free; the strain imposed on the triptycene skeleton by geometric boundary conditions22 is minimised by the inherently small valence angle of Group 15 elements. Similarly the mixed triptycenes 38a–c were obtained from 27 and the corresponding trichloride.18,20 So far we have been less successful in preparing triptycenes with transition metals in the bridgehead positions. Reactions of 17 or 20 with CpTiCl3 gave decomposition products only; with CpZrCl3 1H-NMR spectroscopy indicated the formation of up to 50% of 39 (Scheme 18) but the compound decomposed above 25 °C and was not isolated in pure form.23 Mg 2 CpZrCl3 3/4 4 17 Scheme 18 4.5 Possible applications We have also started to prepare dimetallatriptycenes with unsaturated organic substituents as they may be expected to transfer optical or electrical information along chains containing these triptycenes.Such triptycene containing polymers may have a rod-like rigid structure and conjugative interaction will not be blocked by the saturated Group 14 bridgehead atoms because their distance in the triptycene skeleton is shorter than Chem. Soc. Rev. 1999 28 17–23 22 As As 36 AsCl3 As 3 E Ge Ph 38a E = P 38b E = As 38c E = Sb Cp Zr Zr Cp 39 the sum of their van der Waals radii and especially the heavier metal atoms are known to transfer interactions of this kind.The SiH functionality of 34 (Schemes 16 and 19) offered a number of routes to functionalisation at the bridgehead. Thus platinum catalysed hydrosilylation of phenylacetylene with 34 gave 40; it may be considered as a model for conjugation through a double bond if the reaction is extended to alkynyl substituted triptycenes (such as 43 vide infra) which offer the construction of unsaturated poly-triptycene chains. Ph Ge ClMg 3 27 Ph Ge Si Cl 41 Me3SiCºCLi Ph Ge Si C C 42 Me3Si C C Scheme 19 On the other hand transformation of 34 with N-bromosuccinimide (NBS) in carbon tetrachloride gave interestingly not the expected silyl bromide but the chloride 41 instead. Compound 41 was substituted with lithium trimethylsilylacetylide to give 42 which was deprotected to furnish the triptycylacetylene 43.Copper(I) catalysed oxidation of the latter gave 44 which in its turn may stand as a model for rod-like oligomers if one starts from 9,10-dialkynyltriptycenes. The synthesis of a repeating unit for this strategy is illustrated in Scheme 20. First the disubstituted acetylene 45 was synthesised to furnish a spacer between two triptycene units. Ph HSiCl3 Ge Si H 34 H2PtCl4 PhCºCH NBS CCl4 Ph Ge Si Ph 40 Ph Ge KOH Si C C 43 H Cu(I)/O2 Ph Ge Si 44 2 Reaction of 17 with 45 gave 46 (19% yield) which was deprotected in 80% yield to furnish 47;18 it remains to be seen if this strategy will furnish oligomeric or polymeric diacetylenes 48 on oxidative coupling.3/4 H 5 Conclusions Starting from (relatively) simple organolithium chemistry coincidental circumstances and a strong interest in difunctional organometallic species have led the author into the interesting field of 1,2-metallated benzene derivatives which show a great variety of structures and reactions. Thus the 1,2-dimetallated benzenes may be dimers (M = Zn Al? Ga In) trimers (Zn Hg) or tetramers (Mg). The latter o-phenylenemagnesium has been applied for the synthesis of a number of novel 9,10-dimetallaanthracenes and 9,10-dimetallatriptycenes. Especially the triptycenes exhibit fascinating structural aspects,22 and they may have promise for the preparation of compounds with interesting material properties.It is obvious that the work described here — though mostly performed in high vacuum glass apparatus — did not develop in Mg GeCl + 2 3 Me3Si 4 45 17 SiMe3 H Ge Ge 1) OH– Ge Ge 47 46 Me3Si Ge 2) Cu+/O2 Ge ?? n 48 Scheme 20 a scientific vacuum; many others have made important contributions to the area of polyfunctional organometallics. As a thorough overview of the entire field is beyond the scope of the present report the interested reader is referred to more comprehensive literature. 24–27 6 References 1 Ch. Elschenbroich and A. Salzer Organometallics A Concise Introduction VCH Weinheim 2nd edn. 1992. 2 K. Ziegler and H. Colonius Liebigs Ann. Chem.1930 479 135. 3 G. Wittig and M. Leo Ber. Dtsch. Chem. Ges. 1931 64 2395. 4 G. Wittig M. Leo and W. Wiemer Ber. Dtsch. Chem. Ges. 1930 64 2405. 5 W. Tochtermann Liebigs Ann./Recueil 1997 I. 6 G. Wittig R. Ludwig and R. Polster Chem. Ber. 1955 88 264. 7 G. Wittig and F. Bickelhaupt Chem. Ber. 1958 91 865. 8 U. Schümann and E. Weiss Angew. Chem. 1988 100 573. 9 L. Vecchiotti Ber. Dtsch. Chem. Ges. 1930 63 2395. 10 D. S. Brown A. G. Massey and D. A. Wickens Inorg. Chim. Acta 1980 44 L193. 11 G. Wittig and F. Bickelhaupt Chem. Ber. 1958 91 883. 12 A. D. Vreugdenhil and C. Blomberg Recl. Trav. Chim. Pays-Bas 1963 82 453; 461. 13 F. Bickelhaupt Angew. Chem. 1987 99 1020; Angew. Chem. Int. Ed. Engl. 1987 26 990. 14 F. Bickelhaupt J. Organomet. Chem. 1994 475 1. 15 N. J. R. van Eikema Hommes F. Bickelhaupt and G. W. Klumpp Recl. Trav. Chim. Pays-Bas 1988 107 393. 16 M. A. M. G. Tinga G. Schat O. S. Akkerman F. Bickelhaupt E. Horn W. J. J. Smeets and A. L. Spek J. Am. Chem. Soc. 1993 115 2808. 17 M. Schreuder Goedheijt T. Nijbakker O. S. Akkerman F. Bickelhaupt N. Veldman and A. L. Spek Angew. Chem. 1996 108 1651; Angew. Chem. Int. Ed. Engl. 1996 35 1550. 18 M. A. Dam PhD Thesis Vrije Universiteit Amsterdam 1997. 19 M. A. Dam T. Nijbacker B. T. de Pater F. J. J. de Kanter O. S. Akkerman F. Bickelhaupt W. J. J. Smeets and A. L. Spek Organometallics 1997 16 511. 20 N. Rot PhD Thesis Vrije Universiteit Amsterdam 1998. 21 N. P. McCleland and J. B. Whitworth J. Chem. Soc. 1927 2753. 22 M. A. Dam F. J. J. de Kanter F. Bickelhaupt W. J. J. Smeets A. L. Spek J. Fornies-Camer and C. Cardin J. Organomet. Chem. 1998 550 347. 23 N. Rot and M. Oberhoff unpublished results. 24 K. Nützel in Methoden der Organischen Chemie (Houben-Weyl) ed. E. Müller Thieme Stuttgart 1973 vol. 13/2a p. 47. 25 W. E. Lindsell in Comprehensive Organometallic Chemistry eds. G. Wilkinson F. G. A. Stone E. W. Abel Pergamon Oxford 1982 vol. 1 p. 155. 26 W. E. Lindsell in Comprehensive Organometallic Chemistry II eds. E.W. Abel F. G. A. Stone G. Wilkinson Pergamon/Elsevier Oxford 1995 vol. 1 p. 57. 27 G. S. Silverman and P. E. Rakita Handbook of Grignard Reagents Marcel Dekker New York 1996. Review 8/04191A 23 Chem. Soc. Rev. 1999 28 17–23

ISSN:0306-0012    

DOI:10.1039/a804191a

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Polarity-reversal catalysis of hydrogen-atom abstraction reactions concepts and applications in organic chemistry Brian P. Roberts Christopher Ingold Laboratories Department of Chemistry University College London 20 Gordon Street London UK WC1H 0AJ Received (in Cambridge) 8th June 1998 The rates and selectivities of the hydrogen-atom abstraction reactions of electrically-neutral free radicals are known to depend on polar effects which operate in the transition state. Thus an electrophilic species such as an alkoxyl radical abstracts hydrogen much more readily from an electronrich C–H bond than from an electron-deficient one of similar strength. The basis of polarity-reversal catalysis (PRC) is to replace a single-step abstraction that is slow because of unfavourable polar effects with a two-step process in which the radicals and substrates are polaritymatched.This review explores the concept of PRC and describes its application in a variety of situations relevant to mechanistic and synthetic organic chemistry. 1 Introduction Within the context of the reactions of electrically-neutral free radicals the term ‘polar effect’ is used to describe the influence on the activation energy of any charge transfer which may occur on proceeding from the reactant(s) to the transition state. The dependence of reactivity and selectivity in radical chemistry on such polar effects has been recognised for more than 50 years and was emphasised by Walling in his seminal monograph which was published in 1957.1 The transition state for the hydrogen-atom transfer reaction shown in eqn.(1) may be represented in valence-bond terms as a hybrid of the structures 1a–d and within a series of reactions Brian Roberts born 1944 in London received his BSc degree from UCL in 1965. After working with Professor Alwyn G. Davies for his PhD (1968) he was subsequently Turner and Newall Research Fellow at UCL where he was appointed Lecturer in 1972. The year 1972–73 was spent on leave of absence at Chelsea College London working with Professor M. John Perkins and he spent a period as Visiting Research Scientist at the NRCC in Ottawa working with Dr Keith U. Ingold (1980). Dr Roberts is now Reader in Chemistry at UCL and his research centres on the chemistry of free radicals in solution.He is a recipient of the Meldola Medal (1973) and the RSC award in Organic Reaction Mechanisms (1996). [El1 H El2 ] •‡ . . d+ [El1 H Nuc] . d– . Nuc [Nuc H El2 ]•‡ . . H-El1 + Nuc• + H-El2 –H (1) 1a (2) d+ FAVOURED (3) El· + H–Nuc —? El–H +Nuc· ® Nuc· + H–El —? Nuc–H +El· © d– A· + H—B —? A—H + B· [A· H–B]‡ Ô [A–H B·]‡ Ô [A� H· B+]‡ Ô [A+ H· B�:]‡ 1b 1c 1d for which the overall enthalpy change is similar the activation energies would be expected to decrease as the contributions from the charge-separated structures 1c and 1d increase. If the structure 1c is more important than 1d the radical A· may be described as electrophilic and B· is said to be nucleophilic while if structure 1d is the more important A· is nucleophilic and B· is electrophilic.For such a series of reactions the activation energy is predicted to decrease as the electronegativity difference between A· and B· increases. If El· and Nuc· represent electrophilic and nucleophilic radicals respectively the hydrogen-atom abstraction reactions (2) and (3) should be favoured because of polar effects while reactions (4) and (5) will not be favoured. El1· + H–El2 —? El1–H +El2· (4) Nuc1· + H–Nuc2—? Nuc1–H +Nuc2·©DISFAVOURED (5) ® Our own interest in the exploration and exploitation of polar effects arose out of a research programme designed to investigate the properties of boron-containing isoelectronic analogues of well-known carbon-centred radicals and this work began with a study of the borane radical anion H3B·2 (which is isoelectronic with the methyl radical H3C·)2 and of various ligated boryl radicals of the types L?.BH2 and L?.BHR in which L is a phosphine an amine or a sulfide.3 Amine– alkylboryl radicals (R3N?. BHR) are isoelectronic analogues of secondary alkyl radicals (R3C–CHR) and are very readily . generated by hydrogen-atom transfer to tert-butoxyl radicals from the corresponding amine–alkylborane complex [eqn. (6)].4 Although radical philicity is clearly a relative attribute,5 a radical that has a high ionisation energy (IE) and a high electron affinity (EA) will usually exhibit electrophilicity while a low IE and EA will usually confer nucleophilic properties. In general radicals that have a high Mulliken electronegativity [(IE + EA)/ 2] will be electrophilic and those with a low electronegativity will be nucleophilic.6 Alkoxyl radicals are thus electrophilic while amine–boryl radicals have particularly low ionisation energies (5.47 eV has been calculated for H3N?.BHMe)5 and are very nucleophilic accounting for the high rate of reaction (6) which is an example of the general type shown in eqn. (2); for reaction (6) the charge-separated structure [ButO– H· RHB +/NR3] (cf. 1c) is an important contributor to the transition state. EPR studies have shown that amine– alkylboryl radicals rapidly abstract hydrogen from acetonitrile [eqn. (7)] while the corresponding abstraction by tert-butoxyl radicals [eqn.(8)] is very sluggish.4 The cyanomethyl radical derived from acetonitrile is electrophilic and these results can be Chem. Soc. Rev. 1999 28 25–35 25 understood in terms of polar effects since reactions (7) and (8) constitute examples of the general processes shown in eqns. (3) and (4) respectively. Consistent with these observations the overall abstraction of hydrogen from acetonitrile by the electrophilic alkoxyl radical is promoted by a small amount of amine–alkylborane through the sequence of rapid reactions (6) and (7) which replace the relatively inefficient single step (8).7 The polarity of the radical that abstracts hydrogen from the acetonitrile is thereby reversed (from an electrophilic alkoxyl radical to a nucleophilic amine–boryl radical) thus facilitating the overall transfer of hydrogen and for this reason the process is referred to as polarity-reversal catalysis.4,7 ButO· + R3N?BH2R —? ButOH + R3N?ÿBHR (6) R3N?ÿ BHR + CH3CN —? R3N?BH2R + H2 ÿCCN (7) (8) ButO· + CH3CN —? ButOH + H2 ÿCCN 2 Polarity-reversal catalysis (PRC) The principle underlying polarity-reversal catalysis of hydrogen-atom transfer is generalised in Scheme 1.The lack of El1· + H–El2 ––– slow?H–El1 + El2· uncatalysed reaction El1· + H–Nuc –––? fast H–El1 + Nuc·ß¶ catalytic cycle Nuc· + H–El2 –––? fast H–Nuc + El2·f fast El1· + H–El2 + H–Nuc catalyst –––––––––––? H–El1 + El2· overall reaction Nuc1· + H–Nuc2 ––– slow?H–Nuc1 + Nuc2· uncatalysed reaction Nuc1· + H–El –––? fast H–Nuc1 + El·ß¶ catalytic cycle El· + H–Nuc2 –––? fast H–El + Nuc2·f + H–El catalyst fast Nuc1· + H–Nuc2 ––––––––––?H–Nuc1 + Nuc2· overall reaction Scheme 1 stabilising charge-transfer in the transition state for the direct abstraction shown in eqn.(4) is overcome by including an hydridic catalyst H–Nuc when the single-step process is replaced byycle of two hydrogen-atom transfer reactions both of which benefit from favourable polar effects. Similarly the slow direct abstraction reaction (5) is promoted by a protic catalyst H–El.5† Of course the activation energy for hydrogenatom transfer depends on factors other than polar effects6 and careful consideration must be given to the strengths of the bonds involved when choosing a suitable polarity-reversal catalyst the activation energy for an endothermic reaction cannot be less than (DH° + RT) no matter how favourable are the polar factors! Reference to Fig.1 clarifies the situation for reaction (4) and its polarity-reversal-catalysed equivalent. Because of the exponential dependence of the rate constant on the activation energy for a reaction two steps with low activation energies can lead to a faster overall reaction than is achieved in a single-step process which has a much higher activation energy. Ideally the overall enthalpy change associated with an uncatalysed exothermic reaction should be partitioned so that both steps of the catalytic cycle are themselves exothermic. For example § The enantioselectivity factor is the rate constant for abstraction from the faster reacting enantiomer relative to the rate constant for abstraction from the less-reactive enantiomer.Chem. Soc. Rev. 1999 28 25–35 26 [El1 H El2 ] •‡ . . E El1• + H-El2 d+ d+ [El1 H Nuc] . d– . + H-Nuc d– [Nuc H El2 ]•‡ . . H-El1 + Nuc• + H-El2 –H-Nuc Fig. 2 EPR spectra obtained when tert-butoxyl radicals are generated in the presence of bis(2-cyanoethyl) ether at 257 °C (a) in the absence of a catalyst and (b) in the presence of Me3N?BH2Thx. H-El1 + El2• Fig. 1 Schematic potential energy diagram illustrating the principle of PRC for promotion of a hydrogen-atom transfer of the type shown in eqn. (4) by an hydridic catalyst H–Nuc. abstraction of hydrogen from acetonitrile by the tert-butoxyl radical [eqn.(8)] is exothermic by ca. 47 kJ mol21.8 Calculations5 indicate that DH(B–H) in an amine–alkylborane is ca. 432 kJ mol21 and the two steps in the cycle for the catalysed abstraction [eqns. (6) and (7)] are exothermic by ca. 8 and 39 kJ mol21 respectively. The control of regioselectivity that can be exercised using PRC is clearly illustrated by the normal and catalysed reactions of tert-butoxyl radicals with bis(2-cyanoethyl) ether as studied by EPR spectroscopy.4 The transient radical products of elementary reactions can be detected in solution in steady-state concentration (ca. 5 3 1027 mol dm23) during continuous photochemical generation of the reactant radicals directly in the microwave cavity of the EPR spectrometer.9 Thus when a cyclopropane solution containing di-tert-butyl peroxide and bis(2-cyanoethyl) ether was irradiated with UV light at 257 °C the EPR spectrum of the a-alkoxyalkyl radical 2 [eqns.(9) and (10a)] was observed (Fig. 2a).5 The H–COR and H–CCN bonds are of similar strength,8 but a-alkoxyalkyl radicals are nucleophilic (the corresponding cation is relatively stable) while acyanoalkyl radicals are electrophilic and polar effects direct abstraction to the H–COR group. However in the presence of 10 mol% trimethylamine–thexylborane (Me3N?BH2Thx)‡ as an hydridic polarity-reversal catalyst the EPR spectrum of the a-cyanoalkyl radical 3 was detected to the exclusion of that of 2 [eqn. (10b)] (Fig. 2b). Now the tert-butoxyl radical reacts more readily with the amine–borane than with the ether on ‡ The 1,1,2-trimethylpropyl (‘tert-hexyl’) residue (Me2CHCMe2–) is commonly referred to as the thexyl group (Thx).ButOOBut uncatalysed –ButOH ButO• (NCCH2CH2)2O with Me3N BH2Thx catalyst account of the very favourable polar effects for the former reaction and then the highly nucleophilic amine–boryl radical abstracts hydrogen selectively from the H–CCN group to yield 3 and regenerate the catalyst. The rate constant for abstraction of hydrogen from Me3N?BH2Thx by the tert-butoxyl radical has been estimated5 to be 4.7 3 107 dm3 mol21 s21 at 284 °C and if the Arrhenius A-factor is assumed to be 109 dm3 mol21 s21 the corresponding activation energy would be ca. 5 kJ mol21. 3 Fig. 3 EPR spectra obtained when tert-butoxyl radicals are generated in the presence of cholesteryl acetate at 233 °C (a) in the absence of a catalyst and (b) in the presence of Me3N?BH2Thx.H3COCH2C a-C–H group. In particular while Me more reactive than MeCH of the decreasing strength of the a-C–H bond along the series CH3CO2Et > MeCH2CO2Et > Me2CHCO2Et coupled with the increased steric protection that a-methylation affords to an 2CHCO2Et is 1.1 times 2CO2Et towards the amine–boryl radical Me3N?ÿ BHMe it is 9.2 times less reactive than the propanoate towards the more sterically-hindered Me3N?ÿ BHThx. A similar trend was found for amine–boranecatalysed hydrogen abstraction from the two types of a-C–H group present in 3-methylbutan-2-one [Me2CHC(O)CH3] where Me3N?ÿ BHThx shows a strong preference for abstraction from the less-hindered methyl group while Me3N?BHBu ÿ abstracts the more weakly bound but less accessible tertiary ahydrogen atom.10 Competitive hydrogen-atom abstraction from cyclopenta- 1,3-diene and from cyclohepta-1,3,5-triene for which the strengths of the C–H bonds are fairly similar has been examined.12 The cyclopentadienyl cation is antiaromatic while the cycloheptatrienyl cation is a stabilised aromatic species and it has been shown that electrophilic tert-butoxyl radicals 3 Amine–boranes as hydridic polarity-reversal catalysts Extensive EPR studies of elementary reactions have been used to explore the utility of amine–boranes as hydridic catalysts for the overall transfer of electron-deficient hydrogen to electrophilic alkoxyl radicals.Thus in the presence of a catalytic amount of amine–alkylborane hydrogen is rapidly and selectively abstracted from a C–H group a to the carbonyl substituent in esters lactones ketones imides acetic anhydride and related compounds.5,10 For example at 252 °C the uncatalysed reaction of tert-butoxyl radicals with methyl methoxyacetate 4 gives a mixture of radicals resulting from competitive abstraction of hydrogen from the ether 2CH group and from the a-CH2 group while in the presence of Me3N?BH2Thx only abstraction from the latter was detected. The uncatalysed reaction of tert-butoxyl radicals with cholesteryl acetate 5 gives a mixture of radicals resulting from unselective abstraction from the cholesteryl moiety (Fig.3a) but abstraction from the acetyl group to give the radical 6 was not detectable by EPR spectroscopy. However in the presence of Me3N?BH2Thx abstraction from the electron-deficient a-CH3 group occurs selectively and 6 is the only radical detected (Fig. 3b). With tetrahydro-4H-pyran-4-one 7 tertbutoxyl radicals abstract hydrogen mainly adjacent to the endocyclic oxygen atom while in the presence of an amine– alkylborane catalyst abstraction takes place exclusively a to the carbonyl group as judged by EPR spectroscopy. Reaction of tert-butoxyl radicals with an equimolar mixture of tert-butyl methyl ether and diethyl malonate at 284 °C afforded only the EPR spectrum of the tert-butoxymethyl radical (Fig.4a) while the only spectrum observed in the presence of Me3N?BH2Thx was that of H ÿ C(CO2Et)2 resulting from abstraction of electron-deficient a-hydrogen from the malonate (Fig. 4b).5 PRC by Me3N?BH2Bu has also been used to generate the radicals R ÿ C(CO2Et)2 (R = H alkyl or CO2Et) for kinetic studies of their addition reactions with alkenes and with tert-butyl isocyanide by a-hydrogen-atom abstraction from the corresponding esters in the presence of tert-butoxyl radicals.11 The selectivity of amine–borane-promoted hydrogen-atom transfer to the tert-butoxyl radical depends on the nature of the catalyst because it is the amine–boryl radical that is responsible for hydrogen abstraction. This is clearly illustrated by the relative reactivities of CH3CO2Et MeCH2CO2Et and Me2CHCO2Et towards catalysed abstraction of hydrogen from their a- C–H groups (Table 1).5,10 The data can be understood in terms hn (9) (10a) NCCH • 2CHOCH2CH2CN 2 C8H17 O 2ButO• 4 (10b) 5 X = H 6 X = unpaired electron • NCCHCH O OCH3 3 H2C O 7 2OCH2CH2CN O O X Chem.Soc. Rev. 1999 28 25–35 27 Fig. 4 EPR spectra obtained when tert-butoxyl radicals are generated in the presence of an equimolar mixture of tert-butyl methyl ether and diethyl malonate at 284 °C (a) in the absence of a catalyst and (b) in the presence of Me3N?BH2Thx. The central multiplet in the spectrum of ButOCH ÿ 2 is broadened as a consequence of restriction of rotation about the ÿ C–O bond. Table 1 Relative rates of a-hydrogen-atom transfer from esters to the tertbutoxyl radical catalysed by amine–boranes at 284 °C Ester reactivity (per molecule) Amine–borane catalyst CH3CO2Et MeCH2CO2Et Me2CHCO2Et Me3N?BH3 Me3N?BH2Me 3N?BH2Bus 7.2 7.3 2.4 0.5 6.3 6.7 4.9 4.6 (1) (1) (1) (1) Me Me3N?BH2Thx abstract hydrogen much more slowly from the diene than from the triene as would be expected on the basis of polar effects.However in the presence of Me3N?BH2Thx as hydridic polarity-reversal catalyst hydrogen abstraction (now by the nucleophilic amine–boryl radical) takes place exclusively from the diene because the cyclopentadienyl anion is aromatic while the cycloheptatrienyl anion is not the structure [Me3N?B + HThx ÿ H R–] contributes to the transition state only when R is cyclopentadienyl.In fact in its reaction with cyclopentadiene alone the tert-butoxyl radical prefers to add to the ring to give the cyclopentenyl adduct 8 [eqn. (11a)] (see Fig. 5a) rather than to abstract hydrogen to give the cyclopentadienyl radical. However in the presence of an amine–alkylborane as polarity-reversal catalyst only the cyclopentadienyl radical was detected by EPR spectroscopy [eqn. (11b)] (see Fig. 5b).12 H OBut • no catalyst (11a) 8 ButO• + with Me3N BH2Thx catalyst • (11b) - ButOH An interesting example of PRC is provided by the reactions of the primary and secondary amine–boranes RNH2?BH3 and Chem. Soc. Rev. 1999 28 25–35 28 Fig.5 EPR spectra obtained when tert-butoxyl radicals are generated in the presence of cyclopenta-1,3-diene at 2116 °C (a) in the absence of a catalyst 3N?BH2Thx. and (b) in the presence of Me R2NH?BH3 with tert-butoxyl radicals.13 Although the EPR spectra of the corresponding aminyl–borane radicals R ÿ NH?BH3 and R2 ÿN ?BH3 were observed as the ultimate reaction products it is clear that these electrophilic species are formed indirectly through the intermediacy of the less stable but nucleophilic amine–boryl radicals RNH2?BH ÿ 2 and R2NH?ÿBH2. Polar effects direct abstraction by ButO· initially to the B–H group and the amine–boryl radicals formed then rapidly abstract hydrogen from the parent amine–borane to give the thermodynamically more stable isomeric aminyl–borane radicals [e.g.eqn. (12)]. The amine–borane is here serving as a polarity-reversal catalyst for hydrogen abstraction from itself! t Me NHABH3 2 Bu O· -ButOH · · (12) [Me NHABH ] 2 2 3 2 R R O El (13) ButOH OOBut H El A 2 +Me NHABHA3 Me NABH 2 -Me NHABH3 3.1 Radical-chain reactions PRC has been used to control reactivity and selectivity in radical-chain reactions for functionalisation a to an ester carbonyl group.14 Thus in the presence of quinuclidine–borane as catalyst methyl acetate dimethyl malonate triethyl methanetricarboxylate and ethyl cyanoacetate (H–El) each react with allylic tert-butyl peroxides at 30 °C in benzene solvent to give products resulting from 2,3-epoxypropylation at an a-C–H group as generalised in eqn.(13).14 The propagation stage of this radical-chain process is illustrated in Scheme 2 for the reaction of methyl acetate with tert-butyl 1,1-dimethylallyl peroxide. The function of the amine–borane is to increase the rate of overall hydrogen-atom transfer from the ester to the tertbutoxyl radical and to direct abstraction exclusively to the electron-deficient a-C–H group (Scheme 2 steps a and b); no epoxypropylation at the ester-methyl group was detected. ButOH BH R3N 3 BH • R3N ButO• 2 a O MeCO2Me d R3N BH3 c • MeO2CCH2 b •CH2CO2Me OOBut Scheme 2 OOBut In the absence of the amine–borane such epoxypropylation reactions are sluggish and require a much higher temperature. For example when a dilute solution of allyl tert-butyl peroxide 9 in methyl acetate as solvent was heated in an autoclave at 140 °C for 10 h the epoxypropylation product consisted of 10 and 11 in the ratio of 7:3 reflecting the low selectivity with which the tert-butoxyl radical abstracts hydrogen from the two types of C–H bond present in the ester.15 O OOBut MeO2CCH2 O 9 11 Related chain reactions of esters with vinylic epoxides to yield allylic alcohols are also catalysed by amine–boranes (e.g.Scheme 3).15 Again the role of the amine–borane is to promote MeO2CCH2 El • MeC(O)OCH2 El 10 O CH(CO2Me)2 (14) O O H2C(CO2Me)2 • El OH Scheme 3 Reagents and conditions ButOOBut + UV light initiator amine– borane catalyst 30 °C.regioselective abstraction of hydrogen from the a-C–H group of the ester but now steps c and d in Scheme 2 are replaced by the addition of the electrophilic carboxyalkyl radical to the vinyl epoxide to give an intermediate oxiranylcarbinyl radical which then undergoes rapid ring opening [eqn. (14)]. The allyloxyl O• radical thereby produced goes on to abstract hydrogen from the amine–borane catalyst to give the allylic alcohol and regenerate the chain-carrying amine–boryl radical. 4 Thiols and selenols as protic polarity-reversal catalysts As part of a Faraday Society Discussion in 1953 Barrett and Waters reported that thiols catalyse the radical-chain decarbonylation of aldehydes [eqns. (15) and (16)].16 In the general discussion that followed this paper F.R. Mayo suggested an explanation for the catalysis based on the key role of polar effects. Mayo pointed out that the chain-propagating abstraction of hydrogen from an aldehyde by an alkyl radical [eqn. (16)] does not benefit from favourable polar effects in the transition state because both the alkyl radical and the acyl radical are nucleophilic the reaction is thus an example of the general type represented in eqn. (5). Mayo proposed that the catalysis of the overall hydrogen transfer reaction (16) through the cycle of reactions (17) and (18) could be understood because the thiyl radical is electrophilic. In the general terminology adopted here the thiol is acting as a protic polarity-reversal catalyst for reaction (16) and this type of catalytic cycle can be used with advantage in several classes of radical-chain reaction.(15) (16) (17) (18) R ÿC NO —? R• + CO R• + RCHO —? RH + R ÿC NO R• + XSH —? RH + XS• XS• + RCHO —? XSH + R ÿC NO 4.1 Hydroacylation of alkenes The intermolecular radical-chain addition of an aldehyde to an alkene to give a ketone (hydroacylation) was first studied halfa-century ago and the simple addition of primary aldehydes (RCH2CHO) to electron-deficient alkenes (e.g. a,b-unsaturated ketones and dialkyl maleates) can give good yields of adducts. However a major problem with the propagation stage of the radical-chain pathway [reactions (19) and (20)] remains the inefficiency with which the acyl-radical adduct 12 abstracts hydrogen from the aldehyde [reaction (20)].17 (19) C• RC(O) C • RC O C C 12 (20) H C RC(O) C C• RC(O) C RCHO • RC O In view of the foregoing discussion it would be anticipated that thiols would act as polarity-reversal catalysts for radicalchain hydroacylation reactions provided that loss of the thiol by addition to the alkene does not cause problems and it has been shown that thiols do indeed promote the addition of primary aldehydes to a variety of alkenes under mild conditions.17 While thiol catalysis is effective for the hydroacylation of electronrich -neutral and -poor alkenes it is most efficient for addition to electron-rich double bonds.For example the addition of butanal to isopropenyl acetate at 60 °C in the presence of ditert-butyl hyponitrite as initiator and methyl thioglycolate (MeO2CCH2SH) as catalyst affords the adduct 13 in 80% yield (see Scheme 4) while a similar reaction in the absence of thiol gives only 8% yield.Such addition of an aldehyde to an enol derivative provides a non-ionic route to acylated or silylated aldol adducts.17 O OAc MeO2CCH2SH Pr C• PrCHO OAc O • MeO2CCH2S• Pr OAc O MeO2CCH2SH Pr 13 Scheme 4 4.2 Dehalogenation deoxygenation and desulfurisation by silanes A trialkylsilyl group shows many properties in common with those of an acyl group. Both are p-acceptors the corresponding 29 Chem. Soc. Rev. 1999 28 25–35 radicals are both nucleophilic and the Si–H bond in R3SiH is weaker than many aliphatic C–H bonds as is the aldehydic C–H bond in RCHO.Currently-quoted bond dissociation enthalpies (in kJ mol21) are MeC(O)–H 374 Et3Si–H 398 and (Me3Si)3Si–H 351.8,18 For comparison DH(MeS–H) is 365 kJ mol21 and DH(Bu3Sn–H) is 308 kJ mol21.8 The removal of a functional group G from an organic compound R–G and its replacement by hydrogen to give R–H is a basic transformation of considerable importance in synthetic organic chemistry. Tributyltin hydride is pre-eminent amongst reagents for the homolytic reductive removal of functional groups and such reactions follow the chain mechanism generalised in eqns. (21) and (22) [reaction (21) is sometimes a stepwise addition–elimination process]. However for practical (21) (22) Bu3Sn• + R–G —? Bu3Sn–G + R• R• + Bu3Sn–H —? R–H + Bu3Sn• and ecological reasons it would be desirable to use simple readily available silanes in place of trialkyltin hydrides.The corresponding propagation cycle using triethylsilane is shown in eqns. (23) and (24) and while reaction (23) is generally faster than its tin counterpart reaction (24) is relatively slow (23) Et3Si• + R–G —? Et3Si–G + R• R• + Et (24) 3Si–H —? R–H + Et3Si• at moderate temperatures because of the greater strength of the Si–H bond as compared with the Sn–H bond. As a consequence reductions using simple silanes are not generally viable under mild conditions. Although reaction (24) is usually exothermic in common with the corresponding abstraction of hydrogen from an aldehyde [eqn. (16)] it does not benefit from favourable polar effects because both the alkyl radical and the silyl radical are nucleophilic.This analysis suggests that the overall hydrogen-atom transfer shown in eqn. (24) should be promoted by a protic polarity-reversal catalyst and it has been shown that thiols can serve in this capacity.19 The trialkylsilane –thiol couple acts an effective replacement for tributyltin hydride for the reduction of alkyl halides (bromides and chlorides — the latter are not usually reduced efficiently by the tin hydride) dialkyl sulfides and the S-methyl dithiocarbonate (xanthate) esters derived from primary and secondary alcohols. For example ethyl 4-bromobutanoate 14a was reduced to ethyl butanoate 14b in essentially quantitative yield by four equivalents of triethylsilane in refluxing cyclohexane in the presence of dilauroyl peroxide as initiator and tert-dodecanethiol (mixture of isomers) as polarity-reversal catalyst.19 Reduction of cholestanyl xanthate 15a with triethylsilane in refluxing octane with di-tert-butyl peroxide initiator and tert-dodecanethiol catalyst afforded cholestane 15b in 94% isolated yield.However appreciable yields of cholestane were obtained from the peroxide-initiated reduction by triethylsilane in the absence of a thiol catalyst and it seems likely that this reduction is promoted by a thiol formed in situ from the xanthate.19 C8H17 XCH2(CH2)2CO2Et 14a X = Br 14b X = H X H 15a X = OC(S)SMe 15b X = H 4.3 Hydrosilylation of alkenes Hydrosilylation of alkenes [eqn.(25)] is an important method for the formation of Si–C bonds and such addition reactions can Chem. Soc. Rev. 1999 28 25–35 30 proceed by a radical-chain mechanism [eqns. (26) and (27)] or under the influence of various transition metal catalysts in particular rhodium palladium and platinum complexes. However radical-chain hydrosilylation of alkenes using trialkylsilanes has not found much use in synthesis because the hydrogen-atom abstraction step [eqn. (27)] is relatively slow at moderate temperatures and competing telomerisation of the alkene can also be a problem. Again reaction (27) should be subject to PRC by thiols and provided addition of the catalyst to the alkene can be suppressed thiols should therefore catalyse the radical-chain hydrosilylation of alkenes.This has been realised in practice20,21 and when addition of the thiol catalyst to the alkene was a problem this could usually be overcome by adding the former slowly to the reaction mixture using a syringe pump. For example a good yield of the triethylsilane adduct 17 (25) + R3SiH C C H C C R3Si (26) + C• C C C R3Si • R3Si + + C• C (27) C C H R R3SiH R3Si R3Si 3Si • was obtained by hydrosilylation of diethyl allylmalonate 16 at 60 °C using tert-dodecanethiol as protic polarity-reversal catalyst. However the potential difficulty caused by loss of the thiol by addition to the alkene is highlighted by the very low yield of silane-addition products obtained from the corresponding reaction of diethyl diallylmalonate 18.20 Here it appears that addition of the thiyl radical to one of the double bonds is rendered effectively irreversible by the rapid 5-exo-cyclisation of the adduct radical and it is evident that for the thiol catalysis to be successful any addition of the thiyl radical to the alkene should be reversible under the reaction conditions.H H (EtO2C)2C (EtO2C)2C (EtO2C)2C SiEt3 18 17 16 Thiol catalysis of hydrosilylation is more effective for the addition of arylsilanes than trialkylsilanes presumably because the weaker Si–H bond in the former results in more rapid abstraction of hydrogen by the thiyl radical to form the corresponding silyl radical.21 Furthermore methyl thioglycolate (MeO2CCH2SH) and triphenylsilanethiol (Ph3SiSH) are generally more efficient hydrosilylation catalysts than tertdodecanethiol again probably because of an increase in the rate of abstraction of hydrogen from the silane by the thiyl radical this time because of an increase in the strength of the S–H bonds and in the electrophilicities of the thiyl radicals involved.21–23 Thiols have also been shown to catalyse the addition of tris(trimethylsilyl)silane [(Me3Si)3SiH] to alkenes.23 Intramolecular radical-chain hydrosilylation leading to the cyclisation of alkenyloxysilanes is also catalysed by thiols.22 For example the allyloxydiphenylsilane 19 (see Scheme 5) underwent almost quantitative cyclisation to give the oxasilacyclopentane 20 at 60–65 °C in the presence of di-tert-butyl hyponitrite as initiator and tert-dodecanethiol as catalyst; no cyclisation took place in the absence of thiol.The propagation stage of the chain reaction is shown in Scheme 5 and involves 5-endo-cyclisation of the intermediate silyl radical 21. The attempted thiol-catalysed tandem cyclisation of the allyloxysilane 22 failed presumably for the same reason as did • SiPh2 XSH O H SiPh2 O • 19 SiPh2 XS• O XSH SiPh2 O 20 SiPh2 H SiPh2 O O 24 22 (28) + C C XS C• XS• R + + (29) R3SiH 3Si • XS• 21 Scheme 5 the intermolecular addition of triethylsilane to diethyl diallylmalonate 18 because addition of thiyl radicals to either end of the diene system is rendered irreversible by the rapid 5-exocyclisation of the adduct radical formed.For effective catalysis of hydrosilylation the conditions must be chosen such that the major fate of the thiyl radical is to be converted into a silyl radical which then adds irreversibly to the CNC group. Both reactions (28) and (29) are potentially reversible and detailed H SiPh2 O 23 C XSH kinetic analysis of catalysed hydrosilylation reactions is a complex problem. The efficiency with which the thiyl radical is converted into the silyl radical will depend on the structures of the particular radicals involved and on the natures and relative concentrations of the thiol silane and alkene. The kinetics and thermodynamics of the thiyl radical–silane reaction (29) are of critical importance for catalysed hydrosilylation and also for the successful use of the silane–thiol couple in other situations.Accepted values for DH(Et3Si–H) and DH(MeS–H) at the time the silane–alkanethiol couple was first introduced for reduction19 indicated that the abstraction of hydrogen from triethylsilane by an alkanethiyl radical was exothermic by ca. 7 kJ mol21 but currently promulgated values (see above) imply that it is endothermic by ca. 33 kJ mol21,8 as has been noted by Zavitsas.24 However in our view since the alkanethiol– triethylsilane couple functions effectively at moderate temperatures reaction (29) is very unlikely to be endothermic in the forward direction by more than 10–20 kJ mol21 when X = R = simple alkyl. It seems likely that the S–H bond in an alkanethiol may be stronger and/or the Si–H bond in a trialkylsilane may be weaker than the most recently proposed values.Alternatively some of the alkanethiol may be converted to R3SiSH in which the S–H bond is probably stronger,19 under the reaction conditions.23 Another drawback with the silane–thiol system is illustrated by the low yield of intramolecular hydrosilylation product obtained from the tert-dodecanethiol-catalysed cyclisation of the allyloxysilane 23 to give the oxasilacyclopentane 24.22 It was thought likely that the alkanethiyl radical abstracts hydrogen from the allylic C–H groups in 23 (in particular the C–H group adjacent to oxygen) to give a stabilised allylic radical incapable of propagating the chain in competition with the desired abstraction from the Si–H group.This interpretation was supported by the observation that the corresponding cyclisation of 19 (see Scheme 5) which lacks such allylic C–H groups was inhibited by a small amount of allyloxytrimethylsilane (Me3SiOCH2CHNCH2) which does possess them. The silanethiols Pri 3SiSH and Ph3SiSH turned out to be much more effective catalysts for the cyclisation of 23 and reasonably good yields of 24 were obtained in their presence. It was suggested that a silanethiyl radical abstracts hydrogen more rapidly and/or selectively from the Si–H group in 23 than does an alkanethiyl radical. Thiol-catalysed cyclisation of homoallyloxysilanes was also successful and again silanethiols were generally the most successful catalysts.For example the but-3-enyloxysilane 25 underwent radical-chain cyclisation to give a 72:28 mixture of the oxasilacyclohexane 26 and the oxasilacyclopentane 27 arising from competitive 6-endo- and 5-exo-ring closure respectively of the intermediate silyl radical 28.22 • SiPh2 SiPh2 H SiPh2 O O O SiPh2 O 27 28 26 25 Finally cyclisation of the homoallyloxysilane 29 gives only the oxasilacyclohexanes 30 and 31 because 5-exo-cyclisation of the intermediate silyl radical is retarded by the methyl group on the double bond and the cis:trans ratio in the product depends on the nature of the thiol catalyst. The less stable transisomer predominates because equatorial attack of the thiol on the intermediate oxasilacyclohexyl radical 32 is favoured over axial attack which would incur a repulsive steric interaction with the axial phenyl group attached to silicon.However it was also shown that the trans-isomer was converted to the more stable cis form in the presence of initiator and Ph3SiSH at 65 °C although almost no isomerisation took place in the presence of MeO2CCH2SH. H SiPh2 SiPh2 SiPh2 O O O 31 30 29 HSX SiPh2 O Me Ph Si O Me Ph 32 33 Presumably the relatively electrophilic Ph3SiS· abstracts hydrogen reversibly from the activated C–H bond adjacent to oxygen allowing isomerisation to take place via the intermediate radical 33.22 4.4 Reductive carboxyalkylation of alkenes Depending on the nature of the substituents present a silyl radical generally abstracts bromine from an alkyl bromide much more rapidly than it adds to a terminal alkene.18 Abstraction of bromine from an a-bromoester would be expected to be still more rapid and the resulting a-carboxyalkyl radical should add relatively rapidly to an electron-rich alkene to give a nucleophilic carbon-centred radical which in turn should abstract 31 Chem.Soc. Rev. 1999 28 25–35 hydrogen relatively rapidly from a thiol all because of favourable polar effects in the respective transition states. This analysis suggests that inclusion of an a-bromoester in a reaction system designed originally for thiol-catalysed hydrosilylation of an electron-rich alkene could result in interception of the silyl radical by the halogenoester and lead to reductive carboxyalkylation of the alkene through the chain-propagation cycle shown RHal Ph3SiHal R• Ph3Si• EDG XSH Ph3SiH EDG XS• • R EDG XSH R Scheme 6 in Scheme 6 (EDG = electron-donating group).Such reactions have been shown to provide viable methods for C–C bond formation and specially-reactive chlorides such as dimethyl chloromalonate may be used with advantage in place of the corresponding bromides.25 For example reductive carboxyalkylation at 60 °C of the enol acetate 34 with dimethyl chloromalonate and triphenylsilane in the presence of Ph3SiSH as protic polarity-reversal catalyst affords the adduct 35 in good yield (86%). O OAc OAc Pri (MeO2C)2CH Me H Ph 35 34 36 H 3)3 OSi(SiMe Pri Pri Pri 3)3 OSi(SiMe3)3 OSi(SiMe H • Me Me Me Ph H H Ph Ph H 38 37b 37a 4.5 Hydrosilylation of ketones The radical-chain hydrosilylation of ketones with tris(trimethylsilyl) silane initiated by di-tert-butyl hyponitrite at 30 °C is evidently catalysed by tert-dodecanethiol although this was not stated explicitly.26,27 Presumably the mechanism is analogous to that of the thiol-catalysed hydrosilylation of an alkene; the chain-carrying a-siloxyalkyl radical formed by addition of (Me3Si)3Si· to the carbonyl-oxygen atom is nucleophilic and abstracts hydrogen more rapidly from the thiol catalyst than from the silane.For example addition of (Me3Si)3SiH to the ketone 36 gives a 12.6:1 mixture of the adducts 37a and 37b and the predominance of the former reflects the preference for the thiol to attack the intermediate radical 38 from its less-hindered bottom face.The isomer ratio should depend on the nature of the thiol but this was not investigated. Assuming that the thiol catalyst is the sole or major hydrogen-atom donor the results of experiments involving the (Me3Si)3SiD–RSH couple appear to require that H/D exchange between the thiol and the silane [cf. eqn. (29)] is rapid under the reaction conditions. 4.6 Applications to other reactions PRC has been applied to the radical-chain addition of tributyltin hydride to terminal alkynes.28 For example the reaction of Chem. Soc. Rev. 1999 28 25–35 32 excess Bu3SnH with methyl propiolate 39 to give the product of double hydrostannylation 40 is catalysed by p-methoxythiophenol.The vinylstannane MeO2CCHNCHSnBu3 is formed first but addition of Bu3Sn· to this to give the radical 41 is highly reversible. In the presence of the arenethiol the adduct radical 41 (which is probably relatively nucleophilic by virtue of the presence of the two b-Bu3Sn substituents) is rapidly and irreversibly trapped to give 40 and p-MeOC6H4S· which then goes on to abstract hydrogen from the tin hydride and regenerate the thiol catalyst. In the absence of the thiol trapping of 41 by the tin hydride is inefficient. SnBu SnBu 3 3 H H C • MeO2CCH C MeO2CCH2 MeO2CC CH SnBu3 SnBu3 39 41 40 Most arenethiols are ineffective as polarity-reversal catalysts for hydrogen-atom transfer from silanes to alkyl radicals because the ArS–H bond is appreciably weaker than that in an alkanethiol and the equilibrium shown in eqn.(29) lies far to the left. In fact thiophenol inhibited the trace amount of addition of PhMe2SiH to isopropenyl acetate that was observed in the absence of any thiol.17,21 An exception was provided by 2,4,6-tris(trifluoromethyl)thiophenol which did catalyse the hydrosilylation of isopropenyl acetate.17 Presumably here the S–H bond is strengthened by the presence of the three electronwithdrawing CF3 groups on the ring. The Si–H bond in a silane is much stronger than the Sn–H bond in the corresponding tin hydride accounting for the efficacy of simple arenethiols as protic polarity-reversal catalysts for hydrogen transfer from the latter.Crich and his co-workers have reported the use of benzeneselenol as a polarity-reversal catalyst for the abstraction of hydrogen from tin hydrides by carbon-centred radicals.29–31 The electronegativity of selenium is only marginally less than that of sulfur and PhSe· is expected to exhibit electrophilic properties like PhS·. The Se–H bond is much weaker than the S–H bond and both enthalpic and polar factors favour abstraction of hydrogen from the selenol by a nucleophilic alkyl radical a process which is extremely rapid at room temperature and significantly faster than the direct abstraction of hydrogen from the tin hydride.32 However polar effects favour abstraction of hydrogen by PhSe· from the tin hydride and the catalytic cycle involved is shown in eqns.(30) and (31). It was pointed (30) R• + PhSeH —? RH + PhSe• (31) PhSe• + Bu3SnH —? PhSeH + Bu3Sn• out that undesired radical rearrangement processes which are sufficiently rapid to proceed in the presence of tin hydride alone can be suppressed in the presence of PhSeH (added as such or formed in situ by reduction of PhSeSePh) because the precursor radical is trapped by the selenol before the rearrangement can take place.29 PRC by benzeneselenol of the overall transfer of hydrogen from tributyltin hydride to relatively unreactive radicals such as allylic and cyclohexadienyl radicals has proved useful in increasing the efficiency of chain reactions involving these species.31 The much greater rate at which a nucleophilic carboncentred radical R· abstracts hydrogen from PhSeH as compared with Bu3SnH has also been exploited in experiments designed to measure the rate of rearrangement of R· using the so-called radical-clock method.30 The ‘clock reaction’ is the trapping of the alkyl radical R· by a constant catalytic quantity of the selenol under pseudo-first-order conditions,32 obviating the need to work with a large excess of the tin hydride in order to achieve simple reaction kinetics.Exchange reactions of the type shown in eqn. (32) between (32) R1• + R2–H —? R1–H + R2• one alkyl radical and a hydrocarbon to give a similar alkyl radical generally have relatively large activation energies and are very slow at moderate temperatures. In the high-temperature pyrolyses of hydrocarbons in the gas phase it has been demonstrated that the inclusion of HCl HBr or H2S can accelerate reactions of the type (32) and modify the end-product distributions from chain reactions that involve this elementary step.33 This phenomenon was referred to as ‘hydrogen transfer catalysis’ although the probable part played by polar effects was not discussed.Thermoneutral or nearly thermoneutral hydrogen-atom transfer between two strongly nucleophilic carbon radicals [cf. eqn. (5)] should be particularly susceptible to PRC by a protic catalyst of the type H–El and similarly transfer between two electrophilic carbon radicals [cf. eqn. (4)] should respond well to PRC by an hydridic catalyst H–Nuc. It has been reported that radical-induced racemisation of (R)- tetrahydrofurfuryl acetate 42 at 60 °C can be induced by certain thiols (XSH) through the chain mechanism shown in Scheme 7.34 The thiol is here acting as a polarity-reversal catalyst for the H O AcO AcO XS• H XSH O O ( R)-42 AcO XSH XS• AcO ( R)-42 O H ( S)-42 AcO AcO AcO O O O O (33) + + H H 43 42 43 Scheme 7 thermoneutral transfer of hydrogen between the nucleophilic radical 43 and the parent ester 42 [eqn.(33)]. The nature of the AcO group X in the thiol is crucially important in determining catalyst efficiency and racemisation was most rapid when X was an electron-withdrawing group. Triphenylsilanethiol was the most effective of the thiols investigated while simple alkanethiols were very inefficient catalysts and it was thought that the p-electron withdrawing silyl substituent increases both the strength of the S–H bond in the thiol and the electrophilicity of Ph3SiS• compared with the situation when X is an alkyl group.It was pointed out that a-alkoxyalkyl radicals similar in structure to 43 are involved in the radical-induced strand cleavage of DNA under anaerobic conditions as initial products of hydrogen abstraction from the 4A-position and after strand cleavage a similar a-alkoxyalkyl radical centre is generated in the oligonucleotide fragment. By promoting the reaction of this fragment radical with undamaged DNA appropriate thiols might serve as polarity-reversal catalysts to amplify the radicalinduced damage caused intentionally to the DNA in tumour cells in vivo during radio- or chemo-therapy.34 (34) 5 Enantioselective hydrogen-atom abstraction Reactions that involve enantioselective atom transfer are relatively rare.Enantioselective hydrogen-atom transfer from and to carbon mediated by a homochiral radical Z*• and the closed-shell molecule Z*–H is generalised in eqn. (34). This Z*· + HCabc >—— Z*–H + ·Cabc reaction proceeds through the diastereoisomeric pair of transition states 44a and 44b and it is the energy difference between a •‡ •‡ Z* Z* H H C a C c b b c 44b 44a these two structures that determines the enantioselectivity of the hydrogen-atom transfer. If the compound Z*–H is a polarityreversal catalyst the possibility of catalytic enantioselective hydrogen-atom transfer arises in chain and non-chain processes that are promoted by PRC.Enantioselective hydrogen transfer under conditions of PRC was first reported in 1991,35 when it was shown that partial kinetic resolution of methyl 2-phenylpropanoate 45 could be achieved during UV photolysis of di-tert-butyl peroxide at 283 °C in the presence of the initially-racemic ester and a catalytic amount of the homochiral amine–borane complex 46. In this non-chain process it is the homochiral amine–boryl radical 47 that is responsible for hydrogen abstraction from the a-C–H group in 45 although the enantioselectivity was small and (R)- 45 was only ca. 2.4 times more reactive than the (S)- enantiomer. Me Me Me X BH BH NCH2CH2N 2 Ph C CO2Me Me Me H 45 46 X = H 47 X = unpaired electron The elementary enantioselective hydrogen-atom abstraction step has been studied in isolation using the EPR technique and high enantioselectivity factors (s)§ were obtained for abstraction from the a-C–H groups of dimethyl 2,2-dimethyl- 1,3-dioxolane-4,5-dicarboxylate 48 by the amine–boryl radical 47 and by related species which contain substituted isopinocampheyl groups.36 It was found that (S,S)-48 was 21 times more reactive towards 47 than (R,R)-48 at 285 °C.Under the experimental conditions (photolysis of ButOOBut in the presence of the ester and catalyst in an inert solvent) the radical 49 and its antipode go on to dimerise and disproportionate. MeO2C MeO2C MeO2C O O O O O O MeO MeO MeO 2C 2C 2C 49 ( S,S)-48 ( R,R)-48 ‡• L S M B C H L M S 50 Partial kinetic resolution of a number of racemic chiral esters and of camphor was brought about by photolysis of ButOOBut in their presence along with a polarity-reversal catalyst of the type exemplified by 46.37 However the values of s were generally small (@5) and the enantiomeric excesses (ees) of the residual substrates were also relatively small (although of course this depends on the amount of substrate consumed).The 33 Chem. Soc. Rev. 1999 28 25–35 large value of s shown by 48 in its reaction with 47 allowed the racemic ester to be successfully resolved. Thus after 75% of the initially-racemic tartrate had been consumed during photolysis of ButOOBut in the presence of the ester and 46 as polarityreversal catalyst at 290 °C the ee of the remaining tartrate was 97% in favour of (R,R)-48.The sense of the observed enantioselectivity could often be understood on the basis that the more stable of the two transition states 44a and 44b is that in which there is a staggered arrangement of the groups attached to boron and to carbon such that long-range torsional- and steric-interactions are minimised as shown in structure 50 (L M and S are large medium and small groups).37 However in general the transition state energies appear to be determined by a subtle interplay of steric stereoelectronic and electrostatic interactions together with the effects of hydrogen-bonding in appropriate systems.38 Ab initio molecular orbital calculations for the prototypical abstraction of hydrogen from acetaldehyde by the ammonia–boryl radical [eqn.(35)] showed that the (35) H3N?ÿBH2 + CH3CHO —? H3N?BH3 + •CH2CHO optimum transition-state geometry is influenced by (i) stereoelectronic factors arising from the need to delocalise the unpaired electron and negative charge onto the carbonyl group (ii) electrostatic interactions between the dipolar CNO and N?B groups and (iii) the existence of hydrogen-bonding between one NH group and the carbonyl-oxygen atom.38 Attempts to discover homochiral amine–borane complexes which are generally efficient polarity-reversal catalysts for the kinetic resolution of chiral carbonyl compounds by enantioselective hydrogen abstraction from a-C–H groups have not been very successful to date.38,39 The polycyclic amine–borane 51 shows high thermal- and air-stability and when it was used as a catalyst for the kinetic resolution of O-trimethylsilylpantolactone 52 at 274 °C lactone with an ee of 84% in favour of (R)-52 was isolated after 71% of the substrate had been consumed (s = ca.5).39 Et Et N O O H BH2 H Me3SiO 52 51 Partial kinetic resolution of racemic trans-2,5-dimethyl- 1-phenyl-1-silacyclopentane 53 [the (2R,5R)-enantiomer is shown] has been brought about by its radical-chain reaction with a deficiency of alkyl bromide in the presence of the homochiral silanethiol 54 as polarity-reversal catalyst.40 Enantioselective abstraction of hydrogen from 53 by the silanethiyl radical 55 takes place as part of the chain-propagation cycle although the optical purities of the residual silane [(2R,5R)-enantiomer in excess] and of the bromosilane 56 [the (2S,5S)-enantiomer shown was in excess] obtained by this route were very low.Ph Ph Ph Ph Si Si Si Si Br S SH H 56 55 54 53 In the above examples of enantioselective hydrogen-atom transfer from and to carbon the transition state 44 is approached from the left-hand side of eqn. (34) i.e. by the reaction of a homochiral radical Z*· with the chiral substrate. More recently the combination of an achiral silane and a homochiral thiol as Chem. Soc. Rev. 1999 28 25–35 34 polarity-reversal catalyst has been utilised to bring about enantioselective hydrosilylation and enantioselective carboxyalkylation of prochiral alkenes.21,23,25 In these reactions the transition state 44 is approached from the opposite direction i.e.by the interaction of a prochiral carbon-centred radical ·Cabc with a homochiral hydrogen-atom donor Z*–H. Radical-chain hydrosilylation of a number of prochiral terminal alkenes [eqn. (36)] has been carried out at 60 °C in the presence of a homochiral thiol as polarity-reversal catalyst.21,23 R1 R1 R3Si R*SH (36) * R3SiH H catalyst R2 R2 57 The stereogenic centre in the adduct 57 is set when the intermediate b-silylalkyl radical abstracts hydrogen from the thiol and the most successful catalysts for inducing asymmetry in the adducts were derived from carbohydrates by introduction of an SH group at the anomeric position.For example addition of triphenylsilane to the methylenelactones 58 and 59 afforded the adducts 60 and 61 in good chemical yield and moderate to high enantiomeric purity [eqn. (37)] when the pyranose thiols R R Ph3SiH 60 oC R R (37) Thiol 63 catalyst * Ph3Si O O O O 60 R = Me (76% ee) 61 R = Ph (95% ee) 58 R = Me 59 R = Ph 62 and 63 were used as catalysts (5 mol% based on alkene).21,23 With the b-glucose thiol 62 as catalyst the adduct 60 was obtained with a 50% ee and this was raised to 76% ee by using the b-mannose thiol 63. Corresponding hydrosilylation of the diphenyl analogue 59 afforded the adduct 61 with an ee of 87% using the b-glucose thiol and with an ee of 95% (isolated chemical yield 90%) using the b-mannose thiol as catalyst.Evidently the extra bulk provided by the gem-b-diphenyl groups in the intermediate radical 64 is responsible for the increase in ee over that obtained with the dimethyl analogue. As discussed previously for the silanethiols and methyl thioglycolate the high chemical yields obtained using these carbohydrate thiols as catalysts are probably a result of the relatively high strength of the S–H bonds and the relatively high electrophilicities of the corresponding thiyl radicals as compared with simple alkane-thiols and -thiyl radicals as a result of the presence of several electronegative oxygen atoms in the molecules.23 R AcO AcO OAc O O R SH SH AcO AcO AcO AcO Ph3Si O O OAc 64 63 62 R R (MeO2C)2CH * O O 65 Enantioselective reductive carboxyalkylation (cf.Scheme 6) of the methylenelactone 58 has also been carried out using the thiols 62 and 63 as catalysts.25 The hyponitrite-initiated reaction of 58 with dimethyl chloromalonate and triphenylsilane at 60 °C in the presence of the b-glucose thiol 62 gave the compound 65 with an ee of 24% and this value was raised slightly to 27% by using the b-mannose thiol 63 as catalyst. 6 Concluding remarks The key role played by polar effects in free-radical chemistry is well established. The basic idea behind PRC is also not new having been put forward by Mayo in 1953 to explain the catalysis by thiols of the radical-chain decarbonylation of aldehydes as reported by Waters and co-workers.16 However it was some 34 years before the generality of the principle of PRC was recognised and applied in a variety of situations.7 Aside from applications to different types of reaction (e.g.catalytic epimerisation at selected chiral carbon centres in molecules that possess several such centres) future progress in this area is likely to focus on the development of new hydridic and protic catalysts H–Nuc and H–El in which the strengths of the bonds to hydrogen are tailored to requirements and the search for generally-applicable homochiral catalysts H–Nuc* and H–El* designed to give greater chiral discrimination in enantioselective hydrogen-atom transfer. 7 Acknowledgements I would like to express my gratitude to all my co-workers past and present especially Dr Hai-Shan Dang who have contributed so much to our work in the area of polarity-reversal catalysis.8 References 1 C. Walling Free Radicals in Solution John Wiley & Sons Inc. New York 1957. 2 J. R. M. Giles and B. P. Roberts J. Chem. Soc. Perkin Trans. 2 1982 1699 (editorially-corrected version J. Chem. Soc. Perkin Trans. 2 1983 743). 3 J. A. Baban and B. P. Roberts J. Chem. Soc. Perkin Trans. 2 1987 497 and earlier papers cited therein. 4 V. Paul and B. P. Roberts J. Chem. Soc. Perkin Trans. 2 1988 1183. 5 V. Paul B. P. Roberts and C. R. Willis J. Chem. Soc. Perkin Trans. 2 1989 1953. (See also B. P. Roberts and A. J. Steel J. Chem. Soc. Perkin Trans.2 1994 2411 for the revised calculated ionisation energy of H3N?BHMe.) ÿ 6 B. P. Roberts and A. J. Steel J. Chem. Soc. Perkin Trans. 2 1994 2155; B. P. Roberts J. Chem. Soc. Perkin Trans. 2 1996 2719. 7 V. Paul and B. P. Roberts J. Chem. Soc. Chem. Commun. 1987 1322. 8 CRC Handbook of Chemistry and Physics ed. D. R. Lide 78th edn. CRC Press Boca Raton 1997. 9 A. G. Davies D. Griller and B. P. Roberts J. Chem. Soc. (B) 1971 1823; J. A. Baban and B. P. Roberts J. Chem. Soc. Perkin Trans. 2 1981 161. 10 P. Kaushal P. L. H. Mok and B. P. Roberts J. Chem. Soc. Perkin Trans. 2 1990 1663. 11 V. Diart and B. P. Roberts J. Chem. Soc. Perkin Trans. 2 1992 1761. 12 V. Paul B. P. Roberts and C. A. S. Robinson J. Chem. Res. (S) 1988 264. 13 I.G. Green and B. P. Roberts J. Chem. Soc. Perkin Trans. 2 1986 1597; J. N. Kirwan and B. P. Roberts J. Chem. Soc. Perkin Trans. 2 1989 539. 14 H.-S. Dang and B. P. Roberts J. Chem. Soc. Perkin Trans. 1 1993 891. 15 E. Montaudon F. Rakotomanana and B. Maillard Tetrahedron 1985 41 2727. 16 K. E. J. Barrett and W. A. Waters Faraday Discuss. Chem. Soc. 1953 14 221. [See also E. F. P. Harris and W. A. Waters Nature (London) 1952 170 212.] 17 H.-S. Dang and B. P. Roberts J. Chem. Soc. Perkin Trans. 1 1998 67 and earlier papers cited therein. 18 C. Chatgilialoglu Chem. Rev. 1995 95 1229. 19 S. J. Cole J. N. Kirwan B. P. Roberts and C. R. Willis J. Chem. Soc. Perkin Trans. 1 1991 103 and earlier papers cited therein. 20 H.-S. Dang and B. P. Roberts Tetrahedron Lett. 1995 36 2875. 21 M. B. Haque and B. P. Roberts Tetrahedron Lett. 1996 37 9123. 22 Y. Cai and B. P. Roberts J. Chem. Soc. Perkin Trans. 1 1998 467. 23 M. B. Haque B. P. Roberts and D. A. Tocher J. Chem. Soc. Perkin Trans. 1 1998 2881. 24 A. A. Zavitsas J. Chem. Soc. Perkin Trans. 2 1998 499. 25 H.-S. Dang K.-M. Kim and B. P. Roberts Chem. Commun. 1998 1413. 26 B. Giese W. Damm J. Dickhaut F. Wetterich S. Sun and D. P. Curran Tetrahedron Lett. 1991 32 6097. 27 B. Giese M. Bulliard J. Dickaut R. Halbach C. Hassler U. Hoffmann B. Hinzen and M. Senn Synlett 1995 116. 28 J.-C. Meurice M. Vallier M. Ratier J.-G. Duboudin and M. Pétraud J. Organomet. Chem. 1997 542 67. 29 D. Crich and Q. Yao J. Org. Chem. 1995 60 84. 30 D. Crich X.-Y. Jiao Q. Yao and J. S. Harwood J. Org. Chem. 1996 61 2368. 31 D. Crich and X.-S. Mo J. Org. Chem. 1997 62 8624. D. Crich and J.-T. Hwang J. Org. Chem. 1998 63 2765. 32 M. Newcomb Tetrahedron Lett. 1993 49 1151. 33 C. Rebick in Frontiers of Free Radical Chemistry ed. W. A. Pryor Academic Press New York 1980 pp. 117–137. 34 Y. Cai and B. P. Roberts Chem. Commun. 1998 1145. (See also M. S. Akhlaq H. P. Schuchmann and C. von Sonntag Int. J. Radiat. Biol. 1987 51 91.) 35 P. L. H. Mok and B. P. Roberts J. Chem. Soc. Chem. Commun. 1991 150. 36 P. L. H. Mok and B. P. Roberts Tetrahedron Lett. 1992 33 7249. 37 P. L. H. Mok B. P. Roberts and P. T. McKetty J. Chem. Soc. Perkin Trans. 2 1993 665. 38 H.-S. Dang V. Diart B. P. Roberts and D. A. Tocher J. Chem. Soc. Perkin Trans. 2 1994 1039. 39 H.-S. Dang V. Diart and B. P. Roberts J. Chem. Soc. Perkin Trans. 1 1994 1033 (corrigendum p. 2511). 40 H.-S. Dang and B. P. Roberts Tetrahedron Lett. 1995 36 3731. Review 8/04291H 35 Chem. Soc. Rev. 1999 28 25–35

ISSN:0306-0012    

DOI:10.1039/a804291h

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Fluorous phase separation techniques in catalysis Elwin de Wolf,a Gerard van Kotena and Berth-Jan Deelman*b a Debye Institute Department of Metal-Mediated Synthesis Utrecht University Padualaan 8 3584 CH Utrecht The Netherlands b Elf Atochem Vlissingen B.V. PO Box 70 4380 AB Vlissingen The Netherlands Received (in Cambridge) 20th July 1998 Fluorous solvents have limited miscibility with conventional organic solvents. Combined with the fact that compounds functionalized with perfluorinated groups often dissolve preferentially in fluorous solvents this can be used to extract fluorous components from reaction mixtures. This review discusses the application and potential of fluorous phase separation techniques for the recovery of soluble metal catalysts. 1 Introduction Perfluorocarbon fluids especially perfluoro-alkanes ethers and amines have some unique properties which make them attractive alternatives for conventional organic solvents.Among others they are extremely inert apolar and thermally stable allowing vigorous reaction conditions to be employed.1 In addition these fluids and other compounds containing Elwin de Wolf graduated from Utrecht University in a subject in heterogeneous catalysis in 1997. Currently he is studying for his PhD at the Department of Metal-Mediated Synthesis of Utrecht University where he is working on the synthesis of novel ligands for fluorous phase catalysis. Berth-Jan Deelman obtained his PhD degree from Groningen University in 1994 with Professor J.H. Teuben at the Department of Organic and Molecular Inorganic Chemistry.After a post-doctoral period at the University of Sussex (UK) working with Professor M.F. Lappert on the development of new catalysts for olefin polymerization he accepted a research position at Elf Atochem Vlissingen B.V. As part of a collaboration between Elf Atochem and Utrecht University he is currently Elwin de Wolf Berth-Jan Deelman perfluoroalkyl chains often have low surface energies.2 Technical application of perfluorocarbon fluids benefits from the fact that they are available in a broad range of boiling points that they are generally not miscible with water non-toxic and even biocompatible.3 Use of perfluorocarbons however also may have some disadvantages. Well known are the C1- and C2-fluorocarbons (freons) which are greenhouse gases and have become because of their inertness a major environmental problem.These compounds are thought to be responsible for depleting the stratospheric ozone layer. Their boiling points however are much lower than the boiling points of the higher perfluoroalkanes. As a consequence of this higher perfluoroalkanes have lower vapor pressures and might therefore cause less environmental problems than their smaller chain analogues. About the impact however of longer perfluoroalkanes on the ozone layer and the greenhouse effect less is known. Because of their unique properties perfluorocarbons and perfluoroalkylated compounds have been applied as corrosion a part-time lecturer at Utrecht University and is involved in a joint research program on the development of novel ligand systems for application in solution phase catalysis.Gerard van Koten obtained his PhD from Utrecht University (with Professor G.J.M. van der Kerk) during his stay in the Laboratory for Organic Chemistry (TNO) in Utrecht (1967–1977). After a period in the Inorganic Chemistry Department at the University of Amsterdam where he was promoted to Professor (1984) he went back to Utrecht University (Debye Institute) in 1986 to become Professor of Organic Chemistry. He has been Visiting Professor in Strasbourg Heidelberg and Sassari. Research interests comprise the organometallic chemistry of late (Ni Pd Pt Ru) and early transition metals (Ta La Lu) as well as of Cu Li and Zn and the development and use of chelating arylamine (‘pincer’) and aminoarenethiolate bonded organometallic complexes as catalysts for homogeneous catalysis in particular for fine-chemical synthesis.The preparation and use of the first examples of homogeneous dendrimer catalysts demonstrate his interest in supramolecular systems with (organometallic) catalytically active functionalities. Gerard van Koten 37 Chem. Soc. Rev. 1999 28 37–41 and oxidation inhibitors for engine oils hydraulic fluids lubricants and greases,4 in polymers (e.g. the well-known PTFE)5 and dendrimers6 with special surface properties in perfluorinated monolayers and coatings,7 supports for immobilization of biomolecules in biosensors8 and in preparation of liposomes vesicles micelles emulsions and bilayers.9 Biphasic mixtures of perfluorocarbon and hydrocarbon fluids have also been employed as media for carrying out suspension polymerization of water sensitive monomers (e.g.lauryl methacrylate and styrene).1 Perfluorocarbon fluids are usually not miscible with common organic solvents and as a result they display good separation properties with these solvents. Combined with the preferential solubility of perfluoroalkyl-substituted compounds in fluorous ( = perfluoro function containing) solvents this feature can be employed for the selective extraction of fluorous compounds from organic reaction mixtures which is the subject of this review. Although immobilization of biomolecules on fluorous supports has been successfully employed in affinity chromatography and it has been shown that fluorous enzymes immobilized on fluorous supports retain ca.70% of their original activity,8 the potential of these so called fluorous phase separation and immobilization techniques in transition metal catalysis has only recently received attention after the first report by Horváth et al.10 A brief review on this subject has appeared recently.11 Below an overview is given on the application of fluorous phase immobilization and separation methods in catalysis. For an overview on steric and electronic effects of perfluoro substituents in organometallics the reader is referred to an early review by Hughes et al.12 2 Potential of fluorous phase separation techniques for catalysis In catalysis homogeneous catalytic systems are often preferred over heterogeneous ones because of their better product and substrate selectivity.A general problem in homogeneous catalysis however is separation and recycling of the catalyst. This has led to the development of several supported catalytic systems e.g. immobilized versions of homogeneous catalysts on inorganic supports13 and systems connected to polymers or dendrimers,14 with the combined advantages of both homogeneous and heterogeneous catalysis. Another elegant solution for this separation problem is the aqueous biphasic Ruhrchemie/ Rhône-Poulenc process.15 In this process a water soluble version of the conventional Rh–PPh3 catalyst is used i.e.TPPTS–Rh (TPPTS = P(m-C6H4SO3Na)3). The catalytic process is performed under biphasic conditions with the aqueous phase containing the catalyst and the organic phase containing the products. The catalyst can be easily removed from the products by phase separation. In this way losses of rhodium are kept below 1026 mg kg21 of product produced. Despite the advantages of aqueous biphasic systems in catalysis they also have some disadvantages. Some reactants or catalysts hydrolyze when exposed to water resulting in decreased performance for these systems. Furthermore due to the two phase nature of the system the catalyst is not homogeneously mixed with the products. Therefore the reactants have to cross the phase boundary which could lead to mass flow limitations resulting in considerably lower reaction rates as compared to single phase homogeneous systems.This effect is enhanced by the often low solubility of organic substrates in water.10 The special physical properties of perfluorinated compounds described above and the problem associated with aqueous biphasic catalysis inspired Horváth et al. to use fluorous biphase systems in rhodium catalyzed hydroformylation.10 Here the fluorous phase as an alternative to the aqueous phase denotes Chem. Soc. Rev. 1999 28 37–41 38 a solvent which is rich in C–F bonds. Below a certain temperature the fluorous phase does not mix with an organic phase containing the reactants and products. The principle of Fig. 1 Principle of fluorous biphase catalysis.fluorous biphase catalysis can be depicted as follows (Fig. 1). At room temperature the system consists of a fluorous phase containing a fluorous phase soluble catalyst and a hydrocarbon phase containing the reactants. Above a certain temperature the two phases mix to form one phase allowing efficient homogeneous catalysis to proceed. Catalyst recovery can then be achieved by cooling of the reaction mixture below the temperature where phase separation occurs. Alternatively if the phase transition temperature of a certain fluorous biphasic system is too high or if desirable for other reasons the catalytic reaction can also be performed under biphasic conditions. For reactions which cannot be performed in an aqueous biphasic system e.g.due to low solubility of reactants in the aqueous layer diffusion limitations or water sensitive components a fluorous biphasic system could be an alternative. Perfluoro solvents do not mix with water and can only contain water at the ppm level. Fluorous phase separation techniques are of course not limited to applications in catalysis but can also be used in organic synthesis. Fluorous synthesis developed and recently reviewed by Curran,16 is based on separation of fluorous reactants and by-product from the desired organic products. Because of its simplicity and speed fluorous extraction of fluorous side products from the desired organic product is also an attractive purification technique for application in combinatorial synthesis.A library of compounds can be synthesized much faster if the compounds can be purified quickly without the use of time consuming separation techniques like crystallization distillation or filtration. 3 Applications of fluorous phase separation techniques in catalysis To render a catalyst preferentially soluble in a fluorous phase it is usually functionalized with one or several perfluoroalkyl groups also sometimes referred to as pony tails. Most often perfluorohexyl (C6F13) and perfluorooctyl (C8F17) groups are used. Branched perfluoroalkyl groups are less common. The length and number of the perfluoroalkyl groups are important because they influence the solubility of a perfluoroalkylated compound in a fluorous solvent.17 However in many cases the distribution of a fluorous catalyst in a fluorous biphasic system has not been optimized.The strongly electron-withdrawing properties of perfluoroalkyl functions could have a dramatic effect on the catalytic activity when compared to the nonsubstituted system. This can be easily understood from the known high electronegativity of the CF3 group (3.5 according to Pauling18) the reversed polarization of the C–I bond in perfluoroalkyl iodides (which renders the iodine atom rather than the a-carbon susceptible to nucleophilic attack19) and the remarkable increase in N1s and Zn2p3/2 ionization potential of Zn–porphyrin complexes upon perfluoroalkylation.20 The strong electron-withdrawing nature of perfluoroalkyl groups was also observed in (trifluoromethyl)cyclopentadienyl transition metal complexes.12 Although it is difficult to predict the exact effect of electronwithdrawing perfluoroalkyl groups on catalytic activity since this will depend on the intimate details of each catalytic cycle involved it can be easily understood that a decrease of electron density on the metal center is expected to have a significant effect on for example the delicate s-donation/p-back donation balance in bonding of catalytically important substrates such as CO H2 and olefins.In practice the above problem has been solved by using ethylene or propylene spacers that insulate the strongly electron-withdrawing perfluoroalkyl tail from the remainder of the transition metal catalyst. Another type of spacer which has been developed in our own laboratory is the 2CH2CH2SiMe2- moiety.21 CF3]3}3]10,22 Fluorous biphasic catalysis was first demonstrated in hydroformylation of alk-1-enes using [HRh(CO){P[CH2CH2(CF2)5- in the presence of an excess of P[CH2CH2(CF2)5CF3]3.The ratio of normal to branched aldehyde (n/i) obtained was comparable to that for non-fluorous [HRh(CO){PPh3}3] (n/i = 2.9) in a conventional solvent. This value was slightly higher than that obtained for the perprotio analogue [HRh(CO){P[(CH2)7CH3]3}3] under non-fluorous conditions (n/i = 2.3). Although selectivity was higher the activity of the fluorous catalyst under biphasic conditions was an order of magnitude lower which was explained by the lower solubility of CO and H2 in the fluorous biphasic solvent system.Another aspect relevant to recycling of the catalyst is the amount of catalyst that leaches into the organic phase. It was found that losses of rhodium amounted to 4.2% after 9 cycles (corresponding to 0.6 mg of Rh per kg of product) for the above fluorous catalyst. At this point this fluorous hydroformylation system cannot compete yet with the aqueous biphasic system. As mentioned earlier the losses of rhodium in the aqueous biphasic system are less than 1026 mg of Rh per kg of product. Although this was not quantified further judging from the increase in activity and drop of selectivity after each catalyst recycle the authors suggest that the system most probably suffers from leaching of the fluorous phosphine as well.Closely related fluorous rhodium complexes trans- [ClRh(CO){P[CH2CH2(CF2)5CF3]3}2] trans-[ClRh(CO)- {P[C6H4(CF2)5CF3-4]3}2],23 and iridium complex 124 have been prepared although no catalytic activity of those complexes was mentioned. For 1 a phase distribution constant > 300 in CO [CF3(CF2)5CH2CH2]3P Ir P[CH2CH2(CF2)5CF3]3 Cl 1 favor of the fluorous phase was found indicating that efficiency of catalyst recovery by a single phase separation could be higher than 99.7%. A fluorous analogue of Wilkinson’s catalyst [ClRh{P[CH2CH2(CF2)5CF3]3}3]25 was found to be active in hydroboration of alkenes (turn-over number (TON) = 300–9000 conversion 80–90%). From our own laboratory the synthesis of 2a and 2b was reported which catalyze the Kharasch addition of CCl4 with methyl methacrylate (Scheme 1).21 In CCl4 selectivity and NMe2 2a R = -SiMe2CH2CH2(CF2)5CF3 2b R = H Ni Cl R NMe2 activity of 2a were found to be comparable to non-perfluoroalkylated 2b.Efficient recycling of 2a by fluorous extraction was however hindered by the limited solubility of 2a in fluorous solvents which is probably due to the relatively small size of the perfluoroalkyl portion of the catalyst. O O + CCl4 Cl3C OMe OMe Cl 3a M = Co ArF (CF2)7CF3 ArF = N N (CF2)7CF3 M Ar ArF N N 3b M = Mn Cl (CF2)7CF3 ArF ArF = Cl (CH2)2OCH2RF RFCH2O(CH2)2 N N N N RFCH2O(CH2)2 (CH2)2OCH2RF RF = CF3 – –q = 3.38; p = 0.11 2 Scheme 1 Several studies on the application of fluorous catalysts for oxidation catalysis have appeared.For instance Pozzi et al. prepared among others fluorous tetraarylporphyrin complexes 3a and 3b.26,27 The perfluoroalkyl groups were introduced by F cross coupling of the corresponding iodoaryl derivatives and perfluoroalkyl iodides or using FITS [(perFluoroalkyl)phenyl- Iodonium TrifluoromethaneSulfonates] reagents respectively. Cobalt complex 3a was found to have good activity and selectivity in fluorous biphasic epoxidation and it was shown by UV–VIS spectroscopy that 3a was completely partitioned in the fluorous phase. Manganese derivative 3b was found to be an active epoxidation catalyst under aqueous biphasic conditions. Copper and cobalt complexes of fluorous tetraazacyclotetradecane 4 are catalysts for alkane and alkene oxidation by t- 4 CF2(OCFCF2) q(OCF2) pOCF3 BuOOH and O2 under fluorous biphasic conditions.28 Selectivities of 80% towards ketone moderate to high yields but modest turn-over numbers (18–330 mol mol21 of catalyst) were found.It was mentioned that the Cu and Co complexes of 4 were only present in the perfluoro solvent (UV–VIS). When the fluorous layers were used for a second run the catalytic activity was retained very well except for the Cu-catalyzed oxidation of cyclohexene where only 50% of the activity was retained. Vincent29 et al. described Mn and Co complexes of 1,4,7-[C8F17(CH2)3]3-1,4,7-triazacyclononane and their activity in fluorous biphasic oxidation of cyclohexene to cyclohex- 2-en-1-ol and cyclohex-2-en-1-one in the presence of O2–t- BuOOH.Overall yields were high however selectivities were poor. Most importantly catalyst recovery was successful. Further examples of fluorous oxidation catalysts are Ru and Ni complexes of the fluorinated acetylacetonate anion ([(C7F15)C(O)CHC(O)(C7F15)]2) reported by Klement.30 They are active in catalytic oxidation of aldehydes sulfides and epoxidation of cycloalkenes under biphasic conditions. Catalyst recovery was claimed to be easy and no leaching was observed Chem. Soc. Rev. 1999 28 37–41 39 which was concluded from the fact that the organic layer was colorless. 6H4C6F13)3}4] turned A fluorous palladium complex [Pd{P(C out to be active in cross-coupling of arylzinc bromides and aryl iodides.31 The catalyst was recycled several times without a significant drop in reaction yields.The activity of the catalyst was higher than the activity of its non-perfluoroalkylated analogue. This was explained by the lower electron density on the phosphorus atoms which favors reductive elimination in these cross-couplings. A few variations on the use of fluorous catalysts and fluorous phase separation techniques were reported. For example a fluorous phase soluble polymer has been used to remove reagents catalysts or ligands from a non-fluorous reaction mixture.32 Also a fluorous analogue of Wilkinson’s catalyst,33 [RhCl{P[m-C6H4(CH2)2(CF2)5CF3]3}3] was employed in hydroformylation with supercritical CO2 as solvent. Here the perfluoroalkyl tails serve to increase the solubility of the catalyst in this very apolar medium.A conversion of 92% and high selectivity for the n-aldehyde (82%) was reached which is comparable to the selectivity for the non-perfluoroalkylated Wilkinson’s catalyst (92%).22 4 Phase behavior of fluorous biphasic systems and phase distribution of fluorous catalysts The special properties of perfluorinated compounds are mainly due to the high electronegativity of the fluorine atom (4.0 on the Pauling scale) and the larger van der Waals radius of fluorine (1.47 Å) as compared to hydrogen (1.2 Å). Although the C–F bond is highly polar,34 perfluoroalkanes are apolar media and the miscibility with organic solvents is generally low.35 Because fluorine atoms are very difficult to polarize the van der Waals interactions between perfluoroalkanes are weak compared to those in alkanes.The weaker van der Waals interactions result in lower boiling points for perfluoroalkanes compared to the corresponding normal alkanes.36 Also it is unfavorable for an alkane to mix with a perfluoroalkane because in terms of energy the gain in van der Waals interactions between perfluoroalkane and alkane molecules upon mixing is not compensated for by the loss of alkane–alkane and perfluoroalkane –perfluoroalkane van der Waals interactions. This leads to low miscibility of organic and perfluoro solvents. Fluorous biphase catalysis is based on this low miscibility with other organic solvents. One of the most extreme examples is PTFE which only dissolves in its lower oligomers i.e.long chain perfluoro-n-alkanes.5 The miscibility and critical temperature (Tc) for fluorous biphasic systems i.e. the temperature above which the two liquids are miscible in all ratios can be determined from a phase diagram like the one depicted in Fig. 2.36 From this diagram it can be concluded that also at temperatures below but close to Tc substantial amounts of perfluorosolvent are dissolved in the organic layer. This could be a potential problem since this may result in some solubility of fluorinated compounds in the organic layer at temperatures close to Tc leading to e.g. catalyst loss during phase separation in fluorous biphase catalysis. In general Tc is close to the phase separation temperature of biphasic systems consisting of equal volumes of each phase.36 Tc can also be predicted using the Hildebrand–Scatchard Theory also called Regular Solution Theory35,36 using eqns.(1) and (2). Here R is the universal gas K 2 1 (1) Tc ª (2) (n +n ) 4R 2 2 1 40 K = (d -d ) constant (cal mol21 K21) vi the molar volume (cm3 mol21) Tc the critical temperature (K) and K (cal cm23) is a measure of the Chem. Soc. Rev. 1999 28 37–41 Fig. 2 Phase diagram of perfluoromethylcyclohexane and benzene; x = mole fraction of benzene (redrawn from data in ref. 36). interaction energy between unlike molecules relative to that of like molecules. The weaker the interaction between two unlike molecules the higher the value for K.Sufficient large values for K correspond to limited miscibility of the biphasic system. The Hildebrand parameter di (cal1/2 cm23/2) of a solvent is defined as the square root of the enthalpy of vaporization (DHi v) divided by the molar volume (vi) (eqn. (3)). / )1 / 2 d i = (DHi v n i (3) Values of d and Tc for specific biphasic systems have been tabulated by Scott35 and Lo Nostro.36 Large differences in Hildebrand parameters correspond to large values for K resulting in low miscibility. Eqn. (1) shows that a large K also corresponds to a high critical temperature. From eqns. (1) and (2) it can be calculated that two liquids are miscible at room temperature when |d12d2| is less than 3.5 cal1/2 cm23/2 for an average molar volume of 100 ml.The Hildebrand parameters for perfluorocarbons are very low (5.7–6.1) compared to hydrocarbons ( > 7.0) corresponding to the fact that perfluorocarbons are not completely miscible with organic liquids. For some commonly used solvents d-values are as follows nhexane (d = 7.3) cyclohexane (d = 8.2) and toluene (d = 8.9). Especially solvents which have a d-value around 9 are suitable for use in fluorous biphasic separations since they give phase separation at room temperature. In general higher polarity solvents also have higher d-values and consequently give good phase separation with perfluorocarbons. Explanations given in the literature for experimental deviations from the Hildebrand–Scatchard theory are mainly speculative. For example interpenetration of hydrocarbon molecules molar volume changes the polarity of the C–F bond and differences in ionization potential were mentioned.35,36 Better correlations with experimental data are sometimes obtained with the more complicated Flory–Huggins Theory or Theory of Reed.36 All these models predict the miscibility of two liquids.Especially relevant for homogeneous catalysis under fluorous biphasic conditions is the partitioning of perfluorinated ligands and metal complexes across a certain fluorous biphasic or multiphasic solvent system. To avoid leaching of catalyst during product separation the catalyst should be present in the fluorous phase only. However no models are available which predict the phase distribution of fluorous components in fluorous biphasic systems.A fundamental understanding of factors which govern this distribution however is essential for further development of fluorous phase separation and immobilization techniques in catalysis and synthesis. Also the role of micelle formation in fluorous biphasic catalysis needs to be considered. For aqueous biphasic systems the role of micelle formation by e.g. water soluble surfaceactive phosphines [P{C6H4(CH2)m-4-C6H4SO3Na-4}3] has been proved.37 It was found that both reaction rate and selectivity of oct-1-ene hydroformylation are improved by micelle formation probably due to the larger contact surface between the two phases resulting in less diffusion limitation. Furthermore it has been reported that addition of amphiphiles to an aqueous chiral rhodium hydrogenation catalyst results in an increase of both the activity and enantioselectivity of the catalyst.38 The amphiphile which forms micelles in water probably increases the solubility of the substrate in water which increases the activity of the catalyst.It is also known that fluorous surfactants i.e. molecules containing a fluorous part and a hydrophilic part can form micelles in water above a certain concentration (the critical micelle concentration cmc).39 However no studies dealing with micelle formation and the effect on activity and selectivity in reactions carried out under fluorous multiphasic conditions have appeared. 5 Conclusions and perspectives Since the first report by Horváth et al.in 1994 on the application of fluorous multiphasic separation techniques in catalysis a number of catalysts have been made fluorous phase compatible by the introduction of perfluoro functions. In cases where comparison has taken place the activities and selectivities of those catalysts do not differ very much from the non-fluorous analogues. However often comparisons with non-fluorous systems have not been made making it difficult to evaluate the effect of the perfluoroalkyl tails and the fluorous biphasic solvent system on the catalyst. Under fluorous biphasic conditions the activity of the fluorous catalyst can be lower compared to both fluorous and non-fluorous single phase conditions as a result of mass transport limitations. Also other important questions remain to be clarified.For instance the influence of size structure and number of perfluoro tails attached to a catalyst or other fluorous component will undoubtedly have a significant influence on the phase distribution of these species in fluorous multiphasic systems. However data on phase distribution of fluorous compounds and the amount of catalyst leaching due to non-zero solubility in the non-fluorous phase are extremely scarce. Also the knowledge of factors which determine the absolute solubility of fluorinated species in perfluorinated solvents is limited. This is a serious handicap in evaluating the possibilities for practical application of fluorous phase separation techniques relative to e.g. aqueous biphasic techniques for catalyst recycling and will have to be addressed in future studies.The distribution is also important from an economic point of view. A fluorous catalyst will be more expensive than a non-fluorous catalyst. However if the catalyst can be fully recovered it will be cheaper in the long term. Also only a limited amount of synthetic methods for connection of perfluoro functions to catalysts have been reported although a multitude of synthetic routes for perfluoro functionalization is available in the literature. Thus far the types of perfluoro functions employed are mainly limited to perfluoroalkyl groups (with or without a 2CH2CH2- spacer) and there is clearly a need for introduction of other perfluorinated moieties with more diverse structures.Finally examples of fluorous phase catalysis now include catalytic hydroformylation hydroboration C–C coupling and oxidation but many more fluorous multiphasic catalytic processes could be envisioned. Especially in the area of asymmetric catalysis fluorous phase extraction techniques could have potential for the recycling of the often expensive chiral ligands and/or catalysts. These techniques will undoubtedly be developed in the near future. It will be clear that the full potential of fluorous separation and immobilization techniques can only be evaluated completely when further information regarding the above aspects has been obtained. Review 8/05644G 6 Acknowledgements We thank Elf Atochem and the Dutch Ministry for Economic Affairs for financial support.Chem. Soc. Rev. 1999 28 37–41 7 References 1 D. W. Zhu Synthesis 1993 953. 2 L. A. Halper C. O. Timmons and W. A. Zisman J. Colloid Interface 3 J. G. Riess and M. Le Blanc Pure Appl. Chem. 1982 54 2383 and 4 C. Tamborski C. E. Snijder and J. B. Christian US Patent 4454349 5 P. Smith and K. H. Gardner Macromolecules 1985 18 1222 and 6 K. Lorenz H. Frey B. Stühn and R. Mülhaupt Macromolecules 1997 Sci. 1972 38 511. references cited therein. 1984; Chem. Abstr. 1984 101 152079z. references cited therein. 30 6860. 7 R. Banga and J. Yarwood Langmuir 1995 11 4393. 8 R. K. Kobos J. W. Eveleigh and R. Arentzen Trends Biotechnol. 1989 7 101. 9 T. H. Maugh Science 1979 206 205. 10 I. T. Horváth and J. Rábai Science 1994 266 72.11 B. Cornils Angew. Chem. Int. Ed. Engl. 1997 36 2036. 12 R. P. Hughes Adv. Organomet. Chem. 1990 31 183. 13 M. G. L. Petrucci and A. K. Kakkar Advan. Mater. 1996 8 251 and references cited therein. 14 J. W. Knapen A. W. van der Made J. C. Wilde P. W. N. M. van Leeuwen P. Wijkens D. M. Grove and G. van Koten Nature 1994 372 659. 15 W. A. Hermann and C. W. Kohlpaintner Angew. Chem. Int. Ed. Engl. 1993 32 1524. 16 D. P. Curran Angew. Chem. Int. Ed. Engl. 1998 37 1174 and references cited therein. 17 R. P. Hughes and H. A. Trujillo Organometallics 1996 15 286. 18 J. E. Huheey J. Phys. Chem. 1965 69 3284. 19 T. Umemoto Chem. Rev. 1996 96 1757. 20 J. G. Goll K. T. Moore A. Ghosh and M. J. Therien J. Am. Chem. Soc. 1996 118 8344.21 H. Kleijn J. T. B. H. Jastrzebski R. A. Gossage H. Kooijman A. L. Spek and G. van Koten Tetrahedron 1998 54 1145. 22 I. T. Horváth G. Kiss R. A. Cook J. E. Bond P. A. Stevens J. Rábai and E. J. Mozeliski J. Am. Chem. Soc. 1998 120 3133. 23 J. Fawcett E. G. Hope R. D. W. Kemmitt D. R. Paige D. R. Russell A. M. Stuart D. J. Cole-Hamilton and M. J. Payne Chem. Commun. 1997 1127. 24 M. A. Guillevic C. Rocaboy A. M. Arif I. T. Horváth and J. A. Gladysz Organometallics 1998 17 707. 25 J. J. J. Juliette I. T. Horváth and J. A. Gladysz Angew. Chem. Int. Ed. Engl. 1997 36 1610. 26 G. Pozzi F. Monatari and S. Quici Chem. Commun. 1997 69. 27 G. Pozzi J. Colombani M. Miglioli F. Monatari and S. Quici Tetrahedron 1997 52 6145. 28 G. Pozzi M. Cavazzini and S. Quici Tetrahedron Lett. 1997 38 7605. 29 J.-M. Vincent A. Rabion V. K. Yachandra and R. H. Fish Angew. Chem. Int. Ed. Engl. 1997 36 2346. 30 I. Klement H. Lütjens and P. Knochel Angew. Chem. Int. Ed. Engl. 1997 36 1454. 31 B. Betzemeier and P. Knochel Angew. Chem. Int. Ed. Engl. 1997 36 2623. 32 D. E. Bergreiter and J. G. Franchina Chem. Commun. 1997 1531. 33 S. Kainz D. Koch W. Baumann and W. Leitner Angew. Chem. Int. Ed. Engl. 1997 36 1628. 34 B. E. Smart in The Chemistry of Functional Groups Supplement D eds. S. Patai and Z. Rappoport John Wiley & Sons Ltd. Chichester 1983 pp. 603–653. 35 R. L. Scott J. Phys. Chem. 1958 62 136 and references cited therein. 36 P. Lo Nostro Adv. Colloid Interface Sci. 1995 56 245 and references cited therein. 37 H. Ding B. E. Hanson T. Bartik and B. Bartik Organometallics 1994 13 3761. 38 G. Oehme E. Paetzold and R. Selke J. Mol. Catal. 1992 71 L1. 39 W. Guo T. A. Brown and B. M. Fung J. Phys. Chem. 1991 95 1829. 41

ISSN:0306-0012    

DOI:10.1039/a805644g

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Asymmetric Claisen rearrangement Hisanaka Ito and Takeo Taguchi* Tokyo University of Pharmacy and Life Science 1432-1 Horinouchi Hachioji Tokyo 192-03 Japan Received (in Cambridge) 7th July 1998 Development of the asymmetric Claisen rearrangement is one of the challenging tasks in synthetic organic chemistry. There have been numerous reports of the asymmetric Claisen rearrangement based on the intramolecular chirality transfer using chiral substrates. On the other hand reactions of achiral substrates with an external chiral activator have been studied during the last decade. In this review article recent advances in the asymmetric Claisen rearrangement are described. 1 Introduction Recent advances of asymmetric reactions in synthetic organic chemistry have been well documented and excellent enantioselective reactions successfully established.Among these epoxidation of allylic alcohols dihydroxylation of a carbon–carbon double bond hydrogenation of multiple bonds (CNC CNO and CNN) and carbon–carbon bond forming reactions through the addition of organometallics to a carbonyl functionality are the typical reactions in which a high level of asymmetric catalysis has been realized.1 The Claisen rearrangement has been one of the most useful tools for the formation of carbon–carbon bonds.2 Since the discovery by Claisen in 1912 development of the Claisen rearrangement the [3,3] sigmatropic rearrangement of allyl vinyl ethers which now involves a variety of modified variants has made this reaction widely applicable to the synthesis of organic molecules in particular to natural product synthesis.3,4 An extension of the Claisen rearrangement to an asymmetric version is however a challenging task.The Takeo Taguchi was born in Sapporo Japan in 1947. He received his BS (1969) and his PhD degrees (1974) from the Tokyo Insitute of Technology. He joined the faculty of TIT as an Assistant in 1974. He was appointed Lecturer (1976) Associated Professor (1983) and Professor (1989) at Tokyo University of Pharmacy and Life Science (formerly Tokyo College of Pharmacy). He was a postdoctoral fellow at UCLA with D. J. Cram from 1981 to 1982. His present research interests include the development of new synthetic methodologies via early transition-metallic compounds via electrophilic iodination of C–C multiple bond and for organic fluorine compounds.Hisanaka Ito Takeo Taguchi 2 LU 3 O HO 3' asymmetric Claisen rearrangement can be classified into two types a diastereoselective reaction or an enantioselective reaction (Scheme 1). The diastereoselective Claisen rearrangement of a substrate having a chiral center has been extensively studied. Although the substrate can have a chiral center at the 1-position in the rearrangement system or at any other position near to the rearrangement system usually a chiral center at the 1-position is quite efficient for the intramolecular chirality transfer and such a rearrangement system can be constructed by using a chiral secondary or tertiary allylic alcohol.These chiral alcohols are available in enantiomerically enriched forms for example by asymmetric reduction of an a,b-unsaturated carbonyl compound kinetic resolution of a racemic alcohol by using the Sharpless epoxidation and by addition of a vinylmetal to a chiral carbonyl compound. While this type of internal chirality transfer reaction has been successfully employed for the total synthesis of natural products this extensive literature will not be discussed here. Asymmetric induction based on the incorporation and subsequent cleavage of a chiral auxiliary within an otherwise achiral substrate may be regarded as a second group of diastereoselective Claisen rearrangements. As an alternative a chiral auxiliary may be temporarily incorporated on the achiral substrate not only to induce a chirality transfer but also to activate the reaction.In such a system the chiral auxiliary can be detached from the product so readily for example by simple workup procedure that this process does not involve formal introduction and removal of the chiral auxiliary. The most advanced variant of this concept is the enantioselective reaction involving a chiral Lewis acid as mediator. An enantioselective reaction employing a catalytic amount of chiral Lewis acid has Hisanaka Ito was born in Kawasaki Japan in 1967. He received his BS (1989) and his PhD degrees (1994) with Professor Taguchi from Tokyo University of Pharmacy and Life Science (formerly Tokyo College of Pharmacy).He joined the Sagami Chemical Research Center with Dr Susumu Kobayashi as a Research Chemist in 1994. He returned to Professor Taguchi’s group as Assistant (1996). His present research interests include the development of new synthetic methodologies via organozirconium compounds via Claisen rearrangement and for organic fluorine compounds. 43 Chem. Soc. Rev. 1999 28 43–50 2' 3,3-sigmatropic rearrangement 1' 3' O 1 3 2 Claisen rearrangement allyl vinyl ether system asymmetric Claisen rearrangement • diastereoselective version R* O R* = substituent at chiral center X* X* O O * * X* = chiral auxiliary • enantioselective version X* O X* = chiral reagent Scheme 1 not however been reported because the resulting carbonyl group in the product shows a higher Lewis basicity than the ethereal oxygen in the substrate.That is the Lewis acid coordinates to the product more tightly than to the starting material. We will review here recent advances in the asymmetric Claisen rearrangement of allyl vinyl ether derivatives. Regarding the aza- and thia-Claisen rearrangements readers are recommended to see a recent review by Enders et al.5 The Claisen rearrangement of the parent substrate allyl vinyl ether is a symmetry-allowed and concerted pericyclic reaction involving a suprafacial pathway which proceeds through a high preference for a chairlike transition state. In the boatlike transition state an anti-bonding interaction between LUMOC(2) and HOMOC(2’) makes it an unfavorable pathway.Introduction of a substituent onto the carbon chain of the rearrangement system is possible at every position resulting in different effects involving the reactivity and the stereochemical outcome of the rearranged products. In the Claisen rearrangement in particular those of acyclic systems the relative stereochemistry (syn/anti) at the newly formed adjacent chiral centers is highly controlled by the geometry of the double bonds (E/Z) when substituted at the 3- and 3’-positions. If the substrate is a chiral molecule due to a substituent at the 1-position this chirality can be transferred to the 3- and/or 3’-position through the highly ordered transition state. These variations are illustrated in Scheme 2 in those cases where the rearrangement proceeds via a chairlike transition state.2 Asymmetric aliphatic Claisen rearrangement We summarize here the asymmetric aliphatic Claisen rearrangements the diastereoselective Claisen rearrangements of the substrates having a chiral auxiliary and the enantioselective Claisen rearrangements mediated by chiral reagents. Chem. Soc. Rev. 1999 28 43–50 44 O R* O * * O * * O * * 2' 1' 3' O O 3 1 2 [3,3] 2 LUMO 3 chairlike TS O HOMO 3' 2' 2 3 boatlike TS unfavorable interaction O 2' X X X X R1 R1 R1 R1 O O O O R2 R2 R2 R2 X X X X R1 R1 R1 R1 O O O O R2 R R R2 R R R2 R2 Scheme 2 2.1 Asymmetric Carroll rearrangement The [3,3]-sigmatropic rearrangement of the allylic ester of a bketo acid giving rise to a g,d-unsaturated ketone via decarboxylation of the rearranged b-keto acid is known as the Carroll rearrangement.3 In particular this has been an important process in a commercial scale production of geranylacetone and b-ionone (Scheme 3).O O O H O OH OR' ( i-PrO)3Al O or R O O R O O –CO2 CO2H R R Scheme 3 In 1995 Enders et al. reported the highly efficient asymmetric Carroll rearrangement to generate a chiral quaternary carbon by employing (S)- or (R)-1-amino-2-methoxymethylpyrrolidine (SAMP or RAMP respectively) as a chiral auxiliary (Scheme 4).6,7 The SAMP (or RAMP) hydrazone 1 is readily prepared by treating the allylic ester of a b-keto acid with SAMP in the presence of toluenesulfonic acid.A good chirality transfer could be realized in the dianion-mediated rearrangement which may involve control of the enolate geometry and a conformationally restricted transition state due to the formation of the intramolecularly chelated intermediate 4. For example dianions of SAMP hydrazones 1 generated by treatment with 2.6 equiv. dianionic version OMe OMe N O N N N R2 O3 1) 2.6 eq. LDA O R1 R1 R1 2) LiAlH4 R1 R2 HO 2 OMe Li O R2 Li N O N R1 O R1 R1 R1 1 HO R2 R1 = (CH2)3 R2 = c-Hex 77% 93% de >97% ee R1 = (CH2)4 R2 = i-Pr 66% 89% de >98% ee 4 OMe OMe TBS N O N N N R2 1 O R1 R1 R1 R1 R2 HO 6 3 Lewis acid-mediated version 1) TBSOTf Hünig base 2) LiAlH4 R1 = (CH2)3 R2 = n-Bu 68% 20% de 57% ee 5 Scheme 4 of LDA at 278 °C in THF–N,N,N’,N’-tetramethylethylenediamine rearranged at room temperature and the following reduction of the rearranged acid with lithium aluminum hydride provided the b-hydroxyhydrazones 2 in good diastereomeric excess (88–98% de 59–83% yield).b-Hydroxyketones 3 (82 to > 98% ee) were obtained by treating the b-hydroxy hydrazones 2 with ozone. The Carroll rearrangement of SAMP hydrazones 1 via their silyl ketene acetals 6 also proceeded at room temperature in the presence of a Lewis acid such as tert-butyldimethylsilyl triflate (TBSOTf) while this process showed low diastereoselectivity and low asymmetric induction.Thus a mixture of the SAMP hydrazones 1 1.3 equiv. of TBSOTf and 1.5 equiv. of Hünig base was allowed to react at room temperature leading to the rearranged products which in turn reacted with lithium aluminum hydride to give b-hydroxy hydrazones 5 as a mixture of the four possible diastereomers in moderate selectivity. Interestingly the major diastereomer had the opposite configuration at the newly formed chiral center to that from the dianion-mediated reaction. Enders has also applied this process to the total synthesis of (2)-malyngolide (9) (Scheme 5)8 using the dianionic asymmetric Carroll rearrangement as a key step. Upon treatment of the RAMP hydrazone 7 with 2.4 equiv. of lithium 2,2,6,6-tetramethylpiperidide (LiTMP) the rearrangement proceeded smoothly and the following reduction with lithium aluminum hydride gave the b-hydroxy hydrazone 8 with high diastereoselectivity ( > 96% de) in moderate yield (57%) which was converted in 6 steps to (2)-malyngolide (9) (96% ee).2.2 Asymmetric Eschenmoser–Claisen rearrangement The [3,3]-sigmatropic rearrangement of ketene N,O-acetals first developed by Eshenmoser,3 involves several variants to generate the rearrangement precursors such as the reaction of OMe OMe Li Li N N O N O N LiAlH 2.4 eq. LiTMP 4 O O O OMe 6 steps O N N OH OH (–)-malyngolide (9) 8 Scheme 5 O O 3% 95% CONMe2 NMe2 O O 97% 5% CONMe2 NMe2 Scheme 6 Ar* Ar* Li Ar* H N N N LDEA O O O THF 11 10 78% anti/ syn 98 2 94% de for anti OH O Zn O 11 O 12 I OMe = Ar*-NH2 NH2 12 Scheme 7 7 57% >96% de an allylic alcohol with an ynamine or the dimethyl acetal of an N,N-dimethyl carboxamide.The stereochemical outcome of the rearrangement of the dimethyl acetal of an N,N-dimethylpropionamide with (E)- and (Z)-crotyl alcohol under thermodynamic conditions was explained by an axial orientation of the C(1)- methyl group of the ketene N,O-acetal based on the assumption of a chairlike transition state in which (Z)-ketene N,O-acetals are predominant (Scheme 6).9,10 NMe2 NMe2 Metz et al. have reported a chiral auxiliary induced imidate Claisen rearrangement based on an axially chiral binaphthylamine derivative 12 (Scheme 7).11,12 The requisite I2 KI EtOH H2O lithium enolate was prepared by treatment of the imino ether 10 with lithium diethylamide (LDEA).Although the Claisen 45 Chem. Soc. Rev. 1999 28 43–50 rearrangement of the corresponding N-TMS azaenol ether did not proceed the lithium azaenolate rearranged at 0 °C to afford the g,d-unsaturated amide 11 with high 2,3-anti selectivity (anti/syn 98:2) and excellent chiral induction (94% de for the anti-isomer). This high selectivity was explained by considering the chairlike transition state the preferable (Z)-configuration of the azaenolate and an efficient chiral environment constructed by the chelation of the methoxy group to lithium atom in the azaenolate intermediate. Conversion of the rearranged amide to carboxylic acid and quantitative recovery of the chiral auxiliary 12 were both efficiently achieved by iodolactonization of the amide 11 followed by the reductive olefination of the iodolactone with zinc powder.During these procedures no epimerization was observed. The asymmetric amide acetal Claisen rearrangement using a chiral 2-substituted or 2,5-disubstituted pyrrolidine was reported by Welch et al. (Scheme 8).13 Treatment of the proline- O CF3SO3Me N OBn 13 Scheme 8 derived N-propionyl amide 13 with methyl trifluoromethanesulfonate gave the iminium salt 14 which was then reacted with lithium alkoxide of allylic alcohols to result directly in the formation of the rearranged amide 15 with moderate chiral induction (65% de).In these reactions rearrangement proceeded at room temperature in a stereospecific manner with respect to the configuration of the carbon–carbon double bond of the allylic alcohol. 2.3 Asymmetric Ireland–Claisen rearrangement The Ireland–Claisen reaction the rearrangement of a ketene silyl acetal derived from an allylic ester first reported in 1972 has been widely applied in organic synthesis in particular to bioactive and natural product synthesis.3 The versatility of this variant possibly involves the use of a substrate composed of a stoichiometric assembly of allylic alcohol and acid components relatively low temperature of the rearrangement process an efficient control of ketene silyl acetal geometry a highly reliable and predictable transfer of stereochemistry from the starting material to the product through a highly ordered transition state.As for the asymmetric versions there have been reported a number of examples using substrates having a chiral center in allylic alcohol component which showed high degree of chirality transfer. Development of new variants of the Ireland–Claisen rearrangement will be focused upon here. Kallmerten et al. reported the diastereoselective asymmetric Ireland–Claisen rearrangement using a chiral phenethyl ether as an auxiliary (Scheme 9).14 The auxiliary was connected to the a-position of the allylic ester moiety in the substrate 16. Although the high syn selectivity could be achieved the Chem. Soc. Rev. 1999 28 43–50 46 OMe –OTf N + OBn 14 OLi O N OBn 15 65% de O 3 O CH H MeO Ph O R 3 17 KHMDS TMSCl CH2N2 O CH H + O O 3 Ph O CH H R MeO Ph R 18 R = Me; 50% de R = Ph; 72% de O OTMS OH O H HO O R p-stacking interaction 16 1) separation 2) KOH MeOH 3) H2 Pd-C NHTs O NHTs Ph HN Ph CO HN 2Me O O O 21 20 85% 65% de 19 Scheme 9 intramolecular chirality transfer was not so high (R = Me; 50% de Ph; 72% de).In the case of the substrate having a phenyl group in the allylic moiety slightly higher diastereoselectivity was rationalized by the p-stacking interaction between both phenyl groups as shown below. Chromatographic separation of the resulting diastereomers 17 18 and removal of the chiral auxiliary by hydrogenolysis gave the hydroxy acid 19 with high enantiomeric purity.Kazmaier reported the diastereoselective Ireland–Claisen rearrangement of allylic glycinate 20 having an N-protected amino acid as a chiral auxiliary (Scheme 10).15 To achieve 1) 3.5 eq. LDA 1.2 eq. ZnCl2 10% Pd(PPh3)4 2) CH2N2 Scheme 10 control of enolate geometry the allylic ester 20 was treated with an excess LDA in the presence of a metal salt such as zinc chloride tin(ii) chloride or cobalt(ii) chloride to form the chelated enolate. The palladium(0) catalyzed Claisen rearrangement of the chelated enolate proceeded through intermolecular allylic alkylation via a p-allylpalladium intermediate to give the dipeptide derivative 21 in 85% yield with moderate diastereoselectivity (65% de).In 1991 Corey and co-workers reported the first enantioselective Ireland–Claisen rearrangement of achiral allylic esters 23.16 They employed the chiral boron reagent 22 which has the C2-symmetric bissulfonamide for the formation of a boron enolate within a chiral environment (Scheme 11). It is noteworthy that the geometry of the boron–enolate could be controlled by the choice of tertiary amine and solvent system. Thus on using triethylamine and a toluene–hexane system reaction of (E)-crotyl propionate 23 with the chiral boron reagent 22 led to the selective formation of (E)-enolate 24 and the following Claisen rearrangement proceeded at 220 °C for 14 days to afford the anti isomer 26 with excellent diastereoand enantioselectivity in 65% yield (96% ee).On the other hand (Z)-enolate 25 could be selectively formed by using diisopropylethylamine in dichloromethane and this rearranged OBL* O 2 L*2BBr Et3N O HO O toluene–hexane O 24 26 65% 96% ee OBL*2 O 23 L*2BBr i-Pr2NEt O HO CH2Cl2 27 75% >97% ee Ph Ph N O2S N B Br CF3 CF3 CF3CF3 O H 1) ( S S)-22 O HO 24 23 OH H 25 SO2 L*2BBr ( S S)-22 Scheme 11 to the syn isomer 27 in 75% yield ( > 97% ee). The high diastereoselectivities (anti 26 vs. syn 27) are rationalized by the efficient control of the enolate geometry and the preferred chairlike transition state. In the case of allyl propionate the enantioselectivity of the rearranged product markedly decreased.The chiral bissulfonamide could be easily recovered from the reaction mixture. The efficiency of this rearrangement method was further demonstrated by the application to the synthesis of natural products (+)-fuscol (25)17 and dolabellatrienone (28).18 In the synthesis of fuscol (25) the Claisen rearrangement of 3-methylenebutanoic acid geranyl ester (23) by using chiral boron reagent (S,S)-22 and triethylamine in toluene gave the compound 24 after lithium aluminum hydride reduction of the carboxy group of the rearrangement product (Scheme 12). Et3N toluene 2) LiAlH4 and separation of diastereomer major diastereomer >99% ee H (+)-fuscol (25) Scheme 12 Although the diastereoselectivity was not so high (3 1) the enantioselectivity of the major product 24 was excellent ( > 99% ee).The enantioselective synthesis of the marine diterpenoid dolabellatrienone (28) was achieved by the rearrangement of the achiral 15-membered lactone 26 leading via a ring contraction to the stereochemically defined 11-membered carbocyclic acid 27 (Scheme 13).18 In this reaction the choice of amine was found to affect the chemical yield of the rearrangement product. Treatment of the lactone 26 with the chiral boron reagent (S,S)- 22 and pentaisopropylguanidine at 278 °C to form the boron– enolate and the following rearrangement at 4 °C for 48 h gave the acid 27 in excellent diastereo- and enantioselectivity (86% yield > 96% de > 98% ee).Construction of the a-iso- ( S S)-22 O N i-Pr CO2H ( i-Pr)2N H O N( i-Pr)2 27 26 86% >98% ee O H dolabellatrienone (28) Scheme 13 propylidene cyclopentanone skeleton by several steps led to the first enantioselective synthesis of this diterpene 28 by which the absolute stereochemistry was unambiguously revised from that previously reported. Kazmaier et al. described a highly efficient enantioselective Ireland–Claisen rearrangement of the chelated enolate of Ntrifluoroacetylglycinate 29 in the presence of a chiral bidentate ligand (Scheme 14).19–21 The best results were obtained in the HN CO2H CF3 5 eq. LHMDS 1.1 eq. Al(O i-Pr)3 2.5 eq. quinine O O 30 HN CF3 98% 98% de 86% ee O O CF HN 3 CO2H 5 eq.LHMDS 1.1 eq. Al(O i-Pr)3 2.5 eq. quinidine 29 O 31 96% 98% de 86% ee Scheme 14 combined use of a lithium hexamethyldisilylamide (LHMDS)– aluminum isopropoxide system and quinine or quinidine as the chiral amino alcohol ligand. Thus the chelated enolate generated by treating 29 with 5 equiv. of LHMDS in the presence of 1.1 equiv. of aluminum isopropoxide and 2.5 equiv. of quinine underwent the rearrangement at 278 °C–room temperature to give the (2R 3S)-g,d-unsaturated amino acid 30 in good yield with high diastereo- and enantioselectivity (98% de 86% ee). With the same substrate the antipode 31 could be obtained in almost the same degree of chiral induction by using quinidine as a chiral ligand (98% de 86% ee).2.4 Asymmetric Claisen rearrangement The asymmetric carbanion-accelerated Claisen rearrangement of the substrate having a phosphonomethyl substituent at 2-position was reported by Denmark et al. in 1987.22,23 The accelerating effect of phosphorus-stabilized carbanions in the rearrangement is so remarkable that the reaction proceeds at 220 to 20 °C within a short period while the parent compound rearranged at much higher temperature. They designed a chiral 1,3,2-oxazaphosphorinane moiety not only as a chiral auxiliary but also as an anion stabilizing agent. Typical examples are shown in Scheme 15. When the compound 32 was treated with KH–DMSO in the presence of LiCl the reaction proceeded under mild conditions to give the g,d-unsaturated ketone 34 with high diastereoselectivity (80% de) in 78% yield.The use of lithium as the counter cation is crucial for the coordination to construct an efficient chiral environment and in the absence of 47 Chem. Soc. Rev. 1999 28 43–50 O O KH LiCl P O DMSO N t-Bu cis-32 O KH LiCl P O O DMSO N t-Bu trans-33 Bn N O O n-BuLi P N Bn 36 Scheme 15 LiCl chiral transfer was not observed (0% de 62% yield). Under the thermal conditions in the absence of KH–DMSO and LiCl the rearrangement of 32 took place above 100 °C and the diastereoselectivity was found to be relatively poor (32% de). They also examined the 1,3,2-diazaphospholidine derivatives 36 having the C2-symmetric chiral auxiliary.24 However the diastereoselectivity was not as high as observed in the formation of 34 or 35.Very recently we examined the enantioselective Claisen rearrangement of difluorovinyl allyl ether derivatives 39 having a phenol moiety (Scheme 16).25 By treating substrate 39 OH R O R O OH n-BuLi F CF3 neutral workup F 39 Ph Ph N N O2S B 38 SO2 Tol Tol OH O Br ( S S)-40 R Et3N F F R = TMS 60% (from 38) 85% ee L* L* B R O O F 41 F Scheme 16 preparated by dehydrofluorination of the trifluoromethylated ether 38 with chiral boron reagent (S,S)-40 the Claisen rearrangement proceeded smoothly to give the b-substituteda, a-difluoroketone 41 with moderate to good enantioselectivity. The formation of a covalent bond between a chiral boron reagent and the phenolic hydroxy group and following coordination of ethereal oxygen to the boron atom should result in an efficient chiral environment.Chem. Soc. Rev. 1999 28 43–50 48 O O P O N t-Bu 34 78% 80% de O P O O N t-Bu 35 71% 84% de Bn O N O P N Bn 37 39% 75 25 In 1990 Yamamoto et al. reported the first example of a chiral Lewis acid catalyzed enantioselective Claisen rearrangement (Scheme 17).26–28 They used 1.1–2 equiv. of the modified Ph TMS O 42 Ph TMS O 43 R R O X X A chiral binaphthol–aluminum complex [(R)-45] as an activating reagent. In the case of the silylated (E)-substrate 42 the Lewis acid promoted Claisen rearrangement proceeded smoothly via the chairlike transition state to give the silyl ketone 44 with high enantioselectivity (88% ee).The chiral aluminum reagent 45 efficiently discriminated between the two enantiotopic chairlike transition states. In the case of the (R)-45 reagent the transition state A should be more favorable as compared with transition state B. It is noteworthy that the absolute stereochemistry of the product from (Z)-substrate 43 was the same as that of (E)- substrate 42. They explained this result by the fact that the boatlike transition state C is favorable in the case of (Z)- substrate 43 due to the steric repulsion between the allylic 1.1–2 eq. ( R)-45 99% 88% ee 1.1–2 eq. ( R)-45 64% 58% ee Si t-BuMe2 O Al Me O Si t-BuMe2 ( R)-45 R O O X C B Scheme 17 Ph TMS O 44 O R X D substituent and bulky silyl group in the chairlike transition state D.The substrate having a germyl group instead of a silyl group underwent the rearrangement with a higher enantioselectivity. Further development of chiral Lewis acids for the enantioselective Claisen rearrangement by the same group have been studied based on the monomeric aluminum alkoxide composed of an extremely bulky and axially chiral phenol derivative (Scheme 18).29 To this end C3-symmetric aluminum tris- R R (Ar*O)3Al CHO O 47 46 R = t-Bu 70% 91% ee F OH Ar*OH = 48 Scheme 18 phenoxide derived from the naphthol derivative 48 was found to work nicely for the rearrangement of non-silylated achiral allylic vinyl ethers 46 to give the g,d-unsaturated aldehydes 47 having a newly formed chiral center at the b-position in high enantioselectivities (61–92% ee).On the basis of X-ray analysis of the aluminum trisphenoxide–DMF complex they discussed the mechanism of chiral induction in the rearrangement reaction. 3 Asymmetric aromatic Claisen rearrangement Only a few examples of an asymmetric aromatic Claisen rearrangement have been reported. For example (R)-(E)-ether 49 upon heating rearranged to give a mixture of (S)-(E)-50 and (R)-(Z)-51 in a ratio of 82 18 although the optical purity of each product was not reported (Scheme 19).30 OH OH O 200 °C + 51 49 50% 82 18 50 Scheme 19 In general several problems found in the aromatic Claisen rearrangement have retarded the development of its asymmetric variant.4 These involve the ortho para-selectivity of migration abnormal Claisen rearrangement based on the 1,5-sigmatropic hydrogen shift and the rearrangement via an allylic cation mechanism resulting in a loss of regioselectivity and stereospecificity with respect to the geometry of the parent allylic moiety particularly in the case of Lewis acid-mediated conditions.Therefore finding an efficient activating agent for aromatic Claisen rearrangement is an important goal. Very recently Trost et al. succeeded in achieving an asymmetric aromatic Claisen rearrangement by an intramolecular chirality transfer of the para-protected chiral substrate (Scheme 20).31 They synthesized chiral allyl aryl ethers H O OH 10% Eu(fod)3 H CHCl3 50 °C OMe OMe 53 52 79% 97% ee 97% ee Scheme 20 such as 52 by enantioselective O-alkylation of hydroquinone monomethyl ether using a chiral palladium catalyst.Upon treatment of the substrate 52 with a catalytic amount (10% mol) of Eu(fod)3 as Lewis acid an excellent intramolecular chirality transfer could be achieved. We reported the first enantioselective aromatic Claisen rearrangement using an O-allyloxy phenol derivative and a chiral boron reagent (Scheme 21).32 Upon treatment of the Ph Ph SO N N 2 O2S B Tol Tol OH O Br OH OH ( S S)-40 Et3N ( S)-55 ( E)-54 89% 94% ee OH O OH OH ( S S)-40 Et3N ( R)-55 ( Z)-54 92% 95% ee Ph Ph OO S N N O B Ar S O O O R 56 Scheme 21 substrate (E)-54 with 1.5 equiv.of the chiral boron reagent (S,S)-40 and triethylamine at low temperature a boron– phenoxide intermediate was formed which underwent the ortho-rearrangement at 245 °C to give the product (S)-55 in 89% yield with excellent enantioselectivity (94% ee) without the formation of by-products such as the para-rearranged or abnormal Claisen products. Reaction of (Z)-54 and (S,S)-40 gave the rearrangement product having the (R)-configuration with excellent selectivity (R-55 92% 95% ee). The selective ortho-rearrangement observed in the boron-mediated reaction could be rationalized by considering the catechol borate structure in the product since para-rearrangement takes place via Cope-rearrangement of the ortho-rearranged dienone intermediate which in the present case readily isomerizes to the catechol form.Furthermore the catechol borate structure in the product would be unfavorable for a 1,5-hydrogen shift leading to abnormal Claisen rearrangement. The importance of the boron–phenoxide intermediate should be also noted for the rearrangement reaction to proceed regioselectively and to achieve a high enantioselectivity since the substrates without a 49 Chem. Soc. Rev. 1999 28 43–50 phenolic hydroxy group (allyl phenyl ethers and allyl orthomethoxyphenyl ethers) or an ortho-hydroxymethylated substrate did not rearrange when treated with the boron reagent at room temperature.The mechanism of asymmetric induction can be explained by considering the 5-membered cyclic intermediate 56 formed by a covalent bond between the boron reagent and the phenolic hydroxy group and following coordination of the ethereal oxygen to the boron atom. In this cyclic intermediate one of the arenesulfonyl groups would effectively shield one face of the benzene ring therefore the allylic moiety approaches its other face resulting in the enantiotopic facial selection of the allylic double bond. 4 Conclusion The diastereoselective Claisen rearrangements of the substrates having chiral auxiliary and enantioselective variants of achiral substrates were well studied during the last decade as summarized in this article. Numerous successful achievements have been reported in the aliphatic Claisen rearrangement while only a limited number of examples of enantioselective aromatic Claisen rearrangements have appeared.Development of an enantioselective Claisen rearrangement by using a catalytic amount of a chiral source still remains as one of the future objectives in this field. 5 References 1 R. Noyori Asymmetric Catalysis in Organic Synthesis John Wiley New York 1994. 2 L. Claisen Ber. Dtsch. Chem. Ges. 1912 45 3157. 3 P. Wipf in Comprehensive Organic Synthesis ed. B. M. Trost and I. Fleming Pergamon Press Oxford 1991 vol. 5 p. 827. 4 S. J. Rhoads and N. R. Raulins Org. React. (NY) 1974 22 1. Chem. Soc. Rev. 1999 28 43–50 50 6 D. Enders M. Knopp J. Runsink and G. Raabe Angew.Chem. Int. Ed. 5 D. Enders M. Knopp and R. Schiffers Tetrahedron Asymmetry 1996 7 1847. Engl. 1995 34 2278. 7 D. Enders M. Knopp J. Runsink and G. Raabe Liebigs Ann. 1996 1065. 8 D. Enders and M. Knopp Tetrahedron 1996 52 5805. 9 W. Sucrow and W. Richter Chem. Ber. 1971 104 3679. 10 P. A. Bartlett and W. F. Hahne J. Org. Chem. 1979 44 882. 11 P. Metz and B. Hungerhoff GIT Fachz. Lab. 1996 40 690. 12 P. Metz and B. Hungerhoff J. Org. Chem. 1997 62 4442. 13 J. T. Welch and S. Eswarakrishnan J. Am. Chem. Soc. 1987 109 6716. 14 J. Kallmerten and T. J. Gould J. Org. Chem. 1986 51 1152. 15 U. Kazmaier J. Org. Chem. 1994 59 6667. 16 E. J. Corey and D.-H. Lee J. Am. Chem. Soc. 1991 113 4026. 17 E. J. Corey B. E. Roberts and B. R. Dixon J. Am. Chem. Soc. 1995 117 193. 18 E. J. Corey and R. S. Kania J. Am. Chem. Soc. 1996 118 1229. 19 U. Kazmaier and A. Krebs Angew. Chem. Int. Ed. Engl. 1995 34 2012. 20 A. Krebs and U. Kazmaier Tetrahedron Lett. 1996 37 7945. 21 U. Kazmaier Liebigs Ann./Recueil 1997 285. 22 S. E. Denmark and J. E. Marlin J. Org. Chem. 1987 52 5742. 23 S. E. Denmark G. Rajendra and J. E. Marlin Tetrahedron Lett. 1989 24 S. E. Denmark H. Stadler R. L. Dorow and J.-H. Kim J. Org. Chem. 25 H. Ito A. Sato T. Kobayashi and T. Taguchi Chem. Commun. 1998 26 K. Maruoka H. Banno and H. Yamamoto J. Am. Chem. Soc. 1990 30 2469. 1991 56 5063. 2441. 112 7791. 27 K. Maruoka H. Banno and H. Yamamoto Tetrahedron Asymmetry 1991 2 647. 28 K. Maruoka and H. Yamamoto Synlett 1991 793. 29 K. Maruoka S. Saito and H. Yamamoto J. Am. Chem. Soc. 1995 117 1165. 30 H. L. Goering and W. I. Kimoto J. Am. Chem. Soc. 1965 87 1748. 31 B. M. Trost and F. D. Toste J. Am. Chem. Soc. 1998 120 815. 32 H. Ito A. Sato and T. Taguchi Tetrahedron Lett. 1997 38 4815. Review 7/06415B

ISSN:0306-0012    

DOI:10.1039/a706415b

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Planar chiral arene chromium(0) complexes potential ligands for asymmetric catalysis Carsten Bolm and Kilian Muñiz Institut für Organische Chemie der RWTH Aachen Professor-Pirlet-Straße 1 D-52074 Aachen Germany. E-mail Carsten.Bolm@rwth-aachen.de Received (in Cambridge) 22nd July 1998 Planar chiral (h6-arene) tricarbonyl chromium(0) complexes are well-studied compounds in synthetic organic chemistry. However the number of investigations concerned with their application in asymmetric catalysis is rather low. This review aims to summarise the main approaches toward the stereoselective synthesis of planar chiral derivatives of (h6-benzene) tricarbonyl chromium(0) and to present recent applications of such complexes as ligands in enantioselective catalytic processes.1 Introduction The first aimed synthesis of (h6-benzene) tricarbonyl chromium( 0) from benzene and hexacarbonyl chromium(0) was achieved in 1958 by Natta.1 The general structure of such compounds is a characteristic half-sandwich complex in which the six p-electrons of the arene are bound to the central chromium atom in a h6-fashion. The remaining three carbonyl ligands are coordinated in such a way that the molecule’s overall structure is pseudo tetrahedral. Such arrangement has been compared to a ‘three leg piano stool’. For various reasons these complexes have found general synthetic interest their preparation is mostly easier than the synthesis of the related cyclopentadienyl transition metal complexes the products are usually crystalline which makes Carsten Bolm born in 1960 studied chemistry in Braunschweig (Germany) and at the University of Wisconsin Madison (USA).He completed his doctorate with Professor M. T. Reetz in Marburg (Germany) on chiral catalysts for enantioselective C– C coupling. After a postdoctoral stay with Professor K. B. Sharpless at the Massachusetts Institute of Technology Boston (USA) he worked with Professor B. Giese in Basel (Switzerland) obtaining his habilitation on enantioselective catalysis with chiral ligand–metal complexes. In 1993 he became Professor of Organic Chemistry at the University of Marburg (Germany). Since 1996 he has a chair of Organic Chemistry at the RWTH Aachen (Germany). His awards include the Heinz-Maier-Leibnitz prize the ADUC-Jahrespreis for habilitands the annual prize for Chemistry of the Akademie der Wissenschaften zu Göttingen the Otto-Klung prize and the Otto-Bayer award.Carsten Bolm Kilian Mu� niz Zn (CH3)2N H3C ZnO O H R Cr (CO)3 them easy to purify and as a consequence of the metal complexation of the arene several transformations become feasible which cannot be carried out on the metal-free arene ring.2,3 Furthermore both the chromophoric character of the complexes and the characteristic proton NMR shifts of the hydrogen atoms of the complexed arenes allow an unambiguous detection of such complexes and among other things have made them suitable markers in bioorganometallic chemistry.4 It was the fact that those derivatives which bear an unsymmetrically 1,2- or 1,3-disubstituted arene ligand are no longer superimposable with their mirror image that made them interesting for modern organometallic chemistry.Enantiomers like A and ent-A have been termed planar chiral compounds.5,6 B B D2 MLx A A D1 Cr Cr Cr CO CO CO OC OC OC OC OC OC A B ent-A The stereochemical descriptor for an element of planar chirality is usually determined following the rules introduced by Schlögl:7 the arene ring is monitored from that side which is not Kilian Muñiz born in 1970 studied chemistry at Universität Hannover (Germany) and at Imperial College London (UK). In 1996 he completed his Diploma (Universität Hannover) with a research project on Fischer carbene complexes at Universidad de Oviedo (Spain) under the joint guidance of Professor J.Barluenga and Professor H. M. R. Hoffmann. Currently he is working on his PhD in the group of Professor C. Bolm at RWTH Aachen (Germany). His research interests are in the area of transition metal chemistry mainly in the synthesis of enantiomerically pure organometallic compounds and their application in asymmetric transformations. 51 Chem. Soc. Rev. 1999 28 51–59 coordinated to the chromium moiety. The priority of the substituents is then determined employing the Cahn–Ingold– Prelog (CIP) rules. If the shortest path from the substituent displaying highest priority to the following one is clockwise the absolute configuration is denoted as Rp and the opposite case is referred to as Sp.For example if one imagines a priority order for the two substituents of complex A as A > B the absolute planar chirality is Sp. There exists also a different procedure for the stereochemical assignment consisting of an extension of the CIP system which results in opposite planar chiral descriptors.6 Throughout this text we shall employ the original rules by Schlögl. Chromium complexes of planar chirality have extensively been applied as stoichiometric auxiliaries and/or suitable starting materials for asymmetric synthesis of biologically interesting substances.3,8 An aspect that so far has not been exploited in detail is the use of those arene complexes that display two donor functionalities D1 and D2 and when complexed to other transition metals are thereby able to serve as bidentate ligands in organometallic complexes of the general type B.In order to determine the status quo of such complexes in asymmetric catalysis we will first describe general synthetic routes toward optically active 1,2-disubstituted (h6-arene) chromium tricarbonyl complexes (hereafter referred to as ‘ACTCs’) and then present an overview on their application in catalytic processes. Ph Ph 1) N Li 2 2) ClSi(CH3)3 1) 2 2) E+ Cr (CO) 3 6 7 a E = SiMe b E = SnBu Scheme 1 2 General synthetic routes The use of chiral compounds as stoichiometric or substoichiometric auxiliaries requires extensive knowledge of efficient routes for their preparation.Generally two different approaches toward optically active ACTCs can be distinguished which consist of the use of either external or internal chiral auxiliaries leading to enantiomeric or diastereomeric products respectively. In this review the synthesis of the former will be discussed first. Attention will only be paid to those reactions that give direct access to such complexes in a stereoselective manner thus excluding all kinds of enzymatic or chemical resolutions. OCH3 Cr (CO)3 1 O P Ph Ph Cr (CO) Chem. Soc. Rev. 1999 28 51–59 52 An elegant example of an external chiral auxiliary approach is based on the use of C2-symmetric lithium amide bases such as 2 (Scheme 1).9 It is assumed that prior to the lithiation a precoordination of the lithium amide to the methoxy substituent enables a highly selective differentiation between the two enantiotopic hydrogen atoms ortho to the methoxy group of complex 1.This method is restricted in the sense that only certain donor functionalities are tolerated as substituents in the starting material but anisole complexes like 1 give the corresponding 2-functionalised products 3 with very high enantioselectivities. ACTC 3 can be further functionalised by a second lithiation at the remaining ortho-position followed by reaction with electrophiles like aldehydes giving rise to potential bidentate compounds. It should also be pointed out that 3 has emerged as a suitable starting material for the synthesis of natural products like (+)-ptilocaulin.10 A related study was reported for the N,Ndiisopropylcarbamate of tricarbonyl (h6-phenol) chromium(0).In this case chiral lithium base 4 proved slightly superior to 2 and various ortho-substituted products with enantioselectivities in the range of 64–73% ee were obtained.11 When DMF was used as electrophile a single recrystallisation of the resulting product led to a significant increase of the enantiomeric excess (95%). The fact that such functionalisaon is not necessarily restricted to aromatic positions was shown for the chromium complex of isobenzofuran. When chiral base 5 was employed deprotonation of this prochiral substrate was found to be highly selective giving after quenching with appropriate electrophiles the (R,Rp)-configuration products in up to 99% ee.9 Here chiral base 2 was less effective leading to a product with 73% ee only.An interesting further extension has been described for ACTC 6. Stereoselective deprotonation employing 2 as chiral base allowed the synthesis of the planar chiral derivatives 7a,b in high yields and with enantiomeric excesses of up to 73%. For the present case it is noteworthy that product 7a has an Sp configuration which is opposite to the Rp configuration obtained for the functionalisation of 1. Further modification of 7b enabled the synthesis of complexes like 8 which by recrystallisation was obtained in nearly enantiomerically pure form (95% ee).12 Due to its hydroxy functionality and phosphino group 8 can be expected to be an interesting ligand precursor in future asymmetric processes.Ph Li N 4 OCH3 Ph Si(CH3)3 Ph Cr (CO)3 Ph N Li N Li Ph 5 Ph H OH E O P P 3 Ph Ph Ph Ph Cr (CO)3 3 8 3 3 Another enantioselective synthesis of planar chiral complexes is a singular sequence which is based on a tandem reaction of nucleophilic addition and hydride abstraction.13 This approach is limited to those arene chromium complexes which bear strong acceptor substituents like imines hydrazones or oxazolines. By virtue of its relatively low electron density the aromatic ring is accessible for regioselective nucleophilic addition at the ortho-position. The mechanism is believed to be initiated by precoordination of the lithium species to the external chiral ligand 10.A subsequent nucleophilic addition of this chirally modified lithium reagent to the arene occurs with an efficient stereodifferentiation between the two enantiotopic ortho-carbons of 9 to give the intermediary h5-cyclohexadienyl complex 11. This complex can either react further with electrophiles to yield chiral cyclohexadienes or as is of interest in the present case can be oxidised by hydride abstraction to yield the planar chiral arene complex 12 (Scheme 2). Following O Ph-Li N OCH3 Ph Cr (CO)3 Ph OCH3 9 10 Scheme 2 this sequence enantioselectivities of up to 98% ee were obtained for phenyllithium as nucleophile. It should also be mentioned that a diastereoselective version of this sequence is known where a chiral (S)-1-amino-2-methoxymethylpyrrolidine (SAMP)-hydrazone was employed as internal auxiliary.13 The use of external chiral auxiliaries can be extended further to the application of asymmetric catalysis with chiral ligands. In doing so planar chiral compounds can be obtained from prochiral precursors like the dichlorobenzene complex 13 (Scheme 3). Under catalytic conditions using 10 mol% of palladium and 12 mol% of ferrocene (S,Rp)-PPFA planar chiral complex 14 is formed in acceptable yield with a moderate enantioselectivity of 44% ee. The same catalytic system was also reported to be suitable for the cross-coupling reaction of 13 with arylboronic acids although enantioselectivities again remained below 70% ee.14 On the other hand a structurally related prochiral ACTC bearing 1,3-dichloro-2-methylbenzene as the arene ligand gave a coupling product with only 8% ee under these conditions.Among the other approaches of employing internal chiral auxiliaries which have received much attention the diastereoselective complexation of chirally modified arenes is one of the most effective ones. An appropriate chromium tricarbonyl precursor is directed by an asymmetric donor group in the side chain of the arene.8,15 Such groups should effect a precoordina- Li Ph H O N Cr (CO)3 11 Ph3C Ph O N Cr (CO)3 12 Cl Cl CH2=C(CH3)B(OH)2 Cl PdCl(p-C3H5) 2 Cr (CO)3 Cr (CO)3 13 14 N(CH3)2 Ph2P Fe Ph (CH3)3 Si O O N B 3 16 15 ( S Rp)-PPFA Scheme 3 (CH3CN)3Cr(CO)3 Scheme 4 ( S Rp)-PPFA tion to the chromium moiety and thus enable a substantive differentiation of the two diastereotopic faces of the arene ring.This principle has widely been used in the synthesis of planar chiral arene complexes bearing benzylic alkoxy functionalities. 8 Since most of these complexations are carried out under relatively harsh thermal conditions in order to allow the substitution of three carbonyls by the arene ring equilibration can take place resulting in the formation of diastereomeric mixtures. As a consequence of the demand for milder reaction conditions tricarbonyl(h6-naphthalene)chromium(0) (‘Kündig’s reagent’) has emerged as a suitable chromium tricarbonyl precursor:3,8 this complex is known to be relatively labile because of a possible ring slippage via h6?h4 coordination.This process creates a free coordination site at the resulting unsaturated 16 e-chromium centre and allows a selective coordination of the incoming arene upon further displacement of the naphthalene. An interesting complementary approach was recently reported for an optically active Lewis acid complex (Scheme 4) Ph 3)3 (CH Si B N Cr (CO) 2-trimethylsilylborabenzene was coordinated to an enantiomerically pure 2-phenyloxazoline derived from (S)-leucinol affording the atropisomeric Lewis acid–base adduct 15. Complexation of the borabenzene’s p-system onto the chromium tricarbonyl fragment proved to be completely chemoselective and resulted in the formation of the planar chiral Lewis acid complex 16 as a single diastereoisomer.16 To date this example represents the only enantiopure planar chiral (h6-arene) chromium tricarbonyl complex derived from a heteroarene.An alternative route that does not rely on the complexation of an existing pre-manufactured arene ring onto the chromium tricarbonyl moiety is based on a benzannulation reaction of Fischer carbene complexes. In this procedure which is also known as the ‘Dötz reaction’ an a,b-unsaturated carbene an alkyne and one molecule of carbon monoxide undergo a formal [3 + 2 + 1]-reaction within the coordination sphere of the metal to yield the arene ligand. In the synthesis of planar chiral compounds this approach is unique in the sense that only Fischer carbene complexes of pentacarbonylchromium(0) yield those products in which the metal is attached to the resulting arene group.Chem. Soc. Rev. 1999 28 51–59 53 Diastereoselective Dötz reactions have been described for three possible routes a first report in 1994 introduced a sequence based on achiral chromium carbene complexes.17 Here the stereochemical course of the reaction was governed by sterically demanding a-chiral propargylic (prop-2-ynylic) ethers and led to high diastereoselectivities of up to 92% de. A more general approach is given for the use of those Fischer carbene complexes bearing chiral substituents like for example chiral pool derived alkoxy groups.For the reaction of the (2)-menthoxy carbene complex 17 with tert-butylacetylene a 10:1 ratio of the two resulting diastereoisomers 18a and 18b was reported (Scheme 5).18 O (OC)5Cr 17 OR O Cr (CO)3 18a R = Si(CH3)3 Scheme 5 It was suggested19 that an E/Z-isomerisation due to rotation around the carbene carbon–oxygen bond would prevent the formation of an even more selective arrangement during the first two steps of the benzannulation. Indeed complete diastereoselection was never observed. In a complementary approach a series of chiral cyclohexenyl carbene complexes of chromium was submitted to a benzannulation. 19 Again diastereoselectivities remained below 90% and although a detailed discussion based on kinetic studies and theoretical calculations was made the exact stereochemical course of the Dötz reaction still remains to be further elucidated in order to allow the achievement of complete stereoselection.Most approaches for the synthesis of planar chiral ACTCs still rely on diastereoselective transformations and use the directing power of internal chiral auxiliaries. Thus enantiomerically pure substituents are capable of directing a base to selectively remove one of the two diastereotopic hydrogens in the ortho-position to create the element of planar chirality. This process is generally termed directed ortho-metalation20 and it is closely related to the one discussed above for the use of chiral bases which were employed to differentiate the deprotonation of enantiotopic hydrogens.Most suitable ortho-directing groups consist of saturated oxygen or nitrogen functionalities like tertiary amines ethers or acetals which all display free electron pairs. Hard Lewis acids of this type have been proved to be very efficient for precoordinating and selectively directing the incoming alkyllithium base. Thus a diastereomerically pure lithiated intermediate is generated which upon quenching with electrophiles yields the 1,2-disubstituted arene chromium complexes displaying a plane of chirality. Originally this concept had been developed for the synthesis of optically pure ferrocene derivatives like PPFA but was adapted successfully to the chemistry of ACTCs by the group of Davies and others starting complex 19 is easily accessible from Chem.Soc. Rev. 1999 28 51–59 54 ClSi(CH3)3 RO O Cr (CO)3 18b commercially available a-phenylethylamine by the Eschweiler –Clarke reaction followed by simple complexation with hexacarbonylchromium(0). This synthesis is even more convenient than the one for the corresponding N,N-dimethyl-aferrocenylethylamine which involves a resolution step. Selective deprotonation of ACTC 19 is then accomplished by means of tert-BuLi at low temperature resulting in the formation of lithiated intermediate 20. The observed stereoselectivity has been explained by the assumption that during the deprotonation the most favourable conformation is obtained for a transition state of minimized steric interaction in which the benzylic methyl group points away from the bulky chromium tricarbonyl moiety below the aromatic plane (Scheme 6).N(CH3)2 Cr (CO)3 19 O O Cr (CO)3 22 Subsequent treatment of 20 with electrophiles like TMSCl yields planar chiral ACTC 21 as a single diastereoisomer.21 A related system was presented for the chromium complexes of N,O-dimethylephedrine and N,O-dimethylpseudoephedrine. 22 Again complete diastereoselection was reported assuming the high stereoselectivity to derive from a favourable N,O-chelatisation of the metal in the lithiated intermediate. Another example for an ACTC bearing an ortho-directing group is given with tartrate derived acetal 22 which undergoes ortho-functionalisation with up to 94% de.23 Several similar systems are known that are formally all based on chiral pool derived acetals of the chromium complex of benzaldehyde.24 We have recently reported on the use of chromium complex 23 prepared from benzylated SMP.25 Here the reaction is tert BuLi OCH3 OCH3 Scheme 6 N(CH3)2 Li Cr (CO)3 20 ClSi(CH3)3 N(CH3)2 Si(CH3)3 Cr (CO)3 21 N OCH3 Cr (CO)3 23 O S Ph Cr (CO)3 24 assumed to proceed via an intermediate lithium species in which the metal is coordinated to both the amine and the oxygen of the alkoxy group giving rise to a highly favoured chelate structure.Consequently diastereoselectivities of up to 97% de have been achieved in reactions with a variety of electrophiles.It was further demonstrated from the use of sulfoxide derived ACTC 24 that the stereogenic centre of the ortho-directing group need not necessarily be a chiral carbon atom. Again diastereomeric excesses were excellent with up to 99% de. Interestingly upon using 2.5 equivalents of base both orthohydrogen atoms can be removed and subsequent treatment with an electrophile now furnishes the second diastereoisomer with opposite planar chirality. This can be explained by the assumption that the second lithiation on the remaining originally less favoured ortho-position yields a non-chelated aryllithium which then undergoes a faster reaction with the electrophile. To date these sequences represent the only examples for a directed ortho-metalation in which both diastereoisomers can be obtained selectively by simple control of the amounts of base added.26 Having generated the element of planar chirality via diastereoselective processes with the help of internal control one can further manipulate the resulting diastereomeric products.For instance removal of the original chiral ortho-directing group leaves an enantiomeric arene chromium complex with only planar chirality. This goal has been found to be of particular difficulty. Thus in most cases the removal of chiral diols from their respective acetals was not successful particularly when compounds with ortho-substituents other than TMS or methyl had been prepared.23,24 An elegant procedure for the mild removal of a tertiary amino group was reported by Gibson née Thomas.27 In the presence of dimethyldioxirane (25) amine complexes like 21 are smoothly converted into their corresponding N-oxides at low temperature which upon warming undergo a Cope elimination (Scheme 7).O O N(CH3)2 O N(CH3)2 Si(CH3)3 Si(CH3)3 Cr (CO) Cr (CO)3 3 26 21 Si(CH3)3 Cr (CO)3 27 25 Scheme 7 The resulting enantiopure ortho-substituted styrene complexes like 27 can then be transformed further. For the present discussion in this review however it is usually more interesting to maintain the former ortho-directing group as a potential donor atom for coordination of transition metals. This approach has been pursued for those orthodirecting groups based on amines like 19 or 23.We have recently shown that 23 is a suitable starting material for the synthesis of a series of potential ligands and ligand precursors bearing both the heteroatoms of the SMP moiety combined with phosphino or hydroxy groups created in the ortho-functionalisation step.25 CH3 N PPh2 N(CH3)2 PR2 PPh2 Cr (CO)3 Cr (CO)3 29 30 a R = Ph b R = C6H11 3 Catalyses While the use of achiral ACTCs in catalytic hydrogenation and isomerisation reactions is well-studied,28 planar chiral derivatives for asymmetric catalysis have so far received only scant attention. This is even more surprising taking into account their structural relationship to the great number of successfully employed ferrocene ligands. It is evident that due to their structural properties ACTCs cannot serve as ligands in oxidation chemistry since under normal oxidative conditions decomposition of the complex will take place.However they have been employed in the two other areas of asymmetric catalysis C–C-bond formation and multiple bond reduction. In the area of C–C-bond formation two palladium catalysed processes involving planar chiral diastereomerically pure ACTCs are known asymmetric cross couplings and nucleophilic substitutions. For the purpose of an efficient coordination to palladium four bidentate ACTCs have been synthesised complex 28 N(CH3)2 PPh2 Cr (CO)3 28 represents a derivative of 19 and was obtained via the above described ortho-functionalisation.29 Diaminophosphine 29 was prepared in an analogous manner using (R)-N,N,N’-trimethyl- N’-(a-phenylethyl)ethylenediamine as starting material.30 ACTCs 30a,b were synthesised similarly starting from the chromium complexes of (S)-a-phenylethyl methoxymethyl ether followed by a two step sequence of diastereoselective ortho-phosphorylation and nucleophilic displacement at the benzylic position.30 Using 28 as the ligand in palladium catalysed cross-coupling reactions between 1-phenylethyl zinc chloride and vinyl bromide [eqn.(1)] the coupling product was generated with an enantiomeric excess of up to 61%. Enantioselectivities were lower for the corresponding magnesium reagent and for the use of 2-phenylvinyl bromide. Alternatively nickel(II) chloride could be used as the metal source and the complex was then generated in situ albeit without achieving a change in enantioselectivity (61% ee) and by getting slightly lower chemical yields.29 Another well-studied reaction which relies on palladium catalysis the Tsuji–Trost reaction consists of an allylic alkylation via nucleophilic displacement.Here complexes 29 and 30a were employed as ligands and again the catalytically active species was formed in situ.30 For the standard system the reaction of 1,3-diphenylacetoxypropene with sodium dimethyl malonate an enantiomeric excess of 94% in favour of the (S)- enantiomer was achieved with 29 [eqn. (2)]. Under the same conditions catalysis relying on ligand 30a led to a moderate ee of only 61% and the reaction needed to be carried out at a much lower temperature (278 °C) in order to obtain at least 86% ee.Due to the change in absolute configuration of ligand 29 compared to 30 the product now was of R-configuration. 55 Chem. Soc. Rev. 1999 28 51–59 Br ZnCl H3C OAc NaCH(CO2CH3)2 PdCl(p-C3H5) 2–L* In a different reaction for the formation of C–C bonds aldehydes can be converted into chiral secondary alcohols by means of dialkylzinc reagents in the presence of substoichiometric amounts of suitable ligands. For this type of catalysis three structurally related catalyst precursors have been described a whole series of these compounds could again be obtained from ACTC 19 via an ortho-functionalisation using ketones or aldehydes as electrophiles to give 31a–c and 31d,e N(CH3)2 OH R2 R1 Cr (CO)3 31 a R1 R2 = H b R1 R2 = Et c R1 R2 = Ph d R1 = Et R2 = H e R1= Ph R2 = H N OCH3 CR2OH Cr (CO)3 33 a R = H b R = Ph respectively.31 For the latter case the creation of the new stereocentre at the benzylic position was reported to be highly selective yielding diastereomerically pure compounds.In the diethylzinc addition to benzaldehyde [eqn. (3); R = Et) 5 mol% of ACTCs 31b or c led to the formation of (S)- 1-phenylpropanol with very high enantioselectivities. On the other hand the hydroxymethyl compound 31a that lacks sterically demanding substituents in the benzylic position led to only 15% ee. Introduction of additional stereocenters in the benzylic position was found to have no beneficial effect and the enantiomeric excesses of products from catalyses with ACTCs 31d and e were identical or slightly lower.However a change Chem. Soc. Rev. 1999 28 51–59 56 PdCl2–L* (1) or NiCl2–L* CH3 CO2CH3 H3CO2C (2) N HO Cr (CO)3 32 OH CHO R ZnR –L* 2 (3) O Ni(acac)2–Zn(C2H5)2–L* C2H5 O (4) of absolute configuration at this stereogenic centre from (S) to (R) resulted in a dramatic decrease in enantioselectivity. Switching planar chirality by the use of a complex in which the chromium moiety was attached to the other face of the arene ring than in 31b also decreased the enantiomeric excess of the catalysis product (29% ee). Thus it has to be concluded that in order to achieve excellent enantioselectivities an efficient internal cooperation of all stereoelements of the ligand is essential.For the present system ACTC 31b evidently displays an optimum combination of planar and central chirality and a transition state C was proposed containing seven- and sixmembered chelate rings. H Zn (CH3)2NH3C ZnO O H R Cr (CO)3 C A different ligand type was presented with ACTC 32.32 This complex was derived from the enantiopure chromium complex of indan-1-one and its synthesis involved a nucleophilic addition of 2-(lithiomethyl)pyridine to the carbonyl function. In a similar manner three other structurally related complexes were synthesised. Amongst the four different ligands complex 32 performed best yielding (S)-1-phenylpropanol with 70% ee from the reaction of benzaldehyde with diethylzinc [eqn.(3); R = Et]. It is noteworthy that a dramatic effect was observed when the uncomplexed ligands were employed. In the case of 32 the arene ligand alone catalysed the formation of a product with only 10% ee. Finally ACTCs 33a,b were obtained from 23 as described above.25 They were employed in the asymmetric addition of dimethyl or diethylzinc to benzaldehyde [eqn. (3); R = Me or Et] and ferrocenecarbaldehyde. Catalyst loading had to be 10 mol% in order to obtain acceptable conversion and enantioselectivities (eemax 86%). It was observed that complex 33b performed much better than 33a which is in accord with the observation made for the series of ACTCs 31.31 A related catalytic reaction is the nickel catalysed asymmetric conjugate addition of dialkylzincs to Michael acceptors such as chalcones [eqn.(4)]. For this purpose ACTC 31e performed best. Using 5 mol% of nickel salt and a high ligand loading of 50 mol% the product was obtained in 90% chemical yield having an enantiomeric excess of 62%. However these figures imply that this catalytic system is rather ineffective. Even when both the nickel salt and the ligand precursor were used in stoichiometric quantities an ee of only 78% was obtained.31 For asymmetric catalytic reductions two different approaches are known. The first is based on borane mediated hydrogen transfer and has been extensively investigated. ACTCs 34a,b and 35a,b are readily prepared starting with a 3 R R (CO) Cr OH NH 34 a R = H b R = Me 3 R R (CO) Cr O N BH 36 a R = H b R = Me complexation of (S)-indoline-2-carboxylic acid with the chromium tricarbonyl fragment followed by hydride reduction or alkyl addition to the resulting complex to give the final amino alcohols.33 When treated with a solution of borane in THF a catalyst like 36 is proposed to be formed.With 10 mol% of BH3–oxazaborolidine catalyst 36a an enantiomeric excess of 50% was obtained in the reduction of acetophenone [eqn. (5)]. O BH –L* (CO)3 Cr OPCy N PCy2 37 Use of a bulkier substrate like 9-acetylphenanthrene led to a rise in enantioselectivity (62% ee). However acceptable enantiomeric excesses could only be obtained with stoichiometric amounts of 34.For the reduction of acetophenone it was further found that catalyst precursor 35a with opposite (Sp)-planar chirality but identical central chirality gave a significantly lower R R OH NH Cr (CO)3 35 a R = H b R = Me OH 3 (5) O 2 O O Rh(L*)O2CCF3 2 OH (6) O O enantioselectivity (25% ee). Relating this drop in ee to the change from (Rp)- to (Sp)-configuration suggests an internal mismatched case of stereoelements for (S,Sp)-35a. Interestingly for the other two diastereomers 34b and 35b which in their side chains bear two additional methyl groups the reaction outcome was found to be the reverse. Here a higher enantiomeric excess was obtained in catalyses with the diastereomer having (Sp)-configuration (39% ee for ACTC 35b compared to 20% ee for 34b).All these results have been explained by an assumed transition state D which is depicted for ACTC 34a representing the most efficient catalyst precursor Cr(CO)3 N O B H3B O D within this set. In D the bulky tricarbonylchromium(0) moiety enables an appropriate distinction between the methyl and the phenyl group by minimising steric interactions with the carbonyl ligands after coordination of acetophenone and borane onto the oxazaborolidine catalyst. For ACTC 35a the chromium moiety is located on the opposite side of the arene leading to a lower selectivity in the coordination step. For the other pair of diastereomers it was suggested that in the case of 34b the additional two methyl groups would cause unfavourable interactions with the chromium moiety leading to a highly strained overall complex.Thus 35b which does not suffer from such interactions is a better catalyst precursor. Another catalytic reaction where ACTCs have been used is the carbonyl reduction with molecular hydrogen catalysed by a complex derived from a ligand bearing a phosphite and a phosphinite group and rhodium salts. Ligand 37 was obtained by conversion of 34a into the corresponding diphosphine derivative.34 Interestingly the analoguous complex bearing two diphenylphosphinite groups could not be synthesised since conversion stopped after generation of the aminophosphinite. ACTC 37 was employed as ligand for an in situ complexation to rhodium sources [Rh(COD)Cl]2 and [Rh(COD)OCOCF3]2.The complex obtained from [Rh(COD)OCOCF3]2 was used as catalyst for the asymmetric hydrogenation of dihydro-4,4-dimethylfuran-2,3-dione [eqn. (6)] and a product with an enantiomeric excess of > 99% was obtained. Interestingly the enantiomeric excesses for the two catalysts derived from either ACTC 37 or its uncomplexed arene were found to be equally high indicating that in this case the phosphite/phosphinite structure dominates the course of the catalysis. However when the substrate was N-benzylbenzoylformamide catalysts derived from 37 and the respective Rh(i)-source performed better than the ones prepared from the uncomplexed arene. This indicates that either the additional stereoelement or the electronic change resulting from the complexation has a significant influence on the catalysis.3.1 Further modifications of the catalyst structures One of the most interesting features of chromium arene complexes relies on the possible electronic and steric tuning by substitution of one of the carbonyl ligands by a different donor molecule as for example tertiary phosphines (E?F Scheme 8). Such an exchange has been carried out on complexes 29 28 Chem. Soc. Rev. 1999 28 51–59 57 D2 D2 PR3 hn D1 D1 Cr Cr OC CO CO R3P OC OC F E Scheme 8 31b–d and 34 leading to the corresponding phosphine or phosphite complexes 38 39a,b 41 and 42 respectively. A N(CH3)2 N CH3 N(CH3)2 PPh2 PPh2 N(CH3)2 PPh2 Cr L(CO) Cr (CO)2PPh3 2 38 39 a L = PPh3 40 b L = P(OMe)3 R R OH N(CH3)2 OH NH R2 R1 2 Cr (CO)2PR3 Cr (CO) P(OPh)3 41 42 large influence on the electronic properties of the complex as a whole can be expected since phosphorus ligands are weaker pacceptors than the carbonyl group.35 As a general consequence the electron density of the arene ligand and thereby the basicity of the diphenylphosphino group should be increased.This assumption is in agreement with the results observed in the catalysed asymmetric Tsuji–Trost alkylation with 38 as ligand which leads to a product with lower ee compared to the one obtained from a reaction with parent complex 29. This suggests that by the diminished p-acceptor character of the coordinating phosphino group a less efficient chiral recognition of the enantiotopic carbon centres in the intermediate h3-allyl complex results.For the catalytic system derived from 28 the following observations were made ACTC 28 itself is the best ligand for the asymmetric cross-coupling described above [eqn. (1)] giving the desired product with an enantiomeric excess of 61%. Catalysis with the corresponding uncomplexed arene 40 as ligand yields a product with only 40% ee indicating that either the electronic consequences resulting from the complexation of 40 to the Cr(CO)3 fragment or the additional element of planar chirality have a decisive influence on the reaction outcome. Use of phosphine and phosphinite substituted complexes 39a and 39b also affords products with lower enantiomeric excess (37 and 17% ee respectively).29 Thus electronic reasons cannot be solely responsible for the course of the catalysis since substitution by the weaker p-acceptor PPh3 leads to a lower decrease in selectivity.It is therefore reasonable to argue that the element of planar chirality has a decisive influence on the catalysis and that the observed differences in the enantiomeric excesses are due to steric interactions to a significant extent. A positive influence of such a substitution was reported for the diethylzinc addition reactions.31 Thus complexes of the Chem. Soc. Rev. 1999 28 51–59 58 general structure 41 were found to be superior to their tricarbonyl parent complexes regardless whether the substitution of the carbonyl was accomplished by PPh3 or P(OPh)3.This indicates that the additional steric bulk of these ligands rather than the electronic tuning was the reason for this beneficial effect. Another substitution of this kind was reported for the oxazaborolidine system.33 Thus ACTC 42 yielded a product with only 18% ee in the borane mediated reduction of acetophenone. This result was explained assuming disfavoured steric interactions resulting in a partly reverse coordination preference. However electronic reasons have to be taken into account here also because an increase of the electron density of the arene will undoubtedly affect the electronic nature of the amine and thereby decrease the Lewis acidity of the whole complex.While so far a lot of data are available for ferrocene ligands structural elucidation of the corresponding ACTCs has not been carried out extensively. Moreover most catalysts and catalyst precursors have been formed in situ and no insight into their respective structures has been gained (for example the catalysts in the alkylzinc and the borane chemistry the rhodium based hydrogenation catalysts and most complexes involving palladium as the metal centre). The only exceptions are complexes 43a–c which were obtained from simple complexation of N(CH3)2 PdCl2(NCCH3)2 Ph2P 28 39a,b PdCl2 Cr L(CO)2 43 a L = CO b L = PPh3 c L = P(OMe)3 Scheme 9 palladium(II) chloride with ligands 28 and 39a,b.Although they were characterised in the usual way their solid state structures were not reported.29 It is even more surprising that there exist only two X-ray crystal structures for ACTCs that have been employed in asymmetric catalysis. Such structural determinations were carried out for 30a30 and 31e.31 Finally with the exception of the NMR data for 43a–c29 no investigations on solution structures of chiral catalysts derived from ACTCs have been reported in the literature either. 4 Conclusion We have presented a brief overview on the most efficient routes for the preparation of optically pure ACTCs. It has been shown that although a variety of different synthetic strategies exists there is still a need for more general approaches that allow the efficient preparation of a greater number of complexes.Such novel routes must also lead to the development of suitable syntheses of enantiopure ACTCs with sole planar chirality. So far asymmetric catalyses with such ligands have not been reported. In contrast optically active ACTCs with both planar and central chirality have been extensively tested and their capability to serve as ligands for asymmetric transformations has widely been demonstrated. However we believe that the full potential of such ligands has not been exploited in depth yet. We expect that ACTCs will play a much broader role in the near future. In order to achieve this goal it will be necessary to study mechanistic details of the catalyses and to pay much more attention to the structural elucidation of respective transition metal complexes in which ACTCs act as (multidentate) ligands.5 Acknowledgements K. M. acknowledges the hospitality of Dr S. E. Gibson during an ERASMUS stay at Imperial College in 1993/4. We are grateful to the Deutsche Forschungsgemeinschaft (DFG) within the Collaborative Research Center (SFB) 380 ‘Asymmetric Synthesis by Chemical and Biological Methods’ and the Fonds der Chemischen Industrie for financial support. We also thank Professor Dr A. Salzer Dr C. Ganter and D. Vasen for several discussions. 6 References 1 G. Natta R. Ercoli and F. Calderazzo Chim. Ind. (Milan) 1958 40 287. 2 M. F. Semmelhack in Comprehensive Organometallic Chemistry II vol. 12 eds. E. W. Abel F.G. A. Stone and G. Wilkinson Pergamon New York 1995 p. 979 and references therein. 3 L. S. Hegedus Transition Metals in the Synthesis of Complex Organic Molecules University Science Books Mill Valley 1994 ch. 10 and references therein. 4 G. Jaouen and A. Vessiéres Acc. Chem. Res. 1993 26 361 and references therein. 5 V. I. Sokolov Chirality and Optical Activity in Organometallic Compounds Gordon and Breach Science Publishers New York 1990. 6 A. Solladié-Cavallo in Advances in Metal Organic Chemistry vol. 2 ed. L. S. Liebeskind JAI London 1989 p. 99. 7 K. Schlögl Top. Stereochem. 1967 1 39. 8 M. Uemura in Stereochemistry of Organometallic and Inorganic Compounds vol. 5 ed. P. Zanello Elsevier Amsterdam 1994 p. 507 and references therein. 9 R.A. Ewin A. M. Macleod D. A. Price N. S. Simpkins and A. P. Watt J. Chem. Soc. Perkin Trans. 1 1997 401 and references therein. 10 K. Schellhaas H.-G. Schmalz and J. W. Bats Chem. Eur. J. 1998 4 57. 11 E. P. Kündig and A. Quattropani Tetrahedron Lett. 1994 35 3497. 12 A. Ariffin A. J. Blake W.-S. Li and N. S. Simpkins Synlett 1997 1453. 13 A. Fretzen and E. P. Kündig Helv. Chim. Acta 1997 80 2023. 14 M. Uemura H. Nishimura and T. Hayashi J. Organomet. Chem. 1994 473 129. 15 A. Alexakis P. Mangeney I. Marek F. Rose-Munch E. Rose A. Semra and F. Robert J. Am. Chem. Soc. 1992 114 8288. 16 J. Tweddell D. A. Hoic and G. C. Fu J. Org. Chem. 1997 62 8286. 17 R. P. Hsung and W. D. Wulff J. Am. Chem. Soc. 1994 116 6449. 18 K. H. Dötz and C. Stinner Tetrahedron Asymmetry 1997 8 1751. 19 R. P. Hsung W. D. Wulff and C. A. Challener Synthesis 1996 773. 20 V. Snieckus Chem. Rev. 1990 90 879. 21 J. Blagg S. G. Davies C. L. Goodfellow and K. H. Sutton J. Chem. Soc. Perkin Trans. I 1987 1805. 22 S. J. Coote S. G. Davies C. L. Goodfellow K. H. Sutton D. Middlemiss and A. Naylor Tetrahedron Asymmetry 1990 1 817. 23 Y. Kondo J. R. Green and J. Ho J. Org. Chem. 1993 58 6182. 24 J. W. Han S. U. Son and Y. K. Chung J. Org. Chem. 1997 62 8264 and references therein. 25 C. Bolm K. Muñiz and C. Ganter New J. Chem. in the press. 26 S. G. Davies T. Loveridge and J. M. Clough J. Chem. Soc. Chem. Commun. 1995 817. 27 P. W. N. Christian R. Gil K. Muñiz-Fernández S. E. Thomas and A. E. Wierzchleyski J. Chem. Soc. Chem. Commun. 1994 1569. 28 M. Sodeoka and M. Shibasaki Synthesis 1993 643. 29 M. Uemura R. Miyake H. Nishimura Y. Matsumoto and T. Hayashi Tetrahedron Asymmetry 1992 3 213. 30 Y. Hayashi H. Sakai N. Kaneta and M. Uemura J. Organomet. Chem. 1995 503 143. 31 M. Uemura R. Miyake K. Nakayama M. Shiro and Y. Hayashi J. Org. Chem. 1993 58 1238. 32 S. Malfait L. Pélinski and J. Brocard Tetrahedron Asymmetry 1996 7 653. 33 G. B. Jones S. B. Heaton B. J. Chapman and M. Guzel Tetrahedron Asymmetry 1997 8 3625. 34 C. Pasquier S. Naili L. Pelinski J. Brocard A. Mortreux and F. Agbossou Tetrahedron Asymmetry 1998 9 193. 35 C. A. Tolman Chem. Rev. 1977 77 313. Review 8/01291A 59 Chem. Soc. Rev. 1999 28 51–59

ISSN:0306-0012    

DOI:10.1039/a801291a

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

The tethered nitrogen in natural products synthesis Spencer Knapp Department of Chemistry Rutgers University 610 Taylor Road Piscataway New Jersey 08854-8087 USA. Email knapp@rutchem.rutgers.edu Received (in Cambridge) 10th September 1998 A variety of nitrogen-containing natural products including aminosugars and aminocyclitols have been synthesized by routes that feature the intramolecular delivery of a temporarily-tethered nitrogen nucleophile to an electrophilic site. This general tactic for amino group introduction frequently provides entropic advantages as well as improved site selectivity and stereoselectivity compared with the corresponding intermolecular approach. An occasional additional benefit is that the resulting cyclized products can be more easily manipulated toward the desired target than the corresponding free amino compounds.These aspects are illustrated in a discussion of several natural products syntheses from the author’s laboratory. ‘Help from without is often enfeebling in its effects but help from within invariably invigorates’ Samuel Smiles Self-Help 1859 1 Introduction There are many ways in which the synthetic chemist can view potential targets for synthesis. One approach which is certainly not favored by everyone would be to look for interesting patterns of functionality in a group of natural products judge whether existing methodology is fully up to the task and then perhaps set about devising new methods for the installation of this functionality. If the methods have worth they ought to be applicable to the syntheses of the very natural products that inspired their development.Spencer Knapp was born in 1951 in Baytown Texas and raised in Tallmadge Ohio. As a Fellow of the Ford Foundation Six- Year BA-PhD Program he received his BA in 1972 and his PhD in 1975 from Cornell University both under the mentorship of Jerrold Meinwald. Following an NIH postdoctoral fellowship at Harvard University with E. J. Corey he joined the faculty of Rutgers University in 1977. His research interests include the synthesis of natural products enzyme inhibitors and complex ligands and the development of new synthetic methods. He has received an NCI Young Investigator Award an American Cyanamid Faculty Award and a Hoechst-Celanese Innovative Research Award and serves on the Editorial Board of the Journal of Carbohydrate Chemistry.OH Nature has provided chemists with an amazing variety of densely functionalized small molecules that contain amino and hydroxy groups sprinkled about an otherwise simple carbocyclic or heterocyclic framework. A few examples relevant to the present discussion are shown in Fig. 1. This list includes SCH OCH3 HO NH2 HN OCH3 CH3 (+/–)-sporamine 1 O HO HN HO H H H3C O HO HO SCH3 NH SCH lincomycin 3 2 3 HO OH HO (+)-mannostatin 5 Fig. 1 Synthetic targets with cis vicinal amino alcohol functional grouping. representatives from both the aminosugar (e.g. lincomycin 3) and aminocyclitol (e.g.valienamine 4) families of natural products. One could easily add representatives from other classes of natural products such as alkaloids amino acids and sphingosines. Clearly no synthesis of these molecules should be contemplated without paying attention to the problem of how to introduce amino and hydroxy groups with control of site and stereochemistry. A recurring ‘functional grouping’ or pattern of functionality that can be identified in these molecules is the vicinal amino alcohol and for each molecule in Fig. 1 an example of this functional grouping has been boxed-in to show how it occurs in context. Apart from this context however one can consider general methods of introducing vicinal functionality and the Chem.Soc. Rev. 1999 28 61–72 trehazolin 6 O H O H HO 3 H3C HN O HO (CH3)2N OH (–)-methyl ravidosaminide 2 OH CH3 H N HO NH HO H CH3 N 2 CH2CH2CH3 OH (+)-valienamine 4 HO N HO HO O O HO O OH HO OH n-Pr OH 61 most obvious is in many ways also the best electrophilic functionalization of an alkene. An old example1 will suffice to show the power of this approach (Scheme 1). Trans-addition to NCO I I 7 8 achiral mono-functional N C O I CH3OLi (cat.) CH3OH 74% 9 OCH N C O ether 3 HN I O chiral difunctional face-differentiated site-differentiated 10 Scheme 1 Alkene trans-functionalization. cyclopentene (7) of iodonium isocyanate generated in situ from iodine and sodium cyanate occurred by way of a cyclic iodonium ion 8 to give the trans-iodo isocyanate 9.The isocyanato group was converted to the methyl carbamate (10) by addition of methoxide. Displacements of iodo may be contemplated for the further synthesis of derived vicinal functionality. The overall transformation is from the achiral monofunctional cyclic alkene 7 to the chiral difunctional facedifferentiated and site-differentiated iodo carbamate 10 and this represents a tremendous increase in the complexity or level O [O] 12 (and diastereoisomer) Z Z N •• N 11 15 (and diastereoisomer) Scheme 2 Synthesis of trans vicinal amino alcohols by attack of an external nucleophile on a 3-membered intermediate.‘Z’ represents an amino activating or protecting group; ‘O’ represents an epoxidizing reagent. Z N HO Y E O 19 18 O Y H2N E N Z 23 Scheme 3 Synthesis of cis vicinal amino alcohols by attack of a tethered internal nucleophile on an electrophilic intermediate. ‘Y’ represents the tether; ‘Z’ represents an amino activating or protecting group; ‘E’ represents an electrophilic reagent such as iodonium. (single stereoisomer and regioisomer) 22 Chem. Soc. Rev. 1999 28 61–72 62 of functionality or functional group content of the cyclopentane. Cyclic vicinal amino alcohols come in two types trans and cis. One can imagine functionalizing a cyclic alkene from opposite faces to arrive at the former and from the same face to arrive at the latter.The trans introduction of amino and hydroxy groups can be accomplished in a straightforward manner (Scheme 2) ring-opening of an epoxide 12 by a nitrogen nucleophile or ring-opening of an aziridine 15 by an oxygen nucleophile. Of course the use of an amino protecting activating or precursor group such as N-sulfonyl or azido may be desirable and similarly for the hydroxy. Note that unless the cyclic system 11 is symmetric or has special features that differentiate the two alkene sites and the two alkene faces mixtures of diastereoisomers and regioisomers (relative to a preexisting stereogenic center or substituent say) may be expected. Introducing the cis vicinal amino alcohol functionality is typically more complicated.Two possibilities are displacement of a trans-difunctional precursor such as 13 or syn-addition of both the nitrogen and oxygen atoms by way of a reagent like O3OsNNHR.2 One can still expect site- and face-differentiation to be a problem. In the late 1970’s when we first began thinking about approaches to potential synthetic targets such as those in Fig. 1 it was clear that there was a need for additional methods tailored specifically for the cis vicinal amino alcohol functional grouping. It was also clear that the concept of neighboring group participation3,4 could be put to use for managing the activation and site and stereochemistry of introduction of the nitrogen or oxygen nucleophile. Scheme 3 shows this idea in generalized form.The basis of the neighboring group participation approach to the synthesis of vicinal amino alcohols is to use the hydroxy group as a handle (that is directing group) onto which a Z HO NH NH HO Z NH 2 2 14 13 (and regioisomer) Z OH H2N R-OH HN OR 17 16 (and regioisomer) Z NH2 Y N E HO O E 21 20 (single stereoisomer and regioisomer) HO Y O E H2N N E Z 25 24 nitrogen nucleophile can be tethered (19) or alternatively to use the amino as a handle for tethering a nucleophilic oxygen (23). Intramolecular delivery of the nitrogen (or oxygen) nucleophile to an electrophile-activated alkene affords excellent control of the site and stereochemistry of the eventual amino (hydroxy). For most alkene cyclizations involving participation across five atoms the nucleophilic nitrogen (oxygen) will be delivered kinetically at the proximal carbon and cis to the directing group (20 and 24).Although special cases could arise where steric hindrance CNC polarization reversibility or other effects alter this result it is nevertheless possible to design and execute synthetic routes with intramolecular participation as the key element for installing the cis,vicinal amino alcohol functional grouping. Furthermore recognizing the appropriate allylic alcohol (amine) precursor for this sequence serves as a tremendous simplification of the synthetic planning and for some targets the appropriate allylic precursor is actually known in optically pure form or is easily envisioned as the product of a short sequence.This review describes the use of neighboring group participation for the synthesis of the naturally occurring cis vicinal amino alcohols shown in Fig. 1. For most examples this involves the intramolecular delivery of an O-tethered nitrogen nucleophile. Within this narrowly defined operational format however there has been considerable opportunity for the author’s research group to develop new cyclization chemistry and to take advantage of some of the special features of the cyclized but not yet deprotected intermediates. As a result this approach has proven to be surprisingly versatile and fully complementary or perhaps even superior to many other methods for introduction of nitrogen into organic substrates.One might ask When is an intramolecular method preferred to intermolecular N-substitution? The answer seems to be whenever the latter fails or importantly whenever an efficient route can be devised. The literature concerned with amino alcohol synthesis by Ncyclization methods has been reviewed previously,3,4 and there are numerous contributions over the last twenty years whose description can fill chapters. At the risk of slighting some excellent work by imaginative researchers the author would venture a personal short list of some of the important early contributors to ideas in this field. Overman5 demonstrated in 1974 that trichloroacetimidates could be used as a source of Otethered nitrogen in an aza-Claisen process. Iodocyclization of these derivatives was later used by Fraser-Reid6,7 and Cardillo8,9 to synthesize ristosamine and daunosamine.Roush,10 Kishi,11 and Vasella12 used N-benzylcarbamate cyclizations to synthesize sphingosines ceramide and 2-amino-2-deoxy-dpentitols from epoxy alcohols themselves the products of Sharpless asymmetric epoxidation. Other investigators have developed analogous tethered-N-delivery processes but not many of them have been applied in natural products syntheses. Of course there are many excellent synthetic methods that still await their most elegant applications but as a general rule the use of a particular synthetic method in a complex synthesis further validates it in the eyes of the organic synthesis community. One need only think of reactions such as the Wittig reaction and hydroboration to appreciate how synthetic applications reflect the power and usefulness of certain transformations.2 Carbonimidothioate cyclizations model studies13,14 Amide carbonyls and in fact many other types of carbonyls have long been recognized as excellent O-participating groups especially through five-membered ring transition states.15 This is due to the presence of a considerable amount of electron density on the carbonyl oxygen as represented by the resonance structure 27 in Fig. 2. The amide nitrogen is only weakly Lewis O O N X N X R R 26 27 amides nucleophilic on oxygen O •• N X R 28 imino ethers nucleophilic on nitrogen Fig. 2 Nucleophilicities of amides vs.imino ethers. basic. In contrast the nitrogen atom of an imino ether as represented by 28 is fully capable of participating as a Lewis base through its lone pair of electrons. Furthermore N-alkylated salts of imino ethers are easily converted to N-alkylated amides by hydrolysis. Therefore the device of using an imino ether as an N-participating group could allow the intramolecular nucleophilic introduction of amino in protected form to an electropositive site five or six atoms away. Model studies based on cyclohexenol 29 of the use of iodo (N and O)-cyclizations to introduce cis,vicinal functionality are shown in Schemes 4 and 5. Condensation (Scheme 4) of the sodium salt of 29 with benzyl isothiocyanate led to the ambident anion 30 (anionic character at both N and S) which reacted with iodomethane exclusively on sulfur (the softer atom) to afford the S-methyl carbonimidothioate derivative 31.This rather exotic but easily prepared imino ether derivative cyclized under treatment with iodine to give the iodo iminium salt 32 which in turn hydrolyzed upon quenching to give the iodo oxazolidinone 33. One can see buried within this structure the protected cis vicinal amino alcohol functional grouping (boxed) here flanked by an iodo substituent that offers opportunities for further transformation. For example the iodo was displaced under silver-assisted solvolysis conditions to give the trans-hydroxy oxazolidinone 34 and this could be oxidized to the corresponding ketone 35 and then reduced from the less hindered side to give the trans-hydroxy oxazolidine 36.The iodo could also be eliminated as H–I under E2 conditions by treatment with the amidine base DBU (1,8-diazabicyclo[5.4.0]undec-7-ene) affording the unsaturated oxazolidinone 37. Note that the transformations subsequent to iodocyclization benefit from the presence of the cis-fused oxazolidinone which not only ties up the otherwise reactive hydroxy and amino groups but also lends some face- and site-bias to the nearby cyclohexyl carbons. Scheme 5 shows a complementary route to fused oxazolidinones based on an O-cyclization. Reaction of cyclohexenol (29) with carbimidoyl chloride 38 led to the imino ether derivative 39 which on heating rearranged smoothly to carbamate 40 by a [3.3] sigmatropic pathway.Iodocyclization of 40 occurred with O-participation to give an iodo iminium intermediate 41 (compare 32) and hydrolysis upon quenching led to the trans-iodo oxazolidinone 42. As before the iodo could be replaced by hydroxy (43 or 45) or keto (44) or eliminated to the alkene (46). These model studies published during 1982–1984,13,14 established the synthetic links between a cyclic allylic alcohol precursor and a variety of derived amino alcohol amino diol and other functional groupings that might appear as part of interesting synthetic targets. The model transformations turned out to hold up well to the much more rigorous test of multistep synthesis with polyfunctional substrates. Chem. Soc. Rev. 1999 28 61–72 63 OH 1.NaH THF 2. PhCH2NCS 29 I (isothiocyanate condensation) SCH3 O N I 32 29 1. I2 OH CH2Ph oxidize to ketone 91% Cl I O N CH2Ph 43 3 Synthesis of the aminocyclitols 2-deoxyfortamine fortamine and sporamine16,17,18 2-Deoxyfortamine (51 Scheme 6) fortamine (53 Scheme 7) and sporamine (1 Scheme 8) are aminocyclitol components of the antibiotics istamycin A fortamicin A and sporaricin A respectively. They are structurally analogous but differ in the nature of the substituents at C-1 and C-2. Thus divergent synthetic routes could perhaps be developed from a common intermediate already bearing appropriate functionality at C-3 to C-6. As events transpired that common intermediate is the Chem.Soc. Rev. 1999 28 61–72 64 H3C S Na CH2Ph CH2Ph N O N O 3. CH 4. I2 3-I (S-alkylation) I 31 30 O O O O S 3 N aq. Na2SO3 (–HI –CH3SH) AgOCOCF CH3NO2 N CH2Ph 90% overall 75% OH I 34 33 O O O O reduce from less hindered face N N CH2Ph CH2Ph 96% OH O 36 35 O O base (–HI) N CH2Ph 33 100% 37 Scheme 4 Carbonimidothioate iodocyclization model studies. OCH3 I CH2Ph N O OCH3 1. KH 2. heat CH2Ph N 96% overall (–H2 –KCl) 3 39 38 40 I Ag I O 2. aq. Na2SO3 (–HI –CH3OH) O OCH •• OH2 3 96% overall O 55% N N CH2Ph CH2Ph 41 42 O reduce from less hindered face oxidize to ketone O O O O 92% 88% N (iodocyclization) CH2Ph O OCH N CH2Ph O N CH2Ph CH2Ph AgOCOCF3 H2O CH3NO2 OH 45 44 O base (–HI) O 42 100% N CH2Ph 46 Scheme 5 [3,3]-Sigmatropic rearrangement and iodocyclization model studies.epoxy oxazolidinone 50 and an efficient synthesis of this compound in optically active form was devised (Scheme 6). The three aminocyclitols bear vicinal amino alcohol functional groupings both trans and cis in several places so the synthetic planning essentially became an exercise in matching the epoxide-based methods and the new iodocyclization methods to this functionality. 3,4-Epoxycyclohexene (47 Scheme 6) preferentially opens with amine nucleophiles at the allylic C–O bond and this allowed the straightforward synthesis16,17 of resolved allylic carbamate 48.Iodocyclization and quenching as previously described (Scheme 5) gave the iodo oxazolidinone 49 which intermediates 44% overall O 47 (racemic) 50 O O N H3C 52 aq. quench 75% overall four steps including separation of diasteriomeric 1. base (–HI) 2. epoxidation O O 89% overall N H3C 50 1. PhSeNa (nucleophilic attack by PhSe–) 2. oxidative elimination of PhSeH 94% overall OH OCH3 O O N H3C 57 carbonimidothiate formation with t-BuNCS and CH3I O OCH3 OCH Scheme 6 Synthesis of 2-deoxyfortamine. O N Br was dehydroiodinated as in the model studies and then epoxidized from the less-hindered (convex) face. Site-selective reaction of the epoxide 50 with azide anion reduction of the azido to amino and then deprotection led to the simplest aminocyclitol target 2-deoxyfortamine 51.In this route and in those that follow the cis vicinal amino alcohol substructure was formed by a cyclization reaction whereas the trans vicinal amino alcohol substructures were made by SN2 epoxide opening with nitrogen nucleophiles. Scheme 7 shows the transformation of epoxy oxazolidinone 50 to fortamine (53).16 Another site-selective nucleophilic opening of the epoxide ring this time with the sodium salt of selenophenol led to a hydroxy selenoether which in turn was converted to allylic alcohol 52 by the dependable oxidative synelimination method (with loss of PhSeOH). Epoxidation syn to the allylic hydroxy another well-precedented reaction was 90% overall Scheme 8 Synthesis of sporamine.followed by site-selective azide attack on the epoxide and conversion as before to the aminocyclitol product 53. Once again the epoxide-opening reaction proved its merit for trans vicinal amino alcohol preparation. Not only were these reactions highly site-selective but the site could also be predicted (and the synthesis planned or at least rationalized) based on the simple principle that the ring openings tend to occur in trans-diaxial fashion from the predominant half-chair conformation. For epoxides 50 and 54 this is displayed as Fig. 3. The sporamine synthesis (Scheme 8)18 used allylic alcohol 52 as the starting point for a carbonimidothioate cyclization.tert- Butyl isothiocyanate was employed to allow for convenient removal of the N-alkyl substituent. For this case a bromocyclization was carried out [bis(collidine)bromonium perchlorate is an effective source of Br+] and aqueous quenching gave rise to H3CO 1. NaN3 (nucleophilic attack by N3 –) 2. H2 Pd-C (reduction of azido to amino) 3. 4 M HCl heat (hydrolysis of oxazolidinone) 4. aq. NaOH (to neutralize) 96% overall O 3 OH O O N H3C OCH3 52 Scheme 7 Synthesis of fortamine. H3C O O N H3C 55 t-Bu 1. trifluoroacetic acid (removal of N- t-Bu) 2. n-Bu3SnH (reduction of bromide) 3. aq. NaOH (hydrolysis of oxazolidinones) Chem. Soc. Rev. 1999 28 61–72 I O iodocyclization and quench O N O 89% overall N CH3 OCH3 H3C OCH3 48 (resolved) HO NH2 2 HN CH3 OH 1.syn-epoxidation 2. NaN3 HO 3. H2 Pd-C HN 4. 4 M HCl 100 °C 5. aq. NaOH OCH3 CH3 53 fortamine S ClO4 – SCH3 t-Bu N O 56% overall bis(collidine) bromonium perchlorate N t-Bu O (bromocyclization) O N Br H3C OCH3 3 56 HO NH2 HN CH3 49 OH OCH3 51 2-deoxyfortamine NH2 OH O OCH OH OCH3 1 (+/–)-sporamine 65 N3 – attack favored here N3 – attack favored here O O H3C O H3C N O N H3CO H3CO O O OH 50 54 Fig. 3 Pseudo-axial attack by azide to afford a trans-diaxial product. bromo oxazolidinone (57). Three operations finished the route SN1 acidolysis of the N-tert-butyl radical-based reductive debromination and basic hydrolysis of the two cis-fused oxazolidinones.3. aqueous quench O H3C O OCH3 OH N O (epoxidation and methanolysis) H CH3 61 Scheme 9 Synthesis of methyl ravidosaminide. 4 Synthesis of the amino sugar methyl ravidosaminide19 Ravidosamine 3,6-dideoxy-3-(N,N-dimethylamino)altropyranose (absolute configuration unknown) is the aminosugar component of the aromatic C-glycoside antibiotic ravidomycin (it also lacks the O-4 acetyl of the antibiotic). The cis,vicinal amino alcohol functional grouping at C-3,4 (shown as the methyl glycoside 2 Scheme 9) suggested that a carbonimidothioate iodocyclization could be employed for its synthesis in a manner analogous to that used previously on carbocyclic systems.Furthermore one of the N-methyls could be introduced as methyl isothiocyanate and the other could come from reduction of the oxazolidinone carbonyl. In this way ravidosamine or a simple derived glycoside should serve as an interesting aminosugar target to further test the generality of the tethered nitrogen delivery method. There are of course differences between substituted cyclohexyl substrates and pyranosides that could obstruct the direct transfer of a successful reaction from one realm to the other. One important difference is that the pyranose skeleton is more electron-rich at the anomeric carbon but more electron-poor at other carbons compared with the cyclohexyl skeleton both characteristics attributable to the presence of the ring oxygen.Thus SN1 or solvolytic substitutions are generally easier to carry out at C-1 of the pyranose (resonance effect) but slower at C-2 (inductive effect). SN2 reactions can be difficult at secondary carbons for both classes of substrates because of the steric hindrance imposed by the various substituents. With additional electron-withdrawing substituents SN1 reactions are further inhibited for both classes. The intramolecular delivery of a nucleophile can in principle help compensate for the reduced reactivity but intermolecular displacements subsequent to the cyclization are still subject to the usual limitations. The appropriate saccharidal allylic alcohol substrate for carbonimidothioate cyclization is 58 (Scheme 9) available from H3C 1.carbonimidothioate formation 2. iodocyclization O HO OEt 58 m-chloroperoxybenzoic acid CH3OH 66 61% overall O Chem. Soc. Rev. 1999 28 61–72 commercial triacetyl-d-glucal in several steps.19 Anionic condensation with methyl isothiocyanate followed by S-methylation iodocyclization and quenching as before led indeed to the iodo oxazolidinone 59. The stereochemistry is presumably a result of kinetically favored pseudo-trans-diaxial addition of the electrophile and nucleophile. At this point replacement of iodo by hydroxy would complete the installation of functionality on the pyranose ring. The reaction designed to accomplish this functional group exchange had been worked out in model studies on a cyclohexyl substrate (Scheme 4).14 However reaction of iodo oxazolidinone 59 with silver trifluoroacetate in nitromethane gave no trans-hydroxy oxazolidinone; in fact it gave no reaction at all at least up to 90–100 °C by which temperature slow destruction of starting material set in.The model displacement had occurred at 0 °C perhaps with N-participation accounting for the clean stereochemical result. The reduced reactivity of the pyranoside substrate 59 relative to the cyclohexyl model 33 can be attributed to the two additional electron-withdrawing oxygens two atoms away from the carbon undergoing displacement. Thus an alternative to the direct or even N-assisted solvolytic displacement of iodide had to be found.Sometimes in synthesis one has to jettison hard-won functionality in order to open up new pathways to the target and this was the case for 59 (Scheme 9). Treatment of 59 with zinc powder gave the glycal 60 the result of a reductive vicinal elimination. In spite of the fact that stereochemical and functional complexity at C-2 was lost it could be recovered immediately by epoxidizing 60 in methanol to give the methyl glycoside 61 as a mixture of anomers. Alternatively alkene 60 was hydroxylated with osmium tetroxide and the resulting vicinal diol was converted to the same methyl glycoside under acidic methanolysis conditions. Both oxidative transformations occur from the less-hindered exo-face of glycal 60 as the cisfused oxazolidinone again serves as a stereochemical directing element.Lithium aluminum hydride reduction of the oxazolidinone ring of 61 led to the target aminosugar 2. O N CH 59 LiAlH (reductive cleavage of oxazolidinone) 5 N-Benzoylcarbamate cyclizations synthesis of aminosugars20,21 An often-used strategy for the synthesis of aminosugars from carbohydrate starting materials is simply to replace one of the hydroxys of the carbohydrate with an amino group usually by means of an SN2 substitution on a sulfonate ester or epoxide with a good nitrogen nucleophile such as azide anion. One can take advantage of the wide variety of enantiomerically pure and stereochemically rich carbohydrate starting materials that are commercially available or made by literature procedures in a few steps.The obvious drawback of this approach is that substitutions on secondary carbons of pyranosides and furanosides for reasons mentioned in the previous section can be H3C H3C O O O N O 90% CH3 I 3 60 H3C OH OEt 4 O Zn EtOH (reductive elimination) HO OCH3 87% overall NMe2 2 methyl ravidosaminide quite low-yielding. Iodocyclization of an unsaturated imino ether offers one alternative to intermolecular substitution as the nucleophilic nitrogen is well-positioned prior to alkene activation for participation at the intended carbon. However this requires an allylic alcohol precursor. A more general strategy would be to arrange for intramolecular delivery of a tethered nitrogen anion in an SN2 fashion to a sulfonate or epoxide carbon center.Both kinds of electrophiles can be made from poly-hydroxy starting materials by well-precedented transformations. Our candidate for the tethered nitrogen nucleophile for intramolecular ‘SN2’ is the anion of an N-benzoylcarbamate prepared simply by reaction of benzoyl isocyanate with an appropriately positioned hydroxy (Scheme 10).20,21 Because the reaction with benzoyl isocyanate occurs rapidly under mild and neutral conditions sensitive substrates such as epoxy alcohols and diol monotriflates can be used and many of these are available in enantiomerically pure form. As a model cinnamyl alcohol 62 was converted to its N-benzoylcarbamate derivative epoxidized and then treated with sodium hydride to generated the sodium salt of the N-benzoylcarbamate anion 64.Intramolecular cyclization occurred at the proximal epoxide carbon to give the N-benzoyloxazolidinone 65 which subsequently underwent N?O benzoyl migration to produce the final product 66. Although the benzylic carbon might have also undergone substitution in an intermolecular displacement the intramolecular reaction using the five-atom tether was highly site-selective. McCombie co-discovered the N-benzoylcarbamate cyclization,22 and several other examples of this cyclization to afford protected amino diols have been reported more recently.23–29 Pyranoside substrates can also be used for a direct sugar-toaminosugar transformation (Scheme 11).20,21 Thus methyl 4,6-O-benzylidene-a-d-glucopyranoside 67 a trans-vicinal diol was selectively converted to its 2-O-trifluoromethanesulfonate ester by reaction with triflic anhydride.Monotriflation of diols under these conditions tends to occur 1. PhCO-NCO condensation 2. m-CPBA epoxidation Ph HO 62 O O Na Ph HO OCH less reactive more reactive 86% overall COPh N H Ph H O 65 O O O HO 67 base-induced cyclization 3 85% N O benzoyl migration Scheme 10 N-Benzoylcarbamate cyclization. 66 1. (F3C-SO2)2O sulfonation 2. PhCO-NCO condensation 77% overall O COPh N Ph O O O O OCH 69 Scheme 11 Sugar to aminosugar transformation. preferentially at the hydroxy flanked by a cis,vicinal heteroatom here the methoxy oxygen presumably because of intramolecular H-bonding and an accompanying increase in the electron density at the hydroxy oxygen.The less-reactive hydroxy here O-3 is now available for condensation with benzoyl isocyanate under conditions that do not disturb the heat- and base-sensitive 2-O-triflate. The resulting N-benzoylcarbamate 68 was cyclized as its sodium salt to afford the Nbenzoyloxazolidinone 69 and deprotection by basic hydrolysis led to the aminosugar derivative 70. The overall change is hydroxy-to-amino at a defined site C-2. Two other examples of pyranoside diol mono-triflate-mono- N-benzoylcarbamates were successfully cyclized (71 and 72 Fig. 4) but two other examples failed (73 and 74). In the latter Ph Ph O O O OCH3 Tf-O O O NHCOPh 71 (cyclizes) O Ph O O Tf-O O OCH3 O NHCOPh 73 (no cyclization) Fig.4 Additional examples of N-benzoylcarbamate cyclizations. COPh O NH H Ph O O H 63 O NH O H Ph H O COPh Ph O O basic hydrolysis Ph 90% O HO O 70 3 NaH (base-induced cyclization) 88% O O O O O OCH SO2CF3 NH Ph 68 Chem. Soc. Rev. 1999 28 61–72 O O OCH3 O Tf-O O O NHCOPh 72 (cyclizes) Ph O O O OCH3 O O-Tf NHCOPh 74 (no cyclization) O Ph Na N H Ph O O H 64 O O 3 NH2 O OCH3 67 cases nearby axial substituents (shown in boxes) can be blamed for blocking the approach of the N-benzoyl nitrogen.Two solutions presented themselves the offending axial substituent can be removed as in 72 or it can be used itself to deliver the tethered nitrogen nucleophile. Scheme 12 shows the application of the latter idea to the hydroxy-to-amino transformation of 3,5-dibenzoylmannopyranoside 75. base-induced cyclization OH 79 Me2N HO CH 85 6 Synthesis of the antibiotic lincomycin30 Lincomycin (3) is an antibacterial used to treat infections in humans and animals. It consists of two components joined by an amide link an aminooctose thioglycoside termed methyl thiolincosaminide (86) and l-trans-n-propylhygric acid (87 Scheme 14). Coupling of 86 and 87 to regenerate 3 has been achieved so that the synthesis of 86 constitutes a formal synthesis of the antibiotic.From the standpoint of aminosugar synthesis the structure of 86 presents a particularly challenging less reactive HO HO 80 O H 68 more reactive PhCOO OH O OCH3 O O HO OCH3 NaH Me2N-CN (alkoxide addition to CN) H3C PhCH2O PhCH Chem. Soc. Rev. 1999 28 61–72 65% N H O 1. PhCO-NCO 2. triflation O 75 PhCO N PhCOO O PhCOO 65% OCH3 77 H3C H HO PhCO-NCO H O O 80 Scheme 13 Attempted N-benzoylcarbamate cyclization for lincomycin. Na HO PhCOO 3 O O NH PhCH2O OCH3 3 2 HO H O Scheme 12 N-Benzoylcarbamate cyclization to non-vicinal site. -a,D-talopyranoside H3C H O NaH H O O H O 2O O H O PhCH2O Me2N N O H H PhCH2O PhCH 82 H 2O OCH HO H 3C HO HO SCH 2O OCH3 86 methyl lincosaminide Scheme 14 Synthesis of lincomycin.H feature the location of the amino group C-6 is extremely hindered and intermolecular introduction of a nitrogen nucleophile at this position can be expected to be very difficult. There are other challenges presented by a synthesis of 86 to be sure but it seemed to be a good aminosugar test case for our methods of tethered nitrogen delivery. We selected a strategy that asked the axial C-4 hydroxy to deliver the tethered nitrogen nucleophile to C-6 over a sixmembered ring transition state. Commercially available methyl a-d-galactopyranoside (79 Scheme 13) was protected at C-2,3 and chain-extended at C-6 to produce the epoxy alcohol 80 after several steps.Condensation of 80 with benzoyl isocyanate was successful but treatment of the N-benzoylcarbamate 81 with sodium hydride as before (Schemes 10–12) caused reversion to 80. We infer that the reactivity of this anion is not sufficient to overcome the steric barrier to closure at C-6; one evidently needs a more reactive and less-hindered nitrogen nucleophile. After experimenting with several nitriles and other potential tethers,31 we found conditions for inducing the intramolecular epoxide opening (Scheme 14). Condensation of the sodium salt O PhCOO N COPh O CF3SO2 O H O PhCOO OCH3 AcO AcO 76 NHAc OAc O OCH3 68% hydrolysis and acetylation PhCONH reverts to 80 PhCH2O Me2N N H3C O H H O O PhCH2O PhCH2O OCH3 83 78 methyl 4-amino-4-deoxy- N O O O-tetraacetyl O OCH3 H CH3 N PhCH2O OCH3 Me2N N O O H H3C PhCH2O PhCH 84 HN HO HO H O CH 81 H3C known coupling to 87 3 H n-Pr HO H N HO 3 HO SCH3 87 n-Pr 3 lincomycin of epoxy alcohol 80 with N,N-dimethylcyanamide gave rise to an isourea anion (82) which evidently closed at C-6 to give a dihydrooxazine 83.This intermediate immediately rearranged however to the more stable C-6,7 oxazoline (85) by way of the tetrahedral intermediate 84. The overall result was not only effective delivery of the tethered nitrogen to C-6 but also creation of a well-protected form of the C-6,7 amino alcohol.This proved to be critical for subsequent transformations at C-1 to install the axial methylthio group (85 ? 86). It is also a demonstration of another advantage that intramolecular introduction of the nitrogen substituent can bestow. Intermolecular attack (by azide or phthalimide for example) leads simply to a protected amine; intramolecular introduction of amino leads to an intermediate with two protected groups that can be deprotected simultaneously or otherwise manipulated to advantage. There are further examples of these features in the syntheses in the sections below. toluene reflux [3,3] rearrangement NCS O BnO 7 Synthesis of the aminocyclitol valienamine and related pseudo-sugars32 For the remaining sections we return to the aminocyclitol arena.Valienamine (4) is a seven-carbon unsaturated amino tetrol that shows a-glucosidase inhibitory activity. It is also found as a component of several other pseudo-oligosaccharide glycosidase inhibitors wherein the amino is linked to C-4 of a glucopyranoside. Both of the carbon-bonds to this nitrogen are secondary congested and difficult to form by intermolecular reactions such as halide displacements reductive amination and epoxide BnO D-glucose BnO SBn CH2Ar O N BnO BnO OBn 90 89 BnO BnO 93 AcO N AcO OAc AcO AcO O 1. Luche reduction 2. Mitsunobu inversion 3.base hydrolysis 88 62% overall OMe Ac O AcO OMe 95 N-linked pseudo-disaccharide Scheme 15 Synthesis of valienamine analogues. 71% overall OBn BnO BnO condensation and rearrangement as before 52% overall aminolysis. Our previous work with carbonimidothioate chemistry6 indicated that even bulky N-substituents such as tert-butyl are tolerated and the conditions are mild enough that functionalized N-substituents might also be compatible. We therefore set out to synthesize 4 and some related compounds and also to use the tethered nitrogen delivery methods for linking the Nsubstituent to the aminocyclitol (Scheme 15).32 The conversion of d-glucose to the cyclohexenone 88 (Scheme 16) followed a literature route.Selective reduction of the carbonyl from the less-hindered face and then inversion of the resulting allylic alcohol by the Mitsunobu protocol led efficiently to the protected conduritol derivative 89. A word about the Mitsunobu inversion sequence is appropriate here.33 This procedure which involves treating an alcohol with diethyl azodicarboxylate triphenylphosphine and benzoic acid and then hydrolyzing the resulting (inverted) allylic benzoate is particular efficient in the cyclohexenol cases we and others have studied. In fact it is so dependable that we can incorporate into our synthetic planning the preparation of the wrong allylic alcohol knowing that the stereochemistry can be ‘fixed’ at a later point in the route. A couple of extra steps are spent but they are very high-yielding and lead to the desired product with greater stereoselectivity.Of course one frequently makes compromises of this type in organic synthesis — the use of protecting groups is a prime example — but these tactics simply add to our capabilities until we are able to find a more direct solution. Conversion of 89 to an S-benzylcarbonimidothioate (90) set the stage for [3,3]-sigmatropic rearrangement (reminiscent of Scheme 5) to the allylic amine derivative 91 which was deprotected to afford 7-nor-valienamine (92). When the OH BnO carbonimidothioate condensation BnO OBn Ar = p-(MeO)C6H4 Bn = CH2Ph 89 AcO O deprotection and acetylation NHAc AcO OAc SBn N OBn CH2Ar 91 92 7- nor-valienamine tetraacetate BnO O SBn N BnO OBn 80% O BnO BnO BnO OMe 94 OBn O N BnO BnO BnO OMe BnO did not rearrange (N is too hindered) OBn 96 62% overall deprotection and acetylation SBn O Chem.Soc. Rev. 1999 28 61–72 69 BnO HO iodocyclization as before BnO HO 65% overall BnO OH 98 97 (resolved) BnO BnO O BnO H BnO I syn elimination of HOI BnO BnO N Ar O O 100 101 Scheme 16 Synthesis of valienamine. carbonimidothioate was prepared with the glucose-derived isothiocyanate 93 as the condensation partner the [3,3] rearrangement led after deprotection to a pseudo-disaccharide of 7-nor-valienamine 95. This is a fairly complex and hindered N-substituent but there are limits.The corresponding carbonimidothiate 96 derived from a protected 4-isothiocyanatoglucose did not rearrange but instead underwent elimination to form a cyclohexadiene. For the synthesis of valienamine itself (Scheme 16),32 we prepared the cyclohexenediol 97 as the enantiomer shown and converted it by a multistep route capped with a Mitsunobu inversion procedure to the allylic alcohol 98. Iodocyclization of the carbonimidothioate derived from p-methoxybenzylisothiocyanate led to the iodo oxazolidinone 99. At this point syn-elimination of H–I to the alkene is required. Previous model studies (Scheme 4) include an E2 dehydroiodination promoted by DBU but this is almost certainly a trans-vicinal process and it is blocked in the case of 99.The syn elimination has only limited literature precedent — Reich had shown that oxidation of iodides to the corresponding iodoso derivatives (see 100) led to some syn elimination of HOI accompanied by other cationic processes such as substitution. 34 It is here that the role of the research director as opposed to the poor soul carrying out the actual experiments comes into prominence in a way that can be compared to the role of a priest pastor or other mediator of faith. The syn oxidative dehydroiodination had not to the research director’s knowledge been used in a synthetic route nor did it appear especially promising because of the cationic side reactions. However heavily oxygenated substrates such as carbohydrates and cyclitols and 99 do not readily give SN1 or solvolytic products for reasons already alluded to (Section 4).Hence the syn elimination pathway might be more prominent for these substrates. More worrisome however was the prospect that the product alkene would be more susceptible to oxidation than the starting material so that at temperatures appropriate for elimination over-oxidation at the alkene site might take preference to iodoso formation. But in this case the research director reasoned the product alkene would be flanked by three electron-withdrawing allylic heteroatoms which ought to significantly reduce the reactivity of the alkene toward epoxidation. As often happens in synthesis there is no appropriate model for this elimination reaction that is easier to make than the actual substrate 99 except perhaps the racemate.Thus faith (there is no better word for it) is required on the part of the experimentalist that an investment of many synthetic steps will be rewarded by a successful elimination reaction deep into the route that is without adequate precedent. The same experimentalist should not be reminded at this point that Chem. Soc. Rev. 1999 28 61–72 70 BnO I BnO m-CPBA (oxidation of iodo to iodoso) 71% BnO N Ar O O 99 Ar = p-(MeO)C6H4 AcO AcO deprotection and acetylation as before AcO 47% overall N NHAc Ar O OAc 102 (+)-valienamine tetraacetate O there might well be similar cases of misplaced faith that have never been brought to light.As it happened oxidation of iodo oxazolidinone 99 with m-chloroperoxybenzoic acid at 210 °C for two and a half days led to the formation of the protected valienamine derivative 101 in good yield and deprotection and acetylation gave the desired target as its peracetate 102. 8 Synthesis of the mannosidase II inhibitor mannostatin A35 By the early 1990’s tether-based aminocyclitol synthesis had been brought to a point where the mere sighting of a cis,vicinal amino alcohol functional grouping reflexively triggered the author’s urge to try another carbonimidothioate iodocyclization. Mannostatin (5) is a powerful and selective inhibitor of Golgi processing mannosidase II. Its cis vicinal amino alcohol substituents at C-4,5 are flanked by a C-1 trans-methylthio group (see 108 Scheme 17).The allylic alcohol precursor (103) was obvious and if fact known in optically pure form. What wasn’t known was an electropositive ‘methylthio’ electrophile that could initiate cyclization. Our experiments to develop such a reagent were unsuccessful but Fuchs later found that CH3S– OSO2CF3 functioned extremely well in this role.36 Nevertheless a carbonimidothioate iodocyclization was carried out to give the iodo oxazolidinone 105 in excellent overall yield and the oxazolidinone nitrogen was oxidatively dealkylated as before. Replacement of iodo by methylthio with retention of configuration at C-1 was achieved under basic conditions (NaSMe in dimethylformamide solution). We had hoped in another instance of faith that the oxazolidinone nitrogen of 106 would participate in this displacement so that the methylthiolate would then open an N-acylaziridine intermediate.14,37 Whether or not this was actually the course of the mechanism the trans-methylthio oxazolidinone 107 formed with apparently complete stereoselectivity and was obtained in satisfying yield.Basic hydrolysis of the oxazolidinone followed by acidic removal of the cyclohexylidene ketal gave mannostatin A as its hydrochloride 108. 9 Synthesis of the trehalase inhibitor trehazolin and a related glucosidase-inhibiting trehazoloid pseudo-disaccharide38 Was everything worth knowing known about carbonimidothioate iodocyclizations? Perhaps but the structure of the HO O O Ar = ( p-MeO)C6H4 O H N O I O O 106 O HO HO Ar BnO 90% overall I AcO (reductive elimination) 112 Ar carbonimidothioate condensation 103 (from D-ribonolactone) NaSCH3 DMF 90% N N SN2 epoxide hydrolysis BnO O displacement with retention O seven steps OH 109 D-ribonolactone O O OAc O O OH 115 trehazolin aminocyclitol (117 Scheme 18) was thoughtprovoking.Here the cis,vicinal amino alcohol functional grouping is improperly oriented for a Markonikov iodocyclization onto a trisubstituted alkene and anti-Markovnikov iodocyclizations were unknown or at least very rare. And yet the appropriate allylic alcohol precursor 110 was available in optically pure form by a short literature sequence.What did the anti-Markovnikov cyclization have going for it? Kinetic iodocyclizations of various types (iodolactonizations,39 iodolactamizations, 40 etc.3,4) usually favor fused mode over bridged mode and five-membered rings (5-exo) over six-membered rings (6-endo). Furthermore heavily oxygenated ring systems as we have seen are not ‘normal’. The two allylic heteroatom substituents distal to the cyclizing iminoether might be expected to modify the normal polarizability of the trisubstituted alkene such that development of partial positive change at that (distal) disubstituted alkene carbon is somewhat discouraged. Thus it was another instance of faith but not an instance of acting without guidance.116 Scheme 18 Synthesis of the trehazolin aminocycitol. CH2Ar O N MeS N iodocyclization and quench O O 85% overall O O O 105 104 O H N O deprotection SCH anti-Markovnikov iodocyclization 3 95% overall O O 107 Scheme 17 Synthesis of mannostatin. OH BnO carbonimidothioate iodo cyclization BnO 82% O O Me Me Ar = ( p-MeO)C6H4 110 exclusively anti-Markovnikov iodo cyclization O Ar N Zn THF Mitsunobu inversion O BnO 83% overall OAc Ar 114 113 O Ar N O deprotection and acetylation BnO HO 89% 60% overall HO CH2Ar HO Ar I Me 111 Chem. Soc. Rev. 1999 28 61–72 OH Iodocyclization of the carbonimidothioate derived from 110 gave the desired iodo oxazolidinone 111 resulting from anti- Markovnikov addition.Modification of the previous (Scheme 9) reductive vicinal elimination reaction entailed prior conversion of the acetonide to vicinal diacetate 112 which was then reduced in situ with zinc metal to afford the unsaturated oxazolidinone 113. Mitsunobu inverson and syn-epoxidation directed by the nearby hydroxy were followed by hydrolytic conversion to the trans,vicinal diol 116 and then deprotection and acetylation gave the trehazolin aminocyclitol as its hexaacetate 117. The coupling of an appropriately protected form of the trehazolin aminocyclitol (118) with a 4-isothiocyanatoglucose derivative 119 to produce the (1?4)-trehazoloid pseudodisaccharide glucosidase inhibitor 122 is shown in Scheme 19.The oxazolidinone ring serves triple duty here in that it forms as the result of the iodocyclization survives the installation and protection of the remaining hydroxys and then falls away obligingly following N-dealkylation and basic hydrolysis. ceric ammonium nitrate (removal of N-p-methoxybenzyl) I 92% O NH3 Cl SCH3 OH HO 108 mannostatin•HCl O N O Ac2O H2SO4 O O Me O syn epoxidation N 90% O OH NHAc OAc AcO AcO OAc AcO 117 trehazolin aminocyclitol 71 H O N 1. O-benzylation 2. oxidative N-dealkylation O BnO 116 86% overall BnO OBn BnO 118 HO BnO HO HO N deprotect BnO BnO HO NH O 71% overall BnO 121 Condensation of the liberated amino with isothiocyanate 119 led to a thiourea 120 which was cyclized by treatment with freshly prepared dry yellow mercuric oxide to produce the cisfused isourea 121.(one is tempted to devise a direct cyclization route to this isourea!). Deprotection gave the (1?4)-trehazoloid pseudo-disaccharide glucosidase inhibitor 122 which was of interest as an analogue of the naturally-occurring trehalase inhibitor trehazolin (6). Soon afterwards the natural product itself was also synthesized following a similar route employing the corresponding 1-isothiocyanatoglucose derivative.41 10 Summary and concluding remarks Tethered nitrogen delivery is an effective strategy for natural products synthesis. In addition to the expected entropic advantage of using an intramolecular displacement one gains excellent site- and stereoselectivity in the placement of the eventual amino group.Furthermore the cyclization initially produces a cis,vicinal amino alcohol functional grouping wrapped in a heterocyclic cloak that can direct and facilitate subsequent transformations. The dependability of these transformations late in the route benefits both the planning and the completion of the synthesis. Exploring tethered nitrogen delivery has given the author and his coworkers opportunities to enjoy functionality-centered (not just skeleton-centered) and method-based (not just target-based) natural products synthesis. Samuel Smiles the original self-help guru who advocated perseverance and courage as the route to success would certainly have approved.Scheme 19 Synthesis of trehazoloids. 122 (1 4) trehazoloid pseudodisaccharide 8 A. Bongini G. Cardillo M. Orena S. Sandri and C. Tomasini 7 H. W. Pauls and B. Fraser-Reid J. Chem. Soc. Chem. Commun. 1983 1031. Tetrahedron 1983 39 3801. 9 G. Cardillo M. Orena S. Sandri and C. Tomasini J. Org. Chem. 1984 49 3951. 10 W. R. Roush and M. A. Adam J. Org. Chem. 1985 50 3752. 11 N. Minami S. S. Ko and Y. Kishi J. Am. Chem. Soc. 1982 104 1109. 12 B. Bernet and A. Vasella Helv. Chim. Acta 1986 69 368. 13 S. Knapp and D. V. Patel Tetrahedron Lett. 1982 23 3539. 14 S. Knapp and D. V. Patel J. Org. Chem. 1984 49 5072. 15 S. Winstein L. Goodman and R. Boschan J. Am. Chem. Soc. 1950 72 16 S.Knapp M. J. Sebastian and H. Ramanathan J. Org. Chem. 1983 48 2311. 4786. 17 S. Knapp M. J. Sebastian H. Ramanathan P. Bharadwaj and J. A. Potenza Tetrahedron 1986 42 3405. 18 S. Knapp and D. V. Patel J. Am. Chem. Soc. 1983 105 6985. 19 S. Knapp G. S. Lal and D. Sahai J. Org. Chem. 1986 51 380. 20 S. Knapp P. J. Kukkola S. Sharma and S. Pietranico Tetrahedron Lett. 21 S. Knapp P. J. Kukkola S. Sharma T. G. Murali Dhar and A. B. J. 1987 28 5399. Naughton J. Org. Chem. 1990 55 5700. 22 S. W. McCombie and T. L. Nagabhushan Tetrahedron Lett. 1987 28 5395. 23 M. E. Jung and Y. H. Jung Tetrahedron Lett. 1989 30 6637. 24 R. D. Larsen P. Davis E. G. Corley P. J. Reider T. R. Lamanec and E. J. J. Grabowski J. Org. Chem. 1990 55 299. 25 M. Yamamoto M. Suzuki K. Kishikawa and S. Kohmoto Synthesis 1993 307. 26 M. E. Jung and Y. H. Jung Synlett 1995 563. 27 I. Cabanal-Duvillard J.-F. Berrien J. Royer and H.-P. Husson Tetrahedron Lett. 1998 39 5181. 28 T. H. Taleb R. A. Al-Qawasmeh C. Schreoder and W. Voelter Tetrahedron 1995 51 3141. 29 W. A. Nugent J. Am. Chem. Soc. 1998 120 7139. 30 S. Knapp and P. J. Kukkola J. Org. Chem. 1990 55 1632. 31 L. M. Engelhardt B. W. Skelton R. V. Stick D. M. G. Tilbrook and A. H. White Aust. J. Chem. 1990 43 1657. 32 S. Knapp A. B. J. Naughton and T. G. Murai Dhar Tetrahedron Lett. 1992 33 1025. 33 D. L. Hughes Org. React. 1992 42 335. 34 H. J. Reich and S. L. Peake J. Am. Chem. Soc. 1978 100 4888. 35 S. Knapp and T. G. Murali Dhar J. Org. Chem. 1991 56 4096. 36 C. Li and P. L. Fuchs Tetrahedron Lett. 1994 35 5121. 37 S. Knapp and A. T. Levorse J. Org. Chem. 1988 53 4006. 38 S. Knapp A. Purandare K. Rupitz and S. G. Withers J. Am. Chem. Soc. 1994 116 7461. 39 M. D. Dowle and D. I. Davies Chem. Soc. Rev. 1979 171. 40 S. Knapp Adv. Heterocycl. Nat. Prod. Synth. 1996 3 57. 41 S. Knapp and A. V. Purandare unpublished results. 11 References 1 A. Hassner M. E. Lorber and C. Heathcock J. Org. Chem. 1967 32 540. 2 L. Guigen H-T. Chang and K. B. Sharpless Angew. Chem. Int. Ed. Engl. 1996 35 451. 3 K. E. Harding and T. H. Tiner Comprehensive Organic Synthesis eds. B. M. Trost and I. Fleming Pergamon Press Oxford 1991 vol. 4 part 1.9 p. 363 and references therein. 4 G. Cardillo and M. Orena Tetrahedron 1990 46 3321 and references therein. 5 L. E. Overman Acc. Chem. Res. 1980 13 218. 6 H. W. Pauls and B. Fraser-Reid J. Org. Chem. 1983 48 1392. Chem. Soc. Rev. 1999 28 61–72 72 S NH HN 1. basic hydrolysis HgO BnO OH 2. coupling with 95% BnO S OBn C BnO OBn O 120 N BnO BnO OMe 119 90% overall OH N N O NH O O NH HO OH O HO OMe HO 6 HO trehazolin OH Review 7/06417I

ISSN:0306-0012    

DOI:10.1039/a706417i

出版商:RSC    

年代:1999

数据来源: RSC