REDUCTION BY METAL-AMINE SOLUTIONS APPLICATIONS IN SMTHESIS AND DETERMINATION OF STRUCTURE By A. J. BIRCH and HERCHEL Smmr (CHEMISTRY DEPARTMENT MANCHESTER UNIVERSITY) WORK since the last reviews of the subject 1 y has been concerned chiefly with the exploitation of metal-nmine solutions in synthesis and in the investigation of natural products. Theoretical developments have been mainly incidental and have added little to what was already 1mown.l We are here concerned with the practical aspects but an appreciation of the theoretical background is essential for use of the reagents to the best purpose since variations in technique are possible. So references to theoretical aspects are made below where necessary but overlapping with earlier reviews has been avoided as far as possible. 1.Reduction by metal-ammonia and metal-amine solutions (a) partial or coniplete saturation of a wide variety of unsaturated substlances including polycyclic aromatic compounds dienes and trienes which are conjugated or are rendered so under the alkaline conditions of the reaction and in some cases of simple olefins ; and ( b ) reductive fission of alkyl aryl or diary1 ethers and sulphides and the hydrogen01 ysis of various groups attached to nitrogen oxygen and sulphur. Which particular reagent is used depends on the nature of the substrate and on how far if is desired tha,t reduction shall proceed. Metal and an 66 Acid ’’ in Liquid Ammonia.-The reagents so far ex- amined consist of an alkali metal and an “acid ” such as methanol or ammonium chloride in liquid ammonia sometimes with co-solvents such as ether or tet,rahydrofuran.The reagents are powerful if the “ acid ” is an alcohol and are then capable of reducing a terminal double bond or an isolated benzene ring. In this they differ fi.0~11 ammonia reagent’s lacking the alcohol although solutions of lithium in ethylamine are also capable of reducing benzene rings and terminal double bonds (see below). The alcohol also has the effect of buffering the reaction mixture preventing accumulation of strongly basic NH,. This explains the comparative simplicity of the results since base-catalysed rearrangements of double bonds are usually avoided. Reduction of a benzene ring leads to the cx6-dihydro- derivative unlike the lithium-ethylamine reagent where the as-dihydro- derivative which is probably formed initially is rearranged to the conjugated c+?-dihydro-derivative which is then rapidly reduced further.As will be seen similar results to the latter are obtained with calcium hexammine and by reduction with sodium and ethanol in liquid ammonia followed by Two main types of reduction are observed Birch Quart. Rev. 1950 4 69. ?Watt Cherti. l i e u . 1950 46 317. B 17 18 QUARTERLY REVIEWS an excess of sodium in a r n m ~ n i a . ~ The xb-hydrogen atoms added by the metal-alcohol-ammonia reagent avoid carbon atoms carrying dimethyl- amino- alkoxy- or alkyl groups in that order and are attracted to positions carrying carboxyl groups. The latter effect outweighs the others ; carboxyl groups labilise 0- and p-methoxyl groups in terms of the following equilibria to hydrogenolysis. The requirement for an added source of protons is now interpreted ROH Ar + E + Ar-* + HOR + Ar*H + OR- -$ ArH- - ArH + OR- The anion-radical Ar*- formed initially must add a proton in order that reduction may be completed; it appears not to be sufficiently basic to abstract this proton from ammonia and requires a more acidic proton source.If the acidity of this source is high as with ammonium salts the predominant reaction is evolution of gaseous hydrogen unless the substance is very rapidly and readily reduced ; alcohols seem to have about the optimum pK for the reduction of benzene rings. The reduction with sodium and methanol or ethanol in liquid ammonia of o-xylene,B naphthalene,' tetralin,8 arid 1 4-dihydronaphthalene has been subjected to a rigorous re-examination. The results confirm the originally defined reducing properties of the reagent.Diphenyl has been reduced in both rings giving a product showing no selective ultraviolet absorption which is either compound (I) or more probably compound (II).9 (1 ) rn (u) An important modification in technique has been the use of lithium instead of sodium or potassium for the reduction of aromatic rings. The greater solubility of lithium in liquid ammonia enables larger proportions of co-solvents to be used without the formation of two-phase systems ; consequently difficulties arising from the low solubility of substrates in ammonia systems can be overcome. Higher yields of reduction products may be associa3ted with the high concentration of metal and also with the higher normal reduction potential of lithium in ammonia (-2.99 v) compared with that of potassium or sodium (-2.59 and -2.73 v respectively).1° An earlier device for increasing the solubility of phenyl ethers in ammonia systems viz.the formation of glyceryl or 2-hydroxyethyl ethers,llP l2 enables the cheaper sodium or potassium to be used. This technique has not yet been fully investigated but in reduction of 3-2'-hydroxyetIhoxyoestra- a Birch J . 1946 693. Ref. 1 p. 88. Birch J . Roy. Inst. Ghem. 1957 80 100. Huckel and Worffel Ghem,. Bey. 1955 88 338. Kuckel and Schlee ibicl. p. 346. Idem ibid. p. 2098. Huckel and Schwen ibid. 1936 89 150. lo Wilds and Nelson J. Arney. Chem,. SOC. 1953 75 5360. l 1 Birch and Mukherji J . 1949 2531. l2 Birch J. 1960 36'1. BlRCH AND SMITH REDUCTION BY METAL-AMINE SOLUTIONS 19 1 3 5-trien-17p-01 l3 the yield of 19-nortestosterone is of the same order as by the lithium method.14 I n cases where there are serious losses of alkoxyl groups by reductive fission the method may be superior in that it inhibits hydrogenolysis through alkoxide formation by the side-chain hydroxyl group.It is possible also that reduction is facilitated by cyclic donation of a proton to an anionic intermediate. Terminal double bonds may be reduced by the alcohol-containing reagent. This has been observed with various dialkylallylamines 15 and with hex- 1 -ene which has been converted into hexane in 41% yield by two atomic propor- tions of sodium and methanol in liquid ammonia.l6 I n the last case no reduction occurs in the absence of an alcohol or when ammonium bromide is used as proton donor. Of particular interest is the observation that 2-cyclopropylpent-1 -ene gives 2-cyclopropylpentane l6 and no ring-open products with the sodium-ammonia-methanol reagent whereas methyl cyclopropyl ketone affords a mixture of methyl propyl ketone and pentan-2-01 with sodium and ammonium sulphate in liquid amrn0nia.l' A further illustration of the control exerted by the acid strength of the proton donor over the reduction products is the ultimate formation of aldehydes rather than alcohols when ammonium acetate is substituted for ethanol as the proton source in the sodium-ammonia reduction of amides.18 Alcohol production is ascribed to the ethoxide-catalysed decomposition and further reduction of the intermediate 1 -amino-alcohol (aldehyde-ammonia).The buffering of the medium by use of the more acidic ammonium acetate as proton source avoids this decomposition and the amino-alcohol is converted into the aldehyde during the working up 0- H+ R.CO.NH -+ ReAH-NH -+ RCH(OH)*NH -+ RCHO Job;,- NH +- R*CHO -+ R*CH,*OH Aldehydes may also be ultimately obtained by the sodium-ammonia-alcohol reduction of amidines (even arylamidines) presumably via the 1 1- diamines.ls This result is due to the lower acidity of NH than of OH and also to absence of hydrogenolysis of the C-N bonds.I n contrast to the reductive fission of C-0 bonds which in general occurs readily in benzyl ethers and in aryl acetals 19 and ketals,20 C-N bonds conjugated with aromatic nuclei are not split because nitrogen is less electrophilic than oxygen. Pyridine compounds are reduced more readily than the hydrocarbons Birch and Bauer unpublished work.l4 Wilds and Nelson J . Amer. Cheni. SOC. 1953 75 5366. lGKing J. 1951 898. l 7 Volkenburgh Greenlee Derfer and Boord ibid. 1949 71 3595. lQ Birch Hextall and Sternhell ibid. 1954 7 256. 2o Pinder and Smith J . 1954 113. Greenfield Friedel and Orchin J . Arner. Chein. SOC. 1954 76 1258. Birch Cymerman-Craig and Slaytor Austral. J . Chem. 1955 8 512. 20 QUARTERLY REVIEWS of the same ring size because the heteroatom is better able than carbon to stabilise a negative charge. Pyridine and quinoline compounds for example readily give 1 4-dihydro-derivatives ; 22 di- tri- and tetra-meric dihydro- compounds are also produced. Reduction can be effected even in the absence of added proton sources but the products may then contain larger amounts of polymers.Thiophen with sodium and ethanol in liquid ammonia gives a complex mixture of 2 3- and 2 5-dihydrothiophen but-2-ene-l-thiol but-l- and -2-ene and hydrogen sulphide.22 Presumably reduction occurs to 2 3- and 2 5-dihydrothiophen which then undergo further reactions because of the known ready reductive fission of C-S bonds. Pyrrole and furan rings appear to be unaffected. Sodium Potassium or Lithium in Liquid Ammonia.-In a number of instances little difference would be expected from the " buffered " and protonated reagents mentioned above. For example (-)-a-phellandrene (mentha-1 5-diene) is reduced by sodium or by sodium and ethanol in liquid ammonia to the same mixture of (-)- (60%) and (+)-menth-l-ene (40y0),23 showing a large proportion of ccp-reduction.The reduction of other 1 3-dienes to the 1 4-dihydro-derivativesY e.g. 2-methyl- 2 3-dimethyl- and 1 1 3-trimethyl-butadiene to the but-2-enes in yields of 98-99 93-94 and 72% respecti~ely,~~ should not be altered by the presence of alcohols. In other cases if addition of protons to the intermediate anions is avoided by ensuring the absence of proton sources other tha,n ammonia the occurrence of further reduction is inhibited by the negative charges present. Proton-addition occurs during the working-up. Sodium-ammonia reduction of dipheiiyl has been shown to be analogous to that ofnaph- thalene,25 two atoms of sodium being added to give a deep red sodium salt decomposed by ammonium chloride t,o 1 4-dihydr0diphenyI.~ The phenyl group here exerts an effect similar to that of carb0xyl,~7 as is to be expected from its ability to stabilise an adjacent anionic charge.Similarly fluorene yields an unstable dihydrofluorene of undetermined constitution which readily disproportionates to fluorene and 1 4 11 12-tetrahydro- fluorene.28 In related work it was shown that cyclopentadiene and indene give cyclopentene and indane respectively .s The reduction of a/?-unsaturated ketones to the saturated ketones of which examples are given below also illustrates the protective effect of a negative charge in an intermediate in permitting eventual isolation of the saturated ketone. Similar reactions have been carried out with unsaturated esters and acids. The protection of ally1 alcohols against hydrogenolysis 21 lief. 2 1). 362 ; Birch iinpublislietl work. z 2 S. P. Birch and McAllnn Natiwe 1950 165 S99.23 Birch unpublished work. 2 4 Levinn Svarclienko Kostin Treschova and Okinievicli Sborkin obshchei Khinc. 25Ref. 1 p. 81. 2G Benkesei. Arnold Lainbert and Thomas J . Amer.. Chew&. Soc. 195.5 77 6042. ?' Ref. 1 p. 86 ; Birch Hextall and Stcrnhell Austral. .7. Chetn. 1954 7 2%;. 28 Huckel and Schwen Rer. 1956 89 481. Akad. Nauk X.S.S.R. 1'353 1 356; Chem. Abs. 195.5 49 829. BIRCH AND SMlTH REDUCTION BY METAL-AMINE SOLUTiONY 21 and of acetylenes against reduction ca,n be achieved by initial formation of the salts. Lithium in Alky1amines.-Solutions of lithium in amines of low molecular weight such as methylamine ethylamine and the propylamines constitute reducing agents of very great power if little selectivity. The amines are in general more powerful solvents for organic substances than ammonia and have higher boiling points (C2H,*NH, b.p.16.5" ; NH, b.p. -33"). Accordingly their use ma,y avoid a common and serious difficulty often encountered in liquid ammonia reductions namely the low solubility of the substrate in the solvent system. The higher working tempgrature also undoubtedly facilitates the initial steps in the reduction and favours the conjugation and therefore further reduction of the primary products. The annexed examples illustrate typical reductions by these reagents Simple O E t - O E t + OEt QCHiCHiOH -t 0 CH~CH o H~ (44 7o i (24%) 29 26 CH,€H2€ SC *CCH21;CH 3 0 -78"_ CH,CH,CHLCHfCH,kCH3 CH,CH,*CH~CH*[CH,]jCH -k CHi[CH&CH3 oiefins may be saturated tetrasubstituted double bonds being least readily reduced in accord with the view that the process involves initial electron addition.1 A similar reduction of double bonds which must however be terminally situated occurs with sodium and methanol in 1 iquid ammonia (see above).Inhibition of the reduction of di- and tri-substituted double bonds by working at low temperatures (e.g. -78") has been observed.26 3O Isolated benzene rings are reduced to the tetrahydro-state or partly to the hexahydro-state depending on the conditions. Acetophenone gives the allylic alcohol 1-1 '-hydroxyethylcyclohexene whereas its diethyl ketal gives 1 -ethylcycZohexene in agreement with the view that hydrogenolysis of the 2y Benkeser Robinson Sauve and Thomas J . Amer. C!lreni. Soc. 1955 77 323. so Benkeser Rchroll and Sauve ibid. p. 3378. 22 QUARTERLY REVIEWS hydroxyl group in the intermediate benzyl alcohol is inhibited by salt formation.Similarly reduction of phenol (more correctly lithium phen- oxide) is very largely stopped a t the cyclohexanone stage probably by pro- tection of the carbonyl group as the enol anion. I n contrast formation of the phenoxide anion or the acetophenone enol anion 20 is sufficient to inhibit reduction by metal-ammonia-alcohol reagents. Use of excess of lithium favours dealkylation of anisole since the product then consists of phenol and cyclohexanone; the theoretical amount of lithium leads to a mixture of 2 5-dihydroanisole (the initial product) and the conjugated 2 3-dihydro- isomer. Although allyl (and benzyl) alcohols may resist hydrogenolysis owing to salt-formation allyl ethers for which salt-formation is impossible are readily cleaved.The course of the reaction is simila,r to that with the alkali metal-ammonia reagent but alkylamine systems offer the practical advantage of greater solvent power and reactivity. The cleavage of cis-( +)- carvotanacetyl methyl ether with lithium and ethylamine yields ( &)-p menth-l-ene,S1 in agreement with the view that the reaction proceeds through a symmetrical intermediate probably the rnesomeric anion produced together with alkoxide ion by the addition of two electrons.32 It is known that compounds capable of producing very stable anions in a fission reaction are reduced readily,33 so it would be expected that allyl acetates and benzoates should be cleaved more easily t,han the free alcohols provided the first stage is not reduction of the ester-carbonyl group.This has been demonstrated in the steroid series where for example lithium in ethylamine converts 3P-acetoxycholest-4-ene (111) to cholest-4-ene and 4-/3-acetoxycholest-5-ene (IV) and 6-/3-acetoxycholest-4-ene (V) both give the same mixture of cholest-4- and -5-ene.31 There is no evidence of formation of a t)hermodynamically unstable isomer since 3-j3-acetoxycholest-l -ene gives cholest-2-ene and none of the less stable ~holest-l-ene.~l Work on the fission of steroid epoxides has confirmed that the direc- t)ion of reductive ring opening as with propylene oxide,3* is consistent with a potential-determining stage involving the addition of 2 electrons. The 31 Hallswortli Henbest and Wrigley J . 1957 1969. 33 Ref. 1 11. 71. 3 3 Dean and Berchet J .Amel.. Chem. SOC. 1930 52 2823. 3 4 Birch J . Proc. Roy. Soc. New South Wules 1949 83 245. BIRCH AND SMITH REDUCTION BY METAL-AMINE SOLUTIONS 23 plansr geometrical requirements for the transition state 35 in such cases ensure that axial alcohols are produced stereospecifically. Thus 5a 6a- epoxides (VI) are converted into 5a-alcohols and 2a 3a- 7a 8a- and 9a 1 la-epoxides similarly give 3a- 8a- and %-alcohols respectively.36 Lithium aluminium hydride may reduce steroid vic-epoxides similarly,37 but it has no effect on the sterically hindered 7a 8a- and 9cc lla-epoxides thereby illustrating the power and low steric hindrance associated with the metal-amine reagents. Calcium Hexammine.-The reducing capabilities of calcium hex- ammine have been but little investigated during the period since the last review.Reductions are usually carried out with a suspension of the reagent in an inert solvent (e.g. ether dioxan 1 Zdimethoxyethane tetrahydrofuran) so that solubility may usually be achieved by choosing a suitable solvent. Under sufficiently vigorous conditions the reagent saturates the double bond in simple olefins ( 2 5-dimethylhex-2-ene for example gives 2 5-dimethylhexane 38) and reduces isolated benzene rings to the dihydro- or tetrahydro-~tate.~~~ 40 Methoxyl groups may be cleaved from the ring by fission of intermediates e.g. methyl m-tolyl ether gives l-methyl- cycZohexene,3 and 2-methoxynaphthalene gives a mixture of hexahydro- naphthalenes having homoannular conjugated diene systems. 4O Conjuga- tion is probably due to the presence of calcium amide.The Protection of Functional Groups.-The protection of reducible groups in some cases can be accomplished by salt formation to give anions. This has already been illustrated for ally1 alcohols and carbonyl compounds. Ethynyl groups may be similarly protected; thus the sodium salt of undeca-1 7-diyne is converted into undec-7-en-l-yne by sodium in liquid ammonia.41 This method which is simple in operation is only useful when the alcohol enol or acetylene is considerably the strongest acid present. The presence of an acid of comparable strength (e.g. ethanol) permits reduction to occur.20 Conversion of a carbonyl into a non-reducible group can be achieved through formation of an acetal ketal or enol ether provided the alkoxyl groups are not in an allylic or a benzyl position and the enol- ether double bond is unconjugated.Accordingly there are difficulties when the carbonyl group is in the a- or @-position to an aromatic ring.lg *O The problem has been solved for benzaldehyde derivatives by converting them 35 See e.g. Barton and Cookson Quart. Rev. 1956 10 67. 36 Hallsworth and Jenbest J. 1957 4604. 37 E.g. Plattner Heusser and Kulkami Helv. Chirn. Acta 1949 32 265 ; Plattner Furst Koller and Kuhn ibid. 1954 37 258. 3* Kazanskii and Gostunskaya J . Gen. Chem. (U.S.S.R.) 1955 25 1659. 39 Kazanskii and Glushev ibid. 1938. 8 6 4 2 ; Bull. Acud. Sci. [J.R.S.S. 1938 4O Birch and Dunstan unpublished work. *lDobson and Raphael J. 1955 3558. 1062 1065 and earlier papers. 34 QUARTERLY JLEVIEWS into tetrahydroglyoxalines e.g. (VII) which resist hydrogenolysis whilst the aromatic nucleus is reduced.ls The aldehyde group can then be regenerated by acid treatment'.The presence of other reducible groups is undesirable e.g. tetrahydro-1 2 3-triphenylglyoxaline (VII ; R = Ph) undergoes fission l8 because the N-phenyl groups stabilise the negative charge on the intermediate nitrogen anion. Hence dialkyltetrahydro- glyoxalines are generally used. tJnfortunately the method is inapplicable to ketones which fail to react with NN'-dialkyethylenediamines. Stereochemical Aspects of Reduction.-A review 42 of the products formed by the reduction of various multiple bond systems by dissolving metal reagents which act through the forination of intermediate carbanions has indicated that where stereoisomeric products are possible the thermo- dynamically stable ones are usually formed.The rule holds in many cases inter alia the reduction of acetylenes to trans-ethylene~,~~? 43 ketones to secondary alcohols oximes to amines conjugated dienes and trienes to olefins and ctf?-unsaturated ketones esters and acids to the corresponding saturated derivatives although recent work has shown that it is not univer- sally applicable. The case of x/?-unsaturated ketones is of particular importance. The reduction has been interpreted 42 as following the reaction B I l l I l l I I ! " E - I1 +- -_ - C.:C-c=o -j -C-c- c-0 -.+ ('H C (2 0 (VIII) (IX) path delineated. The stability of the en01 anion (IX) in the absence of an excess of '' acid " of comparable strength permits the eventual isolation of the saturated ketone. I n a further discussion 44 of this type of reaction it has been pointed out that (VIIL) should add a proton at the very basic ,8-position by a process involving little activation energy in which case tlhe nature of the product is determined by the most stable conformation of the anion rather t'hm the least hindered approach of the proton donor.This accounts for the fact that when the P-position is capable of yielding stereo- isomers the most stable one is invariably formed. For the cc-position other considerations apply and the nature of' the final product may depend on whether Icetonisation of' the en01 anion is therniodynamically or kinetically controlled. Kinetic control a t this stage has been observed to give the thermodynamically unstable isomer in reduction of the ketone (X) which gives the cis-isomer 4 4 (XI) readily convertible into the trans-isomer.A closely related example is formation of tlhe cis-product (XIII) by lithium- ammonia reduction of the styrene (XIT).45 The mechanism of reduction of 4 2 Barton and Robinson .I. 1954 3045. 43 Ctiinpbell (!hem. Iter.. l!U2 31 57. 4 4 Birch Srnith siid Tlioimton J . 195'7 1330. 4 5 Johnson Arkernim I h s t l i t i i i ~ m c l Dewalt J . driio.. (,'/teiji. Soc. 1956 78 6303. BIRCH AND SMITH REDUCTION BY METAL-AMINE SOLUTIONS 25 such styrene compounds is not clear but if it iizvolves an 8 9-dianion (steroid numbering) the first proton should be added a t the more reactive %position. The formation of the less stable isomer in this case has been ascribed to the influence of the 5a-hydroxyl group which by alkoxide formation induces the negative charge a t C,, to adopt the /3-position.When the less stable isomer is formed there is apparently a requirement that an aromatic ring shall be attached to the ring system since reduction of Ag-octal- 1 -one gives only the trans-decalone. 46 A possible explanation for this is that the aromatic ring reduces the energy difference between the cis- and the trans-form of an anion such as (XIV) and thereby ensures an appreciable concentration of the former a t equilibrium. In a more complex case in the steroid series the aromatic ring may not be necessary provided that approach of the proton-donor required for tlhe formation of the more stable product is sufficiently hindered this may be so in the reduction of 3/3-acetoxyergosta-S 22-dien-7-one 46 (XV).It has recently been shown that the sodium-ammonia reduction of' dideuteroacetylene gives trans-dide~teroethylene,~' in line with the fact that this method of producing trans-olefins from acetlylenes is completely stereo- specifi~.~3 41 It is notewortlhy that equilibrium mixtures contain appreci- able amount's of ~is-ethylenes.~~ 2. Use of metal-amine reagents in synthesis 801u- tions of potassium or sodium in liquid ammonia have found widespread application in synthesis for the reductive removal of unsaturated groups used as protecting agents for amino- imino- hydroxyl and thiol groups. They are of particular importance in peptide synthesis having the obvious general advantage over hydrolytic or catalytic methods for compounds which are labile in mid or contain sulphur. Benzyl toluene-p-sulphonyl and benzyloxycarbonyl groups are all efficiently replaced by hydrogen and (a) Metal-Ammonia Solutions.-(i) Removal of protecting groups.46 Birch Smith and Wilson unpublished work. 37 Rabinowitz and Looney J . Amw. Cheni,. SOC. 1963 75 2662. 4B Kilpatrick Prosen Pitzer and Rossini J . Res. Nut. Bur. Stand. 1946 36 559. 26 QUARTERLY ltEVIEWS cystinyl-peptides are cleaved to cysteinyl derivatives. are summarised as follows These applications (i) RS*CH,Ph -+ K*SH $- PhMe (ii) R.NH.SO,*C,H,Me-p -+ R-NH 1- p-C,H,Me.SH (iii) R*NHCO,CH,Ph + R-NH -t CO I- PhMc (iv) RSaSR -+ %R*SH (v) P,O*CH:,Ph -+ ROH + YhMe The use of reactions (i)-(iv) in the synthesis of sulphur-containing peptides ranging from glutathione 49 to oxytocin 5O has been r e ~ i e w e d . ~ l Cleavage by sodium-ammonia has also been used to remove benzyl groups protecting the hydroxyl groups in serine- and tyrosine-containing peptides 52 and the cyclic amino-groups in histidinyl-peptides.53 Peptide groups are apparently unaffected probably because of salt forniation. After the observation that dibenzyl and diphenyl hydrogen phosphate are converted into inorganic phosphate by sodium in liquid ammonia,54 the reagent has proved of crucial importance for the removal of protecting groups including benzyl tosyl and benzyloxycarbonyl in the synthesis of' various phosphates and pyro- phosphates of biological interest."9 j5 Thus the last stage in the synthesis of the " Acetobacter stimulatory factor " (ASF) involved the sodiuni- aninionia fission of four benzyl groups ;)* (PhCH,O),P( O).O.CH,.CMe,*CII( OCH,Ph) *!CO*?JH.CH,CH,] ,-S*CH,Ph -+ (HO),P( O)-O*CH,*CMe,CH( OH).[CO*NH.CH,*CH,] ,.SH An interesting fission of an allryl aryl ether is the selective demethylntion of the readily available homoveratrylamine [2-(3 4-dimethoxypheny1)ethyl- amine] to 2- (3-hydroxy-4-metJhoxyphenyl)ethylamine which can serve as a starting material for the synthesis of naturally occurring isovanillyl derivatives.The fission is based on the earlier conversion 57 of 3 4-di- me thox yt oluene into 3 - hydroxy- 4- met hoxy toluene. The resis tame of the 4-methoxyl group to hydrogenolysis follows from the proposed mechanism for the fission since the negative charge which must develop on oxygen for demethylation to ensue is destabilised by the electron-releasing para-group. Selectivity is therefore different from that observed in acid-catalysed demethyla tion.A cleavage that has found use in partial synthesis of steroids is t'he 49 Loring and du Vigneaud J . Biol. GlLenb. 1935 111 385. 5O du Vigneaud I-tersler Swan Koberts and Katsoyannis J . A n i e r . Chewr . J ~ O C . 51 du Vignenud " Syinposiun~ on Puptide Chemistry " Cheni. SOC. Sper. I'ubl. No. 3 52 GrasBman Wunsch and Deufel and Grassman Wunsvh and Fries quoted by 53 du Vigneaud and Behrens J . Biol. Chem. 1936.117 2 i ; Katchalski and Patchor- 5 4 Baddiley and Thain J . 1953 1611. 56 Baddiley and Mathias J . 19.54 2801 ; Arris Bnddiley Ijuchtiiiitii and Thain. 5 6 Harnlin and Fischcr t J . A m e r . ('hem. Soc. 1953 75 5119. 5 7 Birch J . 1947 102. 1954. 76 3113. 19x5 p. 49. Grassman Fo~tschr. Chem. org.Natui-sloffe 1956 13 547. nik XIVth Int. Congr. Pure Appl. Choni. 1065. J . 19.56 4968. BIRCH AND SMITH REDUCTION BY METAL-AMINE SOLUTIONS 27 reductive removal of the 12-acetoxy-group in hecogenin to give 1 I-oxotigo- genin.5s The reagent of choice was calcium in ammonia ; the reaction A cC most probably proceeds as annexed The free ketol gave the diequatorial diol whereas the cc-ketol (XVI) and the vinylogous ketol 6P-hydroxytesto- sterone both suffer loss of hydroxyl to give (XVII) 59 and testosterone 60 respectively. - a HO j nu (XVI) - (XVII) wn (ii) Reduction of cca-unsaturated ketones. The reaction which quantita- tively a t least is the most important application of the metal-ammonia reagent is the stereospecific reduction of various steroid ap-unsaturated ketones capable of giving rise to one or two new centres of asymmetry.The conversion of a number of 8-en-ll-ones (XVIII) into the corresponding Sp 9cc-dihydro-derivatives in high yield has proved invaluable in routes for the partial synthesis of cortisone.61 In these cases as in the reduction of 3P-acetoxycholest-8( 14)-en-15-one 6 2 (XIX) proton addition to give the $I@- and the 14cc-compound may be kinetically as well as thermodynamically favoured because of the influence of the bulky p-methyl groups. 58Chapman Elks and Wyman Chern. and Ind. 1965 603; Chapman Xlks 69 Zurcher Heusser Jeger and Geistlich Helv. Chirn. Acta 1954 37 1562. 6o Amendolla Rosenkranz and Sondheimer J . Amer. Chem. SOC. 1954 76 1226. 61 E.g. Sondheimer Yashin Rosenkranz and Djerassi ibid. 1952 74 2696 ; Sondheimer Mancera Rosenkranz and Djerassi ibid.1953 75 1282 ; Schoenewaldt Turnbull Chamberlin Rheinhold Erikson Ruyle Chemerda and Tishlar ibid. 1962 '74 2696; Bladon Henbest Jones Lovell and Woods J. 1954 125. Philips and Wyman J. 1956 4344. 6 2 Barton and Laws J. 1954 62. 28 QUARTERLY REVIEWS (iii) Acyloin condensations. Solutions of sodium in liquid ammonia enable the acyloin condensation to be carried out in a homogeneous solution. The method offers obvious practical advantages over the conventional method using sodium dispersed in boiling toluene and conditions have been found which lead to high yields of ncyloin particularly in intramolecular condensations. Thus dimethyl marrianolate methyl ether (XX) gives the ketol (XXI),637 G 4 which is readily converted into estrone and the method becomes the one of choice for the comuletion of ring D in a steroid synthesis.I None of the ketol (XXI) was oht'ained by the older technique. Similarly OH ( X X I I > dimethyl I1 lhecocholanate (XXII) gave n high yield of the tetra- cyclic 11 12-acyloin.63 The reaction has also been employed to provide 1 6- 0x0 test 0s terone . (b) Metal-Ammonia and Alcohols.-These reagents were originally developed for use in steroid synthesis with the object of preparing cyclo- hexenones from anisole derivatives by the reaction path (i). Chiefly as a result of their tendency to give the thermodynamically most stable products where stereoisomerism is possible they were later used for reductions of types (ii) and (iii). Although reactions (ii) and (iii) do not require alcohols Cii) -C=CCOR - -CH-CH-COR (R = OH ,OAlk,or Alk) I I (iii).& - & better yields have been obtained in iiiaiiy cases in their presence (because there are then fewer side reactions). However fairly slow reactions of type (ii) in presence of large excesses of alcohol normally give the alcohol rather than the ketone. A process of type (i) was applied in 1949 to 3-2'-hydroxyethoxyoestra-l 3 5-trien-17@-01 which gave 19-nortesto- 63 Sheehan Codcrre Cohen arid O'Neill J . Amer. G'hem. Soc. 1952. 74 6155. 6 1 Sheehan Coderre and Cmikshank i6id. 1953 75 6231. liB Adanis Patel Petrow and Stuart-Webb J. 1956 297. BIRCH AND SMITH REDUCTION BY METAL-AMINE SOLUTIONS 6'3 sterone,12 a hormone which had 30% of the androgenic activity of testo- sterone and was the first active androgen to be made by total synthesis.Subsequently a wide range of 19-nor-hormones 66 was prepared by this process some more active than the analogues of the natural series. 19- Nortestosterone ,!I-phenylpropionate (" Durabolin ") 6 7 and 17a-ethyl-19- nortestosterone (" Nilevar ") 68 are recommended for use as anabolic agents. Similar applications of the process to 3 4-disubstituted anisoles have provided essential steps in the total synthesis of steroids terpenes and alkaloids. Thus the anisole derivative (XXIII) gave the ketone (XXIV) 69 which formed tlhe basis of rings B c and D in a total synthesis of ll-oxy- genated steroids ; the anisole (XXV) gave the ketone (XXVI),io used to complete a total synthesis of (j-)-totarol and the anisole (XXVII) gave the ketone (XXVIII) which was converted into yohimbone.M e 0 P O H / @OH doMe&' (XXIII) O (XXIV) H ( x x v ) (xxv I) (XXVII) (xxv I I I> OMe A process of type (ii) proved of primary importance in the Merck total synthesis of cortisone in ensuring the correct trans-configuration at) the C-D ring junction reduction of the acid (XXIX) with potassium and propm-2-01 in liquid ammonia gave the dihydro-derivative (XXX) stereo- ~pecifically.~~ It is of theoretical interest that with an llg-hydroxyl group (axial) a predominant amount of the unwanted product with a l4g-hydrogen atom was obtained. Two possible explanations for this result are that (a) G G E . y . Djerttssi Miraiiiontes and Hosrdmmz J. Amer. Chew. Soc. 1953 75 4440 ; Sandovul Miramontes Hosenkranz Djerassi arid Sondheimer ibicl. 1). 41 17 ; Djerassi Lippriian and Grousnian ibid.1956 78 3479 ; Ringold Rosenkrtinz and Sondheimer ibid. p. 3477. 13' Anon. A4?igeu$. Chem. 1957 69 69. 68 Anon. Che?n. Ettg. News 1956 34 813.1. 69 Stork Loewenthal and Mukharji J. Amer. Chcm. SOC. 1956 78 501. 7o Barltrop and Rogers Chem. and Ind. 1957 20. 7 9 Arth Poos Lukes Rohinson Johns Feurer and Sarett J. -4mer. Chew. *Voc. Swan I. 1950 1534. 193.1 76 171.5. 30 QUARTERLY REVIEWS repulsion between negative charges on oxygen causes the niethoxycarbonyl group to assume the trans-a- (axial)-configuration or ( b ) proton transfer through the solvent from the 1 l/%hydroxyl group results in proton addition to the P-side of the 14-carbani0n.~~ Unquestionably the most impressive application of metal-ammonia reagents in synthesis was afforded by the Wisconsin total synthesis of ~teroids.’~ Processes (i)-(iii) were carried out simultaneously on compound (XXXI) to give the ketones (XXXII) and (XXXIII) having respectively five and six new centres of asymmetry.Thirty-two racemates of ketone (XXXII) and sixty-four of ketone (XXXIII) were therefore possible but in fact the mixture of the two compounds was obtained in 25% yield. (x x x I I I) \ A A A l J (XXX I I> Extension of the synthesis to 11-oxygenated steroids exploited the benzyl alcohol fission in the conversion of the trio1 (XXXIV) into the diol (XXXV). The latter can be reduced in much higher yield than its deoxy- analogue. This difference has been attributed to inhibition of hydrogenolysis of the methoxyl group by formation of the 11-alk0xide.7~ It is however significant that the 11-hydroxy-group is in the axial conformation where it can facilitate proton transfer to the 13-position by a cyclic (probably six- membered ring) mechanism.The reduction of 5-methoxytetralin systems which was a feature of these syntheses had previously presented difficulties 7G due to the fact that 1 4-addition of hydrogen must involve at least one position occupied by a charge-destabilising group. Workable yields were only obtained by use of a technique originally developed for sodium 73Kenner Ann. Kepoyts 1954 51 177. 7 4 Johnson Rogier Szmuszkovicz Hadlei. Ackerman Bhattacharya Bloom Stizlmann Clement J<annister and Wynbei-g J . Amer. Chem. SOC. 1956 78 6289 ; Johnson Bannister Bloom Kamp Pappo Roger and Szmuszkovioz ibid. 1953 75 3275. 75 Johnson Pappo and Johns ibid.p. 6339. 7ti Bivc h MUI-I*~LJJ and Sirlitti J . 19.51 1945. BIRCH AND SMITH REDUCTION BY METAGAMINE SOLUTIONS 31 reductions in the Boots laboratories 7 7 employing a two-phase system made by the addition of 40% or more of ethanol to the liquid ammonia. The resistance of vicinal dialkyl derivatives of anisole to reduction has been exploited to afford a selective reduction of' the 3 4-disubstituted anisole ring in the ether (XXXVI). The product (XXXVII) was used to synthesise ( -J-)-l8 19-bisnor-~-homotestosterone. 78 3. Uses in determinations of structure These uses are based chiefly on fission reactions largely of ethers. The metal-ammonia reagent has proved to be a tool of exceptional power in elucidating the structure and stereochemistry of a1 kaloids containing diary1 ether linkages.These ethers are very resistant to hydrolysis but can be reductively split in high yields with solutions of potassium or sodium in liquid ammonia. The simplest example is that of cularin which gives a 5-hydroxybenzyl-3 4-dimethoxyisoquinoline. 79 The method has been widely employed to determine the structure of biscoclaurin bases which in general are cleaved into two benzylisoquir~oline fragments of known or casily determined structure. The opt'ical rotations of such products give also the relative orientations of the asymmetric centres. Alkaloids contain- ing free phenolic groups are more difficult to cleave because of phenoxide formation. In berbamine (XXXVIII) the free phenolic group protects the adjacent diphenyl ether linkage and the product 81 is a base of the dauricin type (XXXIX).This on methylation can be cleaved into two benzyl- isoquinoline fragments (XXXIXa). Further examples of the use of the reagent are in the cleavage of phaeanthine 00'-dimethylcurine and OO'-dimethylisoch~ndrodendrine.~~ The usual co-solvents for these reactions 7 7 Short unpublished work. 79 Manske J. Ameta. ChewA. SOC. 1950 72 65. 8" Tomita Fujita and Miirni .I. Phawn. Lsoc. J q x m 1951 226. a1 Idem ibid. p. 301. 8 2 Kicid and Walker J. 1934 GO!) ; C'hewi. aiitl fnd. 1953 243. 78 Birch and Smith J. 1956 4009. 3 2 QUARTERLY REVIEWS (largely determined by the solubilities of the alkaloids) are toluene and benzene ; in the presence of dioxan phaeanthine yields in addition to the ‘‘ normal ” benzylisoquinoline cleavage products a third benzylisoquinoline formed by an alternative mode of scission.This result is possibly due to the participation of the more polar solvent in the solvent-induced stabilisa- tion of intermediate anions. Cleavage of pilocereine methyl ether into four isoquinolines enables the skeletal structure to be unambiguously defined as in (XL).83 The directions of fission are all predictable on the basis of the rules already set out.1 Alkali metal-ammonia solutions can effect Emde-type fission of quater- nary ammonium salts 84 and may be the reagents of choice because they ininimise the accompanying Hofmann reaction but this property has not so far been appreciably used in alkaloid investigations. It has recently been shown that axial niethoxycarbonyl groups in steroid or terpenoid structures are converted into carboxylic acid groups by alkali metals in ammonia whereas the corresponding equatorial group undergoes Bouveault-Blanc-type reduction to the hydroxymethyl group.85 The same result is obtained in the presence of alcohols and the reaction which appears to be so specific as to be diagnostic is clearly a hydrogenolysis similar to the fission of aryl methyl ethers.Reductive fission of naturally occurring dlyl alcohols and allyl and henzyl ethers may often be of great assistance. Thus the structure of lanceol (XLI) was confirmed by its reduction,86 with sodium and alcohol in liquid ammonia to the known bisabolene (XLII). As a further illustra- tion the terpenoid side chain of the mould metabolite mycelinnamide was recognised 8 7 as being an allyl ether of the geranyl type hy cleavage to inethylgeraniolene and derivative of p-hydroxybenzoic acid.Information as t o the relative configurations of catechins and epicatechins can be procured by cleavage of the benzyl ether linkages in these compounds. (+)-Catechin itself and ( - )-epicatechin give ensntiomorphous alcohols showing that the 3-hydroxyl groups are in opposed configurations as in (XLIII) and (XLIV),88 i.e. that (+-)- or (-)-catechin and (+)- or (-)-epicatechin have the Same configuration of the 3- hydroxyl group. Sodium-ammonia solutions have heen used to cleave what me probably phenyl and benzyl ether links in various lignins and give notable yields of compounds of low molecular weight. 89 Evidence regarding the stereochemistry of the carbon skeleton of various tetrahydrofuranoid lignans with structures such as (XLV) has been obtained by sodium-ammonia cleavage of the benzyl ether links and 83 Djerassi Figdor Bobbit.t and Markley J .Amel.. Chem. SOC. 1956 78 3861. 81 Clayson J . 1949 2016 ; Haworth Lunts and MrKenna J . 1956 3749 ; see 8 5 Wenkert and Jackson J . Amer. Chem. SGC. 1958 80 2 1 i . 86Birch and Murray J. 1951 1888. 8 7 Birch Massy-Westropp and Rickards J . 1956 3717. 88 Birch Clark-Lewis and Robertson J. 1957 358. 89 Freudenbcrg Engler Flickinger Sobek and Klunk Ber. 1938 71 1810 ; Shory- gina Kefeli and Semechkina Dokludy Aknd. Nauk S.S.S.R. 1949 64 689 ; J . Gen. (%em. ((T.S.S.R.) 1949 19 1558 ; Shorygina and Kefeli ibid. 1950 20 1199 1213; Shorygina and Seiuechltina ibid. 1 %53 23 6 1’7. also ref. 34. BIRCH AND SMITH XEDUCTION B Y METAL-AMINE SOLUTIONS 33 examination of the reactions of' the resulting stereoisomeric alcohols (XLVI) including t'heir cyclisations to the aryltetrahydronaphthalene derivatives (XLVII) some of which are natural lignans.Probable configurations for galbulin galcatin and galbelgin from Himantundra barks can be worked OH M e q o H ow) M e 9 (XLI 1) (XLI I I) HO OH (x LI v) CH2 M r z O ~ ~ ~ ~ H M e Me0 \ CH,CHMe MeO/ MeO~c,J~-J;e 'CHMe MeO/ M e O T e (XLVII) OMe O O M e (XLVI) OMe O O M e (XLV) OMe Me0 \ CH CHiOH I I FH-FH .OH HOCH CH "'"0 '0""' OMe (XLV I I I> (XLIX) ' OMe out.so The structure of the lignaii gmelinol (XLVIII) was filially deter- mined 91 as the result of the ring-fission to the trio1 (XLIX) which on oxida- tion with lead tetra-acetate gave formaldehyde. It has recently been discovered that preliminary reduction with sodium and ethanol in liquid ammonia facilit,ates the hydrolysis of the sugar residue from pyrimidine nucleoside~.~~ It is probable that the reagent by destroying aromaticity converts the nucleoside into an acid-lahile enamine glycoside.It will be obvious from the above brief survey that despite the large volume of work made possible by the introduction of the metal-amine reagents much remains to be done especially in investigating the reactivities of various metal- amine-proton-source coiiibimtions and in explaining the observed reactions. It should also be clear that the variety of reagents available and their great power coupled with the specificity obtainable by an appropriate formulation are amongst the factors which will ensure their continued use in structural and synthetic problems. IVe thank Dr. H. B. Henhest for informing us of results before their tblication. 9'J Bimli Milligan (Mrs.) 14;. Siiiitli. and S~maZke unl'ublishecl work. !I1 Birch Hughes and Sinitli. d t t s t r t r l . J . Clhev). 1954 7 83. !' Uurke J. Org. Chojt. 1933 20 6.13. c