1972 2713Dibenzophospholium Salts and Ylides. Preparation, Properties, andWittig Olefin SynthesisBy Ian F. Wilson and John C. Tebby," Department of Chemistry, North Staffordshire Polytechnic, Stoke-on-Trent ST4 2DEThe spectral properties, hydrolyses, and Wittig olefin syntheses of P-benzoyl- and -alkoxycarbonyl-methylene-P-phenyldibenzophospholes and their salts are compared with those of the corresponding triphenylphosphoniumderivatives. The mechanistic implications of the differences in stereoselectivity and of the similarity of the kineticsof the Wittig reaction for the two series are discussed.OUR previous studies of the dibenzophosphole ringsystems have dealt with the expansion of the hetero-cyclic ring,l the preparation of pentaco-ordinate de-rivatives,2 and the lH n.m.~-.~ and 31P n .m ~ . ~ spectra.We now report the preparation and properties of thesalts and stabilised ylide derivatives of P-phenyldi-benzophosphole.The P-phenyldibenzophospholium salt precursers (1)were prepared from phenyldibenzophosphole and suit-able a-halogenocarbonyl compounds in the usualmanner. The benzoylmethylenephosphorane (2) wasE. M. Richards and J. C. Tebby, J . Chem. SOC. ( C ) , 1971,1064; Chem. Comm., 1967, 967.E. M. Richards and J. C . Tebby, J . Chem. Soc. ( C ) , 1970,1425.prepared in methanolic solution by treatment of com-pound (1 ; R = Ph, X = Br) with sodium methoxide.The ylide precipitated from solution. The methoxy-carbonylphosphorane (3) was converted into the oxideby this reagent and was therefore prepared by use ofsodium hydride or lithium diethylamide as base.Thediphosphorane (4) was prepared from P-phenyldibenzo-phosphole and dimethyl acetylenedi~arboxylate.~The i.r. spectra confirmed the ylide nature of theD. W. Allen, I. T. Millar, and J. C . Tebby, Tetrahedron* D. W. Allen and J. C . Tebby, J. Chem. SOC. ( B ) , 1970, 1627.M. A. Shaw, J. C. Tebby, R. S. Ward, and D. H. Williams,J..Chenz. SOC. ( C ) , 1967, 2442; M. A. Shaw and J. C. Tebby,abad., 1970, 6 .Letters, 1968, 7462714 J.C.S. Perkin Iproducts. They contained the usual low frequencycarbonyl bands, at 1500 for (2) and 1620 cni-l for (3),but the wavenumbers were slightly lower than those ofthe analogous triphenylphosphoranes. This suggestsPh3P =CH-C02Me(6)Ph 3 P = C HBZ( 5 ) R( 3 )that there is less p,-amp; back bonding to phosphorus inthe heterocyclic ylides.The IH n.m.r.spectra of the salts (1) and the ylidessupported their assigned structures. The PCH couplingconstants (see Table 1) of the heterocyclic ylides (2) andTABLE 1PCH coupling constantsTriphenylphosphon-Heterocycle JPUH ium derivative JPCH +++(7; R = Ph) (-) 14.5 Ph,PMe I- (-) 13.6(1; R = OMe, X = C1) (-)14 Ph,P*CH,CO,Me C1- (-)13.6(1; R = Ph, X = Br) (-)12 Ph,PCH,Bz Br- (-)12.5(+)26 Ph,P=CH.CO,Me (6) (+)22(+)28 Ph,P=CH*Bz (6) (+)26(3)(2)(3) were considerably larger than those of the corre-sponding triphenylphosphoranes, (5) and (6) respectively,whereas there was relatively little difference between thetwo series of phosphonium salts.It is already wellestablished 336 that the geminal PCH coupling constantsof phosphonium salts are little affected by the inclusionof the phosphorus atom in a small ring. Thus a changein hybridisation of the phosphorus bonding orbitalswould not be expected to be the cause of the increase inthe coupling constant for the stabilised ylides (2) and(3). The increase is attributed to the increase inenolate character indicated by the i.r. spectra. Thisconclusion is based on the assumptions that the couplingconstants are positive and that the s-character of thebonding orbitals of the a-carbon atom is increased byincreased enolate character. The latter assumption issupported by the larger coupling constant of the keto-phosphorane (5) than of the ester-phosphorane (6).Surprisingly, the ylides (3) and (4) showed only singletmethoxy-group resonances in their n.m.r.spectra. TheL. D. Quin and T. P. Burkett, J . Amer. Chem. SOL, 1970,92, 4303.resonance for (3) was unchanged in the temperaturerange -50 to +60". These results contrast with thoseof triphenyl derivatives and suggest that the heterocyclicylides are restricted to one conformation, probably thatwith the 2 (cisoid) geometry.Previousstudies* of 31P n.m.r. spectra of similar heterocyclicThe 31P n.m.r. data are given in Table 2.TABLE 231P N.m.r. chemical shifts of salts and ylides derived fromtriphenylphosphine or P-phenyldibenzophospholeSP (p.p.m.) A8P+ -27 } 8+ -21.5 } 3.5+ l 2-17 } 0Methiodide (7; R = Rile) aMethiodide (7; R = Ph)Salt ( 1 ; R = Ph, X = Br) -24Ylide (2)Ph,PMe, I- - 19Ph,PMe I- - 26Ph,PCH,Bz Br- - 22Ph3P=CH13z - 17a Ref.9.phosphonium salts possessing at least two aliphaticsubstituents showed that the participation of thephosphorus atom in a five-membered ring produced anappreciable downfield shift (Asp 8 p.p.m.) for the 31Presonance. The monoalkyl salts see (7; R = Ph)show this effect to a much lesser extent. The ylide (2)had a chemical shift identical with that of the triphenyl-phosphorane, but this may well be due to cancelling ofeffects. A small deshielding effect produced by the ringcould be cancelled by shielding produced by improved6 - 6 stabilisation, as might be expected when a smallring is present and as suggested by the predominance ofone conformation in the ester (3).The U.V.spectra were dominated by bands arisingfrom the dibenzophospholium chromophore. Howeverthe spectrum of the ylide (2) contained an additionalinflection at 320 nm which is attributed to the de-localised carbanion.The Wittig reactions (olefin synthesis) of the hetero-cyclic and ' acyclic ' ylides have been compared. Thestereochemical course of the reactions of the benzoylylides with benzaldehyde and acetaldehyde in benzeneand ethanol solutions is summarised in Table 3. Theheterocyclic ylide (2) gave more cis-olefin in all instances.TABLE 3cis-olefin (yo) obtained in Wittig reactions of benzoyl ylidesBenzaldehyde at 80"Ylide Ethanol Benzene Ethanol BenzeneAcetaldyhyde a t 25"21.4 27-6 12.9 22.114.0 11.8 10.1 11.6(2)(6)The kinetics of the reactions with benzaldehyde inbenzene have been studied. The rates for the benzoylylides at 80" were followed by i.r.spectroscopy and the7 I. F. Wilson and J. C. Tebby, J.C.S. Perkin I , 1972, 31;J. C: Tebby in ' Organophosphorus Chemistry,' Chem. SOC.Specialist Report, 1970, vol. 1, p. 2961972 2715rates for the ethoxycarbonyl derivatives at 25" werefollowed by a more accurate titrimetric method. Inboth cases the rates of the heterocyclic and non-hetero-cyclic ylides were similar. The rate constants obtainedby the titrimetric method were 2.8 amp; 0.2 x 10" dm-3mol s-l for the heterocyclic ylide (8) and 1.73 amp; 0.05 xfor ethoxycarbonylmethylenetriphenylphosphorane(9).8 The rates were strongly dependent on the methodof purification of benzene.This has been notedb e f ~ r e . ~ ~ ~The similarity of the rates is in accord with theaccepted mechanism of the Wittig reaction.1deg; Theformation of the betaines (10) (see Scheme 1) is the rate-determining step and is little affected by the participationof the phosphorus atom in a ring; the slight accelerationcan be attributed to increased enolate character in theheterocycles. However the small ring does have aneffect on the cis-trans ratio of the olefins formed in theWittig reaction (see Table 3). The increased proportionof cis-olefin from the heterocyclic ylide is attributed toreduced equilibration of the threo- and erytkro-isomers of(10) caused by an increase in their rates of decom-position via the (now stabilised) oxaphosphetan (1 1).O=CHPh "O-CHPh 1 L + -Ar2PhP =C HR Ar2Ph B-CHR I (lo'CHPh O-CHPhI I ArzPhPO + 11 .I)---CHR A r2PhP-CHR(11 1SCHEME 1The Wittig reactions of another heterocyclic ylide (12)have been studied previously.ll No kinetic data were* Recent work l6 indicates that modest increases in the electro-negativity of the groups attached to phosphorus greatly increasethe rate-determining role of steps (a) and (b).t I t is also possible that stable ylides are intermediates (i.e.precursers to the hydroxyphosphoranes) in the hydrolysis of thesalts described in this paper. However this should not alter therationalisation since we consider that the energy barrier to hy-droxyphosphorane formation should be similar for the salt andthe ylide. This cannot apply to reactive ylides which are notreadily formed from the corresponding salts in dilute aqueousalkali.Further, salts such as triphenylalkylphosphoniumhalides are relatively slowly hydrolysed in dilute alkali, whereasthe corresponding reactive ylides are readily hydrolysed bywater.18 This may be attributed to the ready conversion of thehigh-energy ' reactive ylides into the hydroxyphosphoranes,whereas the corresponding ' low-energy ' salts face a largeenergy barrier.presented but the compound appeared to be less reactivethan the corresponding triphenylphosphorane. Thiswas attributed to greater steric hindrance in (12).The studies of heterocyclic ylides have a bearing on analternative mechanism of the Wittig reaction in which apentaco-ordinate intermediate is formed directly fromthe ylide and aldehyde. This mechanism has been putforward in order to explain the stereospecific formation ofcis-olefin from reactive ylides in non-polar solvents.12However such a mechanism would be expected to bemore applicable to stabilised ylides which have reducednucleophilic character, as suggested by the originalproposers of this mechanism.13 The mechanism hasbeen widely discussed in the literature l4 and also findssupporting evidence in the reactions of betaine ylides.15However the absence of fast reactions for the hetero-cyclic ylides, the increase in the proportion of cis-olefin, and the kinetic evidence of other workers9 showconclusively that the mechanism does not apply tostabilised ylides.The hydrolyses of the heterocyclic phosphonium salts(1) were more ready than the hydrolyses of the corre-sponding triphenylphosphonium salts.In fact hydro-lysis was the major problem in the preparation of theheterocyclic ylides. The rate of hydrolysis of compound(1 ; R = Ph, X = Br) was too great to be measured at 0"(ti 1 min) , whereas the triphenylphosphonium salthad a half life of ca. 2 h; thus the rates of hydrolysisdiffer by a factor of at least 120.The hydrolyses of phosphonium salts are normallysecond order in hydroxide ion in accordance withScheme 2. Step (c), normally the slowest,* is muchH+lR3P0 4- RHSCHEME 2faster (and not ' rate-determining ') when the leavinggroup (R-) is a stabilised carbanion such as the 9-nitrobenzyl ani0n.l' We expect this step to be fasterstill when the leaving group (R-) is an enolate anion, asin the hydrolyses described in this paper.? The secondC.Ruchardt, P. Panse, and S. Eichler, Chem. Bey., 1967,100, 1144.9 A. J. Speziale and D. E. Bissing, J . Amer. Clzem. SOC., 1963,85, 3878.lo A. W. Johnson, ' Ylid Chemistry,' Academic Press, 1966,152.l1 M. B. Hocking, Canad. J . Chem., 1966, 44, 1581.l2 W. P. Schneider, Chem. Comm., 1969, 785.l3 L. D. Bergelson and N. I. Shemyakin, Tefrahedron, 1963,19, 149.l4 R. F. Hudson, Chem. in Byitain, 1971, 7 , 287; I.Ugi,D. Marquarding, H. Kwsacek, G. Gokel, and P. Gillespie, Angew.Chem. Internat. Edn., 1970, 9, 728.l5 J . Reucroft and P. G. Sammes, Quart. Rev., 1971, 25, 135.l6 D. W. Allen, B. G. Hutley, and M. J. T. Mellor, J.C.S.Perkin 11, 1972, 63.l7 G. Aksnes and L. J. Brudvik, Acta Chem. Scand., 1967, 21,745.l 8 R. F. Hudson, ' Structure and Mechanism in Organo-Phosphorus Chemistry,' Academic Press, New York, 1965, 2222716 J.C.S. Perkin Istep (b) is not likely to be affected by the inclusion ofthe phosphorus atom in a small ring and therefore therate enhancement observed for the heterocyclic saltsmay be attributed mainly to an increase in the equili-brium constant of the first step (a), due to stabilisationof the hydroxypho~phorane.~~EXPERIMENTALThe 1i.m.r.spectra were obtained with a Perkin-ElmerR10 instrument, the i.r. spectra with a Unicam SP 200instrument, and the U.V. spectra with a Beckman DB-Ggrating spectrophotometer. The 31P chemical shifts arerelative to 85 phosphoric acid. The salts and ylides werevery susceptible to hydrolysis by moist air and whollysatisfactory analyses were not obtained for some compounds.P- Methoxycarbonylmethyl-P-phenyldibenzophospholiumChloride (1; R = OMe, X = C1).-P-Phenyldibenzophos-phole 2o (2-60 g, 10 mmol) and methyl chloroacetate (1.3 g,12 mmol) in dry benzene (20 ml) were heated under refluxfor 4 h. The white crystalline solid was separated (2 g,55) ; m.p. 184" (from chloroform-ethyl acetate). Themother liquors, when poured into ether, gave an oil, which,when dissolved in chloroform and triturated with ethylacetate, gave a further 200 mg of the sa z (CDC1,) 1.1-2.4 (12H, m), 4.25 (2H, d, Jp= 14 Hz), and 6.35 (3H, s)(Found: C, 68.9; H, 4.9.C,,H,,ClOP requires C, 68.4;H, 4.9).P-Phenacyl- P-phenyldibenzophosfiholium Bromide (1 ;R = Ph, X = Br).-P-Phenyldibenzophosphole (2.60 g,10 mmol) and phenacyl bromide (2.2 g, 11 mmol) in drybenzene (20 ml) were heated under reflux for 4 h. Thecrystalline solid was separated (4.5 g, 98); m.p. 247-249" (from chloroform-ethyl acetate), T (CDC1,) 14-2.6(18H, m) and 4-57 (2H, d, J ~ E 12 Hz), Amax. 333 (c SOOO),275 * (18,000), and 242 t nm (53,000) (Found: C, 68.3;H, 4.7. C,6H,oBrOP requires C, 68-0; H, 4.4).P-Benzoylmethylene-P-phenyldibenzophosph (v) ole (2) .-Thephosphonium salt (1; R = Ph, X = Br) (2.34 g, 5 mmol)was partly dissolved in dry methanol (25 ml).Sodiummethoxide (5.5 mmol) was added under nitrogen. After10 min the pale yellow ylide began to precipitate. After4 h the solid was separated (1.2 g, 63) ; m.p. 20A206"(from chloroform-ethyl acetate), 7 (CDC1,) 1.7-2.8 (18H,m) and 5-6 (lH, d, Jpa 28 Hz), A,, 330 (E 9000), 320infl(9000), 275" (14,000), and 240t nm (54,000) (Found: C,C, 83.4; H, 5.7. C,,H,,OP requires C, 82.5; H, 5.0).Reactions of BenzoylmethylenetriphenylphosPhorane andthe Ylide (2) with Aldehydes.-The solution of the ylide(0.19 g, 0.5 mmol) in the appropriate solvent (2.5 ml) wastreated with the aldehyde (0-6 mmol).The acetaldehydereaction mixtures were left at room temperature overnightand then heated under reflux for 5 min. The benzaldehydereaction mixtures were heated under reflux for 2 h. Thesolutions were analysed by g.1.c. for cis-trans ratio and totalyield of olefin (Pye 104 chromatograph; PEGA 20 mcolumn at 175-225"). The trans-olefins were identifiedby comparison of their retention times with those ofauthentic samples. The yields of olefins were almostquantitative; the cis-trans ratios are presented in Table 3.P-Methoxycarbonylmethylene-P-p henyldibensophosph (v) ole(3).-(a) The phosphonium salt (1; R = OMe, X = C1)* Structured band (8 peaks) centred at this wavelength. t Structured band (3 peaks) centred at this wavelength.(368 mg, 1 mmol) was suspended in dry benzene (10 ml)under nitrogen and treated with lithium diethylamide(1.05 mmol) (from butyl-lithium and diethylamine).Themixture was stirred for 3 h, filtered, and evaporated, andthe resultant orange oil was triturated with ether. Theyellow solid (150 mg, 46) had an i.r. spectrum verysimilar to that of the product prepared by method (b).(b) The phosphonium salt (1; R = OMe, X = C1)(368 mg, 1 mmol) was suspended in dry benzene (10 ml) andtreated with sodium hydride (2.5 g, 10 mmol). Absoluteethanol (0.05 ml) was added and the mixture was stirredovernight. The solution was filtered and the benzeneremoved under vacuum to give a yellow solid (200 mg,65). The i.r. spectrum indicated that a small amount ofP-phenyldibenzophosphole P-oxide was present.The ylidewas so readily hydrolysed in the atmosphere that analysescorresponding to the molecular formula were not obtained ;z (CDC1,) 2.15-2.9 (13H, m), 6.45 (3H, s), and 7-12 (lH, d,JPH 26 Hz)-P-Ethoxycarbony lnzethyl- P-pheny ldi benzophospholiurnChEoride (1; R = OEt, X = C1).-The salt, prepared in asimilar manner to the methyl ester, had m.p. 186", vmX. 700,725, 770, 780, 810, 895, 1025, 1070, 1085, 1115, 1130, 1165,1200, 1320, 1385, 1405, 1450, 1475, 1595, and 1710 cm-l,z (CDCl,) 1.0-2-6 (13H, m, ArH), 4.47 (2H, d, JpH 14.5 Hz,PCH,), 5-94 (2H, q, J H H 7-1 Hz, CH,), and 8.97 (3H, t,CH,) (Found: C, 69.4; H, 4.9. C,,H,,,ClOP requires C,69.0; H, 5.2).P- Ethoxycarbonylmethylene-I?-phmyldibenzophosfiho le (8).-The phosphonium salt (1; R = OEt, X = Br) (1 mmol)was suspended in dry benzene (10 ml) and treated withsodium hydride (2-5 g, 10 mmol) and methanol (0.05 ml).The mixture was worked up as for the methyl ester to givea yellow oil, vco 1630 cm-l, z (CDC1,) 2.2-3.1 (13H, m,ArH), 6.20 (2H, d, J H H 7 Hz, CH,), 7-26br (lH, s, methine),and 8-82 (3H, t, Me).DirnethyE 2,3-Bis-5-phenyldibenzophosph (v) ol- 5-ylidene-succinate.-The diphosphorane (4), prepared from 5-phenyldibenzophosphole and dimethyl acetylenedicarboxy-late, had m.p.100-105deg;, vmax. (KBr) 700, 725, 730, 760,775, 845, 1065, 1085, 1165, 1270, 1360, 1440, 1570, 1585,and 1605 cm-l, T (CDC1,) 2-2-2.9 (26H, m) and 6.24 (6H, s)(Found: H, 5.1.Kinetics of Wittig Reactions.-(a) The reactions of thebenzoyl ylides (2) and (5) with freshly distilled benzaldehydein thiophen-free, sodium-dried benzene at 80" (reflux)under nitrogen were followed by i.r.spectroscopy. Therates of disappearance of the aldehyde carbonyl absorptionsat 1700 cm-l and the rates of appearance of olefin carbonylabsorption a t 1660 cm-l were plotted against time. Theslopes showed that the reaction of the heterocyclic ylide (2)was 2-3 times faster than that of (5).(b) The reactions of the etlioxycarbonyl ylides (8) and(9) with benzaldehyde in benzene a t 25.0 "C under nitrogenwere followed by a titrimetric method. Solutions ofbenzaldehyde ( 0 . 0 8 ~ ) and the ylides ( 0 . 0 8 ~ ) were separatelyallowed to equilibrate a t 25.0" for 2 h. The solutions weremixed under nitrogen. Samples (0.4-0.8 cm3) were takenat intervals, added to excess of hydrochloric acid (25 cm3;0-00016~), and titrated against sodium hydroxide(0.0008~) ., The end points were determined conducti-metrically to an accuracy of amp;Om1 cm3.C,,H,,O,P, requires H, 4.8).F. H. Westheimer, Accounts Chem. Res., 1968, 1, 70.20 H. Hoffman, Chem. Ber., 1962, 95, 2563; L. Horner, G.Mummenthey, H. Moser, and P. Beck, ibid., 1966, 99, 27821972 2717Kinetics of Hydrolysis of Methoxycarbonylnzethyltriphenyl- Samples (1 cm3) of the reaction mixture were added tophosphoniurn Chloride and the HeteroLyciic SaEt (1; R = excess of hydrochloric acid (25 cm3; 0.000625~) and con-OMe, X = Cl).-Solutions of the salts (0.02~) in 50 ductimetrically titrated with sodium hydroxide (0.0032~).aqueous ethanol (5 cm3) at 0' were mixed with sodium The solutions were standardised against borax.hydroxide in 50 aqueous ethanol (5 cm3; 0.02111) a t 0". 2/1189 Received, 25th May, 1972
展开▼
