摘要:

QUARTERLY REVIEWS CHEMISTRY OF SOME NEWER ANTIBIOTICS By N. G. BRINK PH.D. and R. E. HARMAN P1r.D. (MERCK SHARP AND DOHME RESEARCH LABORATORIES RAHWAY NEW JERSEY) PENICILLIN made its appearance less than two decades ago as the first of rz group of agents representative of a wholly new approach to the therapy of infectious diseases. I n the ensuing years the concepts of antibiosis and of antibiotics have become so familiar that it does not seem necessary to devote space here to historical background to definitions or even to a consideration of the place of antibiotics in modern medicine. This Review in keeping with its title will confine itself almost entirely to the purely chemical aspects of the subject. Its primary objective will be to acquaint the readcr with some of the types of molecules encountered among these biologically interesting substances and to give a little insight into the degradative methods and synthetic studies that established their structures.A comprehensive treatment of the subject of this Review could well fill an entire volume therefore many significant antibiotics have been omitted. Those chosen were selected primarily on the basis of chemical interest although their demonstrated or potential usefulness and the nature of their biological action were also considered. The wide diversity of structural types among antlibiotics has been a challenge and a delight to those chemists whose task it has been to carry out the structural studies and in some cases to synthesise the substances. Nature’s infinite variety could hardly be better exemplified than in the marvellous array of atomic groupings represented both by the formuh of NH Me&-?H C02H s\ ”\ $H CO CH / U Me OH ,! a 93 94 QUARTERTAY REVIEWS some of the less recent antibiotics-penicillin ( l ) streptomycin (2) chloro- tetracycline (3) chloramphenicol (4) gramicidin-S (Fj)-a.nd by those newer agents to be discussed below.p-N02C,H,CH - CH*CH,-OH I I OH NH-COCHCl (4) cyclo- (L-Valine-L-ornithine-L-leucine-D -phenylalanine-L-proline) ( 5 ) The macrolide antibiotics Within less than ten years a dozen or more members of a new and important class of antibiotics have been discovered. All are produced by various species of Streptomyces and all have in common a many-membered highly substituted lactone ring-hence the name macrolide-to which is attached a dimethylamino-substituted sugar.I n some of the macrolides the lactone ring is linked to one or two additional sugar residues. This group of substances is of interest because of its rapidly developing chemistry because of the biogenetic implications in the structures involved and because a t least some members of the group are of demonstrated therapeutic usefulness. Magnamych-The determination of the structure of magnamycin (S) chemically the most complex of those macrolide antibiotics that have been adequately characterised was begun by workers in thc Pfizer laboratories and completed by Woodward and his collab~rntors.~ I n the early work on CHMe $*2 H (6) A 4- B = Carimbose B = Mycaminose c = Mycarose this antibiotic the a,!?-unsaturated carbonyl system was detected by spectro- scopic studies as were the hydroxyl groups and the carboiiyl group of the Hcchstein Alurai Messina and Regna J .Amer. Chem. Soc. 1953 75 4684. Woodward Angew. Chem. 1057 69 50. 1 Wagner, RRINK AND HARMAN CHEMISTRY OF SOME NEWER ANTIBIOTICS 95 lactone ring. Alkaline hydrolysis liberated acetic acid isovaleric acid and dimethylamine. Mild acid hydrolysis of magnamycin cleaved the glycosidic linkage between the two sugar moieties and released the isovaleryl derivative of mycarose (6 c ) the nitrogen-free carbohydrate fragment. Mycaminose (6 B) the dimethylamino-sugar was isolated after more drastic hydrolysis with acid. Methanolysis of magnamycin removed mycarose and yielded the crystalline base carimbose (6 A +B) C3,,H4,OI2N. The instability of this and related compounds made it impossible to eliminate mycaminose and retain the central lactone ring intact thus adding considerably to the difficulties of the determination of structure.The observation that carimbose was distinctly more basic than the parent antibiotic suggested that in magnamycin the mycarose moiety was attached through an hydroxyl group adjacent to the dimethylamino-group. The pK shifts observed upon acetylation of magnamycin (from 7.0 to 6.0) and carimbose (from 8.3 to 5.4) indicated that the dimethylamino-group was flanked by two hydroxyl groups. This was confirmed and the exact point of attachment of mycarose placed a t the 4’-position of the mycaminose ring by exhaustive methylation of magnamycin followed by hydrolysis to yield the 2’-methyl derivative of the basic sugar. At this point magnamycin could be formulated as C,2H,,0,-O-sugar.Four of the eight oxygen atoms of the unknown fragment were readily accounted for by the presence of a methoxyl an acetyl and an aldehyde group. Two more were shown to be in a lactone function since a carboxyl group was generated by vigorous alkaline hydrolysis of magnamycin after reduction first catalytic and then with borohydride. Existence of the difficultly reducible lactonic carbonyl group was required also to permit base-catalysed removal of a neighbouring proton thus initiating consecutive elimination reactions which yielded a doubly unsaturated carboxylic acid (see below). The remaining two oxygen atoms were assigned to a structural moiety represented by (7) and corresponding eventually to to C(15). of magnamycin. This last conclusion was based upon ( a ) mild reduction with iodide that removed one oxygen atom and gave a product identical with magnamycin B a natural companion of magnamycin whose absorption spectrum was characteristic of an a18 76-doubly unsaturated carbonyl system (8) ; ( b ) mild catalytic hydrogenation which added four hydrogen atoms to the molecule without loss of oxygen but with formation of one new hydroxyl group (9) ; and (c) mild nitric acid oxidation from which ethylene oxide-cis-dicarboxylic acid (10) was isolated.(8) A series of oxidative and hydrolytic experiments led to the isolation and characterisation of the tribasic acid (11) and hence to assignment of structure 96 QUARTERLY REVIEWS for carbon atoms 1 to 11 of magnamycin. In order to establish the positions of the groups that had suffered elimination in the formation of ( I 1 ) carimbose dimethyl acetal was subjected to vigorous reduction mild hydrolysis and then oxidation.Partial structure (12) was assigned to the product since it yielded a 5-membered lactone only after vigorous hydrolysis had removed the amino-sugar. This series of experiments demonstrated that the O-acetyl group of magnamycin was a t position 3 the glycosidic linkage a t position 5 and the aldehyde group on A further series of oxidative experiments on magnamycin B partially represented by (8) revealed the structure a t carbon atoms 11 to 17. Oxida- tion of magnamycin B followed by alkaline hydrolysis gave crotonic acid from carbon atoms 15 16 17 and the 17-methyl group. Tetrahydro- magnamycin B was converted into the enol acetate (13) and ozoiiised and n-octanoic acid isolated after reduction of the ozonide.It is not possible to detail here the many reaction sequences or t o re- produce the brilliant interpretation of the experimental results which led to the elucidation of the structure of magnamycin. Time devoted by the reader to perusal of the original article Methymycin and neoMethymycin.-Methymycin C,,H,,O,N is structur- ally one of the simplest of the macrolide antibiotics and its constitution (14) was the first of the group to be completely worked The isomeric neomethymycin isolated from the mother-liquors of methymycin is repre- sented by formula (15). Acid-hydrolysis of these antibiotics cleaved the glycosidic linkages ; the aglycones were different but both substances gave the same carbohydrate moiety.The sugar proved t o be desosamine (14 B) which had been isolated previously from other macrolide antibiotics and whose structure was already known.5 An interesting feature of methymycin chemistry is the formation of spiroketals (16a b c). Thus treatment of methynolide (14 A) C,,H,805 the aglycone of methymycin with methanol containing hydrogen chloride gave a compound C,sH,oO (l6a) in which hydroxyl and the conjugated carbonyl groups could no longer be detected in the infrared spectrum but which now contained a methoxyl group. A similar product (16b) was formed when methymycin was treated with hydrochloric acid and (16c) was generated from dihydromethynolide with great ease and rapidity when could hardly be better spent. Djerassi and Zderic J . Amer. Chem. SOC. 1956 78 6390.Djerassi and Halpern ibid. 1957 79 2022. Brockmann and Strufe Chem. Bey. 1953 86 876 ; Clark. Antibiotics and Chemo- therapy 1953 3 663. BRINK AND HARMAN CHEBIISTRY OF SOME NEWER ANTIBIOTICS 97 A = Methynolide B = Desosaminc (160) X = Me0 (16b) X = CL E t &le (16~) X=H the reduced aglycone was placed in methanolic acid. It appears that although steric conditions are not favourable for spiroketal formation while the double bond is present saturation of the bond by addition of methanol hydrogen chloride or hydrogen permits ready spiroketal formation between the carbonyl and t'he two hydroxyl groups. Spiroketals have not been observed in the neomethymycin series nor would they be expected in view of the altered relationship of the hydroxyl groups to the carbonyl group.Pikromycin.-Pikromycin was the first of the macrolide antibiotics to be discovered.6 Despite its rather good activity against a variety of Gram- positive micro-organisms it has proved too toxic to be useful clinically. Pikromycin is very closely related chemically to the isomeric methymycin but its exact structure has not yet been established. Extensive degradative studies both by Brockmann and his co-workers in Germany and by Swiss workers Mild acid-hydrolysis of pikromycin yielded desosamine (14 B ; pikro- cinine) and the anhydroaglycone kromycin in which the newly introduced double bond has been shown to be in the 5 6-position. Dihydropikromycin in which the 8 9-double bond was reduced was readily cleaved also with loss of water to dihydrokromycin (18). Again evidence was available to permit assigning the newly introduced double bond t o the 5 6-position.Further information on the C(l)-C(,)-portion of the pikromycin lactone ring was provided by permanganate oxidation of the antibiotic and isolation of the lactone C,,H,,O (19). It is important to note however that the carboxyl of (19) could have originated in either C(,) of structure (17a) or C(l) of structure (17b). Of the two possible points of attachment of the sugar to the aglycone C(5) seems somewhat favoured both because of the location of the double bond a t 5 6 in krornycin and dihydrokromycin and because the extreme ease of removal of the sugar is suggestive of a linkage @ to the carbonyl group. Should desosamine prove to be attached a t C(3) pikromycin Brockmann and Henkel Natwwiss. 1950 37 138.Brockmann and Oster Chem. Ber. 1957 90 605. Anliker Dvornik Gubler Heusser and Prelog ibid. 1956 39 1785. have led to agreement on formula (17a or b) for pikromycin. 8 Anliker and Gubler Helv. Chim. Acta 1957 40 119 1768. 98 QUARTERLY REVIEWS 0 (17a) R = Desosamine moiety R’ = H (17b) R = H Me R’ = Desosamine moiety Me ?ye y e $HMeCO,H C-CH-CH -CH-CH -0- Me HO Et Ahe (18) (19) would have the same structure as methymycin (14) differing only stereo- chemically from that antibiotic. The elucidation of the structure of the rest of the pikromycin molecule came from other degradative studies. One key reaction sequence involved oxidation of dihydrokromycin (18) to yield propaldehyde (from and the ethyl group) lzevulic acid (from C+&o) and the 10-methyl group) and meso-Ma’-dimethylglutaric acid (derived from C(1)-C(5) and the 2- and 4-methyl groups).Erythromycin.-One of the more complex macrolide antibiotics is erythromycin (20) C37HS7013N. The constitution of this clinically useful antibiotic was established by studies carried out in the Lilly laboratories. lo Erythromycin contains desosamine and in addition the 9-carbon sugar cladinose (20 B). Unlike the other members of this group erythromycin does not possess a carbon-carbon double bond in the lactone ring. Erythro- mycin is inactivated under acid conditions presumably because of irrevers- ible spiroketal formation. It is of interest that erythromycin B C37H670 IzN a related antibiotic which differs from its companion only in the absence of the 12-hydroxyl group,ll has the expected greater acid stability.A = Desosamine B = Cladinose (21) Me2Nn Me 0 OH lo Wiley Gerzon Flynn Sigal Weaver Quark Charwette and Monahan J. Amer. l 1 Wiley Sigal Weaver Monahan and Gerzon ibid. p. 6070 Chenz. SOC. 1957 79 6062 and earlier papers. BRINK ANT) HARMAN CHEMISTRY O F SOME NEWER ANTIBIOTICS 99 Other Macrolide Antibiotics.-In addition to the members of this group already discussed there are a number of representatives about which less is known and reports of the discovery of new inacrolide antibiotics continue to appear with some regularity. Narbomycin C,,Hg707N7 in preliminary studies l2 showed the usual characteristics of these compounds two N-methyl and six or more C-methyl groups infrared absorption indicative of two carbonyl groups one of which was conjugated and hydrolysis to yield n basic sugar in this case desosamine.Oxidation gave the ten-carbon-atom lactone (19) also isolated from pikromycin 8 and methymycin.3 One of the newer antibiotics showing considerable clinical promise is oleandomycin C,,H,,O,,N. It has been described l3 as a polyhydroxy- keto-lactone linked glycosidically to both desosamine and the new sugar L-oleandrose. Other antibiotics which probably belong t o the macrolide group include angolamycin,14 miamycin,15 and the spiramycins A B and C.16 Hydro- lysis of the spiramycins (foromacidines 17) liberated propionic acid from C and acetic acid from B ; no volatile acid was obtained from A. Mycarose and mycaminose the two carbohydrates isolated from hydrolytic degrada- tion of magnamycin have been obtained also from each of the spiramycins along with a third sugar which has been assigned structure (21).Stereochemistry.-Some progress has been made towards working out the stereochemistry of the macrolide antibiotics. For example reference to formula (6) will show that magnamycin possesses 17 asymmetric carbon atoms and an asymmetrically substituted double bond so that (6) can represent 262,144 stereoisomers. Infrared studies showed that the sub- stituents on the double bond were in the trans-configuration ; the hydrogens of the ethylene oxide ring were known to be cis because of the isolation of ethylene oxide-cis-dicarboxylic acid ; nitric acid oxida,tion of magnamycin to yield L-( -)-methylsuccinic acid gave the configuration of the groups about C(lo) ; and the stereochemistry a t the 1’- 2’- 3’- and 4’-positions of the mycosamine moiety was deduced from considerations of the relative basicity of the antibiotic and various pertinent derivatives as well as from other points of chemical behaviour.As Woodward has mentioned, this reduced the number of possible isomers to a mere 4096. Experimental evidence has also been acquired for the stereochemical configurations of most of the asymmetric centres of erythromycin.10 l 2 Corbaz Ettlinger Gaumann Keller Kradolfer Kyburz Neipp Prelog Reusser l3 Els Murai and Celmer Abs. Papers 130th Meeting Amer. Chem. SOC. 1956 l4 Corbaz Ettlinger Gaumann Keller-Schierlein Neipp Prelog Reusser a.nd l5 Schmitz Misiek Heinemann Lein and Hooper Antibiotics and Chemotherapy l6 Paul and Tchelitcheff Bull. SOC. chim. France 1957 443 734 1059. l7 Corbaz Ettlinger Giiumann Keller-Schierlein Kradolfer Kyburz Neipp and Zahner Helv.China. Acta 1955 38 935. p. 1 5 ~ . Zahner Helv. Chim. Acta 1955 38 1202. 1857 ‘7 37. Prelog Wettstein and Zahner Helv. Chim Acta 1956 39 304. 100 QUARTERLY REVIEWS It must be made clear that the points mentioned above refer to relative stereochemical configurations ; some work however has been done on absolute stereochemistry. As has been mentioned magnamycin was degraded to L-( -)-methylsuccinic acid known to be related to L-glycer- aldehyde thus establishing the absolute configuration a t C(lo). A series of degradation reactions on neomethymycin gave the same methylsuccinic acid ; in this case the asymmetric centre of the product corresponded to C(4) of the antibiotic (15). Since both neomethymycin and methymycin had been degraded via the C,,-lactone ( 19) t o meso-cccc'-dimethylglutaric acid the absolute configurations a t both C(4) and C(6) were est'ablished for these two antibiotics.The same conclusions were drawn for the configurations of the two corresponding asymmetric centres of pikromycin and narbomycin since the lactone (19) was also obtained from these antibiotics.18 Biogenesis and the Propionate Rule.-The well-known " isoprene rule " advanced by Wallach in 1887 has been of great use in structural studies in the terpene field. In 1907 Collie l9 suggested that a number of naturally occurring aromatic compounds could have originated through condensations of chains built np of acetic acid units. Today the hiosynthesis of fatty acids from acetate is also firmly established. A series of papers by Birch 2o has shown how helpful an " acetate rule " can be within its proper limits in attacking problems of the structures of natural products.With the elucida- tion of the constitutions of some of the macrolide antibiotics a new " pro- pionate rule " has come into being. Several investigators 2 7 21 have called attention to the likely participa- tion of propionate units as well as acetate fragments in the Liosynthesis of some of the branched long-chain aliphatic acids from tubercle bacilli. Now in the lactone nucleus of erythromycin (20) there is a t hand an example of a natural material built up with perfect regularity from seven three-carbon units. The aglycones of methymycin and pikromycin are constructed partly of propionic and partly of acetic acid units but that of magnamycjn seems to be made up almost entirely of acetate fragments.that biogenetic principles now becoming evident will limit the number of possible structures that members of this group can have and should simplify the task of chemists occupied with such structural problems in the future. Woodward has pointed out Novobiocin Novobiocin is a crystalline antibiotic produced by c2 Streptomyces. It is primarily effcctive against Gram-positive micro-organisms and has proved t o be of considcrable use clinically especially in the treatment of infections caused by penicillin-resistant staphylococci. The structure of novobiocin Djernssi and Halpern J . Amer. Chem. SOC. 1957 '79 3927. l9 Collie I. 1907 91 1806. 2o Birch and Elliott Austral. J . Chem. 1966 9 95 a,nd earlier papers.*l Robinson " The Structural Relations of Natural Products " Clarendon Press London 1055 p. 7 ; Woodward Angew. Chena. 1956 68 19 ; Gerzon Flynn Sigal VT'iley Monaha.n and Quark J . Amer. Chem. Soc. 1956 78 6396. BRINK AND HARMAN CHEMISTRY OF SOME NEWER ANTIBIOTICS 101 complete except for a single point of stereochemistry is represented by formula (22). Studies which led to the elucidation of the constitution of Me M e 2 OH Q OH co AH* (22) the antibiotic were carried out independently and the results reported more or less simultaneously by workers in the Merck 22 and the Upjohn 23 laboratories. Novobiocin has the composition C,,H360,1N2 and is a dibasic acid with ph',' values of about 4.3 and 9.1. It possesses a methoxyl group cz mono- substituted amide group and a t least two C-methyl groups.The two acidic functions appeared to be enolic and phenolic in nature. Novobiocin was hydrogenated under mild conditions to the biologically active dihydro- derivative apparently by saturation of an unconjugated double bond since no change in the ultraviolet absorption spectrum was noted. Preliminary degradative studies indicated that the antibiotic molecule was made up of a sugar attached to an aromatic heterocyclic moiety which was in turn linked to a substituted benzoic acid in an A-B-C arrangement and that reactions could be chosen to rupture either the A-B or the B-C link. As inspection of formula (22) wo~ld indicate acid-alcoholysis liberat'ed a sugar derivative. Less expectedly (see below) treatment of novobiocin with hot acetic anhydride cleaved the amide bond and released the sub- stituted benzoic acid.It is convenient to consider separately the determina- tion of the structures of the three parts. When these were elucidated evidence was a t hand to permit assignment of structure (22) to the parent substance. The substituted benzoic acid moiety. Hot acetic anhydride cleaved z 3 novobiocin to a large carbohydrate-containing fragment of the composition C,,H2,01,N2 and a monobasic acid C,,H,,O (23). Treatment of this acid (23) with osmium tetroxide and then periodate yielded acetone ; deacetyla- tion gave a product with the properties of a substituted p-hydroxybenzoic I - OAc (23) (24) Ho2cqJMe 2 2 Folkers et al. J . Arner. Chem. SOC. 1955 77 6404 ; 1956 78 1770 3655 4126 ; 2 3 Hoeksemn et al. Antibiotics and Chemotherapy 1956 6 143 ; J .Arner. Chem. SOL 1958 80 137 140. 1955 77 6710; 1956 78 1072 2019; 1957 79 3789. 102 QUARTERLY REVIEWS acid. The latter compound heated in ethanolic hydrochloric acid cyclised to the known 2 2-dimethylchroman-6-carboxylic acid (24) thus fixing the location of the pentenyl group on the benzene ring. 4-Acetoxy-3-isopentyl- benzoic acid (25) was obtained either by mild hydrogenation of acid (23) or by acetic anhydride cleavage of dihydronovobiocin. The product C,,H,,O,,N (26) men- tioned above was an optically active neutral material which on treatment with methanolic hydrogen chloride yielded an optically inactive amphoteric compound (27) C,,H,O,N. The reactions and properties of this compound indicated that it possessed an aromatic nucleus substituted with a methyl group an amino-group a phenolic hydroxyl group and a strongly acidic enol.Because of the evident relation of compound (27) to 3-amino-4- hydroxycoumarin model studies were undertaken. When 3-benzamido-4- hydroxycoumarin (28) was heated with acetic anhydride the oxazole (29) was formed. This observation appeared to clarify the nature of degradation The aromatic heterocyclic moiety. Me Me OH qv (28) (29) 0-CMe products (26) and (27) except for the location of the methyl substituent and the phenolic hydroxyl group. Alkali-fusion of the substituted coumarin (27) gave 2-methylresorcinol and alkaline hydrolysis afforded 2 4-dihy- droxy-m-toluic acid. With the substituents thus placed the aromatic central moiety of novobiocin was assigned the indicated coumarin structure. Another degradative route in a different laboratory 22 led to an inde- pendent elucidation of the structure of the central portion of the antibiotic.Alcoholysis of dihydronovobiocin gave dihydronovobiocic acid (30). The amide linkage therein was cleaved by reaction with a mixture of acetic acid acetic anhydride and hydrogen bromide and the diacetylated aminohydroxy- coumarin (31) was isolated. Deacetylation with acid converted the product (31) into the “ aromatic ” amine (27) which as mentioned above had been obtained elsewhere by methanolysis of the oxazole (26). Selective deacetyla- tion of the product (31) gave the N-acetyl derivative (32). Both the products (27) and (32) could be reconverted by acetylation into the parent compound (31) showing that no deep-seated changes had occurred during either hydrolytic step.The action of a mixture of hot acetic and hydrochloric acid on dihydro- novobiocic acid (30) resulted in opening of the coumarin ring with only a partial cleavage of the amide linkage and products (33) and (34) were BRINK AND ITARMAN CHEMISTRY OF SOME NEWER ANTIBIOTICS 103 obtained. Because of the aliphatic nature oE the amino-group in (34) and its aromatic character in (27) it was evident that this group had originally been attached to the pyrone ring of the coumarin moiety. Clemmensen reduction of the amino-ketone (34) yielded the known 4-ethyl-2-methyl- resorcinol (35) establishing the relative positions of the substituents in the earlier products of this series. The lack of ester-carbonyl absorption in the infrared spectra of both products (30) and (33) indicated that the benzoic acid moiety of novobiocin was attached to the coumarin structure through an amide linkage.The correctness of the structure assigned to the central heterocyclic moiety of novobiocin has been confirmed by the synthesis 22 of dihydro- novobiocic acid (30). An interesting feature of novobiocin chemistry is the possibility of the existence of tautomeric forms in the central heterocyclic moiety and this became evident during the degradative sequence just described.22 The infrared spectrum of novobiocin (22) contained an absorption band a t 5.92 ,u attributed to the carbonyl group of the unsaturated lactone. Rather surprisingly there was no corresponding carbonyl band in dihydronovobiocic acid (30) but with the conversion of this acid (30) into the diacetylated coumarin (31) the 5.92 ,u band reappeared.Selective deacetylation of (31) by aqueous alkali resulted in a monoacetyl derivative (32) which showed no carbonyl absorption in the 5.5-6-0 ,u region of its spectrum although the band was restored upon reacetylation of (32) to (31). On the basis of these observations novobiocin and the degradation product (31 ) were assigned the indicated coumarin structures while dihydronovobiocic acid (30) and com- pound (32) have been represented as hydroxychromones. It is assumed that the chromone-carbonyl absorption band in the latter compounds has shifted to a longer wavelength where it is masked by the absorption of the amide group. Methsnolysis of novobiocin gave an optically active neutral product (36) * containing a carbamoyl group two methoxyl groups and one free hydroxyl group.The possible presence of a gem.- riimethyl structure was suggested from the Kuhn-Roth C-methyl value of Structure of the sugar. * Since configurations about the glycosidic carbon have not yet bcen assigned structures (36) to (40) have been written to indicate that pure anomers of unknown configurations were used in the experimental work. 104 QUARTERLY REVIEWS 0-30. This compound now designated methyl 3-O-carbamoylnovioside was stable to periodate. After acid hydrolysis the resulting sugar 3-O-carbamoylnoviose (37) consumed one equivalent of periodate with the formation of a mol. of formic acid. The original free hydroxyl group was thus located adjacent to the glycosidic carbon atom i.e. a t position 2 (of the sugar molecule). Either alkaline hydrolysis or acid methanolysis of the sugar (36) gave carbon dioxide and ammonia from the decomposition of the urethane structure and a mixture of anomeric methyl noviosides (38).These gave acetone upon oxidation with chromic acid. The mixture con- sumed one mol. of periodate to give a dialdehyde which yielded glyoxal after mild hydrolysis. Acid hydrolysis of the methyl glycosides (38) afforded noviose (39) which reacted with two equivalents of periodate to form two mols. of formic acid. This series of experiments indicated that the two hydroxyl groups of compound (38) must have been a t positions 2 and 3 and the original carbamoyl group a t position 3. Further evidence bearing on the relation of the two groups a t C(e) and C(3) was provided by the isolation of compound (40) from the methanolysis of the carbamoyl compound (36).This product assigned its cyclic carbonate structure largely on the basis of analytical and infrared data was convertcd by treatmcnt with barium hydroxide into methyl novioside (38). Another sequence of degradative reactions started with the conversion of the glycoside (36) via the diethyl mercaptal(41) into the l-deoxy-derivative (42). Compound (42) was stable to periodate but after removal of the carbamoyl group the new product took up one equivalent of periodate. Prom the reaction mixture were isolated acetaldehyde and after further oxidation of the resulting aldehyde with bromine the (-)-acid (43). Me Me Me Me Me Me M e 0 0 OMe M Z @ o l M e o ( OMe M e 0 OH ? OH 9 OH HO OH HO OH (38) (39) F0 NHz (37) co “JH (36) HO Me CH (SEt) MeoG OH F0 NH (41) HO Me 0 OH $0 (42) NH2 HO Me Met (43) The stereochemistry of the groups a t positions 2,3 and 4 of the sugar moiety has been determined based in part on the rules of optical rotation and the configuration of L-lyxose has been assigned to the carbohydrate fragment .2 4 Thus the structural studies on novobiocin are complete except 24 Walton Rodin Stammer Holly and Follters J . Amer. Ghern. SOC. 1956 78 5454. BRINK AND HARMAN CHEMISTRY OF SOME NEWER ANTIBIOTICS 105 for one point-the configuration about the glycosidic carbon atom of the carbohydrate portion of the molecule. Antifungal antibiotics After the introduction of the broad-spectrum antibiotics particularly the tetracyclines clinicians have noted the occasional occurrence of severe and sometimes fatal super-infections by such organisms as the yeast Candida albicans.The important role of the fungi in these recently observed com- plications of antibiotic therapy has sparked interest in antifungal agents for clinical use in combination with a variety of antibacterial antibiotics. Among antifunga,l agents whose chemistry has been reviewed recently 25 are such diverse structures as the tripyrrylmethene dye prodigiosin the polypeptide fungista.tin the polyenyne mycomycin the sulphur-contain- ing t)hiolutin the complex antimycin A and gliotoxin. Most of these compounds are significantly active against bacteria and other micro- organisms as well a,s yeasts and fungi. This Review will be concerned with a series of antibiotics the aiitifungal polyenes,2G which are active against a wid6 range of fungi and yeasts but do not show antibacterial activity.This class of compound is further characterised by the ultraviolet spectra associated with the polyenic chromophores by a blue-violet coloration with concentrated sulphuric acid by low solubilities and by sensitivity to light and air. Sub-groups of tetra- penta- hexa- and hepta-enes are each characterised by a closely defined series of ultraviolet absorption maxima. The heptaene amphotericin B occurs along with the tetraenic ampho- tericin A of lower activity against fungi. The biologically inactive perhydro- derivative was used in degradative studies 27 because of its more favourable solubility and stability. Amphotericiii B has been assigned the tentative empirical formula C,,H ,3020N ; infrared absorption spectra and behaviour on potentiometric titration suggested the presence of a lactone group in the antibiotic Acetolysis gave two crystalline substances which proved to be acetyl derivatives of an amino-sugar mycosamine (44).This methyl- pentose possesses the D-configuration a t position 6 since periodate cleavage of methyl N-ethylmycosaminide yielded D-methoxy-D'-methyldiglycollic aldehyde (45). The stereochemistry of the other asymmetric centres of mycosamine is unknown and no further information on the nature of the remaining C, moiety of amphotericin B has been reported. A similarity and perhaps also a biogenetic relationship of the polyenic nuclei of these antibiotics to the xanthophylls and carotenoids has been suggested. 25 Duggar and Singleton Ann. Rev. Biochem. 1963 22 459 ; Binkley ibid.1955 26 Ball Bessell and Mortimer J. Gen. Microbiol. 1957 17 96. 27 Walters Dutcher and Wintersteiner J. Amer. Chem. Soc. 1957 79 5076. 24 597. 106 QUARTERLY REVIEWS Fungichromin a pentaene and the hexaene fungichromatin have been obtained from Streptomyces ceZZuZosz. 28 These are pale yellow nitrogen-free polyenes ; little has been reported about their structure. Filipin C3,,H5,,01,, from S. Jilipinensis is another pentaene antifungal agent with properties common to this group of antibiotic^.^^ Light and heat transform filipin into a new substance with neither the ultraviolet absorption nor the biological activity of the parent compound. In alcoholic solution a crystalline transformation product C,,H,,O, is formed ; this too is devoid of antifungal activity but retains the pentaene ultraviolet absorption spectrum.Anti-tumour antibiotics Treatments that man has employed through the centuries in his efforts to cure cancer run the gamut from blood-letting to the use of diets of crab soup. The current experimental era which dates from about 1900 was guided initially by the rapidly developing sciences of bacteriology and immunology. Recent promising advances in the chemotherapy of cancer however may be correlated with studies on cell-metabolism and metabolite- antagonists. Agents have been sought that would selectively inhibit the growth of or actually destroy malignant cells in the body either by inhibiting the synthesis of nucleic acids or by interfering with their utilisation at later stages of the metabolic process. I n the most recent approach metabolic products from fermentation broths have been screened directly for anti- tumour activity.Most of the agents discussed in this section although primarily of interest for their activities against malignant tumours also possess a t least some degree of antibacterial activity. This antibacterial action has frequently been of aid in the isolation of the materials. On this basis the inclusion of a few of these anti-tumour agents in the present Review seems justified. The Actinomycins.-Since the isolation of actinom ycin A by Waksman and Woodruff in 1940 30 a number of closely related antibiotics produced by Streptomyces have been reported. These substances were observed to be highly active against Gram-positive bacteria and somewhat less effective against Gram-negative organisms and fungi; they were later shown to possess a selective cytostatic effect on mammalian tissue.Although a German clinician 31 has described beneficial effects in patients with lymphatic cancers treated with actinomycin C the extraordinary toxicity of these agents renders them of doubtful therapeutic value. Despite this the hope remains that further search among the metabolic products of micro-organisms may lead to products of real utility in anti-tumour therapy. Early studies on the actinomycins were hampered by the quality of the preparations ; the original materials judged homogeneous by classical 28 Tytell McCzLrthy Fisher Eolhofer and Charney " Antibiotics Annual " Medical Encyclopedia Inc. New York 1954-5 p. 716. 29 Whitfield Brock Ammann Gottlieh and Carter J . Amer.Chem. SOC. 1955 '7'7 4799. 30 Waksman and Woodruff Proc. SOC. Exp. BioZ. Med. 1940 45 609. s1 Schulto 2. Krebsforsch. 1952 58 500. BRINK AND RARMAN CHEMISTRY O F ROME NEWER ANTIBIOTICS 107 standards were later shown to be mixtures. These were finally resolved by solvent-partition techniques. 32 To date seven individual compounds have been characterised and structures have been proposed for two. These consist of a polycyclic quinonoid chromophore linked to two cyclic peptide chains. Similarities in physical and chemical properties suggest that all the actinomycins may have the same chromophore and differ only in the nature of the peptide moieties. The elucidation of the structure of actino- mycin C by Brockmann and his collaborators 33 in Germany and of actino- mycin D by Bullock and Johnson 34 in England required investigations in widely diverse areas of organic chemistry.The elegance of the work can only bc suggested in the present Review. The bright red crystalline antibiotic appeared from early studies to be a very weak monoacidic base. A composition C61H93016N12 was suggested by the results of elementary analyses quantitative hydrogenation studies and redox titrations. Quinone functionality was proposed from observations Formula (46) has been provisionally proposed for actinomycin C,. B Me ,Me Ye 8 FH fl 0 CH U b ,CH-O-C-CH -NMe-C-Ct+ NMe / R = -NH-C\H Me MQ 0 5-NH-FH-CO-N-CH-C (46) Et/'Me both on the ready reduction of the antibiotic to the yellow dihydro-derivative and its ready reoxidation to the parent actinomycin and on reductive ncetylation to well-defined polyacetyl compounds.The presence of lactone or ester groups was indicated by the infrared spectrum of actinomycin C and by its consumption of alkali on titration a t elevated temperature. Hydrolysis of the antibiotic gave a mixture of amino-acids ; a cyclic poly- peptide structure was suggested by the absence in the parent compound of acidic or basic functionality other than the very weakly basic nuclear nmino-group. The lactone systems of actinomycin C were hydrolysed in warm dilute inethanolic alkali to give the dibasic actinomycinic acid C,. Controlled ;wid-hydrolysis of this acid attacked the nuclear amino-group first and gave tieaminoactinomycinic acid C,. This was degraded further by acid to a series of deaminoactinocyl peptides and finally with loss of all of the amino- acid residues and of one carboxyl group to the monocarboxy-compound ;Lctinocinin (47).The formation of actinocinin either by the degradation sequence just mentioned or when carried out by direct vigorous acid hydrolysis of actinomycin C, was accompanied by further degradation by decarboxylation to 2-hydroxy-4 5-dimethylphenoxazin-3-one and by 3 2 Roussos and Vining J. 1956 2469. 33 Brockmann et al. Angcw. Chern. 1956 68 70 ; Chem. Ber. 1956 89 1397 and 34 Bullock and Johnson J. 1957 3280 and earlier papers. ea.rlier papers. 108 QUARTERLY REVIEWS hydrolysis to 2 5-dihydroxytoluquinone (48). The structures of the decarboxylated phenoxazinone and of actinocinin itself were established by synthesis. The products were readily prepared by condensations of 2 5- dihydrox yt oluquinone with 3-amino-2- hydroxytoluene and with 2 -amino-3 - hydroxy-4-methylbenzoic acid respectively.Confirmation of the proposed relationship between actinomycin C (46) and actinocinin (47) was secured through synthesis of (49) the dimethyl ester of actinocylbisglycine. This compound prepared by oxidative self-condensation of 2-amino-3-hydroxy- 4-methylbcnzoylglycine methyl ester was nearly identical with actinomycin in its colour reactions and ultraviolet absorption spectrum. Under conditions of alkaline hydrolysis rearrangement of the chromo- phoric structure occurred and actinomycin C yielded the constituent amino- acids and a fragment (50) designated depeptidoactinomycin. This rearrange- ment product has been obtained from all of the actinomycins that have been studied.Notable differences in acid stability colour reactions and absorp- tion spectra indicated that depeptidoactinomycin was fundamentally different in structure from the actinomycin chromophore and this was further demonstrated by the isolation of a dimethylacridine from (60) after zinc dust distillation. The acridone-1 4-quinone formulation of depep- tidoactinomycin was confirmed by synthesis. Studies on the identification and estimation of the amino-acids in hydro- lysates of actinomycin C indicated the presence of two mols. each of L-threo- nine 1,-N-methylvaline sarcosine L-proline and D-a~~oisoleucine. Hydro- lysis of the methyl ketone obtained from actinomycinic acid C (peptide lactones open) by a Dakin-West reaction failed to yield any N-methylvaline and hence it was concluded that residues of this amino-acid must terminate each of the two opened peptide chains.Degradation of actinomycin C with hydrazine gave somewhat more than one mol. of the dioxopiperazine from N-methylvaline and sarcosine and this evidence was interpreted as suggestive of the adjacent disposition of these two amino-acids in the peptide chains. One characterised product of the controlled acid-hydrolysis of actinomycin C was deaminoactinocylthreonine in which only a threonine residue was linked to the nucleus ; thus it was evident that at least one of the peptide chains was linked to the chromophore through threonine. On the basis of these and other observations Brockmann and his group have provisionally advanced a formulation of the peptide portion of actinomycin C as represented in (as) but they have indicated that some points of the structure of the peptide chains remain to be established.Bullock and Johnson have established in somewhat more rigorous BRINK AND HARMAN CHEMISTRY OF SOME TSEWER ANTIBIOTICS 109 CO,H pepii de $H Me \ (50) fashion 34 that actinomycin D has the same structure as that represented in (46) for actinomycin C, except that the two D-alloisoleucine residues have been replaced by D-valine. These investigators discovered that alkaline peroxide cleaved the chromophoric structure and permitted the separation of the two intact peptide chains in each case with the original lactone ring open. One peptide was linked to 7-methylbenzoxazolone-4-carboxylic acid as indicated in (51) and the other to oxalic acid. Mild acid-hydrolysis of the peptide (51) selectively liberated N-mcthyl- valine ; the result of a Dakin-West degradation confirmed the conclusion that N-methylvaline was located a t the carboxyl end of the chain.That ssrcosine was attached to N-methylvaline was indicated by the isolation of the dioxopiperazine of these two amino-acids after high-vacuum pyrolysis of the peptide. Somewhat more drastic hydrolysis of the peptide (51) liberated proline sarcosine and N-methylvaline only traces of valine and 110 threonine. The remaining valine and some threonine were released on more prolonged hydrolysis. Since the peptide itself did not possess a free amino-group it was thus possible to formulate the peptide chain as -NH-threonine-valine-proline-sarcosine-N-methylvaline-CO~H. It was shown in similar fashion that the amino-acids of the second peptide chain attached to oxalic acid were arranged in the same way.Before leaving the chemistry of the actinomycins it is interesting to note that the unusual phenoxazin-3-one structure common to this group of antibiotics appears also in xanthommatin (52) isolated from the eyes of insects According to B ~ t e n a n d t ~ ~ this com- FH2 y02H FHSNH2 pound occurring is the pigments prototype of much of a family importance of naturally in the &o,E" (52) field of chemical genetics. A series of substi- tuted phenoxazin-3-ones has been synthesised ; 36 the 2-amino- and the 2-NN-diethylglycylamino-derivative were reported to be highly bacteriostatic. Azaserine.-The compound 0-diazoacetyl-L-serine (53) was isolated from filtrates of a Streptomyces 37 and has been designated azaserine.It is moderately active against a number of bacteria and fungi but is essentially without activity upon protozoa and viruses. It is quite effective against 3 5 Butenandt Angew. Chern. 1957 69 16. 36 Yuasa Chem. Abstr. 1954 48 12900. 37 Coffey Hillegas Knudsen Koepsell Oyaas and Ehrlich Antibiotics and Chemo- therapy 1954 4 775. H 110 QUARTERLY REVIEWS some experimental mouse tumours. The light yellow-green crystalline antibiotic was assigned diazo-ester functionality on the basis of its infrared spectrum and failure to react with hydroxylamine. Hydrolysis in hot 2N-fOrmiC acid gave L-serine and glycollic acid. Held in solution a t pH 2 and room temperature azaserine was converted into O-glycollyl-L-serine. Hydro- genation converted azaserine into O-acetyl-L-serine.38 Several syntheses of azaserine have been recorded 39 all involving the preparation and diazotisa- tion of O-glycyl-L-serine. The D-serine analogue similarly prepared showed no anti-tumour activity. A series of publications 4O from Buchanan's laboratory on the mode of action of azaserine deals with its effects in the inhibition of purine synthesis by the enzyme systems of pigeon liver. Esters of serine with various substituted acids have been prepared,41 as have also diazoacetyl esters of a number of hydroxy-amino-acids; none of these azaserine analogues has had significant anti-tumour activity. DON.-6-Diazo-5-oxo-~-norleucine (54) abbreviated DON was also isolated from a Streptomyces. 4 2 Although only weakly active against bacteria and fungi DON has a powerful inhibitory action upon an experi- mental mouse 43 Preliminary structural studies indicated that the antibiotic possessed the composition C6H,03N3 and was an amino-acid with diazoketone function- ality.Periodate cleavage gave L-glutamic acid permitting the formulation of DON as 2- or 4-amino-6-diazo-5-oxohexanoic acid. Under conditions of the Wolff rearrangement DON gave a-aminoadipic acid establishing the correctness of structure (54). The compound was synthesised 44 from L-glutaniic acid a-methyl ester by a series of reactions involving protection of the amino-group conversion of the carboxyl group via the acid chloride into the diazo-ketone and removal of the protective group. Alazopepth-This new anti-tumour antibiotic isolated from an actino- mycete has been only partially characterised.45 It appears to be a structural analogue of azaserine and DON containing 1 mole of a-alanine and 2 moles of an amino-6-diazo-5-oxohexanoic acid.38 Fusari Haskell Frohardt and Rartz J . Amer. Chem. SOC. 1964 76 2881. 3Q Nicolaides Westland and Wittle ibid. p. 2887. 40 Levenberg and Buchtman ibid. 1956 78 504. 41 DeWald Behn Moore Morgan and Renfrew Abs. Papers 129th Meeting Amer. 43 Ehrlich Coffey Fisher Hillegas Kohberger Machamer Rightsel and Roegner 4 3 Clarke Reilly and Stock Abs. Papers 129th Meeting Amer. Chein. Soc. 1956 4 4 DeWald and Moore Abs. Papers 129th Meeting Amer. Chem. Soc. 1966 p. 1 3 ~ . 4 5 DeVoe Rigler Shay Martin Boyd Backus Mowat and Bohonos " Antibiotics Chem. Soc. 1956 1 6 ~ . Antibiotics and Chemotherapy 1956 6 487. p. 12M. Annual " Medical Encyclopedia Inc.New York 1956-7 p. 730. BRINK AND HARMAN CHEMISTRY O F SOME NEWER ANTIBIOTICS 111 Sarkomycin.-Much interest has been excited recently by the dis- covery 46 of a weakly antibacterial antibiotic which has been reported to have significant activity against an experimental tumour in mice and is being studied clinically in cases of inoperable malignancies. * 7 Sarkomycin was isolated in Tokyo from the products of Streptomyces erythrochromogenes. The structure 2-methylene-3-oxocycZopentanecarboxylic acid (55) was assigned on the basis of studies a t the Bristol Lab~ratories.~* Sarkomycin obtained only as an oil of doubtful purity had an absorption spectrum suggestive of the presence of carbon-carbon unsaturation and carbonyl and carboxyl groups. Hydrogenation destroyed the antibacterial but not the anti-turnour activity of sarkomycin and gave dihydrosarkomycin a crystalline optically active keto-acid C,H,,O, which contained one C-methyl group.Wolff-Kishner reduction of dihydrosarkomycin and conversion into the amide gave a product that appeared to be a methylcyclopentanecarboxy- amide. Ozonolysis of sarkomycin gave formaldehyde establishing the exocyclic nature of the double bond. Upon destructive distillation of thc antibiotic the double bond shifted into conjugation with the carboxyl group and a compound identified as 2-inethyl-3-oxocycZopent-l- enecarboxylic acid was obtained. From this and other evidence i t was concluded that the chief active constituent of sarkomycin possessed structure (55). The synthesis of the (&)-isomer of sarkomycin by a Mannich reaction on ethyl 3-oxocycZopentanecarboxylate and subsequent thermal elimination of the /I-amino-group and hydrolysis of the ester has been reported.49 Puromycin.-Despite possession of a considerable range of antibacterial activity puromycin has primarily been of ?Me2 interest because of its action against infections N & ? \ caused by various protozoa and for its ability I 11 sCH t o inhibit an experimental cancer in mice.Degradative studies on puromycin discovered in the Lederle Laboratories in 1952 led to the announcement 60 of its structure as 65) 6-dimethylamino-9-[3‘-deoxy-3’-(p-methoxy-~- HN OH phenylalanylamino) - /3 - D - ribofuranosyl]purine the antibiotic has been recently reviewed by B i n k l e ~ . ~ ~ Some of the synthetic studies by Baker and his collaborators 51 will be presented here since they contribute *:H2 (55) H q<i;C\i/ H O % S / OC-FH-CH I \ /OMe (56) in 1953.The work on the structure of -0 NH2 4 6 Umezawa Yamamoto Takeuchi Osata Okami Yamaoka Okuda Nitta 47 Ishiyama J . Antibiotics (Japan) 1!154 7 A 82 ; 1955 8 A 67 ; Fujii Onizawa 48 Hooper Cheney Cron Fardig Johnson Johnson Palermiti Schmitz and Wheat- 4° Tiki Bull. Chem. SOC. Japan 1957 30 450. 60 Waller Fryth Hutchings and Williams J . Amer. Chem. SOC. 1953 75 2025. 6 1 Baker et al. ibid. 1955 77 5911 and earlier papers. Yagishita Utahara and Umezawa Antibiotics and Chemolherupy 1954 4 514. Shima Okuyama and Okamoto ibid. p. 83. ley Antibiotics and Chemotherapy 1955 5 585. 112 QUARTERLY REVIEWS significantly to our fund of chemical knowledge and our understanding of structure-activity relations.The 6-dimethylaminopurine moiety was constructed without difficulty by a series of well-established reactions starting with thiourea and cyano- acetic ester. I n model studies condensation of a-acetobromoglucose with the chloromercuri-derivative of 6-dimethylaminopurine or its 2 8- bis- methylthio-derivative gave nucleosides which on the basis of their ultra- violet absorption spectra were 7- rather than 9-glucosides. Fortunately coupling of the chloroinercuri-derivative of the 2 -met hylt hio- substituted purine with the acetohalogeno-sugar yielded the desired purine 9-glucoside. Further model studies were required to establish conditions for elabora- tion of the purine 9-glycosides of amino-sugars which had not previously been syiithesised.It was found that the required bromo-sugar derivative was not stable under the coupling conditions but that a-acetochloro-D- glucosamine could be condensed successfully with the methylthio-purine. The product on reductive desulphurisation and 0-deacetylation gave the expected N-acetyl nucleoside. The starting material for the preparation of the ribose fragment required for the synthesis of puromycin was the 3 5-isopropylidene derivative of methyl D-xylofuranoside * (57). This was methanesulphonylated the isopro- pylidene group was removed and the product treated with sodium methoxide to form methyl 2 3-anhydro-~-lyxofurtnoside (58). Oxide cleavage with ammonia was followed by N-acetylation to give 3-acetamido-3-deoxy-~-ara- bofuranoside (59). In the final series of reactions involving an inversion of stereochemical configuration the arabinose derivative (59) was methane- sulphonylated and both sulphonyl groups displaced by the action of sodium acetate in hot " Cellosolve " to yield the 3-acetamido-2 5-di-0-acetyl-3- deoxy-D-ribofuranoside (60).This was then converted into an anomeric mixture of the desired 3-acetamido-1-0-acetyl-2 5-di-0-benzoyl-3-deoxy- D-ribofuranosides (61). Preparation of the chloro-sugar was carried out concurrently with nucleoside coupling by treating the titanium tetrachloride complex of the 1 -acetylribofuranoside (61) with the chloromercuri-derivative of 6-dimethyl- amino-2-methylthiopurine in refluxing ethylene dichIoride solution. The crude nucleoside so produced was desulphurised debenzoylated and de-N-acetylated to yield the aminonucleoside (62) that had been obtained Bz0.H2C ,O Ac t- * The anomers were separated by fractional distillation and carried separately through the synthesis.BRINK AND HARMAN CHEMISTRY OF SOME NEWER ANTIBIOTICS 113 previously by degradation of puromycin. Reaction of the aminonucleo- side with the mixed anhydride (63) of N-benzyloxycarbonyl-p-methoxy- L-phenylalanine and ethyl hydrogen carbonate followed by removal of the benzyloxycarbonyl group by hydrogenolysis gave puromycin. H2N OH (62) The a-anomer of the aminonucleoside (62) was synthcsised by Baker and Schaub by a reaction sequence planned on their postulate that in this type of coupling the bulky purine moiety would always enter the sugar ring from the side opposite the group a t C(2)7 regardless of the relative configura- tions at Cp) and C,,,.Reactions similar to those outlined above were used to prepare the arabofuranoside (64). The nucleoside condensation pro- ceeded normally ; desulphurisation and debenzoylation gave the product (Mi) the 2’-epimer of the a-nnomer of the N-acetyl derivative of (62). Inversion a t 2‘ was effected by rnethanesulphonylation and displacement by acetate ion. BzO.H~C 0 W O A c Ac*H N (64) HomH2T AcHN This summary of the work of Baker’s group must suffice to indicate the excellence of their efforts in the field. Their further contributions include synthesis of the 2’-epimer of the aminonucleoside (62) use of the N-phthaloyl Ijlocking group in the preparation of other aminonucleosides discovery of an tinusual degradation of the purine ring system7 and syntheses of puromycin analogues which lack the sugar amino-group.Analogues of the amino- iiucleoside (62) which have variations in the alkyl groups on the pyrimidine nitrogen have also been made.52 Numerous publications deal with the physiological activity of puromycin and the many related compounds that have been prepared and studied. The interesting structure-activity relations can only be briefly indicated here. The amino-acid-free aminonucleoside (62) was highly effective against the mouse cancer and three to four times as active as puromycin Ii2 Goldman Marsico and Angier J . Amer. Chem. SOC. 1956 78 4173. 114 QUARTERLY REVIEWS against the protozoaii Trypanosoma equiperdum. However in contrast to the original antibiotic it was almost devoid of activity toward the dysentery pathogen Endurndm histolyticu and inactive against bacteria.Inter- estingly its cc-anomer showed neither of the types of activity exhibited by the aminonucleoside itself. Most of the derivatives of puromycin have significant anti-tumour activity in mice. Of the several amino-acid analogues prepared by Baker et al. increased activity has been observed upon replacement of the p-meth- oxyphenylalanyl moiety of the parent compound by L-phenylalanyl glycyl leucyl and glycyl-p-methoxy-L-phenylalanyl residues. 53 It was noted that the D-phenykdanyl analogue was inactive. The p-methoxy-group of puromycin although apparently not necessary for anti- tumour activity does have a specific function against E. histolytica since the L-phenylalanyl analogue was devoid of action against this protozoan.54 Other Nucleoside Antibiotics.-Cordycepin (66) was isolated 55 from the mould Cordyceps militaris (Linn.) Link.It is active against strains of B. subtilis and an avia,n tubercle bacillus but is not tumour-inhibitory. Acid hydrolysis gave adenine and a sugar (-)-cordycepose C,H1,O ; a 9-purine linkage was indicated by the ultraviolet spectrum. Formation of an osazoiie from cordycepose eliminated the possibility of a 2-deoxy- structure and formulation as a 3-deoxypentose was suggested by the stability of the parent antibiotic to periodic acid. Cordycepose was oxidised with bromine to a lactone C5H804 whose phenylhydrazide differed from those produced similarly from the four stereoisomeric straight-chain S-deoxyaldo- pentoses. The usual structure assigned to the sugar on the basis of these data was confirmed by ~ynthesis.~6 Nebularine (67) was isolated 57 by Swedish investigators from the mush- room Aguricus (Clytocybe) nebularis Batsch.It is active against myco- bacteria and is highly toxic in mice but a t high dilution attacks cancerous cells preferentially. The structure assigned to nebularine on the basis of degradative evidence has been confirmed by synthesis. 58 HO*H,C OH (66) H 0 *S02N H (68) 63 Bennett Halliday Oleson and Williams “ Antibiotics Annual ” Medical Ency- 54 Bond Sherman. and Taylor “ Antibiotics Annual ” Medical Encyclopedia 6 5 Bentley Cunningham and Spring J . 1951 2301. 56 Raphael and Roxburgh Chem. und Ind. 1953 1034. 57 Lofgren Liining and Hedstrom Actn Chcm. Xcand. 1954 8 670. 68 Brown and Weliby J. Bid. Chern. 1953 204 1019. clopedia Inc. New York 1954-5 p. 756. Inc. New York 1954-5 p. 751. BRINK AND HARMAN CHEMISTRY OF SOME NEWER ANTIBIOTICS 115 A recently reported product of a Streptomyces nucleocidin possesses broad-spectrum antibacterial activity and is also phenomenally active against T. equiperdum in mice having 4000 times the potency of puromycin against this pathogen. Degradative studies 59 have suggested a sulphamic ester structure for nucleocidin as indicated by the partial formula (68). 59 Waller Patrick Fulmor and Meyer J . Amer. Chem. SOC. 1957 79 1011.

ISSN:0009-2681    

DOI:10.1039/QR9581200093

出版商:RSC    

年代:1958

数据来源: RSC

摘要:

ACTIVE NITROGEN By K. R. JENNINGS B.A. and J. W. LINNETT M.A. D.PHIL. F.R.S. (THE INORGANIC CHEMISTRY LABORATORY OXFORD) WHEN nitrogen is subjected to an electrical discharge a t low pressures a brilliant peach-coloured glow is emitted. I n 1900 Lewis discovered that when the discharge is switched off a golden-yellow afterglow remained for several seconds. In 1911 Strutt named the glowing gas " active nitrogen " and suggested that many of its properties could be attributed to the presence of atomic nitrogen. Since then several other theories have been put forward invoking excited atoms and molecules and even ions to explain the after- glow but during the past ten years the application of modern experimental techniques has shown that the main reactive component in active nitrogen is the nitrogen atom in the ground state.The purpose of this Review is to present the results of recent experiments and to show how these havc led to a more detailed exposition of the atomic theory of active nitrogen. Summary of early work An excellent review of experimental work up to 1946 is givcn in a short book by Mitra in which there are many references to original papers. The more important results are summarised below. Nitrogen can be activated by means of a condensed electrode discharge or by a high-frequency electrodeless dis- charge the latter being preferred since the gas does not become contamin- ated with electrodc materials. Active nitrogen has also been produced by bombarding the gas with electrons of energy greater than 16-3 ev,4 and Stanley has recently used an arc discharge to produce active nitrogen a t pressures as high as 20 cm.Hg. The purity of the gas appears to be important a trace (O.lyo) of an electronegative element such as oxygen enhances the glow. Absolutely pure nitrogen in a baked-out vessel appears not to give the afterglow.6 This is almost certainly due to the rapid removal of nitrogen atoms from the gaseous phase by recombination at the walls. Higher concentrations of impurities inhibit the production of the afterglow since atoms are then removed by chemical reaction. The lifetime of the afterglow is very Production of active nitrogen. Kinetics of decay of the afterglow. P. Lewis Astrophys. J . 1900 12 8. * Strutt Proc. Roy. SOC. 1911 A 85 219 ; 1911 A 86 56 ; 1913 A 88 539 ; Mitra " Active Nitrogen-A New Theory " Association for the Cultivation of 1915 A 91 303.Science Calcutta India 1945. Stanley Proc. Phys. SOC. 1954 67 A 821. 4Kenty and Turner Phys. Rev. 1928 32 799. 6 B . Lewis J . Amer. Chem. Xoc. 1929 51 564. 116 JENNINQS AND LINNETT ACTIVN NITEOGEN 117 dependent upon the condition of the walls of the containing vessel. A coating of metaphosphoric acid prolongs the lifetime of the glow consider- ably but a coating of Apiezon oil rapidly destroys it. The lifetime of the afterglow is also very sensitive to pressure the optimum pressure for the production of an afterglow of long life being between 10-1 and mm. Hg. Rayleigh showed that the rate of decay was proportional to the square of the concentration of the active species and to the concentration of the unexcited nitrogen molecules. He also showed that the process responsible for the afterglow has a negative temperature coefficient since I cc T-0'64 where T is the absolute temperature.Emission spectrum of the afterglow. Until very recently it was thought that this consisted only of selected bands of the First Positive System of the N molecule in the visible region of the spectrum which are normally observed in the positive column of an electrode discharge through nitrogen. These bands are due to a transition from a more highly-excited state (the B state) to a lower excited state (the A state) from which the molecules reach the ground state (the X state) by losing their excess of energy in collisions with other molecules and with the containing walls. Prom the intensity distribution of the vibrational bands within the system it appeared that the process leading to the formation of excited molecules in the B state never produced them with more than 12 vibrational quanta and prefer- entially produced them with 12 11 10 6 4 3 and 2 vibrational quanta.However as will be seen later it has recently been suggested that the bands which were thought to be due to transitions from the lower vibrational levels of the B state may be part of a different system involving neither +,he B nor the A state. Bands due to the NO molecule are usually present in the blue and ultra- violet regions of the spectrum owing to the presence of small quantities of oxygen or compounds containing oxygen. If a stream of glowing gas from the discharge tube is passed between two auxiliary electrodes it is found that the gas has a high electrical conductivity.On increase of the applied voltage a saturation current is obtained the magnitude of which is propor- tional to the area of the cathode.8 This suggested that electrons are emitted from the cathode owing to atomic or molecular bombardment but Ray- leigh used a hot cathode and found that the saturation current decreased rather than increased. He suggested that positive ions in the gaseous phase were responsible for the conductivity but more recent work lo indi- cates that the charged particles are electrons and that their concentration is less than 10-6 of the concentration of the active species in active nitrogen. Several workers have shown that passage of the glowing gas through an ion trap does not affect the glow making it very improbable that charged par- ticles play any part in the production of the glow.Electrical properties of the afterglow. Rayleigh (Strutt) Proc. Roy. Xoc. 1935 A 151 5 6 7 ; 1940 A 176 1. Rayleigh Proc. Roy. SOC. 1942 A 180 140. * Constantinides Phys. Rev. 1927 30 95. loBenson J. Appl. Phys. 1952 23 767. 118 QUARTERLY REVIEWS ExcitatioiL of spectra and chemical reactions. When other substances are introduced into a stream of active nitrogen a luminous zone is very often observed a t the point of mixing. The spectra excited in these glows are either those of the unchanged substance or those of radicals or molecules formed by chemical reaction. Summaries of the many systems investigated are given by Strutt,2 Strutt and Fowler,ll and Willey and Rideal.12 Metallic vapours usually give rise to spectra consisting of atomic lines of the metal accompanied by the formation of the nitride.The reaction with nitric oxide produces mainly nitrogen and oxygen l3 with some nitrogen dioxide as a by-product. Hydrocarbons react to form hydrogen cyanide as the main product the CN bands being prominent in the spectrum of the glow. Neither hydrogen nor oxygen reacts with active nitrogen. Strutt l4 observed the temperature rise when active nitrogen was destroyed on a copper oxide probe and compared it with the temperature rise recorded when the active nitrogen was made to react with nitric oxide before it reached the probe. Since these were similar he concluded that active nitrogen did not contain an abnormally high amount of energy. Willey and Rideul,12 using calorimetric methods estimated that the energy content was about 2 ev/mole of total nitrogen.Since active nitrogen does not react with hydrogen hydrogen chloride or nitrous oxide all with dissociation energies of over 60 kcal./mole they concluded that it contained insufficient energy to do so. Spectroscopic excitation up to about 9.6 ev is known but the above workers suggested that all except 2 ev of this was due to chemiluminescence in nitride formation. Rayleigh15 exposed metal foil to streams of active nitrogen and from the temperature attained by the foil calculated that the energy available in active nitrogen may be as high as 12.9 ev/mole of total nitrogen. It has since been shown however that much of this heating effect was due t o electron bombardment from the discharge so that these results are in~a1id.l~ Wrede,ls using the gauge which now bears his name detected concentrations of up to 30-40% of atoms in active nitrogen produced by a strong condensed electrode discharge.The results of a Stern-Gerlach type of experiment on active nitrogen suggested that only 2P+ atoms were present,17 but no absorption in the region of 1400-1800 A in which both 2P and 2D atoms would absorb could be found,18 suggesting that any atoms present were in the ground state. Many experiments since then have confirmed that the vast majority of atoms present in active nitrogen are in the ground state. Energy measurements. Other results. l1 Strutt and Fowler Proc. Roy. SOC. 1911 A 86 105. l2 Willey and Rideal J. 1927 6G9. l3Spealrnan and Rodebush J. Amer. Chem. Soc. 1935 57 1474. 14Strutt Proc. Roy. Xoc. 1912 A 87 179.l 5 Rayleigh ibid. 1940 A 176 16. I s Wrede 2. Physik 1929 54 53. l7 Jackson and Broadway Proc. Roy. Soc. 1930 A 127 678. l8 Herbert Herzberg and Mills Canad. J. Res. 1937 15 A 35. JENNINGS AND LINNIETT ACTIVE NITROGEN 119 Theories of active nitrogen Attempts to formulate a theory of active nitrogen have been hindered by the uncertainty and divergence of opinion on almost every experimental observation. The purity of the nitrogen and the condition of the walls are very important but not unnaturally this was not always recognised by some of the earlier workers. The importance of the electrical properties has been difficult to assess and only Mitra's theory now abandoned attempted to explain them. Energy considerations have been one of the main sources of trouble. Both the energy content of active nitrogen and the dissociation energy of c 8 P N(4S) C N(*P)+M N(4S) + N('DI+M Energy levels of molecular and atomic nitrogen relative to the energy of the molecule in its ground state.the N molecuIe have been extremely uncertain until quite recently. I n addition the number and complexity of the excited levels of the N molecule make a detailed interpretation morc difficult. Consequently it is not difficult to see why there were three totally different theories of active nitrogen in 1945. (In the following pages the different electronic levels of nitrogen atoms and mole- cules are referred to by their spectroscopic designations. Although these are important in deciding whether or not a particular transition or reaction may occur they may be looked upon for the purposes of this Review as mere labels for the different electronic levels.The energies of these states relative to the ground states are given in Pig. 1. In discussion of vibrational levels v' indicates the vibrational level of the upper electronic state v" that of the lower state.) These are outlined below. 120 QUARTERLY REVIEWS The atomic theory. Rayleigh first suggested that active nitrogen con- tained nitrogen atoms and Sponer l9 later suggested a two-step mechanism M + N + N + M + N,** -3 N,* + hv (afterglow) Since three-body collisions are rare in the gaseous phase a t low pressures this could explain the long life of the glow and it would also explain the kinetics of the decay. Now that the dissociation energy of the N molecule is known to be 9.76 ev,20 this theory can satisfy all energy observations (except Rayleigh’s invalid metal-foil experiments).The work of Wrede 16 and of Herbert Herzberg and Mills 17 suggests that active nitrogen con- tains appreciable quantities of ground-state atoms thus supporting this theory. However it does not explain the selective enhancement of certain vibrational bands in the spectrum of the afterglow nor does it offer any explanation of the electrical properties of active nitrogen. This theory 21 suggests that active nitrogen is it mixture of metastable atoms and molecules which produce the after- glow as follows N(,P) + N,(A3C,-b) + N,(B3ny) + N(4S) Curio and Kaplan’s theory. N,(B3ng) -+ Nz(A3C,+) 4- hv This can explain the selective enhancement of certain vibrational bands since the A state is about 6.2 ev and the 2P state 3-56 ev above the respective ground states and between them they can provide 9.76 ev which is just enough to produce molecules in the twelfth vibrational level of the B state.This theory was preferred to the atomic theory when the dissociation energy of N was thought to be 7.37 ev and was apparently supported by the results of the Stern-Gerlach experiment l7 but not by the vacuum-ultra- violet absorption spectrum.18 The absence from the afterglow of the Vegard-Kaplan bands due to the transition A3&+ + XI&+ suggests that there is no appreciable stationary concentration of molecules in the A state as would be required by this theory. Mitra’s theory. In 1945 Mitra suggested that active nitrogen consisted of a mixture of N2+ ions and electrons which would readily explain the electrical conductivity of the gas.The afterglow would be produced by the reactions N,+ + e -t N -+ N,(B3ng) + N,(A3C,+) N2(B3n,) -+ N,(A3&+) + hw Since this required a three-body collision i t would explain the long life of the afterglow but it is unsatisfactory from the point of view of energy. The reaction N2+ + e can liberate only 15.58 ev and 15-85 ev are required to produce one nitrogen molecule in the twelfth vibrational level of the B state and one in the zeroth vibrational level of the A state. Mitra suggests that the energy deficiency of 0.27 ev may be derived from kinetic energy l9 Sponer 2. PhysiE 1925 34 622. 2o Gaydon “ Dissociation Energies ” Chapman and Hall London 1st edn. 1947 ; 21Cario and Kaplan 2. Physik 1939 58 769. 2nd edn. 1953. JENNINGS AND LINNETT ACTIVE NITROGEN 121 but this seems an improbable sourcc of so much energy (6 kcal./mole).The theory can clearly account for energy contents of up t o about 154 ev/mole and so explain Rayleigh’s metal-foil experiments. Chemical reactions were formulated in terms such as X + e -+ X- followed by N2+ + X- -+ products. However for reasons which are explained later Mitra abandoned his theory in 1953. Sincc 1045 use of new and more refined experimental methods has made possible the elucidation of many of the problems concerning the nature of active nitrogen. Although a theory which will explain all the experimental facts has not yet been formulated it will be seen that a modification of the simple atomic recombination theory accounts for most of them. As early as 1947 Gaydon 2O pointed out that the results of Rayleigh’s metal-foil experiments could probably be explained by assuming that the test surfaces were heated by cathode rays from the discharge as well as by active nitrogen.He considered that Mitra’s theory was untenable since it does not explain why an electric field does not quench the glow and as explained above there was an energy deficiency of 0.27 ev. In 1948 Worley 22 failed to find any absorption in the visible region in the glowing gas using a path length of 13 metres. If this theory were correct absorption by ground-state N2+ ions should occur in this region. Strong evidence against this theory was obtaincd by Benson in 1952.lO He repeated Rayleigh’s metal-foil experiments but placed an earthed aluminium tube bent at a right-angle between the discharge and the metal foil so as to remove electrons coming from the discharge.He was unable to reproduce Rnyleigh’s results obtaining an energy content of only 1/400 of that obtained by Rayleigh. He also showed that the intensity of the glow was unaffected by the removal of charged particles and that a beam of the glowing gas was not affected by a magnetic field. The concentration of electrons in the afterglow was found to be only about 109/c.c. compared with a concentration of active particles of about 1015/c.c. As a result of this it became clear that active nitrogen did not consist of a mixture of N2+ and electrons and the theory was abandoned. This theory requires that the glowing gas should contain appreciable concentrations of excited atoms in the 2P and 2D states and excited molecules in the A state.Worley 22 failed t o find any absorption due to transitions from the A state of the nitrogen molecule suggesting a very low concentration of molecules in this state. The lifetime of the A state with respect to radiation was estimated by Muschlitz and Goodman 23 to be about see. but Lichten 24 considers that they were producing molecules in the aln state and that the true lifetime of the A state molecules is probably as high as 10-1 sec. The lifetime is nevertheless very short if these molecules are to account for an 2 2 Worley Phys. Rev. 1948 73 531. 23Muschlitz and Goodman J . Chem. Phys. 1953 21 2213. 2* Lichten ibid. 1957 26 306. Evidence q a i n s t Mitra’s theory. Evidence against Curio and Kaplan’s theory. 122 QUARTERLY REVlEWS afterglow which may last for several hours.7 The evidence supporting the presence of *X atoms rather 2P or 2D atoms is summarised in the following section.It became unnecessary to postulate the presence of excited atoms when the higher dissociation energy (9.76 ev) for N was accepted during the present decade. Evidence supporting the Presence of Ground-state Atoms in the After- glow.-The Wrede gauge showed the presence of appreciable concentrations of atoms in the afterglow,l6 but the gauge does not distinguish between ground-state and excited atoms. Whereas the Stern-Gerlach experiment l7 suggested that atoms in the ground state were absent from the afterglow the vacuum-ultraviolet absorption spectrum 18 suggested that excited atoms were absent from the afterglow. The first positive evidence of the presence of ground-state atoms in the afterglow was obtained by examining the paramagnetic resonance spectrum of active nitrogen.25 This indicated the presence of 4X atoms only but concentrations of excited atoms of up to 1% of the total atom concentration would not be detected by this method.More recently two m’ass-spectrometric studies of active nitrogen have been made.26 27 In each case the appearance potential of the peak for mass 14 was found to be about 14.8 v very close to the ionisation potential 28 of the nitrogen atom in the ground state (14.545 v). Again small concen- trations of excited atoms would not be detected by this method. An appear- ance potential in the region of 16.1 v was found in the earlier investigation 26 but no satisfactory explanation was given.It has been suggested 29 that this appearance potential is due to the presence of ground-state molecules with about 8 ev of vibrational energy. When 2% of nitrous oxide is added to a stream of the glowing gas a considerable temperature rise occurs a t the point of mixing. Because of the negative temperature coefficient of the process responsible for the afterglow there is an appreciable decrease in the intensity of the afterglow immediately after the point of mixing but the intensity actually increases farther down the tube. This suggests that the temperature increase is not due to a reaction of nitrogen atoms with the nitrous oxide especially since no nitric oxide or oxygen are formed but is caused by the deactivation of vibrationally excited ground-state nitrogen m0lecules.29~ The vacuum-ultraviolet absorption spectrum of active nitrogen has been re-examined re~ently.~O There was a very strong absorption line a t 1200 8 indicating the presence of an appreciable concentration of 4S ground-state atoms.Much weaker lines were found a t 1493 and 1743 8 indicating much lower concentrations of nitrogen atoms in the ,D and the 2P state. 26Heald and Beringer Phys. Rev. 1954 96 645. 26 Jackson and Schiff J. Chem. Phys. 1955 23 2333. 27 Berkowitz Chupka and Kistiakowsky ibid. 1956 25 457. 28 Herzberg “ Atomic Spectra and Atomic Structure ” Dover Publications 2nd 290Kaufman and Kelso J. Chem. Phys. 1958 28 510. 30 Tanaka Chem. Aeronomy Conference Cambridge Mass. 1956. edn. 1944. 29 Evans and Winkler Canad. J. Chem. 1956 34 1217. Proceedings to be published by Pergamon Press.JENNINGS AND LIVNETT ACTIVE NITROGEN 123 From the above evidence it is clear that the only species present in active nitrogen in appreciable concentrations are atoms and molecules in the ground state. Recent work on the afterglow Until very recently the spectrum of the afterglow was thought to consist entirely of selected bands of the First Positive System of nitrogen due to the transition B3n -+ A3C,+. The enhanced bands appeared to originate mainly in the twelfth eleventh and tenth vibrational levels of the B state but bands originating in the sixth fourth third and second vibrational levels appeared to be enhanced to a smaller extent. The latter bands lie predominantly in the photographic infrared region and this together with 28 - 24 - n E20- *$ *b 72 - ./=3 C 3 16 - 0 * & B 0 2 4 6 8 10 72 74 16 78 2 lo 3x Vibrational energy (cm.''I 0 14N 14N 15N15N F I G . 2 Apparent distribution of molecules (at 200") in the digwent vibrational levels of the B311g state a s indicated by the intensity distribution in the afterglow o n the supposition that all bands in the visible and the photographic infrared region are part of the First Positive System. [bfter Kistiakowsky and Warneck rcf. 311 their inherent weakness has made the accurate measurement of their wave- lengths difficult. However it has recently been suggested that the weaker bands are not in fact members of the First Positive System but form a new system arising from transitions between unknown states of the nitrogen molecule.31 By summing intensities of bands originating in the same vibrational level of the B state an estimate of population distributioii for l4NI4N and 15N15N was obtained a t 300° 200" (Fig.2) and 100" K and after dilution with helium. 31 Kistiakowsky and lyarneck J . Chein. PIqs. 1957 27 1417. 124 QUARTERLY REVIEWS The relative band intensities were independent of pressure in the range 2-16 mm. but varied with changes in temperature and addenda. The population curves obtained for each isotope coincided for the v’ = 12 11 and 10 peak but there was some disagreement a t lower u’ values. The 12-11-10 peak became broader and moved to slightly lower v’ values as the temperature was raised or helium was added. This indicates that vibrational energy is lost during the process leading to the formation of B state mole- cules in these levels.The lower peaks behave differently however and the evidence that they were not part of the First Positive System has been presented in ref. 31 as follows (1) The population distributions of the two isotopes disagree a t lower w’ levels to an extent which depends upon the conditions. (2) The ratio of intensities of bands originating in the same w’ level (on the above assignment) varies with conditions. (3) The spectrum includes a t least three bands which fit poorly into the First Positive System. These are a t approximately 6934 7823 and 8949& the nearest bands of the First Positive System being a t 6968 7896 and 8926A approximately. (4) The isotopic shift observed is inconsistent with that expected from the First Positive System assignment the shift in the supposed (0,O) band being 130 cm.-l.Whereas the First Positive System is approximately represented by Y = 9520 + 1733(v’ + 3) - 1460(v” + i) the observed bands are better represented by Y = 6050 + 1376(v’ + 8 ) - 1630(v” + 8). Hence neither the B nor the A state appears to be involved. The ratio of the total intensities in the two parts of the population- distribution curve does not vary with pressure and varies only slightly with temperature. It has been suggested that this indicates that the molecules in each of the upper states involved in these transitions are formed from nitrogen atoms via a common intermediate (see next section). I n addition to the above bands bands of the Lyman-Birge-Hopfield System due to the transition alTl[,-+XIX,+ have recently been observed weakly in the vacuum ultraviolet region.28 No bands due to transitions from levels higher than v’ = 6 are observed.A combined mass-spectrometric and photometric study of the after- glow 27 showed that the intensity of the afterglow was proportional to the square of the concentration of nitrogen atoms in the ground state. This was shown to be true for bands originating in the v’ = 11 level of the B state and also for bands which were thought to originate in the v’ = 6 level but which it is now suggested belong to a different system. The ratio of intensity to the square of the atomic concentration did not alter when oxygen was added to the system indicating that oxygen plays no part in the pro- duction of the afterglow. The ratio increased when helium or argon was added to the nitrogen indicating that these gases were more efficient than nitrogen as third bodies for the recombination of nitrogen atoms.Explanation of the afterglow The main features of the afterglow can be summarised as follows (a) The process giving rise to the afterglow is of the second order with respect to nitrogen atoms and of first order with respect to nitrogen mole- JENNINGS AND LINNETT ACTIVE NITROGEN 125 cules. ( c ) The spectrum of the afterglow consists of two or three band systems. The first two observations are readily explained by assuming that the overall process is of the type N + N + M -+ products. From the observed spectrum it appears that the products are N molecules in the alll, B31-II, and Y states where the Y state is the upper state involved in the proposed new band system.The original atomic theory assumed that the above process occurred in one step giving N molecules in the B state but this theory is incapable of explaining the formation of N molecules in three different electronic states. The possible products arising from the collision of two 4S nitrogen (b) The process has a negative temperature coefficient. FIG. 3 Potential-energy curves for some states of the nityogen molecule which are important in the consideration of the afterglow. atoms must now be considered in some detail The 4 8 atom has the elec- tronic configuration 1s22s22p,2py2pz the three unpaired p electrons having parallel spins. Two such atoms may approach each other along four differ- ent potential-energy paths corresponding to the pairing of 6 4 2 and 0 electrons. The spectroscopic designations of the states of the N molecules formed in following these four paths are respectively lCg+ 3Cu+ 5Zg+ and 7C2(+ and the relative probabilities of these paths’ being followed are in the ratio 1 3 5 7.Whichever path the two atoms follow they will fly apart again at the end of the first molecular vibration unless a collision with a third body prevents this. The probability of the molecule’s losing its excess of energy by radiation is vanishingly small. If the atoms follow paths 1 or 2 a collision with a third body would remove the excess of energy from the newly-formed N molecules thereby stabilising them in the X and A states. The respective paths are labelled 1 2 3 and 4 in Fig. 3 I 126 QUARTERLY REVIEWS On the other hand if the atoms follow path 4 they will fly apart again even in the presence of a third body since no bond is formed between the atoms.The fourth possibility is that the atoms would follow path 3 and form a very weakly-bound N molecule in the 5Zg+ state. Now it will be seen from Fig. 3 that the potential-energy curve of this state crosses that of the B state. The suggestion that these curves cross was first made by Gaydon 32 to explain the predissociation observed in the First Positive System of nitrogen due to the transition B -+ A . Bands due to transitions froin the v’ = 13 to the v’ = 16 level have only one head instead of the usual four and have no fine structure but for v’ = 12 or below or 17 or above the bands have the normal structure. This observation is explained by assuming that the molecules in the levels v’ = 13-16 of the B state can undergo a collision-induced radiationless transition to the ?Zg+ state which would then dissociate into 4X atoms.For a full account of this phenomenon see ref. 20. I n this predissociation the sequence of events can be represented by B(v’ = 13 14 15 16) -+ 5C,+ molecules -+ two 4X atoms. It is clear that the reverse of these processes enables 4 S atoms to produce molecules in the B state and in the nitrogen afterglow it is postulated that two 4X atoms collide on the 5Zg+ potential-energy curve after which a collision with a third body induces a preassociation into the twelfth vibrational level of the B state. A similar phenomenon is thought to explain the spectrum of the AlH molecule.20 Emission from the B state constitutes part of the after- glow and satisfactorily accounts for the cut-off of the v’ levels above 12 and for the enhanced intensity of the bands originating in the v’ = 12 level.The effect of temperature and pressure on the intensity distribution of the afterglow spectrum 5 31 indicates that the preassociation may occur with loss of vibrational energy thus accounting for the enhancement of the v’ = 11 and 10 bands in the spectrum of the afterglow a t room temperatures. Stanley,33 by measuring the intensity distribution a t pressures between 3 cm. and 20 em. has been able to calculate that the relative probabilities of loss by molecules in the eleventh vibrational level of the B-state of one quantum of vibrational energy by collision and energy by radiation are in the ratio lop2 1 per cm. of mercury pressure. At pressures of about 20 cm.fairly strong bands are observed owing to transitions from most vibrational levels below the twelfth. The observation of the Lynian-Birge-Hopfield bands in the afterglow 30 can be explained similarly. A strongly forbidden predissociation occurs in the v‘ = 6 level of the alIII state and Gaydon 2o has suggested that this is due to the process alII -+ 5Cg+ -+ two 4X atoms i.e. a process entirely analogous to that postulated above. Clearly the reverse of thib process can give rise to the Lyman-Birge-Hopfield bands in emission and can account for the fact that no bands were observed originating in v‘ levels above the sixth. The analogy being carried a stage further it seems very probable that 32Gaydon Nature 1944 153 407. 33 Stanley Proc. Roy. Soc. 1957 A 241.180. JENNINGS AND LINNETT ACTIVE NITROGEN 127 the new system could arise in a similar way. The ratio of the total inten- sities of these bands to those of the First Positive System in the afterglow is approximately constant under all condition^,^^ suggesting that both B-state and Y-state molecules are formed from the 5&+-state molecules by radiation- less collision-induced transitions these processes being in competition with each other and with the process forming a*IT molecules. The theory of the afterglow spectrum can be summarised schematically a w g + xlcg+ 7 (Lyman-Rirge-Hopfield Bands) (First Positive Bands) (Proposed New Bands) N(4S) + N(4S) 4- M + N,(5Cg+) + B311f + A3C,+ Y-state + 2-state The essential correctness of this theory is supported by two further pieces of work.First the existence of the 5.C,+ state has been shown by work on active nitrogen at very low temperatures (see next section). The binding energy of the state is 0.13 ev. Secondly the non-appearance of bands from the v' = 13 level of the B-state when I5Nl5N was used34 can readily be explained by the theory. This level lies 0.024 ev below the dissociation energy of nitrogen and so if the B-state molecules were formed in a single stage some of them would be formed in the 'u' = 13 level. On the other hand this level lies 0.1 ev above the zeroth vibrational level of the 5C,+ state and this amount of energy would have to be derived from kinetic energy if molecules in the thirteenth vibrational level of the B state of the 15W5N molecule were to be formed by a collision-induced radiationless transition from the 5Cg+ state.There is overwhelming evidence in favour of the preassociation theory. Further work is desirable on the photographic infrared spectrum of the afterglow to clarify the position with regard to the proposed new system and in particular to obtain more precise information about the states involved. However it seems unlikely that extensive modifications of the preassociation theory will be necessary. Active nitrogen at very low temperatures During the last few years it has been possible to freeze out solids believed to contain atoms from gases which have been passed through a discharge t ~ b e . ~ 5 The gases were passed through a discharge tube a t pressures of about 1 mm. and were activated by a 2450 Mc./s generator. They were then led into a specially-constructed Dewar vessel containing liquid helium a t 4.2" K.When nitrogen was subjected to this treatment most spectacular results were obtained. While the discharge is maintained the solid which condenses in the Dewar vessel emits a bright green glow which becomes yellow-green a t higher flow rates. Brilliant blue flashes are observed on the surface of the 3 4 Kistiakowsky and Wsmeck J . Chem. Phys. in the press. 35 (a) Broida Ann. New York Acad. Sci. 1957 6'7 530 ; ( b ) Herzfeld and Broida P h p . Rev. 1956 101 606; (c) Herzfeld ibid. 1957 107 1239. 128 QUARTERLY REVIEWS vessel which are thought to be due to local warming. If the flow of nitrogen is stopped and the discharge turned off a green afterglow persists for some minutes. When this has died away a blue '' flame " is produced by warm- ing the solid to about 35" K and after this glow has disappeared a much weaker green glow can be obtained by cooling the solid to 4.2" K again.Coloured photographs of these glows are reproduced in ref. 35a. The spectra of these glows have been studied in detail in the region 2200-9000 A the main features being (a) Five blue-green lines in the region 5214-5240 A known as the cc-lines ; ( b ) three diffuse yellow-green lines a t 5549 5616 and 5657 A known as /3-lines ; (c) ten bands stretching from 3572 to 6390 A7 known as A-bands ; ( d ) thirty bands of the Vegard- Kaplan system of nitrogen in the region 2320-4450A.36 These are very sharp and are thought to be due to the transition 2D + 4S of the nitrogen atom. In the gaseous state this transi- tion is strongly forbidden but a close doublet due to it has been observed in the auroral spectrum at 5199A.37 The displacement of the or-lines towards the red end of the spectrum the appearance of 5 lines rather than 2 and the much higher intensity of the lines in the solid have been discussed in terms of crystal-field effects on the nitrogen atom.These are diffuse and appear only when the discharge is on and disappear with no detectable time-lag when the discharge is switched off. Since the transit time from the discharge tube to the trap was only see. the lifetime of the upper state involved in this transition is of that order. As traces of oxygen up to about 1% are added to the very pure nitrogen these lines are enhanced relative to the a-line~,~* and they are thought to be due to the 1S,-+1D2 transition of the oxygen atom.A line due to this transition is observed in the auroral spectrum a t 5577 A. It was a t first thought that these bands were emitted in the blue flashes observed on warming the solid to 35" K ~ ~ but it now seems more probable that they are emitted in the blue flashes observed when the active nitrogen is deposited in the t r a ~ . ~ 6 The bands are attributed to the transition 5Cg+ + A3&+ and are the only bands known which involve the 5Zg+ state. They appear with very little rotational structure since free molecular rotation is very difficult in the solid a t low temperatures. In this experiment the molecules in the 5Cg+ state are formed either in the discharge or from 4X atoms in the trap. An analysis of the bands has provided considerable information about the 5Zg+ state which has been of the utmost value in putting the pre- association theory of active nitrogen on a firmer basis.If we assume that D,(N,) = 9-76 ev then D,(N, 5Cg+) = 0-13 ev i.e. this is a weakly-bound state. The vibrational frequency is only 12.1 cm.-l and the internuclear distance aboui; 1.5 A compared with 1-21 and 1-29 A for the B and the A state respectively. Transitions have been observed from the V' = 0 1 and 2 levels with the intensity maximum a t V" = 4. The a-lines. The p-lines. The A-Bands. 36Peyron and Broida J . Physique 1057 18 593. 37 Benard Ann. Geophys. 1947 3 63. 38Peyron and Broida J . Physique in the press. JENNINCS AND LINNETT ACTIVE NITROGEN 120 The Vegard-Kaplan bands. These bands are observed if very pure nitrogen is used and are enhanced in the presence of argon.36 If as little as 0.01 % of oxygen is present the bands are not observed.They become very intense on warming the solid suggesting that the A-state molecules may be formed from 4 8 atoms in the solid. Further evidence of the presence of nitrogen atoms j n the solid comes from the large and rapid increase in temperature which is observed when the solid is allowed to warm. If it is assumed that the heat released is due to the recombination of nitrogen atoms atom concentrations of up to 3% have been recorded.39 Chemical reactions of active nitrogen Since 1849 Winkler and his co-workers have systematically studied the reactions of active nitrogen with a variety of hydrocarbons and their derivative^.^^ Atoms are produced in a high-voltage condensed electrode discharge and concentrations of about 30% are indicated by Wrede-gauge measurements.All products of thc reaction are frozen out and aiialysed. The experiments are carried out at pressures of the order of 1 mm. Hg and different relative flow rates are used in different experiments. The reaction zone is marked by a lilac flame in the case of hydrocarbons but by an orange flame in the case of halogen compounds. The main product in these reactions is hydrogen cyanide and the yield of this rises linearly with increase in hydrocarbon flow rate finally reaching a maximum whose value depends on the number of nitrogen atoms reaching the reaction zone in unit time. The maximum corresponds to the complete removal of atoms by the hydrocarbon. A troublesome feature of these reactions is the formation of appreciable amounts of polymer in the traps and on the walls of the reaction vessel.In the case of ethylene the elimina- tion of moisture prevented the formation of polymer and when both the nitrogen atoms and ethylene were just completely consumed the products of the reaction were hydrogen cyanide 75 ; ethane 10 ; methane 9 ; acetyl- me 3 ; and cyanogen (CN), 2%. Higher olefins give hydrogen cyanide as the main product with lower olefins as the principal side-products. Satur- ated hydrocarbons are much less reactive than olefins hydrogen cyanide being the only product with methane and ethane at room temperature. Any olefins formed in these reactions are more rapidly attacked by the nitrogen atoms and so are not found as products. Alkyl chlorides resemble olehs in their reactivity the main products being hydrogen cyanide and hydrogen chloride and smaller quantities of nlefins and a polymer containing carbon hydrogen nitrogen and chlorine.No chlorine cyanogen chloride or methane was obtained from any alkyl chloride. It is noteworthy that no detectable quantities of ammonia or hydrazine are formed. Because of this it is postulated that the nitrogen atom alone 39 Minkoff and Scherber J . Chem. Phys. in the press. 4O Evans Freeman and Winkler Canad. J . Chew. 1956 34 1271 ; Dunford Evans and Winkler ibid. p. 1074. 130 QUARTERLY REVIEWS of the electronegative atoms does not attack hydrocarbons by hydrogen- atom abstraction but rather by a direct approach to the carbon atom. The only reaction with ethylene which is favoured energetically is N + C,H --f HCN + CH + 62 kcal.This reaction however involves a change of spin from 3 unpaired electrons in the nitrogen atom to 1 unpaired electron in the methyl radical and the transfer of a hydrogen atom from one carbon atom to the other. It is therefore suggested that the first step in this reaction is the formation of a C2H4N complex of long life which can then rearrange to form hydrogen cyanide and a methyl radical. The methyl radicals are removed by the reaction N+CH,* + HCN+ZH* This reaction is preferred to that giving H on grounds of spin conservation. If the complex reacts with a further nitrogen atom the most likely reaction is the formation of N, the ethylene acting as a third body for atomic recombination. The energy acquired by the ethylene molecule will cause it to be very reactive and it is possible that it dissociates into acetylene and hydrogen.Similar reaction schemes are postulated for the reactions of paraffins and alkyl chlorides the formation of the complex being accompanied by the expulsion of hydrogen and hydrogen chloride respectively. The differences in energy content and ease of formation of the complexes may account for slight differences in products. The main features of the spectra of these flames are bands due to CN CH C, and NH radicals and also due to CCl radicals in the case of alkyl chlorides. The reaction of nitric oxide with active nitrogcn has been investigated twice recently. When 15N0 is added to active nitrogen only 14N0 bands are observedY4l indicating that the excited NO molecules are formed by chemical reaction and not by the excitation of the original NO molecules.The effect of nitric oxide on the colour of the afterglow is explained as follows. The addition of a little of it merely weakens the afterglow by rapidly removing nitrogen atoms N + NO -+ N + 0. Excited NO molecules may then be formed by the relatively slow combination of a nitrogen atom and an oxygen atom in the presence of a third body and the emission from these molecules accounts for the range of colours yellow- pink-blue. The addition of more nitric oxide removes all the nitrogen atoms by the above reaction so that none is left to form excited NO mole- cules and the glow is extinguished. Excess of nitric oxide produces the yellow-green NO + 0 continuum of the air afterglow since fairly high concentrations of oxygen atoms and nitric oxide are present together.The presence of appreciable concentrations of oxygen atoms in the colourless and air aft,erglows has been confirmed by the observation of oxygen atomic lines in absorption in the vacuum-ultraviolet region.30 ‘lKaufman and Kelso J . Chern. Phys. 1957 27 1209. JENNINGS AND LINNETT ACTIVE NITROGEN 131 The reaction of nitric oxide with nitrogen atoms has been studied quanti- tatively by sampling the effluent gases from a reaction vessel with a mass spe~trometer.~~ A plot of the concentration of inflowing nitric oxide against concentrations recorded by the mass spectrometer is shown in Fig. 4. The NO in input(%) FIG. 4 Jhriation of percentages of ( a ) atomic nitrogen (uncorrected) ( b ) nitric oxide ( x lo-]) and ( c ) oxygen as measured by a wmss spectrometer when nitric oxide is added to active nitrogen.[After Kistialtowsky and Volpi ref. 421 rapidity of the reaction is indicated by the sharp break in the concentration curve a-b (cf. conductometric titration) and this enables one to use nitric oxide as a titrating agent for nitrogen atoms. The concentration of nitrogen atoms recorded by the mass spectrometer is estimated to be only about i t h of the true value owing to removal of the atoms a t the walls. This probably also explains the non-stoicheiometric relationship between nitrogen atoms and nitric oxide. A similar titration-like plot was obtained with nitrogen dioxide but the varia,tion of concentrations of oxygen and nitrous oxide with inflowing nitrogen dioxide concentration indicates that the reaction is very complex.Conclusion The achievements of the last ten years' work can be summarised as follows (a) Charged particles play no part in the production of the after- glow. ( b ) The only species present in active nitrogen in appreciable con- centrations are atoms and molecules in the ground state; the former are responsible for the' chemical reactivity. ( c ) The preassociation theory succeeds in explaining experimental observations and is able to account for t,he different band systems in the afterglow. It is reasonable to enquire why nitrogen but neither hydrogen nor )xygen has an afterglow which may last for several hours. Since the iitrogen atom has three unpaired electrons it can give rise to a variety of 42 Kistiakowsky and Volpi J . Chem. Phys. 1957 27 1141.132 QUARTERLY REVIEWS nitrogen molecules by combining with another atom in the ground or the excited states By coincidence the shallow 5&,+ curve crosses other curves allowing preassociation to occur. Since the hydrogen atom has only one electron two such atoms meet either on the X1&+ curve forming a ground- state molecule in the presence of a third body or they meet on the purely repulsive 3&+ curve. No state corresponding to the weakly-bound 5Z:s+ state of nitrogen can be formed. In the case of oxygen the 3P atoms can give rise to several low-lying states the highest bound state of which is the A3C,+ state which has a dissociation energy of about 0.48 ev. This is somewhat analogous to the 5X,+ state of nitrogen but as far as is known no other potential-energy curves cross that of the A3C,+ state so that no process similar to pre- association can occur. The A state of the oxygen molecule is much more stable with respect t o dissociation than is the 5&+- state of nitrogen so that molecules in this state survive long enough to emit the Herzberg bands (A3&+ -+ X3Zg-) in the comparatively short-lived oxygen afterglow.43 43 Broida and Gaydon Proc. Roy. SOC. 1954 A 222 181.

ISSN:0009-2681    

DOI:10.1039/QR9581200116

出版商:RSC    

年代:1958

数据来源: RSC

摘要:

QUANTITATIVE NUCLEAR CHEMISTRY By D. L. RAULCH and J. F. DUNCAN (CHEMISTI~Y I~EPARTMENT UNIVERSITY MELBOURNE AUSTRALL4) (1) Introduction.-The secondary chemical effects of radioactive pro- cesses have been studied frequently during the past decade and the qualita- tive aspects of valency change bond rupture etc. during radioactive decay have been the subject of several recent reviews.1 We will not discuss this work but will review the quantitative aspects of radioactivity in so far as they may be of use in obtaining physicochemical data about chemical reactions. First however some recapitulation of the nature of the nucleus and of nuclear reactions is necessary. (2) Nuclear Models.-There are two theories of nuclear structure in current use.2 I n the liquid-drop model one imagines the nucleus to be like a drop of water from which particles (nucleons) can evaporate when the nucleus is excited.According to this theory the relative energies of nuclei can be represented by a parabolic equation the parameters of which vary systematically with nuclear charge. There are some variations in the region of “magic number’’ nuclei which are much more stable than expected on the liquid-drop model. This suggests a periodic structure of some kind for the nucleus which is the basis of the alternative “ shell ” model. The evidence for the existence of nuclear shells at 2 8 20 28 50 82 and 126 neutrons or protons is now very strong.3 They are analogous to the closed electron shells exhibited by the rare gases. Like the electrons in extranuclear orbitals the nucleons can be excited from one energy level to another by electromagnetic radiation.Emission of nuclear particles may also cause changes in nuclear excitations. The energy level of the product nucleus is characterised by quantum numbers (different from those of the parent nucleus) just as for the excited electronic states of atoms. Nuclear energy levels corresponding to these states have been determined for a large number of nuclei. (3) Nuclear Decay.-The several modes of radioactive decay are well known. Here we summarise the important features for our purpose. (a) a-Decay. This occurs when the energy level of the nucleus is very high by comparison with the minimum of the potential energy parabola (liquid-drop model) for the same value of 2. At lower energies decay by other modes is usually more favourable.The most important aspects of Willard Ann. Rev. Nuclear Sci. 1953 3 193 ; Ann. Rev. Phys. Chem. 1955 6 141 ; McKay Progr. Nuclear Physics 1950 1 168. 2 Coryell Ann. Rev. Nuclear Xci. 1953 2 305 ; Halliday “ Introductory Nuclear Physics ” Wiley N.Y. 1950 287 ; Blatt and Weisskopf “ Theoretical Nuclear Physics ” Wiley N.Y. 1952 Chapters 6 7 and 14. a Flowers Progr. Nuclear Physics 1952 2 235 ; Mayer and Jensen ‘‘ Elementary Theory of Nuclear Shell Structure ” Wiley N.Y. 1955. 133 134 QUARTERLY REVIEWS a-decay are that (i) except for an unimportant fine structure the decay energy of an a-emitter is monochromatic and (ii) the emitted a-particle (;He) is an atomic nucleus of large mass by comparison with the electron and a significant fraction of the mass of the original radioactive nucleus.These features lead to a large recoil energy and complete disruption of any chemical compound of which the radioactive atom is part. Light particles may participate in the following types of transition (i) direct particle emission from the nucleus (ii) internal conversion of a y-ray (iii) K-capture and (iv) positron-electron annihilation. The last three result in loss of planetary electrons emission of X-rays from orbital electron transitions and in Auger electrons. They are monoenergetic transitions but unfortunately the associated electronic effects are usually too numerous to allow any simple chemical interpretation. From the present viewpoint they are unimportant. Decay by P-emission is by far the most important for our purpose. A noticeable feature is that the energies of the emitted particles are not mono- chromatic but spread over a range from zero to Em,,.the maximum P-particle energy equal to the difference between the energies of the product and the parent nuclei. For the law of conservation of energy to hold for /3-decay we must postulate the existence of another particle called the neutrino of almost zero rest mass and zero charge. I n each nuclear transi- tion this particle takes away the “missing ” energy. On account of its low mass the neutrino velocity is very high which together with its lack of charge accounts for the fact that its energy has never been directly measured. It is only very recently that this particle has been detected with any certainty.4 The existence of the neutrino is also expected in positron (P+) decay and in K-capture.I n the last case it is monoenergetic. The quantitative aspects of ,!?-decay are predicted by Permi’s theory,5 in which the distribution function for the shape of the @-particle spectrum is estimated by regarding the electrons and neutrinos as the quantisation of the nuclear force field i.e. the nucleus is regarded as made up of “ nucleons ” which are converted into P- or other emitted particles as a result of the decay of the nucleus from a higher to a lower quantum level (shell model). By using this theory the shape of the P-spectrum can be accurately predicted once the value of Em,,. and the degree of forbiddenness are known. This is very important in a study of the chemical effects of nuclear recoil since the vector sum of the momenta of the neutrino and the recoiling atom must equal the momentum of the emitted P-particle.I n section (4) below the shapes of some recoil spectra calculated from Fermi theory are quoted. The various nuclear processes in which y-ray decay can occur are cosmic processes (> 10 Mev) nuclear reactions induced by projectile bombardment (4-10 Mev) By-decay (0-3 MeV) and isomeric transitions (< 1 MeV). Direct interaction between the y-radiation 4Reines and Cowan Phys. Rev. 1953 92 830. ti Fermi 2. Physik 1934 88 161 ; Konopinski Rev. Mod. Phys. 1943 15 209 (b) Light-particle emission. (c) Emission of y-radiation. Skyrme Progr. Nuclear Physics 1950 1 115. BAULCH AND DUNCAN QUANTITATIVE NUCLEAR CHEMISTRY 135 and orbital electrons usually predominates over other chemical effects ; and in particular bond rupture by y-recoil (see below) is small in most cases of natural radioactivity.In the present context interest in y-decay is mainly confined to its subsidiary role on py decay in which the /3-particle is the more significant. TABLE 1. Types of y-decay Process Cosmic processes high-energy Low-energy nuclear bombard- decay men ts p y -Decay Isomeric transitions 10 4-10 0-3 1 Complete rupture of all chemical bonds Rupture by recoil frequent but not obtained in low-energy y-emission or when successive y-quanta are emitted in oriented directions rela- tive t o nuclear dipole Probability of y-emission depends on vector change in nuclear angular momentum and on quantum num- bers (shell model) N or 2 39-49 or 69-81 in which last nucleon in nearly completed shell fills a different level from that expected.Large spin change neces- sary. Internal conversion often obtained (4) Fundamental Considerations in the Chemistry of Radioactive Pro- cesses.-The physical aspects of radioactivity have now been given in sufficient detail for our purpose. We next consider how the radioactive process itself can affect the chemistry of the emitting atom. When a nucleus emits radiation the following types of extranuclear process are possible (a) Direct interaction between the emitted particle and the atomic electrons. The nuclear particle might " knock-out " some orbital electrons as it passes through them. This is more likely with a-emission than in @-emission but in both cases there is general agreement 6 7 that other effects (b and c below) are more important by a factor of the order of 1000.I n y-emission interaction with the orbital electrons by Compton scattering,* internal conversion etc. can be very efficient but these interactions are of limited interest in the present context. The sudden change in nuclear charge on emission of a nuclear particle may result in considerable disturbance of the atomic electrons. Thus on #l-emission from an atom of atomic number 2 the electrons must take up new equilibrium positions which are nearer the nucleus of the daughter atom (charge 2 + I) than in the parent (charge 2). The atomic electrons will still have the same quantum numbers but they will have different energies. This energy (b) Disturbance of the atomic electron shells. BMigdal J . Phye. (U.S.S.R.) 1941 4 449 ; Feinberg ibid. p. 423. 8 Compton Bull. Nut. Reseurch Council 1922 4 No.2. Winther Kgl. danske Videnekab. Selskab 1952 27 No. 2. 136 QUARTERLY REVIEWS difference may be accounted for in one of two ways. First if the speed of the nuclear particle is slow by comparison with the speed of the orbital electrons the energy difference may be transferred to the emitted particle. This (isothermal) process only takes place in p-emission with energies of less than about 4 kev for an element such as argon. On the other hand if the nuclear particle is so fast that the nuclear process is essentially adiabatic the daughter atom must take up the excess of energy by excitation or ionisation. This is often the case although in p-decay where the spectrum is continuous from zero to En,,,. some of the emittedp-particles will be sufficiently slow to allow isothermal adjustment of the daughter levels.If there is no interaction between the @-particle and the electron shells the average excitation is approximately Thus even for light atoms considerable excitation is possible. On the other hand Feiiiberg and Migdal6 have calculated the probability of ionisation in the K L and M shells using (a) simple coulombic wave functions ( b ) non-relativistic treatment and (c) sudden change of nuclear charge and have found it to be small. Others lo have confirmed their conclusions although there is some disagreement on the precise vnlucs for the ionisation probability. Typical figures are for Ra-E 1.28 x lo-* ( K ) 1-04 x (L). For a particular shell the ionisation probability varies inversely as Z2. Data are not available for higher shells although one would expect that it would be greater than for the K shell.are not in disagreement either with experiment or with the calculations of Serber and Snyder,9 yet the large excitation energies predicted by the latter are difficult to reconcile with experiment. Loss by icnisation of many electrons would be expected from Serber and Snyder's estimates whereas in fact multiple ionisation in only a small fraction of the decay is usually obtained ; K-shell ionisation in the p-emitters 35S 32Y Ra-E and 147Pm is very sma1l.l' Similar measurements on the K-capture isotopes 55Fe and 37A agree with the predictions on ionisation probability,12 and charge measurement from pure p-decay shows no abnormal ionisation (see Table 2). However Snell's recent data on 3H and 85Kr suggest that perhaps 10-20% of the decays might result in multiple ionisation in some cases.More ionisation is to be expected in the outer shells owing to the screening effect of the field of the nucleus but no quantitative theoretical predictions are available yet. These effects may be very important in molecular bond rupture. Even if a constituent atom does not have sufficient recoil energy to break a bond electronic excitation with subsequent bond rupture may occur. This important process has been discussed by Wolfsberg,l5 who has estimated 22-8Z2I5 ev. But although Levinger's estimates Serber and Snyder Phys. Rev. 1952 87 152. lo Levinger ibid. 1953 90 11 ; Schwartz J. Chem. Phys. 1953 21 45. l1Boehm and Wu Phys. Rev. 1954 93 518. l2 Porter and Hotz ibid. 1953 89 903 ; Miskel and Perlman ibid.1954 94 1683. l3 Snell Pleasonton and Leming J. Inorg. Nuclear Chem. 1957 5 112 ; Snell and Pieasonton Phys. Rev. 1957 10'7 740. l4 Wexler Phys. Rev. 1954 93 183 ; Miskel and Perlman ibid. 1953 91 899 Kofoed-Hansen and Nielson Kgl. danske Videnskab. Selskab 1955 29 No. 15. l5 Wolfsberg J. Chem. Phys. 1056 24 24. BAULCH AND DUNCAN QUANTITATIVE NUCLEAR CHEMISTRY 137 TABLE 2. Resultant atomic charge (from ref. 14) (Gaseous prcssure 5 x 10-3-5 x mm. Hg) Parent activity 3H 14c 37A 41A 83Kr* Mixture of gases obtained in nuclear fission (83Kr* 85Kr 87Kr 88Kr 135Xe) Parent molccule Transitiont Average positive charge 0.9 & 0.1 1.0 & 0.2 2.6 - 3.4 1.0 f 0.1 10 f 2 7.7 f 0.4 1.3 Charge expected from nuclear process only -f e - = Internal conversion electron. the probability of C-N bond rupture due to electronic excitation in CH,*NH,+ formed from lPC-labelled ethane (see below).When the nucleus recoils during @-decay it travels with a velocity which may be comparable with but is always rather less than planetary electron velocities. Electron loss may then occur because the nucleus “ leaves behind ” some of its associated electrons. Fission recoil energies (-250 MeV) are large enough l6 to result in an atomic charge of +20-25 units but it is unlikely that cc-particle recoil will cause a loss by this mechanism of more than 0.5 unit of charge and p-recoil will be still less efficient.lO Interaction of the nuclear @-particle with the nuclear radiation field may cause emission of continuous X-rays. The probability of emission decreases with increasing ,&particle energy and is usually of the order of 10-4.The intensity has been measured experi- mentally e.g. for 35S Ra-E and 147Pm with good agreement with theory.ll This effect is ignored in subsequent discussion. It seems likely that excitation is the only purely electronic process which is important in P-decay. Unfortunately the evidence is conflicting and it is very difficult lo predict with accuracy how much excitation will take place in a given case. On the other hand the properties of the daughter atom only deter- mine how the excitation energy is distributed. Thus when either X - tadiation (from planetary electron transitions) or y-radiation (from nuclear transitions) is emitted it is characteristic of the daughter atom. The iiuclear and electron shells must therefore attain equilibrium in a time which is short by comparison with the time for X-ray emission I n the case of &-decay extranuclear processes are certainly significant (c) Electron loss by trailing.(d) Internal “ Bremsstrahlung ”. (e) Summary of electronic processes arising in radioactive decay. sec.). 16 Bohr Php. Rev. 1941 59 270 ; Lassen Kgl. danske Vidanskab. Selskab 1945 23 No. 2 ; Knipp and Teller Phys. Rev. 1941 59 659. 138 QUARTERLY REVIEWS since the daughter products are usually positively charged. Even the rare gas radon forms positively charged polonium atoms whereas it would have a double negative charge as a result of the or-decay alone. Whilst the first two electrons could be lost by a spontaneous exothermic process some other feature must cause the loss of the other two.It might be due to electron “ shaking ” but since the electronic energy levels of an atom of charge 2 - 2 are higher than in an atom of charge 2 electronic transitions between the ground states must be endothermic. The a-particle would thus have to supply the ionisation energy. The charge on an atom which leaves a metallic foil is not germane here since it depends on the relative values of the ionisa- tion potential of the atom and the work function of the metal. Thus in a-recoil of Ra-D from Ra-C’ and of Ra-B from Ra-A positive charges of the order of unity are obtained l7 from platinum and nickel surfaces. A more important outcome of the nuclear event than electronic excitation is the recoil energy imparted to the radioactive atom. The nature of the molecular frag- ment resulting from decay will depend on the type and energy of the nuclear transition in which both charge and momentum will be conserved.The (f) Nuclear recoil. This can be precisely evaluated. TABLE 3. Chemical consequences of nuclear recoil Radiation 0 B- B+ I<-capture y in the presence of i3 Isomeric transition Oxidation state of daughter - 2 to -+2 $ 1 -1 t o o 0 0 0 Recoil energy (eW 105 0-20 0-20 0-20 0-20 0-1 Recoil spectrum Line Spectrum Spectrum Line Line Line Main features 2 > 83. Completerupture of all bonds Neutron - excess isotopes. Some bond rupture pos- sible. High oxidation state for daughter Neutron-deficient isotopes. Annihilation radiation indicates re-oxidation of daughter Neutron-deficient isotopes with insufficient energy for ,!3+-decay. Consider- able bond rupture due to electron deficiency.Auger electrons May interact with electron shells (mainly of other atoms) by photoelectric encounter or Compton effect Internal conversion and bond rupture large for y-energy below 400 kev. Photoelectric encounter and Compton effect also possible I 17McGee Phil. Mag. 1932 13 1 ; Makower and Russ ibid. 1910 20 875 882; 1915 29 253. BAULCH AND DUNCAN QUANTITATIVE NUCLEAR CHEMISTRY 139 U @+ @ - Y possible nuclear reactions are summarised in Tables 3 and 4. In discussing these tables we will confine ourselves in the first instance to those reactions in which there is no direct interaction between the nuclear radiation and the extranuclear electronic shells. Chemical reactions initiated by the radiation are not discussed in any detail. maEa/M E/j2/2fi1c2 + m,EB/M = 536Ep2/Af + 541Ep;Af 1 3 y 2 / 2 M ~ 2 = 53GEy2/M TABLE 4.Recoil energies Radiation I Recoil energy (ev) ma = Mass of a-particle. M = Mass of recoiling atom. c = Velocity of light. m.o = Mass of electron. Energies E a Eb and E are in MeV. (i) High-energy recoil. The two most important processes which produce high-energy recoil fragments (a y emission) also give monoenergetic recoils (or line spectra). All the recoil energy may not however be available f'or bond rupture. It is well known that for a diatomic molecule ED = ERm/(M + m) where ED is the energy available for bond rupture E is the calculated recoil energy and M and m are respectively the masses of the recoiling nucleus and the remaining nucleus (or an equivalent reduced inass for a polyatomic molecule).Nevertheless ED is usually sufficient to cause bond rupture even if the direction of recoil is not along the line of centres of the atoms. If the direction of recoil makes an angle 8 with the line of centres a fraction cos 8 of the energy will appear as vibrational energy and the remainder as rotational energy. On integration it emerges 1)hat 2/3 of the total available energy initially appears as rotational energy and 1/3 as vibrational energy. But the energy is usually so large by com- parison with the spacings between the vibrational or rotational levels that rupture is inevitable. The energy necessary for rotational rupture may be somewhat greater than required for vibrational rupture of a molecule in its I tormal modes since excessive energy in rotational states causes a maximum t,o appear in the Morse curve.But in most cases a substantial degree of I )ond rupture will occur via vibrational states even if the energy were initially present in rotational states. Once bond rupture has occurred there will result one or more molecular ragments in a chemically reactive state. What happens next will depend on the type of system. In liquid systems the initial recoil atom causes secondary ionisation to form other chemically reactive species. But the ccoil atom itself is still mobile and can be used to give further data. This depends on the fact that of the order of 100 molecular collisions are required or a 0.1 MeV recoil atom of mass about 200 to be slowed down to thermal velocities. Stopping by direct nuclear collisions is insignificant and hard Icinetic energy) collisions contribute only at the end of the path.Most 140 QUARTERLY REVIEWS of the slowing down process occurs by " soft " collisions in which the recoiling atom partially penetrates the electron cloud of the atom which it hits. Bohr has shown that the stopping power per unit mass is proportional to the nuclear charge of the retarding medium and proportiona.1 to the density. Thus the range of the recoil atom will depend on the density of the stopping material and on its nuclear composition. Remarkably little work has been done on this application in spite of the low thicknesses which may be measured (< 1 ,ug./cm.2). Almost all examples (see p. 146) are a-induced reactions although n y (Szilard-Chalmers) processes could be used also. We include here (a) processes with a (ii) Low-energy recoil processes.I ' l l I I l l 1 I 1111 I Ill1 I Ill I II I (I 1 I I I I I I I 1 I 1 I I I I I I I 1 I I I 1 I I II II I 1 II I1 I I I I I \ \\\ ' 24 I I I I I I *.& 7 2 3 . 4 5 6 7 Recoil energy (e Vl FIG. 1 Calculated recoil energy (ER) spectrum of 14C for different /3-particle neutrino interactions. continuous energy spectrum the maximum of which is comparable with that of chemical bonds and ( b ) monochromatic recoil processes with energies less than chemical bond energies. In many p-decay processes the maximum recoil energy is greater than the chemical binding energy. Nevertheless because the p-spectrum extends from zero to Em,,. a fraction of the recoils will be too small to cause bond rupture. Fig. 1 shows the recoil spectrum of 14C calculated from Permi theory for Em,,.= 0.158 MeV. For this calculation it is necessary to know the angle between the directions of emission of the neutrino (Y) and the p-particle which has still not been settled beyond (A) Binding energies. BAULCH AND DUNCAN QUANTITATIVE NUCLEAR CHEMISTRY 141 dispute. Experimental study l8 of the systems 6He -+ 6Li + r6 + v and 19Ne -+ 19F + 18 + Y suggests that for allowed transitions the Fermi part of the interaction is scalar while the Gamov-Teller part is tensor. The exact value of the interaction depends on the matrix elements of the nuclei involved in the decay. However in most cases little error is introduced if the above assumptions are used as a working basis for the calculation of recoil spectra. As with a-recoil some of the recoil energy is lost as kinetic energy.But the energy available for removing the atom from its environment (ED) is now of the same order as the chemical binding energy (EB). I n general we may write the residual recoil energy as ERR = ED - EB. If E, > 0 the atom will be removed from its environment. If E, < 0 it will not unless the transition causes extranuclear processes (see above). Provided that correction for the latter can be made measurement of the frequency of obtaining free atoms or of the energy distribution of recoils would allow the 3 a 5 3 P No bond rupture I obtained - €8 Recoil energy available for bond rupture (ED) FIG. 2 Diagrammatic illustration of the proportion of recoils causing bond rupture. chemical binding energy to be estimated (see Table 5 and Fig. 2). Alterna- tively if we calculate the degree of bond rupture for a bond of known energy any excess observed experimentally must be due to electronic excitation or to other extranuclear processes.If we set an arbitrary limit of say 57.5 kcal./mole (2.5 MeV) for the energy necessary to break stable chemical bonds by P-decay and write we may calculate Es the maximum /I-energy necessary to cause molecular disruption. Then by use of the liquid-drop model the value of Zo - 2 necessary to provide the @-energy required for bond rupture by recoil may be estimated where Z, is the (non-integral) value of 2 a t the minimum Allen and Jentschke Phys. Rev. 1953 80 902 ; Rustad and Ruby ibid. p. 880 ; Alford and Hamilton ibid. 1954 95 1351 ; Maxon Allen and Jentschke ibid. 1955 97 109 ; Robson ibid. 1955 100 1933 ; Alford and Hamilton ibid.1957 105 673. K 142 QUARTERLY REVIEWS TABLE 5 . Parent molecule 14co HI4CHO H,C 14CH HCi14CH a4C10 ?AsCl lP4CeC13(s) Daughter molecule NO+ H14NHO+ H,C14NH,+ HC14NH+ 34S0,- 77SeC1,+ 144PrC13( s ) ED (max.) (Xev) 0.158 0.158 0.1 58 0.1 58 0.713 (P+) 0.8 0.35 3.75 3.50 3.50 3.75 8.58 5-86 1-77 Expected behaviour EB = 6.49 ev (N-0) ; 11 ev (N-O+). EB = 3.1 (NO). Bond unbroken Some disruption of the molecule 3.7 (K-H). No C-N bond rupture EB = 4.5 ( G N ) . No C-N bond rupture Production of SO and 0. Some dissociation from electron-deficient bonds No data available. Probably loss of at least one chlorine Lattice energy 25 ev. No loss from solid EB = 3.1 (C-H); 4.5 (C-N); EB = 1.9 (Cl-0) ; 5.18 (S-0). All values of EB ED and lattice energies are quoted in ev/molecule and are taken from Gaydon " Dissociation Energies and Spectra of Diatomic Molecules " Chapman & Hall 1947 and from Szwarc Chem.Rev. 1950 47 75. of the energy parabola. I n Fig. 3 plots of ED versus M are given with the corresponding values of Zo - ZM. From this figure the following conclusions may be drawn (a) Bond rupture is most likely with low M high m and high 2 - Z M . ( b ) Practically no bond rupture by recoil occurs for M and m > 100 and (Erna.& < 0.5 whereas it will be substantial for m > 50 and ED > 2.0. ( c ) For M odd the energy of /3-decay to a stable isotope cannot be greater than that given by 2 - ZM = 1. Hence bond M Fra. 3 P-Energy necessary to rupture a 2.5 ev (57.5 kcal./mole) bond between a radioactive atom of mass M and a n atom of mass m.Values of 2 - ZM are quoted for different types of nuclei where Z is the non-integral charge corresponding to the minimum of the energy-2 parabola and ZB is the charge on the radioactive nucleus. T h e values quoted for Z - Z M are accurate to about j 10%. BAULCH AND DUNCAN QUANTITATIVE NUCLEAR CHEMISTRY 143 rupture will always be small with M large and odd. For M even 2 - Z M < 2 if the atom decays to the most stable isotope of the daughter element. With Z,w also even there is only a small amount of rupture with M > 100 ; if ZM is odd rupture by /?-recoil is unlikely [(Ema.& > > 13 with M > 140 if m < 25 and also if M > 200 for all values of wa. To study nuclear chemical reactions in which (a) the thermodynamic properties of the daughter fragment (see below) or ( b ) rupture by electronic excitation or other extranuclear processes (see above) is of predominating importance the degree of bond rupture by recoil must be reduced to a minimum.This is best achieved by use of the radio-isotope which is closest to the minimum in the energy-2 parabola (usually closest to the most abundant isotope) preferably with M even. Conversely if a high degree of recoil rupture is sought an isotope decaying to a radioactive daughter-preferably from a short-lived emitter with M odd-is likely to be most suitable. Although molecular disruption frequently occurs by internal conversion etc. it may also result directly from the nuclear recoil consequent upon y-decay. In Fig. 4 the recoil energies calculated for y-emission for decay from the first excited states to ground states of even- even nuclei are given.Similar results would be obtained for other nuclei (B) y-Radiation. 0 20 40 60 80 100 720 740 Neutron numbers CM-21 F I ~ . 4 y-Recoil eneigies obtained f r o m first excited states of nuclei with even charge and even mass. 144 QUARTERLY ICEVIEWS with some displacement along the energy axis. Decay by y-emission also occurs from higher excited states but the first excitation level often accounts for the largest and most frequent decay. We may therefore generalise from Pig. 4 and conclude (a) that almost all nuclei with high M (> 95 for even-even types) will have recoil energies well below chemical binding energies (say 2.5 ev) ( b ) that the exceptions will be nuclei with nucleon numbers close to magic numbers which will always give high recoil energies and (c) that when M is low (< 50 for even-even types) the (mono- chromatic) recoil energy will be large enough (> 10 ev) to cause bond rupture in all cases-sometimes it will be large enough to remove the recoiling atom bodily from its environment with an expected range of the order of 0.1-1 yo of the range of a-recoil atoms.Consider now what hap- pens to the product of a nuclear reaction in which bond rupture does not occur. It must be formed initially with the same number of electrons as its parent. One or more of the following processes may subsequently occur (g) Thermodynamic properties of the product. (i) The product may be stable 14c032- (as) -+ B + 14N03- (aq) (ii) The product may be disproportionate Hz35S (g) + + H,35C11 (g) -+ H,+ + 35C1 or H35C1 + H t (iii) The product may react with its environment Molecular reaction i H 2 0 s9Fe3+ (aq) + /3 + 59C04f (aq) .-+ 59C03f + IIt -t $0 Electron exchange 34c1- (g) -+ p+ -I- 34s~- ( g) -+ 2e + 34s (g) A wide variety of chemical entities can result depending on the properties of the species initially formed.But by the proper choice of nuclear systems one may control both recoil effects and the thermodynamic properties of the product sufficiently to allow the consequences of each type of nuclear chemical process (recoil electronic excitation and chemical reaction) t o be distinguished. The remainder of this Review considers the experimental aspects in terms of these three types of reaction. (5) Experimental Methods.-To obtain quantitative data about a nuclear chemical event secondary chemical effects such as ionisation of the medium by the emitted radiation must be eliminated.The radioactive source must usually be in dilute solution or a t a low gas pressure or the daughter species must be separable from the system before secondary processes can occur. When the source is mounted on a solid substrate one of the following methods of preparation may be used This usually gives sources which are several atomic layers thick in which the recoil atom loses most of its energy. Contamination is also difficult to avoid. Electrolytic collection from a gaseous source of recoil atoms such as radon is a convenient method for any decay product (i) Precipitation or evaporation. (ii) Electrolytic collection. BAULCH AND DUNCAN QUANTITATIVE NUCLEAR CHEMISTRY 145 of a gas.l9 Jedrzejowski 2o has shown that such sources may sometimes contain aggregates of atoms which might either (a) impede the recoil atom or (b) be carried off the source with the recoil atom.This methGd is sometimes suitable. Sources of Ra-C on platinum and nickel disks prepared in this manner have been found suitable for a- and @-recoil work (see below). This is by far the most reliable method.21 The supporting material has little effect on the ease of escape of a-recoil atoms although a smooth polished surface is to be preferred to a rough one. On the other hand the presence of impurities aggregates or adsorbed gases can prevent completely the escape of p-recoil atoms. Hence solid @-recoil sources are usually prepared by this method using material which has been previously separated by one of the three methods above.Barton 22 has successfully prepared a source by electrolytic deposition followed by heating to 400-500" to remove gas layers. Even then measurements must be made as rapidly as possible after preparation of the source. As with a-recoil the source material is usually deposited on a metal plate but the nature of the metal its work function smoothness cleanness and surface oxidation may considerably affect the yield and the charge of the low- energy p-recoil fragments. Recent work has shown that the /?-recoil yield of a source of Th-B deposited on tungsten can increase by threefold or more over a temperature range of 80-150" under some condition^.^^ It is therefore of first importance that the chemical properties of the source are fully known before recoil experiments are made.Radioactive recoil atoms are frequent,ly collected on a charged plate placed close to the source. From the yields obtained a t various pressures and a t various distances between the source and collector the range of the recoil atoms may be determined. The nature of the collecting surface materially affects the yield of the low-energy /3-recoil fragments and to a lesser extent that of a-recoil atoms. A smooth aluminium or platinum plate is very efficient. There is some evidence that the accommodation coefficient of elements of high boiling point is very low if the pressure is low (-10-8 mm.). This is associated with the difficulty of forming condensation nuclei on the collecting surface. I n the Reviewers' experience checks should always be made to confirm that collection is near 100 yo.Non-radioactive recoil atoms may be detected by measuring the ionisa- tion produced in a gas. Such methods require a gas pressure of a t least several mm. of mercury and are therefore not suitable for work with p-recoil ;%toms. If the chemical properties of the source are suitable @-recoil atoms are positively charged a feature which has been used by Leipunski Z4 to (iii) Electrolytic deposition from solution. (iv) Vacuum distillation. (v) Detection of recoil atoms. 1QSee Baulch and Duncan Australian J . Chem. 1957 10 112. 20 Jedrzejowski Compt. rend. 1929 188 1043. 2 1 Donat and Philipp 2. Physik 1927 45 512 ; Davies Phys. Rev. 1952 86 976. 2 2 Barton Phil. Mag. 1926 1 835. 23 Baulch Duncan and Kepert forthcoming publication.24Leipunski Proc. Camb. Phil. Soc. 1936 32 301. 146 QUARTERLY REVIEWS detect the positively charged ions from a 14C-preparation by means of an accelerating potential applied to a grid. At present Allen-type electron multipliers 25 are the most convenient and widely used detectors of low- energy positive (recoil) ions. An elegant method of detecting non-radioactive daughters has been used by Wolfgang Anderson and Dodson 26 in the detection of the products of decay of 14CH,*14CH by a double labelling technique. If the specific activity is sufficiently high both carbon atoms of a reasonable number of the ethane molecules will be labelled. These will decay to form the radical l4CH3*NH,+ The frequency of bond rupture may then be determiaed from the number of labelled methylamine molecules obtained.The weight of methylamine is very small (-10-8 g.) although its /I-activity can easily be detected. Chemical losses may be reduced and a t the same time allowed for by addition of inactive methylamine as carrier prior to chemical separation from the active ethane remaining after the experiment. Since the activity of the remaining ethane will be several orders of magnitude larger than the resulting methyl- amine efficient and complete separation of ethane from a pure sample is essential. But after carefully checking this aspect of the problem these workers obtain rupture of the C-N bond in approximately 53% of the cases in which 14CH,*14CH decays by p-emission of one of the carbon atoms. Since a-recoil atoms always have enough energy to overcome all chemical binding forces it is the study of the properties of the recoiling atom itself which is of interest.There are two regions (i) that part of its track during which it is being slowed down to thermal velocities and (ii) the subsequent movement of the free atoms by diffusion. Measurement of the number of a-recoil fragments which penetrate matter affords a convenient means of measuring the density or the thickness of the material. Since very thin layers will stop the recoiling fragment this technique is far more sensitive than the conventional 8-ray thickness methods. The range of an a-recoil atom is -10 pg./cm.2 and it is quite possible to measure variations in recoil yield below 1 pg./cm.2. As an example Gregory 27 has shown that a linear yield of recoil radon atoms with thickness results when barium stearate monolayers of effectively infinite area are progressively deposited over a layer of a thorium salt.The recoil yield falls to zero with an increasing number of monolayers. This system is diEcult to study since loss of thoroii from the solid must be stopped by cooling in liquid air. But other systems such as Th-C" (208T1) recoiling from Th-C (212Bi) are also suitable. A n aqueous solution of Th-C and sodium dodecyl sulphate has been used to study the con- centration of soap molecules adsorbed on the liquid-gas interface by collecting the recoiling Th-C" on a charged plate above the surface.27 There are a number of applications of a-recoil to the measurement of 26Allen Rev. Xci. Inst. 1047 18 739. 26 Wolfgang Anderson and Dodson J . Chern. Phys.1956 24 16. 2 7 Gregory Hill and Moorbath Trans. Faraday SOC. 1952 48 643 ; Aniansaon and (6) Some Experimental Results.-a-Recoil. Steiger J . Chem. Phys. 1953 21 1299. BAULCH AND DUNCAN QUANTITATIVE NUCLEAR CHEMISTRY 147 diffusion coefficients in solids. Thus Hevesy arid Seith 28 determined the diffusion coefficient of bismuth in lead chloride using Th-C. The recoil yield was determined before and after heating the lead chloride on which a source of Th-B/C had been deposited. The diffusion coefficient was calculated from the known recoil range. All previous work has used the Flugge-Zimens empirical stopping-power equation relating the recoil range to the square of the recoil energy. This equation has been recently criticised on experimental and theoretical grounds.lg It has been shown that the first-power dependence of the range on the recoil energy is more nearly correct in agreement with Bohr's theoretical treatment.29 I n its passage through a gas the recoil atom changes its charge several times but since most a-recoil ions are atoms of electropositive metals they have usually been found to be positively charged a t the end of their path.Dee using a cloud-chamber technique finds that 84% of such recoil atoms are positively charged at the end of their path.30 A recoil ion from a 6.54 Mev cc-particle produces a total of some lo3 ion pairs of which perhaps 75% may be in the first half of the range. Since the or-particle produces 2 x 105 ion pairs only about 0.5y0 of the total ionisation is obtained from the recoil atom although this is about 75% of the ionisation obtained in the distance the recoil atom travels from the site of the initial radioactive event Having reached the end of its path the atom will diffuse through the surrounding gas.The yield N from such a system in radioactive equili- hiurn at a distance x can be shown to be 319 32 N = No exp ( -A/D)*x from which .D the diffusion coefficient can be calculated. In the two cases hitherto reported it has been observed that D is about 1000 times smaller than expected from kinetic theory. Wide discrepancies between dif- ferent workers associated with experimental difficulties in preparing the source are common. Thus the yield of Ra-C from sources of Ra-B has been variously reported 22 3 3 ~ 34 as 0.1% Z-6Y0 and 20450%. Almost all workers report a rapid deterioration in ,8-recoil sources sometimes of the order of 50% in 20 min.Earlier work was of course subject to many limitations of technique which do not now apply. This is undoubtedly the main reason for the discrepancies for Sherwin 35 (using 32P on a lithium fluoride surface) has recently shown that recoil sources can be prepared which are quite stable. He used the constancy in the number of recoil tibagments (detected by electron multiplier) as a criterion for the purity of his source since it was only such sources which gave a satisfactory resolution 2f the energies of the emitted recoil atoms. These sources could be kept The reason for this is not understood. P-Decay.-(a) Heterogeneous systems. 28Hevesy and Seith 2. Physik 1929 56 780. 26 Bohr Kgl. danske Videnskab. Selskab 1948 18 NO. 8.30Dee Proc. Roy. SOC. 1927 A 116 664. 31Baulch Duncan and Ryan Australian J . Chem. 1957 10 203. 32 Chamie J . Phys. Radium 1934 5 54 436 ; Chami6 and Tsien San-Tsiang ibid 33Makower and RUSS Phil. Mag. 1910 19 100. 3 4 Muszkat ibid. 1920 39 690. 35Sherwin Phys. Rev. 1948 73 1173; 1949 75 1799. l!Ml 2 46 ; Rutherford " Radioactivity " Camb. Univ. Press 1905 275. TABLE 6. Kecoil in homogeneous phases Decay Chemical comporind (I) 5IMn -+ 51Cr (111) 132Te -+ 1321 (IV) 143La+ 143Ce (V) 144Ce + 144Pr (VI) *]OPb + 21OBi (11) 83Se + 83Br ( a ) Te032- ( b ) TeO,,- LaS+ (aq) Acetonylacetone (A) complex Gaseous Pb(CH,) (a) Mn0,- (4 ( b ) Mn2+ (aq) (c) Mn2+ in acetone-H,O (d) CsMn04(s) ( a ) Se032- or Se0,Z- pH = 7 ( b ) ditto pH = 11 and dioxan-H,O (el MnCO,(s) 2-35 B(+) 1.5 0.36 0.93 0.35 0.06 83.0 24.5 2.0 6.74 1.77 0.097 (MeV) KO y 0.17 0.37 1.1 0.22 7x0 Y E-0 y but e- Several < 0.047 Observnt ion (a) 50:; Cr3; ( b ) 97-100% Cr3+ ( c ) 5-15qA Cr0,2- (d) l000,b CrO,*- ( e ) 30% c1-0,~- (a) 400,A BrO - independent of isotope ( b ) 24-30?/ in oxidised form or species (a) 759,; I- + I + HIO 1476 103- (6) 60qb I- + I + HIO 10,- 11% 10,- 12% 10,- 60% CeIV No disriiption of complex in organic solvents but some in water 8% Molecular disruption.Internal conversion 33%. 3sBurgess and Kennedy J. Chem. Phys. 1950 18 97. 3 7 Burgess Davies Edwards Gest Stanley Williams and Coryell J . Chim. phys. S8Davies J . Phys. Coll. Chem. 1948 52 595. 39 Edwards and Coryell see Walil and Bonner “ Radioactivity applied to Chemis- 39sEdwards Day and Overman J . Chem. Phys. 1953 21 1555.1948 45 165. try ” Wiley N.Y. 1951. Ref. 36 37 38 37 38 37 38 39 39 39a Initial product (I) ( a ) CrO,Z- ( b ) Cr+ ( c ) Cr+ ( d ) Cr0,2- ( e ) Cr+ (11) (a) Br0,- or BrO,- ( b ) Br0,- or Br0,- (111) (a) 10,- ( b ) 1 0 4 - (IV) Ce4+ (V) PrA,+ (VI) Bi(CH,),+ Presumed fate of initial species Considerable disruption of CrOa2 - by recoil Oxidised by air Some oxidation by electron annihilation As expected. Some oxidation by electron annihilation Se0,- and Se0,- take no part in the reaction This confirms the r61e of the solvent. Disruption of CrO,*- prevented in solid state The high-energy recoil could cause rupture of a substantial proportion OF Br-0 bonds and would account for reduction Very similar to previous example but recoil energy is too low to account The chemical reaction appears to be initiated by initial for results.products and/or radiation effects A small amount of this very strong oxidising agent could easily be reduced by oxidation of impurities or exchange with CeUI impurity in La. High recoil energy could also initiate cheinical reaction leading to reduction Complexes of P r I V would be reduced to P+. Results in organic solvents show that (i) internal conversion is chemically insignificant (ii) energy necessary to disrupt complex is greater than maximum recoil energy (1.2 ev/molecule). I n aqueous solution effects are therefore due to chemical reaction between PrIv and CeII1 and/or solvent No disruption due to recoil only shows that (i) Bi(CH,),+ is not thermo- dynamically unstable (ii) Bi-C bond energy > ener-7 available for bond rupture (- 0-03 ev) b ..CQ c c d P M b w 150 QUARTERLY REVIEWS for several days if warmed by a small heater placed nearby but rapid deterioration occurred a t room temperature. Owing to the lack of knowledge of the chemistry of most of the work described before about 1940 no quantitative significance can be attached to most of the results. But by using data similar to those in Fig. 2 the following generalisations can be drawn from previous work (i) The energy necessary t,o remove an atom from the surface of an untreated source decreases with increasing temperature whether the source is prepared in the atmo- sphere or in a high vacuum. For bismuth it is greater than about 50 kcal./mole a t room temperature (zero yield) and perhaps 20-30 kcal./mole a t 110". (ii) This behaviour is a feature of the surface properties which can be altered reversibly by surface ~retreatment.~~j 35 In earlier work on /3-recoil it has not always been possible to distinguish clearly between the effects of radiation-induced reactions recoil processes and the reaction of thermodynamically unstable products.I n solution the emphasis has been on the valency changes of the radioactive atom and unfortunately much of the work has been marred by the presence of y-emitters. This makes the interpretation of the results difficult Thus the yield of daughter product has often not been lOOyo in the valency state expected (see Table 6). But the results may in large measure be explained in terms of the profound effect of recoil. Electronic excitation processes do not apparently need to be invoked to account for most of the observations.At least two unequivocal cases are known [cerium acetonylacetone a.nd Pb(CH,),]. In these cases novel conclusions can quite confidently be made (see Table 6). Likewise the recent work on the rupture of G N bonds 26 in the p-decay of 14CH,J4CH in the gas phase showed that in 47% of the decays the daughter molecules survived as 14CH3*14NH,. This system has been theoretically investigated by Wolfsberg l4 by treating the molecule for simplicity as a diatomic molecule. Exact calculations for rotational and vibrational excitation and dissociation are very difficult since no very good wave functions have been formulated for such a system. Molecular-orbital wave functions are assumed for the easier case of electronic excitation. The probability of electronic non-dissociation is found to be 0.815.When this is combined with the fraction 0.74 of recoils in which rupture of the 2.1 eV C-N bond is not obtained a figure of 0.60 & 0.20 is obtained for the probability of non-dissociation in as good agreement with experiment as can be expected. The analysis implies that perhaps one-third of the activity loss is in this case caused by processes other than recoil i.e. by excitation (as proposed by Wolfsberg) or by subsequent chemical reaction of the daughter 14CH3*NH,+. The emission of ?-recoil atoms from gold films electroplated on nickel has been observed 40 by the reaction 197Au(ny)198A~. The lS8Au recoil atoms were collected on a nickel plate & in. away from the source plate with a yield of 0.1%. Recoil was inhibited because the gold atonis 81 285.(b) Homogeneous systems. y-Recoil. 40 Yosim and Davies J . Phys. Chein. 1952 56 599 ; Magnusson Phys. Rev. 1951 BAULCH AND DUNCAN QUANTITATIVE NUCLEAR CHEMISTRY 151 exist on the surface of the target in clusters. Higher yields are obtained if the gold film is prepared by vacuum distillation and the neutron bombard- ment and recoil collection is also carried out in a vacuum. Recoils by ny reactions from similarly prepared sources may be up to 90-100% efficient. In this case the y-recoil energy is quite large enough to displace the recoiling atom from its lattice position. Unfortunately no attempt has been made t o treat this process quantitatively except for some general predictions of the degree of bond rupture expected in neutron-irradiated alkyl halides.41 Internal conversion of y-rays leads to formation of chemically active species of high charge (80Br* -+ SOBrlO+ ; 83Kr* -+ s3Kr7+) and Cooper 42 has carried out calculations to show how far it is possible for molecules to dissociate after formation of Auger electrons. An experiment of significance is the internal conversion of y-rays from 69Zn and 127Te/12gTe. Diethyl- tellurium has been prepared and kept a t 100" in a glass bulb for several hours. Substantial activity was found on the walls afterwards showing that internal conversion by the 0.10 MeV y-ray was frequent. But with 69Zn the 0.47 Mev y-ray is too energetic for significant conversion and no activity was found on the walls with diethylzinc. This experiment is usually quoted as proving internal conversion with tellurium but two other conclusions also follow that neither (i) electronic excitation processes nor (ii) the y-recoil energy can cause disruption of the diethylzinc.From the latter it follows that the zinc-carbon bond energy is greater than the energy available for dissociation. Since the recoil energy is 15 ev the energy available for dissociation is (15)2/(69 + 2 x 15) i.e. some 2.25 ev (52 kcal./mole). (7) Conclusion.-This Review has been written with the intention of stimulating work in this field. New information is urgently needed to clear up the theoretical background much of which is obscure ; but there is no doubt that the quantitative treatment of the nuclear chemical process can provide new thermochemical information in a variety of systems which are of current interest. One of us (D. L. B.) acknowledges the support of a Graduate Research Studentship from the Australian Atomic Energy Commission. Current work by the Reviewers is also supported by a research contract from the A.A.E.C. with this department (Prof. J. S. Anderson P.R.S.). 41Suess 2. phys. Chem. 1940 B 45 297 312. 42Cooper Phys. Rev. 1942 61 1.

ISSN:0009-2681    

DOI:10.1039/QR9581200133

出版商:RSC    

年代:1958

数据来源: RSC

摘要:

RECENT DEVELOPMENTS IN THE BIOCHEMISTRY OF NUCLEOTIDE COENZYMES By J. BADDILEY D.Sc. PH.D. and J. G. BUCHANAN M.A. PH.D. (KING’S COLLEGE UNIVERSITY OF DURHAM NEW CASTLE UPON TYNE) IT would be difficult and perhaps unnecessary to attempt a rigid definition of the term “ coenzyme ”. The coenzymes were originally understood to include a small group of organic compounds of relatively low molecular weight which are required in catalytic amounts in certain enzymic reactions. These coenzymes possess no enzymic properties themselves but presumably combine with the true enzyme protein to form a complex which is able then to catalyse the overall reaction. This description sufficed for some time and it wa.s assumed that the coenzymes like any other catalysts must actually participate chemically in the reactions which they catalysed.It soon became apparent however that this description was losing its signi- ficance. Most of the metalloporphyrin catalysts usually known as pros- thetic groups when attached to an enzyme protein would fit the above definition. On the other hand the nucleotide coenzymes e.g. cozymase and flavin-adenine dinucleotide are only catalytic in their action in multi- enzyme systems where cyclic processes are able to regenerate the coenzyme continuously. If the enzymes responsible for the processes of coenzyme regeneration are removed or destroyed then the coenzyme must be regarded as a substrate and will be required in stoicheiometric amounts. We now know of nucleotides which although catalytic when accompanied by the necessary regenerating enzymes and substrates could only be described most loosely as coenzymes.In this group are the ‘‘ nucleoside-diphosphate- sugar ” compounds. These include uridine diphosphate glucose where the substrate for the reaction (glucose 1-phosphate + galactose 1-phosphate) is actually a part of the coenzyme molecule. Such compounds are frequently termed “reactive intermediates” but in many respects they must be regarded as coenzymes. These were the two nicotinamide compounds (DPN and TPN) the riboflavin coenzyme (flavin-adenine dinucleotide) and adenosine triphosphatc. The discovery in 1945-46 of coenzyme A by Lipmann and his colleagues was soon followed by the isolation of uridine diphosphate glucose (UDPG) by Leloir. Since that time steadily increasing numbers of nucleotides of the UDPG type have been detected or isolated from animals plants and bacteria.Although the exact nature of the enzymic reactions in which these nucleotides parti- cipate is not always known it is certain that they must be involved in processes similar to those already observed for UDPG. For this reason they may be classified as nucleotide coenzymes. There can be little doubt that the rapid progress in this field which has occurred in recent years is largely a result of greatly improved techniques for the isolation and 162 Until about 1945 four nucleotide coenzymes were known. BADDILEY AND BUCHANAN NUCLEOTIDE COENZYMES 153 separation of nucleotide mixtures ; and of these techniques the most out- standing is that of ion-exchange chromatography accompanied by paper chromatography. In this Review no attempt has been made to treat the subject exhaustively.Several books and review articles which have appeared during the last few years describe both chemical and enzymic or metabolic aspects of the longer-established coenzymes such as the pyridine nucleotides flavin-adenine dinucleotide adenosine phosphates etc. and also coenzyme A. For this reason and for space considerations these members of the group are not included in this Review. It is our purpose to emphasise the direction along which recent developments in the field have been progressing. Particular attention has been devoted here to the group of nucleotides of the uridine- diphosphate-glucose type since this group appears to be expanding rapidly in both numbers and importance. In addition we have included the group of nucleoside-monophosphate-X compounds in which X is a substrate molecule (e.g.acyl adenylates and nucleotide-aniino-acid derivatives). “ Active sulphate ” is sufficiently related to this group to be included. The Nucleoside-Pyrophosphate-Substrate Group of Compounds. Uridine Diphosphate Glucose (UDPG) and Uridine Diphosphate Galactose (UDPGal) The first member of this group of coenzymes uridiiie diphosphate glucose (I) was discovered by Leloir and his collaborators during their investigation of the conversion of cc-D-galactose 1 -phosphate into R-D-glUCOSe 1 -phosphate in gahctose-adapted yeast. The structure was established 2-4 by methods which have proved valuable in later investigations in this group. Acid- hydrolysis liberated uridine-5’ phosphate inorganic phosphate and D-glucose ; very mild acid hydrolysis gave uridine-5’ pyrophosphate (UDP) n Uridine - 5 ’ pyrophosphate OH + glucose Glucose 2-phosphate IH+ Uridine - 5’ phosphate t H,PO Glucose I - & 2-phosphate Caputto Leloir Trucco Cardini and Paladini J.Bid. Chem. 1949 179 497. Cardini Pa,ladini Caputto and Leloir Nature 1950 165 191. Caputto Leloir Cardini and Paladini J. B i d . Chem. 1950 184 333. * Paladini and Leloir Biochem. J. 1952 51 426. 154 QUARTERLY REVIEWS and glucose. Such great lability indicated that UDPG was a derivative of glucose 1 -phosphate. Treatment with ammonia gave uridine-5' phosphate and glucose 1 2-hydrogen phosphate (11) ; further alkaline hydrolysis gave glucose 1 - and 2-phosphate while acid yielded glucose 2-phosphate. The formation of a cyclic 1 2-phosphate of glucose does not necessarily mean that UDPG contains a-glucose 1 -phosphate ; the ,8-anomer should also be capable of cyclisation since the hydroxyl groups a t positions 1 and 2 are then both eq~atorial.~ Proof of the a-configuration of UDPG and of the pyranose ring of the hexose comes from its enzymic '-11 and chemical 12-14 synthesis from &-D-glUCOSe 1-phosphate as well as its polari- metric behaviour on acid hydrolysis.4 Uridine diphosphate galactose (UDPGal) has also been synthesised ~hernically.1~ UDPG is formed enzymic- ally 7-11 by reaction of a-D-glucose 1 -phosphate and uridine triphosphate (UTP) Glucose 1-phosphate + UTP + UDPG + Pyrophosphate The enzyme UDPG pyrophosphorylase is analogous to that catalysing the reaction between diphosphopyridine nucleotide and inorganic pyro- phosphate.l5 UDPG has been shown to act as a coenzyme in the galactose-glucose transformation in the following manner 16 l7 UDPG + M-D -Galactose 1 -phosphate Uridyl + transferase UDPGal + a-D-Glucose 1-phosphate UDPGal + UDPG The enzyme for the first reaction galactose phosphate uridyl trans- ferase,17 l 8 has been shown to be absent from patients suffering from congenital galactoszmia ; 19-20 feeding of galactose leads to an accumulation of a-galactose 1 -phosphate. There is evidence that in galactose-adapted Xaccharornyces fragilis and in some green plants 10 UDPGal may be formed from a-galactose 1-phosphate and UTP. The mechanism of the second Khorana Tener Wright and Moffatt J . Amer. Chem. Soc. 1957 79 430. Brown and Higson J. 1957 2034. Munch-Petersen Kalckar Cutolo and Smith Nature 1953 172 1036.Smith Munch-Petersen and Mills ibid. p. 1038. 7 Trucco Arch. Biochem. Biophys. 1951 34 482. lo Neufeld Ginsburg Putman Fanshier and Hassid Arch. Biochem. Biophys. 1957 l1 Munch-Petersen Acta Chem. Scand. 1955 9 1523. l2 Kenner Todd and Webb J. 1954 2843. l3 Michelson and Todd J. 1956 3459. l4 Chambers Moffatt and Khorana J . Amer. Chem. SOC. 1957 79 4240. Kornberg and Pricer J . Biol. Chern. 1951 191 535. l6 Leloir Arch. Biochem. Biophys. 1951 33 186. 1' Maxwell Kalckar and Burton Biochim. Biophys. Acta 1955 18 444. lS Kalckar Braganca and Munch-Petersen Nature 1953 172 1038. lg Kalckar Anderson and Isselbacher Proc. Nat. Acud. Sci. U.X.A. 1956 42 19 ; Biochim. Biophys. Ada 1956 20 262. 2o Kalckar Science 1957 125 105. 69 602. RADDILEY AND BUCHANAN NUCLEOTIDE COENZYMES 156 reaction catalysed by galactowaldenase,” has aroused much interest.At equilibrium the ratio of the glucose to galactose nucleotide is 3 1 showing the increased stability of the equatorial 4-hydroxyl group in glucose over that of its a.xial isomer. Several mechanisms have been proposed for the trans- formation,21 which requires a formal inversion of the hydroxyl group a t the 4-position in the hexose but it has now been found that the enzyme is stimulated by diphosphopyridine nucleotide (DPN) 22 and that an inter- mediate 4-keto-derivative is involved. 23-25 UDPG has been shown to occur widely in plants and animal^,^^-^^ and it has become apparent that it is involved in reactions other than the galactose-glucose interconversion ; UDPGal and galactowaldenase acre present in large amounts in some yeasts not adapted t o galact0se.~2 It was suggested 33 that compounds of the UDPG type could be concerned in transformation of sugars and their subsequent incorporation into poly- saccharides.At the same time it was suggested that UDPG might be involved in sucrose biosynthesi~.~~’ 34 The first direct evidence for a reaction of this type was given by Leloir and Cabib 35 who showed that the trehalose phosphate of Robison and Morgan 36 was synthesised from UDPG and D-glucose &phosphate by yeast preparations. Leloir and Cardini 37-39 later described two enzymes occurring in plants which catalyse the reactions UDPG + D-Fructose 6-phosphate + Sucrose phosphate + UDP UDPG + D-Fructose + Sucrose + UDP 21 22 2 3 24 25 26 27 28 29 30 31 Leloir A d v .Eruymol. 1953 14 193. Maxwell J . Amer. Chem. Soe. 1956 78 1074. Anderson Landel and Diedrich Biochim. Biophys. Acta 1956 22 573. Kowalsky and Koshland ibid. p. 675. Kalckar and Maxwell ibid. p. 588. Ginsburg Stumpf and Hassid J . Biol. Chem. 1856 223 977. Buchanan Lynch Benson Bradley and Calvin ibid. 1953 203 935. Rutter and Hansen ibid. 1953 202 323. Hurlbert and Potter ibid. 1954 209 1. Hurlbert Schmitz Bruinm and Potter ibid. p. 23. Smith and Mills Biochim. Biophys. Actu 1954 13 386. 32 Mills Smith and Lochhead i b i d 1957 25 521. 3 3 Buchanan Bassham Benson Bradley Calvin Daus Goodman Hayes Lynch Norris and Wilson “ Phosphorus Metabolisrri ” Vol. 11 Johns Hopkins Press Balti- more 1952 p. 440. . 3 4 Buchanan A&. Bioclzem. Biophys. 1953 44 140. 3 5 Leloir and Cabib J .Amer. Chem. SOC. 1953 75 5445. 36 Robison and Morgan Biochem. J . 1928 22 1277 ; ibid. 1930 24 119. 37 Leloir and Cardini J . Amer. Chem. SOL 1953 75 6084. 38 Leloir Cardini and Chiriboga J . Biol. Chem. 1955 214 149. 39 Cardini and Eeloir ibid. p. 157. * We consider that this name originally used by Leloir t o describe the overall process galactose + glucose should be retained for the enzyme rather than “ 4-epimerase ” suggested by Kalckar (ref. 20 ; cf. refs. 25 & 32). 156 QUARTERLY REVIEWS Similar results have now been obtained by Bean and HassidY4O who were able also to show that fructose could be replaced by D-xylulose D-rhamnulose or L-sorbose in the above reaction to give the appropriate disaccharides. With sugar- beet leaves Burma and Mortimer 41 found mainly the reaction leading to sucrose phosphate.An enzyme from pea seedlings which catalyses sucrose synthesis from a-D-glucose l-phosphate and fructose has been studied by Turner.42 It is thought that UDPG may be an intermediate and Cardini has independent evidence of this.43 It has been known for some time that mammary tissue contains UDPG 17 28 44 45 and that galactowaldenase is present. Bovine mammary tissue preparations will coiivert UDPG and a-D-glucose l-phosphate into a-lactose 1 -phosphate by the following pathway 46 Lactose synthesis has also been investigated. UDPG +. UDPGal Galactosyl transferase UDPGal + a-D-Glucose 1-phosphate - Lactose l-phosphate + UDP The reactions have been fully confirmed by using both 32P and 1% labelling. It is not certain whether this is the only pathway by which lactose is ~ynthesised.~' Bean and Hassid 4* have suggested that floridoside 2-O-(a-~-galactosyl)glycerol may arise by reaction between UDPGal and a-glycerophosphate.Glaser 49 recently described the reaction of UDPG with oligosaccharides from cellulose using a preparation from Acetobacter xylinum to give a cellulose-like polymer and a similar system from rat liver can effect a synthesis of glycogen.49a The time is clearly ripe for a close investigation of well-known polysaccharide-synthesising enzymes to find whether they have a firmly bound nucleotide component. Uridine Diphosphate Glucuronic Acid (UDPGA) and Uridine Diphosphate Galacturonic Acid (UDPGalA) Work on the formation of glucuronides 50 in liver homogenafes led to the isolation of UDPGA by Dutton and Storey.51 Since it is formed enzymically by oxidation of UDPG by DPN there is no doubt that it has structure (111).Bean and Hassid J. Amer. Chem. Xoc. 1955 77 5737. 41 Burma and Mortimer Arch. Biochem. Biophys. 1956 62 16. 4 2 Turner Nature 1953 172 1149 ; 1954 174 692 ; Biochem. J . 1957 67 450. 43 Leloir 3rd Internat. Congress Biochem. Brussels 1955 p. 154. 44Caputto and Trucco Nature 1952 169 1061. 4 5 Smith and Mills Biochim. Biophys. Acta 1954 13 587. 46 Gander Petersen and Boyer Arch. Biochem. Biophys. 1956 60 259 ; 1957 69 47 Wood Schambye Peeters and Siu J. Biol. Chem. 1957 226 1023. 48 Bean and Hassid ibid. 1955 212 411. 49 Glaser Biochim. Biophys. Acta 1957 25 436. 40aLeloir and Cardini J. Amer. Chem. SOC. 1957 79 6340. 50 Teague Adv. Carbohydrate. Chem. 1954 9 185.61 Dutton and Storey Biochem. J. 1953 53 xxxvii ; 1954 57 275. 85. BADDILEY AND BUCHANAN NUCLEOTIDE COENZYMES 157 UDPG dchydrogenase has been found both in liver 53 53 and in pea seed- l i n g ~ . ~ ~ Atteiiipts to trap an aldehyde intermediate in the oxidation have at least in the case of the liver been uusuccessful. UDPGA is thought to arise solely by the action of UDPG dehydrogenase on UDPG no UDPGA pyrophosphorylase having been detected.31 The nucleotide has been isolated from mung-bean seedlings 55 56 and from Type I1 and Type I11 pneumo~occi,~~ as well as from liver.31 51 It has been shown that UDPGA will transfer its glucuronic acid residue to form both ether 51 58 59 and ester 6O glucuronides and is evidently the coenzyme for such reactions. UDPGalA has been isolated from a Type I pneumococcus.61 There is none of the glucuronic acid derivative present although both Type I1 and I11 organisms contain it.There appears to be a correlation between the nucleotides present in Type I and I11 organisms and the uronic acid residues in their capsular polysaccharides. This is consistent with the polysaccharide- precursor hypothesis for nucleoside- diphosp hat e- sugar compounds. More direct evidence for the participation of UDPGA in polysaccliaride synthesis has come from studies of 14C-labelled nucleotides. An enzyme from tho Rous chicken sarcoma will convert UDPGA and uridine diphosphate acetylglucosamine (UDPAG ; see below) into polymers having the properties of hynluronic acid.G2 The UDPGA can be replaced by UDPG and DPN. Uridine Diphosphate Acetylglucosamine (UDPAG) and Uridine Diphosphate Acetylgalactosamine (UDPAGal) found an unidentified uridine nucleotide in their This was later identified as UDPAG 6 3 Gentle acid-treatment liberated uridine-5’ pyrophosphate and N-acetyl- D-glUCOSamine.The compound is more stable to alkali than UDPG and the products are uridine-5’ phosphate and N-acetylglucosamine 1 -phosphat!e 5 2 Strominger Maxwell Axelrod and Kalckar J. Amer. Chem. Soc. 1954 76 6411 ; J . Biol. Chem. 1957 224 79. 53 Maxwell Kalckar and Strominger Arch. Biochem. Biophys. 1956 65 2. 5 4 Strominger and Mapson Biochem. J. 1957 66 5G7. 5 5 Solms Feingold and Hassid J. Amer. Chem. SOC. 1957 79 2342. 5G Solms and Hassid J. Biol. Chem. 1957 228 357. 57 Smith Mills and Harper J. Gen. Microbiol. 1957 16 426. 58 Storey and Dutton Biochem.J. 1955 59 279. Isselbacher and Axelrod J. Amer. Chern. SOC. 1955 77 1070. 6o Dutton Biochem. J. 1956 64 693. 61 Smith Mills and Harper Biochim. Biophys. Acta 1957 23 662. G2 Glaser and Brown Proc. Nut. Acad. Sci. U.S.A. 1965 41 263. 63 Cabib Leloir and Cardini J. Biol. Chem. 1953 203 1055. Paladini and Leloir preparations of UDPG from yeast. (IV) * L 158 QUARTERLY REVIEWS an interesting consequence of the lack of a 2-hydroxyl group. The assign- ment of an a-configuration depends on the enzymic synthesis of UDPAG from a-D-N-acetylglucosamine 1 -phosphate by yeast 64 and liver prepara- tions.G5 UDPAG has been detected in plant~,~5 5G animal t i s s ~ e s ~ ~ ~ 31 and mi~ro-organisrns.~7 61 66 It has been noted that labelled hyaluronic acid is formed from labelled UDPGA and UDPAG by Rous sarcoma.G2 Glaser and Brown67 have now found that an enzyme from Neurospora crassa catalyses the incorporation of the N-acetylglucosamine moiety of UDPAG into an insoluble polysaccharide with the properties of chitin.The UDPAG fraction from liver contains UDPAGal 68 (V) ; acid- hydrolysis gives N-acetylgalactosamine together with N-acetylglucosamine. The free hexosamines have been separated by paper chromatography and yield the appropriate pentoses on degradation with ninhydrin. A waldenase enzyme exists for the equilibration of the two nu~leotides.~~~ G8 Lardy has found that both yeast and liver can convert a-D-glucosamine l-phosphate and UTP into uridine diphosphate glucosamine. The bio- chemical significance of this reaction is not yet clear. Uridine diphosphate acetylglucosarnine phosphate and uridine diphos- pha,te acetylgalactosamine sulphate have been isolated from hen's oviduct .G9 Uridine Diphosphate Pentoses Ginsburg Stumpf and Hassid 26 have isolated uridine diphosphate D- xylose (UDPXy) (VI) and uridine diphosphate L-arabinose (UDPAr) (VII) from mung-bean seedlings as well as from other plant sources.10 The nucleotides can be synthesised by reaction of a-lo-xylose 1 -phosphate or a-I,-arabinose l-phosphate with UTP in the presence of plant extracts.6 4 Maley Maley and Lardy J . Amer. Chem. SOC. 1956 '78 6303. 66 Maley and Lardy Science 1956 124 1207. 67 Glnser and Brown Biochim. Biophys. Acta 1957 23 449 ; J . Biol. Chern. 1957 G8 Pontis ibid. 1955 214 195 ; Cardini and Leloir ibid. 1957 225 317. 69 Strominger Biochim.Biophys. Acta 1955 1'7 283. Smith and Mills Biochem. J. 1956 64 5 2 ~ . 228 720. BADDILEY AND BUCHANAN NUCLEOTIDE COENZYMES 159 h . . OH P-D-XylOSe 1 -phosphate is inactive but 8-L-arabinose 1 -phosphate is active in this system. Plants also contain a waldenase 26 70 which catalyses the reaction UDP D-Xy + UDP L-Ax- The reaction is not catalysed by yeast galactowaldenase although a formal similarity between this reaction and that catalysed by galacto- waldenase is apparent. Guanosine Diphosphate Mannose (GDPM) From among the nucleotides from yeast Cabib and Leloir 7 1 9 72 isolated GDPM (VIII). Mild acid-hydrolysis liberated mannose and guanosine-5' pyrophosphate ; further hydrolysis gave inorganic phosphate and guano- sine-6' phosphate. More drastic hydrolysis gave guanine.That the nucleotide is a derivative of a-D-mannose l-phosphate follows from its enzymic synthesis from the latter and guanosine triphosphate (GTP) . 7 3 GDPM is present in yeast 7 2 7* and in hen's 0viduct.7~ It is of interest that yeast contains a ~ylannan.~~ P0ntis,7~ using an improved ion-exchange techniq~e,~' has discovered a 70 Ginsburg Neufeld and Hassid Proc. Nut. Acud. Sci. U.S.A. 1956 42 333. 71 Leloir " Phosphorus Metabolism " Vol. I Johns Hopkins Press Baltimore 1951 7 2 Cabib and Leloir J. Biol. Chem. 1954 206 779. 7 3 Munch-Petersen Arch. Biochem. Biophys. 1955 55 592. 7 4 Pontis Biochim. Biophys. Acta 1957 25 417. 75 Strorninger Fed. Proc. 1954 13 307. 7 6 Haworth Hirst and Isherwood J. 1937 784 ; Haworth Heath and Peat J. 77 Pontis Cabib and Leloir Biochim. Biophys.Acta 1957 26 146. p. 76. 1941 833. 160 QUARTERLY REVIEWS new monophosphate of guanosine. The compound appears to be a phos- phodiester linking guanosine in the 3'-position with an unknown residue. Cytidine Diphosphate Choline and Cytidine Diphosphate Ethanolamine The discovery by Kennedy and Weiss that a cytidine derivative partici- pates in the biosynthesis of lecithin has clarified considerably our under- standing of phospholipid metabolism. Isotope studies 's 79 have shown that choline phosphate enters the lecithin molecule as a unit. I n the presence of a particulate enzyme system isolated from rat liver the incorpora- tion of choline phosphate into lecithin required the addition of an impure sample of adenosine triphosphate (ATP). The activity of this crude nucleotide was not associated with its ATP content but was derived from cytidine-5' triphosphate (CTP) which was present as an impurity.It was shown that both cytidine triphosphate and choline phosphate could be substituted in the multienzyme system from liver by synthetic cytidine diphosphate choline (IX). Similarly CDP-ethanolamine (X) was shown to be an intermediate in the enzymic synthesis of phosphatidyl- ethanolamine.81 n n . . "2N Both CDP-choline and CDP-ethanolamine readily yielded the correspond- ing phosphatides in the presence of particulate systems from liver. The rate of synthesis of lecithin from CDP-choline prepared from labelled choline phosphate was much higher than from CTP and choline phosphate. This and other tracer experiments strongly support the view that CDP- choline is an intermediate in lecithin synthesis.Kennedy and Weiss detected CDP-choline and CDP-ethanolamine in the livers of various animals,sO and the crystalline sodium salt of CDP-choline has been isolated in reasonable amount from yeast.82 Direct comparison of the nucleotide from yeast with synthetic CDP-choline proved their identity. The wide significance of the cytidine coenzymes in phospholipid 78 Kornberg and Pricer Fed. Proc. 1952 11 242. 79 Rodbell and Hanahan J. Biol. Chem. 1955 214 607. 8 1 Kennedy ibid. p. 185. 83 Lieberman Berger and Giminez Science 1956 124 81. Kennedy and Weiss ibid. 1956 222 193. BADDILEY AND BUCHANAN NUCLEOTIDE COENZYMES 161 synthesis is indicated by their participation in cell-free preparations from brain 83 and seminal vesicle.84 Furthermore enzymes which catalyse the synthesis and utilisation of CDP-choline have been detected in liver kidney brain yeast and carrot root.80 According occurs mainly to Kennedy 85 the biosynthesis of phospholipids and through the annexed route.It is not possible to discuss fats here Glycerol +ATP -+ Choline + -1- ATI’ 0- / I 0 CH .O*CO*H the details of all stages in this scheme but a brief outline of the enzymic synthesis of CDP-choline from CTP and the subsequent formation of lecithin is included. The enzymic synthesis of CDP-choline occurs readily in extracts of mammalian tissues to which CTP and choline phosphate have been added. The enzyme is known as phosphorylcholine-cytidyl transferase or PC-cytidyl transferase.86 Reaction occurs according to the equation CTP + Choline phosphate + CDP-choline + Pyrophosphate Magnesium or manganese ions are required to activate the enzyme which is absolutely specific for CTP.Other nucleotides e.g. ATP UTP GTP and inosine triphosphate were without effect. The reaction which is readily reversed is an example of the general type of reaction for the synthesis of unsymmetrically substituted nucleoside pyrophosphate coen- zymes from nucleoside triphosphates. A separate enzyme phosphoryl 83 McMurray Berry and Strickland Fed. Proc. 1956 15 313. 84 Williams-Ashman and Banks J. Biol. Chem. 1956 223 509. a5 Kennedy Ann. Rev. Biochern. 1957 26 119. 8 6 Borkenhagen and Kennedy J . Biol. Chem. 1957 227 951. 162 QUARTERLY REVIEWS ethanolamine cytidyl transferase (PE-cytidyl transferase) catalyses the synthesis of CDP-ethanolamine. In this case the reversible reaction between CTP and ethanolamine phosphate is analogous to the one described above.The mechanism of lecithin synthesis from CDP-choline followed from the recognition 87 that a substrate in this reaction was an ccp-diglyceride. It is now known that the following reaction occurs CH,.OCO.R CDP-choline + R *CO-0- - CH *OH CH,*O*CO -R CMP + CH,*O * P o 0 *CH,CH,-NMe,+ II 0 The enzyme required for this reaction is known as PC-glyceride transferase. An analogous enzyme PE-glyceride transferase catalyses a similar reaction with CDP-ethanolamine. PC-glyceride transferase requires magnesium or manganese ions for activity and is specific towards cc/Ldiglycerides ; tri- glycerides and phosphatidic acids are unaffected. It is also specific for CDP-choline since synthetic UDP-choline ADP-choline and GDP-choline are inactive.Liver cells and presumably cells of other tissues possess an enzyme which regenerates cytidine triphosphate from the monophosphate 88 Cytidine-5' phosphate + ATP f CTP + Adenosine-5' phosphate The reaction proceeds a t the expense of ATP thereby enabling lecithin synthesis to proceed in the presence of only catalytic amounts of CTP and CDP- choline. Although the route for phospholipid synthesis outlined above probably represents the major pathway a t least in mammalian tissue other routes may occur in certain circumstances. For example glycerophosphate formed during glycolysis may well contribute to synthesis of phospholipid and fat in appropriate circumstances. It has also been suggested that the nucleotide cytidine diphosphate glycerol might participate in phospholipid metabolism.However evidence for this is still lacking and other functions for this nucleotide are discussed below. On the other hand the presence of serine and inositol in certain phospholipids suggests that the as yet unknown CDP-serine and CDP-inositol might participate in the synthesis of such compounds. Deoxycytidine diphosphate choline has been isolated from sea-urchin eggs but its biochemical function has not yet been described.S8" The biosynthesis of sphingomyelin is apparently analogous to that of lecithin. Sribney and Kennedy sSb have shown that CDP-choline and 8' Weiss Smith and Kennedy Nature 1956 178 594. 88 Herbert and Potter J. Biol. Ghem. 1956 222 453. 88aSugin0 J . Amer. Chem. SOC. 1957 '79 5074. *8b Sribney and Kennedy ibid.p. 5325. BADDILEY AND BUCHANAN NUCLEOTIDE COENZYMES 163 N-acefyl-DL-threo-trans-sphingosin react together in the presence of liver enzymes to give a sphingomyelin. Sphingosins containing higher fatty acid residues are less reactive. Cytidine Diphosphate Glycerol (CDP-glycerol) and Cytidine Diphosphate Ribitol (CDP-ribitol) These nucleotides were isolated by Baddiley and Mathias from Lacto- bacillus arabinosus. 89 They possess closely similar chemical and physical properties and refined ion-exchange methods were necessary for their separa- t,ion and purificati~n.~~ Both yielded cytidine-5’ phosphate on hydrolysis in acid and contained two phosphate groups to each cytidine residue. The structure of CDP-glycerol (XI) was established by the following observations 91 The venom of the rattlesnake Crotalus atrox which contains a pyrophosphatase and a nucleoside-5’ phosphatase hydrolysed the nucleotide to cytidine (XII) inorganic phosphate and a phosphate of glycerol.Further hydrolysis of the glycerophosphate by the action of prostate phosphomonoesterase gave glycerol and a second inol. of inorganic phosphate. It follows that CDP-glycerol is a derivative of cytidine-5’ pyrophosphate with a glycerol residue on the terminal phosphate group. The location of the phosphate on the glycerol residue was determined by its ready oxidation with periodate to glycollaldehyde phosphate (XIII). It follows that the phosphate residue occupies the a-position. @ CH,*OH CHiOH + H3P04 -I- HO+ H,03P-OCH H,N (XII) I GI ycerol The structure (XI) for CDP-glycerol was confirmed by chemical hydro- lysis.Dilute mineral acid yielded cytidine-5’ phosphate and a mixture of a- and /?-glycerophosphates (mainly a). It would be expected that acid- catalysed migration of the phosphate group on the glycerol residue would occur during the hydrolysis. Hot aqueous ammonia hydrolysed the 89 Baddiley and Mathias J. 1964 2723. 9O Baddiley Buchanan Cams Mathias and Sanderson Biochem. J. 1056 64 599. Baddiley Buchanan Mathias and Sanderson J. 1956 4186. 164 QUARTERLY REVIEWS nucleoOide in a manner analogous to that observed with UDPG. The products were cytidine-5 phosphate and the cyclic glycerol 1 2-hydro- gen phosphate (XIV). The latter was stable to periodate and was hydrolysed by acid to a mixture of a- and /I-glycerophosphate. With alkali it gave more /3- than a-glycerophosphate as is to be expected for a cyclic phosphate of this type.CH,*OH CH ,*OH CMP 4- HO-1- I CH,*OH / \ I (XIV) HO O-CH The configuration of the glycerophosphate residue was deterniiiied eii~ymically.9~ Itl was readily oxidised by diphosphopyridine nucleotide in the presence of glycerophosphate dehydrogenase to give dihydroxya cetone phosphate. This enzyme is known to be specific towards L-a-glycero- phosphate and so the glycerophosphate from CDP-glycerol must possess the L-a-configuration which corresponds with that present in the phospholipids. A synthesis of CDP-glycerol confirms this structure.93 The structure of CDP-ribitol (XV) was determined by methods similar to those used for CDP-gly~erol.~~ With Crotalus atrox venom it was hydro- lysed to cytidine inorganic phosphate and a phosphate (XVI) of ribitol.Prostate p hosp homonoes t erase converted the ri bi t ol phosphate into ri bi t ol and inorganic phosphate. The position of the phosphate group followed from the observation that glycollaldehyde phosphate was formed by oxida- tion with periodate. This could only occur if the phosphate occupied a terminal (primary) position. Hydrolysis of CDP-ribitol with ammonia supports the structure (XV) the compound is more labile than CDP-glycerol under comparable conditions the products being cytidine-5’ phosphate and ribitol 1 2(4 5)-hydrogen phosphate (XVII). The structure of the cyclic phosphate follows from its ready oxidation with periodate and lability towards acid. A synthetic compound prepared by the action of trifluoro- acetic anhydride on ribitol l(5)-phosphate was indistinguishable from that obtained from the nucleotide.Under mild conditions the products were cytidine-5’ phosphate ribitol 1 (5)-phosphate Acid-hydrolysis of CDP-ribitol was unexpectedly complex. 9 2 Baddiley Buchanan and Carss J. 1957 1869. n3 Baddiley Buchanan and Sanderson unpublished work. 9 4 Baddiley Buchanan Carss and Mathias J . 1956 4583. BADDILEY AND BUCHANAN NUCLEOTIDE COENZYMES 165 +O.@ HO OH (XVI I I) +OH CHiO*PO,H CHiO’ ‘OH (xv I) (XVI I) and the isomeric ribitol phosphates which arose through acid-catalysed migration of the phosphate group from the terminal position. However even after short periods of hydrolysis some inorganic phosphate was liberated and more prolonged conditions effected almost complete removal of phos- phate from the ribitol phosphate.The main product 95 of this reaction was 1 4-anhydroribitol (XVIII). No ribitol was formed under the acidic conditions. Ribitol itself gave anhydroribitol but under more vigorous conditions. This reaction which occurs to varying extents with all pentitols and hexitols has been used for the characterisation of ribitol and its phos- phate from CDP-ribit~l.~~ The ease of reaction and nature of products were characteristic in all cases. Straightforward methods could not be applied to the determination of the configuration of the ribitol phosphate residue in CDP-ribitol since ribitol phosphates were hitherto unknown in Nature and consequently no enzymic method was available. By the annexed series of reactions Baddiley Buchanan and Carss 92 degraded a very small sample of the cyclic phosphate (XVII) obtained from the nucleotide by hydrolysis with ammonia to known GO ,H -+OH co ,H CH,*O*PO,H H I I (XVII) -+ -I- 0 0 + ‘ t- 1 / \ C’H,*O OH CO,H (XIX) -+O*PO,H CH,*OH O 5 Baddiley Buchanan and Carss J.1957 4058. 0 6 Idem J. 1957 4138. 166 QUARTERLY REVIEWS compounds which could be determined enzymically. The cyclic phosphate (XIX) of glyceric acid was obtained by oxidation of the ribitol derivative first with periodate then with bromine water. The cyclic phosphate acts as a protecting group in this oxidation and the asymmetry at position 2(4) has been retained. Acidic hydrolysis of this cyclic phosphate gave a mixture of glyceric acid 2- and 3-phosphate. The D-form of both of these esters occur as intermediates in the Embden-Meyerhof scheme of glycolysis in many tissues.A rnultienzyme system from rabbit muscle utilised readily and completely the glyceric acid phosphates obtained in this way from CDP-ribitol. It follows that these were derivatives of D-glyceric acid and so the ribitol phosphate must be that shown in (XVI). As this would be related to D-ribose 5-phosphate (or D-ribulose 5-phosphate) by reduction i t is referred to as D-ribitol 5-phosphate (instead of L-ribitol l-phosphate). Evidence relating to the mechanism of biosynthesis and the function of these nucleotides has appeared recently. An enzyme from Lactobacillus arabinosus catalyses both the pyrophosphorolysis of CDP-glycerol with inorganic pyrophosphate and its synthesis from CTP and a-glycero- phosphate 97 L-a-Glycerophosphate + CTP + CDP-glycerol + Inorganic pyrophosphate CDP-ribitol is probably synthesised similarly from CTP and D-ribitol 5-phosphate.Although the presence of L-a-glycerophosphate in one of these cytidine coenzymes has led to suggestions 91 that it may participate in phospholipid synthesis no evidence has been obtained to support this view. On the other hand Baddiley Buchanan and Greenberg 98 detected a polymeric substance in L. arabinosus which was extracted from the organism with trichloroacetic acid and gave on hydrolysis the products expected from a compound composed of glycerophosphate and ribitol phosphate residues joined together through phosphodiester linkages. It is not yet possible to formulate such a compound accurately since it is not known whether the glycerophosphate and ribitol phosphate residues occur as a mixed polymer or in separate molecules.Glucose and other residues may also be present in the polymer. It is probable that CDP-glycerol and CDP-ribitol parti- cipate in the synthesis of these polymers by successive donation of polyol phosphate residues. The cell walls of L. arabinosus and Bacillus subtilis are now known to contain appreciable amounts (20-30%) of ribitol phosphate and it is likely that this is present as a polymer.gg Very little glycerophosphate was detected in the cell-wall preparations but its presence has been reported in larger amounts in other bacteria in macromolecular structures of uncertain composition. loo The above observations suggest a similarity in the function of the cytidine compounds and the uridine derivatives isolated by Park from 97 Shaw Biochem.J. 1957 66 5 . 6 ~ . Qs Baddiley Buchanan and Greonberg ibid. p. 5 1 ~ . 9Q Baddiley Buchanan and Carss Biochim. Biophys. Acta 1958 27 220. loo Mitchell and Moyle J . Gen. Microbiol. 1951 5 981. BADDILEY AND BUCHANAN NUCLEOTIDE COENZYMES 167 Both types appear to be involved in cell-wall Staphylococci (see below). synthesis. Uridine Diphosphate Acetylmuramic Acid and Related Compounds When Xtuphylococcus aureus was grown in the presence of penicillin acid-labile phosphoric esters accumulated. lol These esters were later shown by Park to be uracil derivatives containing two phosphate groups a pentose and an unidentified sugar.lo2 At least three compounds were detected and two of these contained amino-acids. In later work Park used partition chromatography for their purification.lo3 Cautious acid-hydrolysis liberated a uridine diphosphate from thc three nucleotides. lo* Further hydrolysis yielded uridine-5’ phosphate. The ready hydrolysis of one of the phosphate groups in these compounds suggested the presence of a pyrophosphate group. This was confirmed by electro- metric titration of the diphosphate before and after hydrolysis. It was found that the diphosphate like that from UDPG had two primary and one secondary phosphate acidic group. After hydrolysis an additional secondary phosphate acidic group was liberated. It was also shown that one of the unhydrolysed nucleotides (later iden- tified as UDP-acetylmuramic acid) contained @ HY QH only two primary and no secondary phosphoric CH20*E-O-F*OR acid groups. This suggested that it must bear 0 0 (xx) a substituent on the terminal phosphate of the UDP residue as in (XX).The nucleotide pyrophosphatase from potato hydrolysed UDP-acetylmuramic acid to uridine-5 phosphate thus con- firming the presence of a pyrophosphate group in the molecule. The nature of the group R was not established until later. It was known however that although the nucleotide was non-reducing acid- hydrolysis under conditions which removed the group R caused the appear- ance of a reducing substance. The reducing substance showed reactions characteristic of an N-acetylhexosamine containing a carboxyl group. More recently Park and Strominger lo5 have shown that the N-acetyl- liexosamine present in this and the other two uridine derivatives from 8. aureus is identical with 3-O-l’-carboxyethyl-2-acetamido-2-deoxyglucose (XXI) i.e.the N-acetyl derivative of muramic acid. The structure sug- gested by Strange 106 for muramic acid has been confirmed by synthesis.107 From the above evidence it is clear that all these uridinc derivatives possess the general structure (XX ; R = N-acetylmuramic acid) in which the muramic acid residue is attached to the pyrophosphate residue through it;s reducing group. The configuration of the linkage between amino-sugar imd phosphate is not known. HO 101Park and Johnson J . Bid. Chem. 1949 179 585. l o 2 Park Fed. Proc. 1950 9 213. lo3 Idem J . Bid. Chem. 1952 194 877. lo6 Park and Strominger Science 1957 125 99. lo6 Strange Biochem. J. 1956 64 23~. lo4 Idem ibid. p. 885. lo7 Kent ibid. 1957 67 5 ~ . 168 QUARTERLY REVIEWS HO GH.C)" NHAc X = CHMeC02H utx I) The other uridine derivatives from 8.aweus yield on hydrolysis amino- acids in addition to the products already discussed.lo8 One nucleotide contains a single L-alanine residue and the other contains a peptide composed of L-lysine D-glutamic acid and three alanine residues. The alanine obtained from this peptide consisted of approximately equal amounts of the D- and the L-form (estimated microbiologically). The presence of all three uridine derivatives in small amount in cells which had not been treated with penicillin indicated that they have a function in normal metabolic processes. Although their exact rBle is still not clearly understood it is likely that they are coenzymes concerned with the metabolism of muramic acid. Muramic acid was first characterised as a component of certain peptides from bacterial spores,1Og and occurred in bacterial cell walls.ll0 Moreover both D- and L-alanine D-glutamic acid and L-lysine are also found in considerable amounts in many bacterial cell walls.ll1 It seems likely then that these nucleotides are concerned with the synthesis of cell-wall material.lo5 It is possible to visualise their action as somewhat similar to that of UDPG in glycoside and oligosaccharide synthesis where the nucleotide donates a sugar or in this case more com- plex sugar derivative to a hydroyxl group in another molecule.It is interesting that a t least one other nucleotide of this general type CDP- ribitol is also most probably involved in cell-wall synthesis. Strong evidence for the view that Park's uridine compounds are concerned in cell-wall synthesis comes from the known effects of penicillin on bac- teria.lo5? ll1 The antibiotic seriously affects the walls during the very early stages of its action.This would be consistent with the observation that the accumulation of the uridine compounds is also a primary effect of penicillin action.l12 Further studies on the mechanism of the enzymic processes involving these uridine derivatives should considerably assist our understanding of penicillin action. The Adenosine-5' Phosphate-X Group of Compounds. Acyl Adenylates It is now clear that in several enzymic syntheses acid anhydrides of adenosine-5' phosphate (AMP) are involved. The first of these to be fully authenticated 113 114 occurs during the synthesis of X-acetyl-coenzyme A Io8 Park J. Biol.Chem. 1952 194 897. log Strange and Powell Biochem. J . 1954 58 80. 110 Cummins and Harris J . Gen. Microbiol. 1956 14 583. 112 Strominger J . Biol. Chem. 1957 224 509. 113 Berg J . Amer. Chem. Xoc. 1955 77 3163. 11* Idem J . Biol. Chem. 1956 222 991. Cf. Work Nature 1957 179 841. BADDILEP AND RUClIANAN NUCLEOTIDE COENZYMES 169 from acetate ATP and coenzyme A (CoA). Berg showed that the reaction sequence was as follows Contrary to earlier ATP + Acetate + Rcetyl-AMP + Pyrophosphate Acetyl-AMP + CoA + Acetyl-CoA + AMP Acetyl-AMP (XXII) has been synthesised.l13 116-130 The synthetic com- pound was converted enzyinically into acetyl-CoA. 1131 1 1 * 9 21 Experiments designed to trap acetyl-AMY produced in the above reaction seqnence were u~~successful and it appears that all the acyl-AMP intermediates so far described are very tightly bound to their enzymes ; acetylhydroxamic acid could how- ever be isolated by the addition of hydroxyl- amine to reactions in which CoA was 0 omitted.The formation of hydroxamic acids vxN.> under these conditions is generally regarded N\ N as indicating the presence of acyl anhydrides. n-Butyryl adenylate has been synthesised l 2 O 122 and shown to be an intermediate in formation of butyryl-CoA by the fatty-acid-oxidising eiizyine system o€ liver ; similarly hexanoyl and octanoyl aderiylttte are oxidiaed in liver systems 121 1 2 3 without the addition of ATP. An intermediate in leucine metabolism p-hydroxyisovaleryl-CoA (XXIII) undergoes carboxylation to P-hydroxy-p-inethyIglntaryl-CoA (XXIV) in the presence of hicarhiate and ATP.I2* (XXI I) H2N Ye y 2 Me -$-OH CO .S.CO A (XXIII) 2 2 7 AT P It has been suggested 1z5 that an '' activated " form of carbon dioxide is concerned in this reaction and structure (XXV ; R = H) was proposed fbr it. When silver adenylate was treated with ethyl chloroformate among the products were (XXV ; R = Et) and (XXV ; R = H).125 Experiments 115 Jones Lipmann Hilz and Lynen J. Amer. Chent. SOC. 1953 75 3285. Berg J . Biol. Chem. 1956 222 1015. Avison J. 1956 732. Il8 Stadtman and White J . Amer. Chern. SOC. 1953 75 2022. 119 Baddiley Buchanan and Letters unpublished work. I2O Talbert and Huennekens J . Amer. Chem. Soc. 1956 78 4672. l a l Whitehouse Moeksi and Gurin J . Biol. Chem. 1957 226 813. 122 Peng Biochirn. Biophys. Acta 1956 22 42. l z 3 Jeiicks and Lipmann J .Biol. Chem. 1957 225 207. ly4 Bachhawat Robinson and Coon J . Amer. Chem. Soc. 1954 76 3098 ; J . Biol. lZ5 Coon Fed. Proc. 1956 14 762. Bachhawst Woessner and Coon ibid. 1956 f'hem. 1956 219 539. 15 214. 170 QUARTERLY REVIEWS with synthetic carbonatoadenylate (AMP*CO,) (XXV ; R = H)125912s have shown the following reactions to occur ATP + CO - AMP-CO + Pyrophosphate AMP + CO p-Hydrox y $3-methylglutaryl-CoA It is believed that an activated form of carbon dioxide is also an inter- mediate in propionate rnetaboli~m.l~~9 128 In 1941 Lipmann suggested that protein synthesis from amino-acids might involve carboxyl-phosphate intermediates.lzS Evidence is now accumulating that aminoacyl adenylates are concerned with the formation of several amide linkages. In this connection the enzymic synthesis of hippuric acid and its related p-amino-compound have been investigated.Cohen and McGilvery showed that ATP was required for p-aminohippurate synthesis 130 a>nd it was found that benzoyl-CoA was an intermediate in hippurate ~ynthesis.l~~9 132 Benzoyl phosphate was inactive 131 but the overall reaction CoA c__+ Benzoate + Glycine + ATP Hippurate + AMP + Pyrophosphate has now been dem0nstrated,l3~ indicating the intermediate formation of benzoyl adenylate. Pantothenic acid synthesis has been shown to occur in two stages 134 135 ATP + Pantoate + Pantoyl-AMP + Pyrophosphate Pantoyl-AMP + /?-Alanine + Pantothenate + AMP More directly related to protein synthesis are observations of the " activation " of a number of amino-acids by ATP in the presence of enzymes from widely different sources.It was shown 1369 137 that enzymes from rat liver would catalyse the reaction L-Amino-acid + ATP f L-Aminoacyl-AMP + Pyrophosphate Hydroxamic acids could be formed by the action of hydroxylamine from a number of L-amino-acids. There was evidence that several enzymes were lZo Bachhawat and Coon J . Amer. Chem. Soc. 1957 79 1505. lZ7 Flavin Ortiz and Ochoa Nature 1955 176 823. lz8 Flavin Castro-Mendoza and Ochoa Biochim. Biophys. Acta 1956 20 591. lZ9 Lipmann Adv. Enzymol. 1941 1 99. 130 Cohen and McGilvery J. Biol. Chem. 1947 171 121. 131 Chantrenne ibid. 1951 189 227. 132 Schachter and Taggart ibid. 1953 203 926. 333 Idem ibid. 1954 208 263. 334 Maaa and Novelli Arch. Biochem. Bioplays. 1953 43 236. 136 Maas 3rd Internat. Congress Biochem. Brussels 1955 p.32. 136 Hoagland Biochim. Biophys. Acta 1955 16 288. 13' Hoagland Keller and Zamecnik J. Biol. Chem. 1956 218 345. BADDILEY AND BUCIIANAN NUCLEOTIDE COENZYMES 171 present each being responsible for the activation of separate amino-acids ; that utilising L-methionine has been purified. A similar enzyme was dis- covered in yeast by Berg.l13 138 A range of activating enzymes exists in pancreas and an enzyme specific towards L-tryptophan has been purified considerably. 139 Similar enzymes are present in micro-organisms 140-142 and green plants. 143 Some aminoacyl adenylates have been synthesised chemically and their properties examined.141 1 4 2 9 144 The synthetic compounds exhibit a maxi- mum stability at a slightly acid pH and are readily decomposed by mineral acid or a t a pH greater than 7.The enzymic properties of the compounds giving ATP on the addition of inorganic pyrophosphate together with their behaviour with hydroxylamine leave little doubt that they are the products of “ activation ”. Their precise r81e in protein synthesis remains to be discovered. Adenosine-3‘ Phosphate 5’-Sulphatophosphate (‘ ‘ Active Sulphate ” PAPS) The enzymic synthesis of sulphuric esters of phenols 145-147 has been shown to take place in two stages ; 148-151 the first involves reaction between Q H 0-P- OH t Adenosine - 3 ‘ 5 ‘ 138 Berg J. Biol. Chem. 1956 222 1025. 139 Davie Koningsberger and Lipmann Arch. Biochem. Biophys. 1956 65 21. 140 De Moss and Novelli Bact. Proc. 1955 125 ; Biochim. Biophys. Acta 1955 18 141 Idem ibid. 1956 22 49. 142 De Moss Genuth and Novelli Proc.Nat. Acad. Sci. U.R.A. 1956 42 325. 143 Davis Best and Novelli Fed. Proc. 1957 16 170. 144 Berg ibid. 152. 145 De Meio and Tkacz J. Biol. Chem. 1952 195 176. 146 De Meio Wizerkaniuk and Fabiani ibid. 1953 203 257. 147 Bernstein and McGilvery ibid. 1952 198 195. 148 Idem ibid. 1962 199 745. 149 De Meio Wizerkaniuk and Shreibman ibid. 1955 213 439. 150 Segal ibid.,. p. 161. 151 Hilz and Lipmann Proc. Nut. Acad. Sci. U.S.A. 1955 41 880. 592. 172 QUARTERLY REVIEWS inorganic sulphate and adenosine triphosphate (ATP) to give an intermediate (“ active sulphate ”) which transfers its sulphate group to a suitable sub- strate in the second stage. Active sulphate has been shown by Robbins and Lipmann 152 to be (XXVI). Rye-grass 3’-nucleotidase 153 gives adenosine-5’ sulphatophosphate (APS) (XXVII) which has been compared with a synthetic sample ; 154 mild acid-hydrolysis yielded adenosine-3’ 5’ diphosphate.Adenosine-5’ sulphatophosphnte 1549 155 and active s ~ l p h a t e ’ ~ ~ have both been synthesised chemically. The enzymic synthesis of active sulphate has now been more closely investigated 157-160 and shown to consist of two separable lG0 APS is an intermediate and can be detected when inorganic pyrophosphatase has been added to the appropriate enzyme fraction 160 ATP + Sulphate + APS -/- Pyrophosphate .1 Orthophosphate The enzyme ATP sulphurylase is active with a number of inorganic anions159 elen en ate,^^^ sulphite chromate tungstate and molybdate). In these cases unstable anhydrides are formed and decomposition of ATP to AMP coin- petitively inhibited by sulphate is noted.Active sulphate arises through the action of APS-kinase 159 160 APS -+ ATP -+ PAPS + ADP The reactions described have been found to take place in liver yeast and Neurospora. Active sulphate will transfer its sulphate group to a number of substrates in the presence of sulphokinases.161 Phenols have been used as substrates in the liver system 1469 14*7 1507 1 5 1 9 161 but it is now known that a number of steroids will act as substrates.l6l 162 The enzymic formation of active sulphate and its conversion into chondroitin sulphate in chick-embryo cartilage has recently been rep0rted.16~ Salmon liver is stated to contain an unstable nucleotide bearing a sulphatophosphate group.lS4 In view of the remarkable partial structure suggested for this compound further evidence of its homogeneity is desirable.152 Robbins and Lipmann J. Amer. Chem. SOC. 1956 78 2652; J . Biol. Chena. 153 Wang Shuster and Kaplan ibid. 1954 206 299. 154 Baddiley Buchanan and Letters J. 1957 1067. lb5 Reichard and Ringertz J. Amer. Chem. SOC. 1957 79 2025. 156 Baddiley Buchanan and Letters Proc. Chem. SOC. 1957 147. 15’ Segal Biochim. Biophys. Acta 1956 21 194. 15* Wilson and Bandurski Arch. Biochem. Biophys. 1956 62 503. lb9 Bandurski Wilson and Squires J. Amer. Chem. Soc. 1956 78 6408. I6O Robbins and Lipmann ibid. p. 6409. 161 Gregory and Nose Fed. Proc. 1957 16 199. 162 Schneider and Lewbart J. Biol. Chem. 1956 222 787. 163 D’Abramo and Lipmann Biochim. Biophys. Acta 1957 25 211. 164 Tsuyaki and Idler J. Amer. Chem. SOC. 1957 79 1771. 1957 229 837.

ISSN:0009-2681    

DOI:10.1039/QR9581200152

出版商:RSC    

年代:1958

数据来源: RSC

摘要:

THE STRUCTURE OF CABBONIUM IONS By D. BETHELL and V. GOLD (RING’S COLLEGE UNIVERSITY OF LONDON STRAND W.C.2) I. Introduction Definitions Ternzinology and Nomenclature.-The term ‘‘ carbonium ion ” is generally applied to singly charged organic cations which cannot satis- factorily be represented by conventional formuh in which the quadri- valency of all carbon atoms is preserved. Such species can be considered to be produced by the loss of a negatively charged atom or group (and not of an electron only) from a neutral molecule as for instance the loss of a hydroxide group from an alcohol. Taking a ‘‘ parent ’’ alcohol Et,C*OH for example the simplest bond structure (1) of the resultant ion contains one tervalent positively charged carbon atom. Alternatively the ion could be represented by bond structures such as (2) or (3) which involve either a bivalent hydrogen or a quinquevalent carbon atom and the formation of a ring not present in the alcohol.Formulze of this kind and representations involving varying degrees of participation of these extremes have come to be discussed in recent years (although it should perhaps be pointed out that they are very improbable for the triethylcarbonium ion itself). They are often referred to as ( ( non-classical ” or bridged carboniurn ions in contrast to (‘ classical ” structures such as (1). The validity of these alternatives is discussed in Section IV. Systematic naming of structural formulze such as (2) and (3) has not been attempted. At present the best method of naming a particular ion is by reference to the parent neutral molecule and this may be done by adapta- tion of the carbinol convention.On the basis of this method the ion (l) (Z) or (3) would be termed the triethylcarbonium ion. In some cases it is slightly simpler to name the ion in the same way as the corresponding radical (e.g. phenylcarbonium ion = benzyl cation). Neither kind of name will be used here with structural implications. The name triethyl- carbonium ion may relate to (l) (2) or (3) and moreover it does not rule out the physical existence of more than one (isomeric) ionic structure. The Review will be limited to carbonium ions related to alcohols in the above manner. It excludes acylium ions (RCO+) which are analogously related to carboxylic acids (RCOoOH). Experimental Methods of Investigation.-Some carbonium ions (such as triarylcarbonium ions) are stable species under certain conditions and the It raises some questions of nomenclature.M 173 174 QUARTERLY REVIEWS elucidation of their structure presents a typical problem of structural organic chemistry although the rest'ricted range of conditions in which the ions are stable (frequently in strongly acidic solution) imposes liinitations on the methods that can be applied. The majority of carhonium ions that have been discussed in the chemical literature are however only transient entities. Their existence has been inferred from the study of' the course of organic reactions and their detailed structure is bascd on analysis of the Aner details of the reaction mechanism. Stable ions can be studied by the general methods applicable to ionic solutions such as measurements of electrolytic conductivity and freezing- point depression in appropriate solvents.Absorption spectra of the ions may be obtained and can be applied to the spectrophotometric or colori- metric determination of ionisation constants. For unstable ions the conclusions about structure are based on more indirect experimental evidence. Such evidence relates either to the ener- getics of ion formation derived from studies of mass spectra or to the velocities and detailed rate laws of reactions in which the carboniuni ion is supposed to be an intermediate and to the structure and composition of the products. The stereochemical relation between reactants and products which occupies a position of key importance in this field and rearrangements during a reaction including their elusive forms whose occurrence can be demonstrated by the stratagem of isotopic labelling of definite positions come under this heading.Further it is sometimes possible to draw iiifer- ences concerning the nature of an intermediate carbonium ion from simpler facts of product composition. Evidence from these various lines of investigation-only in so far as it is relevant to the discussion of the structure of carbonium ions-will be con- sidered in the following sections. More general reviews of the chemistry of carbonium ions including the history of the concept and methods of forma- tion have already appeared in these Reviews and elsewhere.1 Special aspects of the subject have recently been surveyed in greater detail.2 II. The Charge on the Carbonium Ion Hybridisation and the Analogy between Boranes and Carbonium Ions.- Compounds of electrically neutral 4-co-ordinated carbon have the tetra- hedral arrangement of valencies associated with sp3-hybridisation of the valency electrons.Carbonium ions can be regarded as compounds of a carbon atom which has given up one of the valency electrons i.e. which are formed by ionisations of the type R3El-X -+ R3C+ 3-X- 1 ( a ) Burton and Praill Q u d . Rev. 1952 6 302 ; ( b ) Leffler " The Reactive Inter- mediates of Organic Chemistry " Interscience Publ. Inc. New York 1956 Chapters v-VII. ( a ) Winstein Experientia 1955 Suppl. 11 p. 137; (b) Cram in Newman's " Steric Effects in Organic Chemistry " John Wiley and Sons New York 1956 Chapter V ; ( c ) Streitwieser Chem. Rev. 1956 56 571 ; (d) Burr " Tracer Applications for the Study of Organic Reactions " Interscience Publ.Inc. New York 1957 Chapter 7. BETHELL AND GOLD THE STRUCTURE OF CARBONIUM IONS 175 If a classical formula such as (l) is adequate for the cation produced the valency electrons would then be hybrids of the sp2-type an arrangement which is expected to lead to a planar trigonal disposition of the bonds attached to the a-carbon atom. The same number of electrons and kind of hybridisation are met in boron and it is therefore to be expected that a carbonium ion Me3C+ would possess the same structure as its isoelectronic boron analogue Me,B. The planar configuration of trimethylboron has been established by electron-diffra~tion.~ The analogy between boranes and carbonium ions cannot be pursued in every detail for various reasons the most obvious of these being the electric charge which is associated with every carbonium-carbon atom.Thus the dimeric carbonium ion (a) which is the analogue of diborane is not a known (4) (5) entity presuma.bly because electrostatic repulsion forces outweigh the stabilisation which might be produced by the hydrogen bridges. By contrast there would appear to be a definite possibility of the existence of entities such as ( 5 ) although they do not seem to have received serious attention so far. An ion such as (1) has no exact parallel in boron chemistry because it contains several carbon atoms only one of which would be described as isoelectronic with boron in a classical representation whereas in a poly- nuclear borane all boron atoms are in the same valency state. On the other hand the tendency of boranes to form hydrogen bridges seems to be related to the non-classical bonding in carbonium ions [as in formula (2)].Both classes of molecular species may be described4 as " electron-deficient ". DebcuZisation of the Charge.-The inherent instability of a charged mole- cule (and hence its tendency to nullify that charge by ionic recombination) is relieved by external and internal dissipation or delocalisation of the charge. External dissipation is achieved by interaction of the ion with the surrounding medium (solvent and oppositely charged counter-ions) (see p. 182). The term " internal dissipation " is intended to cover the various electronic effects which result in delocalisation of t,he charge over a larger part of the ion than a classical formula such as (1) would indicate.Internal Charge Dissipation.-The exceedingly unstable carbonium ion CH3+ is stabilised if methyl groups replace the hydrogen atoms to give Me3C+. Being electron-repelling substituents the methyl groups reduce the positive charge on the centre of the ion and thereby effectively spread that charge (or a t least a large part of it) over the periphery of the ion. This stabilisation can be described in terms of an inductive effect and hyper- conjugation and accounts for the well-known sequence of increasing stability of aliphatic carbonium ions primary < secondary < tertiary. However it has not so far proved possible to make this theory more quantitative. SLevy and Brockway J. Amer. Chem. SOC. 1937 59 2085. Longuot-Higgins Quart. Bev. 1957 11 121. 176 QUARTERLY REVIEWS The occurrence of internal charge dissipation and stabilisation in simple carbonium ions is illustrated by the ionisation potentials of different alkyl radicals (collected by Stevenson and quoted in ref.2c) which are tabulated below. These values relate to separated ions in the gas phase and are therefore not complicated by solvation or other effects of external charge dissipation. They illustrate the increasing charge stabilisation in the sequence Me < E t < Pri < But. Thus even in these simple ions at least a significant portion of the charge must be located on the outside of the ion and not just concentrated on the central carbon atom. In fact the differ- ence in ionisation potentials of Me* and Me,C* (i.e. the enthalpy change of the reaction Me+ + Me,C* -+Me* + Me&+) is of just the order to be expected on the naive electrostatic model of the reaction as the transfer of an electronic charge from a small charged conducting body (of the size of the methyl group) to a larger one (of the size of the tert.-butyl group).? Energetics of ionisation (according to Stevenson 2c) ~ Saturated radicals i Conjugated radicals ~ Ionisation potentials of radicals (kcal.) ~ 312 I 230 AH for gas-phase reac- ' tion RC1+ R+ + C1- (kcal.) * 1 328 1 220 223 Et 201 192 183 171 159 188 ~ 178 ~ 158 161 ~ 152 ~ *Values in the top row are based on heats of formation of the anion and of the corresponding alkane and the appropriate C-H dissociation energy.Values in the bottom row were obtained from appearance potential of R+ in the mass spectrum of RC1 and the electron affinity of chlorine.Stabilisation in Conjugated Ions.-A greater extent of stabilisation of carbonium ions occurs in conjugated systems where it is possible for the n-orbitals of groups attached to the central (or cc-)carbon atom to overlap with the vacant porbital on that atom. Examples of this are in increasing order of stability the ions (CH2-CH-CH2)+ Ph*CH,+ Ph,CH+ and Ph,C+. In fact the great stability of the triphenylcarbonium ion has long been recognised (see ref. 1). The data in the Table illustrate the qualitative difference between saturated and conjugated carbonium ions. It will be noted that the values of the ionisation potentials of the allyl and n-propyl radicals are very close which is consistent with the similar size of and group- ing in the ions formed.Nevertheless the ionisation of allyl chloride is Walsh Discuss. flaraduy Xoc. 1947 2 18. t Since the three carbon valencies are planar both in the radical and in the carbonium the greater difference in stability between Me&+ and Me,C* than between CH,+ and CH,- cannot be attributed to a release of steric strain. BETIIELL AND GOLD THE STRUCTURE OF CARBONIUM IONS 177 energetically much more favourable than that of n-propyl chloride. The enthalpy change in the ionisation RCl-+ R+ + C1- can be divided into three additive contributions corresponding to the hypothetical steps AH, the electron affinity of the chlorine atom is obviously independent of the nature of R. A H represents the ionisation potential of the radical. The results quoted indicate that the difference in the AH values between allyl and propyl chloride for the overall ionisation must be attributed to differences in the first step i.e.to the values of the dissociation energy AH,. Since allyl chloride is in no way abnormally unstable (as a result of steric effects for example) the lower endothermicity of the dissociation of allyl chloride into radicals must be due to the increased stability of the allyl radical. I n other words the same degree of extra stabilisation is found in both the allyl radical and in the allyl cation-to a rough approximation a t any rate. Examination of the values quoted in the Table for the benzyl group leads to the analogous conclusion wix. that benzyl radicals and benzyl cations are additionally stabilised to about the same extent. This stabilisation is therefore not a consequence of the presence of an electric charge but arises from quantum-mechanical deloca,lisation of electrons in the conjugated nuclear framework formed.The phenomenon may be described as mesomerism between the equivalent structures CH z=CH- CH 2* and CHz- CH=CH in the case of the radical or between the structures CH,=CH-CH,+ and +CH,-CH=CH for the carbonium ion whereas resonance structures of this type are obviously impossible for the chloride CH,=CH-CH,Cl. The same number of equiva- lent structures can be written for radical and ion and this simple valency- bond description is therefore consistent with the experimental conclusion that allyl radicals and cations are stabilised by resonance to about the same extent. Huckel’s L.C.A.O. molecular-orbital approximation leads to the same result.The allowed energy levels for the electrons in the allyl framework of carbon atoms are spaced out as shown on the left-hand side of Fig. 1. There are two n-electrons in the allyl cation and three in the radical. The lowest (“ bonding ”) level is therefore doubly occupied both in the cation and in the radical and the middle (“ non-bonding ”) level is singly occupied only in the radical. The highest (“ anti-bonding ”) level will always be vacant. The total n-electron energy of the system is the sum of the energies of the individual electrons e.g. in the case of the radical 2(a + 1842) + a = 3a + 2 8 4 2 . The right-hand side of the Figure shows the spacing of the energy levels in a hypothetical allyl radical in which the double bond occupies a fixed position as in the chloride.The difference in total energy GHUckel 2. Physik 1931 70 204. 178 QUARTERLY FIG. Energy lecels in ally1 Real system CI - pZ/:! - . REVIEWS 1 radical and cation. Localised (hypothetical) system . . g - 6 a a + B 0 Electron occupying level in radical. @ Electron occupying level in cation. u = Coulomb integral. /? = Resonance integral. between the real and the hypothetical system is called the resonance energy. It will be seen by comparison of the two halves of the Figure that this resonance energy will be the same irrespective of the occupation of the middle level ie. it will be the same in radical and cation. This treatment can be extended to all conjugated radicals and cations not possessing odd- membered rings. In every case the highest singly occupied level in the radical is non-bonding and corresponds to the energy u and the resonance energies of radical and cation are identical.From the point of view of structure it is important that the theory allows the average location of the odd electron or positive charge in radical and cation to be calculated. In the valency-bond formulz given above the charge could be placed on either of the terminal carbon atoms but never on the central one. It will therefore be equally divided between the two end carbon atoms. In larger conjugated cations (not possessing odd-membered rings) the charge will always be distributed between the " odd " carbon atoms i.e. carbon atoms 1 3 5 7 etc. if we start numbering a t a terminal carbon. Huckel's method also leads to definite values of charge densities for all odd-numbered carbon atoms and the values are not the same a t non-equivalent positions.(However induction may relay some of the charge also to the " even " positions.) Although this theory gives a remarkably good account of many features of conjugated carbonium ions it is not altoget'her satisfactory in every detail. Thus i t predicts the same value (a) for the ionisation potentials of all conjugated radicals. The Table on p. 176 shows this to be an incorrect conclusion. The basic reason for the discrepancy seems to lie in the neglect of electrostatic repulsion which tends to drive an electric charge to the periphery of the ion but has no corresponding effect on the distribution of the odd-electron spin in the free radical. This charge repulsion can again be relieved more effectively in a larger ion than in a smaller one and accord- ingly the ionisation potentials of conjugated radicals are expected to decrease as the size of the radical (and ion) goes up.This is borne out by the last two ionisation potentials in the Table. The trend is correctly BETHELL AND GOLD THE STRUCTURE OF CARBONIUM IONS 179 reproduced in a modified molecular-orbital theory in which this electrostatic interaction is explicitly taken into account .7 More extensive information about stable conjugated ions is available from studies of ionisation equilibria in solution especially of alcohols in acidic media and of chloride^.^ These investigations have confirmed the order of stability for arylcarbonium ions (primary < secondary < tertiary) in semi-quantitative agreement with Huckel’s molecular-orbital theory,lO and have shown that the stability of the ions is enhanced by electron-repelling substituents in the aromatic nuclei.Perhaps the most spectacular example of the effect of charge delocalisa- tion is the stability of the recently discovered cycloheptatrienyl (tropylium) cation l1 which was predicted by Hiickel in 1931. Tropylium bromide has the properties of an ionic compound. The ions are so stable that the com- pound can be recrystallised from ethanol a solvent with The resonance stabilisation of the tropylium ion is compar- ,.A .,.-. able with that in benzene. Another interesting case is tthe I ...- - ion (6) the high stability of which was also deduced from (6) which most carbonium ions react almost instantaneously. [a]+ \.* molecular-orbital theory l2 before its observation.13 The Extent of Phnarity of Conjugated Carbonium Ions.-The mesomeric stabilisation of a coiijugated system is most effective when all n-bonds lie in the same plane.The interaction energy between different parts of a conjugated system which are not coplanar is expected to diminish as the first or second power of the cosine of the angle of twisting.14 The problem arises in connection with the structure of di- and tri- arylcarbonium ions. Calculations based on known bond-lengths and van der Wads radii indicate that in the planar configuration of the triphenyl- carbonium ion steric interference would occur between the ortho-hydrogen atoms on adjacent rings. Lewis Magel and Lipkin l5 therefore proposed a structure in which each of the benzene rings in the ion is twisted out of the plane which contains the three bonds to the central atom.One estimate 1 6 puts the angle of twist required by this model as high as about 50° with a consequent reduction of the resonance energy. Two isomeric fornis of this structure should exist in one of which all rings are twisted in the same sense as in a propeller whereas in the other (“ distorted helical ”) structure one ‘Hush and Pople Trans. Furadny SOC. 1955 51 600. 8 Williams and Bevan Chem. and I n d . 1955 171 ; Deno Jaruzelski and Schries- heim J . Amer. Chem. SOC. 1955 77 3044; Deno and Evans ibid. 1957 79 5801. 9 Lichtin and Bartlett ibid. 1951 73 5530 ; Liclitin and Glazer ibid. p. 5537 ; Evans Price and Thomas Trans. Paraday SOC. 1964 50 568 ; 1055 51 481. 10 (u) Streitwieser J .Amer. Chem. SOC. 1952 74 5288 ; (b) Gold J. 1956 3344 ; (c) cf. Mason J. 1958 808. llDoering and Knox J . Amer. Chem. SOC. 1954 76 3203. 12Gold and Tyo J. 1952 2184. l 3 Pettit Chem. and Ind. 1956 1306. l4 Pauling and Corey Proc. Nut. Acad. Sci. 1951 37 251 ; Dewar J. Amer. Chem. 15 G. N. Lewis Magel and Lipkin J . Amer. Chem. Xoc. 1942 64 1774. 16 Deno Jaruzelslii and Schriesheim J . Org. Chem. 1954 19 155. SOC. 1952 74 3345 ; cf. Guy J . Chim. phys. 1949 46 469. 180 QUARTERLY REVIEWS ring is twisted in the opposite sense to the other two. It has been possible to interpret the spectrum of the crystal-violet ion in terms of these two forms.15 Simple molecular-orbital calculations of the n-electron energies of mono- di- and tri-arylcarbonium ions have been interpreted lob as implying that resonance does indeed involve all rings and that the twisting does not overwhelmingly reduce the resonance energy.This conclusion is also consistent with measurements of the ionisation of compounds such as (7) in which two of the benzene rings are constrained to approximate ~op1anarity.l~~ Recent studies of the infrared spectra of uc “I (71 crystalline complex halides of the triphenylcarbonium ion have provided strong support for the propeller structure.17’ An alternative structure for di- and tri-arylcarbonium ions has been proposed l8 l6 in which (exactly coplanar) resonance interaction is assumed to involve only one or a t most two of the attached aryl groups. The model was put forward in order to account for the close similarity between the electronic absorption spectra or similarly substituted di- and tri-aryl- carbonium ions and to explain in detail the effect of substituent groups on the basic strengths of the parent alcohols.8 At least some of these phen- omena may be given an alternative explanation in terms of the propeller model.lob Chemical Consequences of Charge Deloca1isation.-Charge stabilisation in a carbonium ion is reflected in the velocity of its formation.The transition state of an SNl reaction such as the solvolysis of an alkyl chloride requires a partial separation of the charges. The partial positive charge will be dis- tributed over the organic portion of RC1 in a similar manner to the distri- bution of the integral charge in the carbonium ion R+. In consequence structural features such as the presence of electron-repelling substituent groups which enhance the stability of the ion will also make the transition state more stable relatively to the starting compound and thus increase the velocity of ionisation.Thus the rates of unimolecular solvolyses of alkyl chlorides in formic acid l9 follow the expected sequence MeCl < EtCl < PrWl < ButCl. Of course experimental data of this kind determined in the main by Hughes and Ingold and their preceded measure- ments of ionisation potentials of radicals and allowed correct conclusions about the stability of carbonium ions the electronic effects of substituents and hence also about the charge distribution to be drawn. However t’he need to consider solvation forces and the possible release of steric strain during an ionisation slightly complicates the interpretation of the phen- omena and has on occasions given rise to controversy The rate for ally1 chloride appears to be somewhat anomalous since in l7 ( a ) Bartlett BuZZ.SOC. chim. France 1951 18 C 100 ; ( b ) Sharp and Sheppard Newman and Deno J . Amer. Chem. SOC. 1951 73 3644 ; Branch and Walba J. 1957 674. ibid. 1954 76 1564. l9 Bateman and Hughes J . 1940 945. 2o For a summary see Ingold “ Structure and Mechanism in Organic Chemistry ” G. Bell and Sons London 1953. BETHELL AND GOLD THE STRUCTURE OF CARBONIUM IONS 181 spite of the mesomeric effect it is only about 25 times greater than that for n-propyl chloride .21 The charge delocalisation also has some more straightforward chemical consequences. Thus the distribution of the charge between positions 1 and 3 in an allyl cation implies that ionic recombination (or further reaction of the ion with another species) may involve either of these positions and in a substituted allyl ion may lead to a product with the original position of the double bond or to the rearranged product,22 as shown Anionotropic rearrangements analogous to this allylic rearrangement are also possible in larger conjugated systems.It must however always be borne in mind that the above unimolecular (XNl’) reaction via carbonium ions is not the only mechanism by which rearrangement may take place.23 It is also expected that the mesomerism in the allyl cation which lowers the double-bond character of the initial double bond would make the system less rigid and may permit geometrical isomerisation e.g. (cis) X \ -C - C H=CH / C- CYH 1’ ‘CH H CH,X \ ,:c=c; R (trans) - .(Re a r r a ng e d) This problem has not been fully studied but there is some evidence for easy cis-trans-interconversion in conjugated carbonium ions.24 It also follows that a carbonium ion which is more effectively stabilised by charge delocalisation will be less reactive towards a nucleophilic entity. This phenomenon is exemplified by the inertness of the tropylium ion mentioned above and in the relative reactivities of different diarylcar- bonium ions as alkylatiiig agents towards anisole the more stable diphenyl- carbonium ion is less reactive t h m the 4 4‘-dichlorodiphenylcarbonium Further stable ions appear to be more ‘‘ discriminating ” their 21 Vernon J. 1954 423. 2 2 E.g. Catchpole and Hughes J. 1948 4 ; for an extensive review see DeWolfe 2 3 de la Mare England Fowden Hughes and Ingold J.Chim. phys. 1948 45 236. 240roshnik Karmas and Mebane J . Arner. Chem. Suc. 1952 74 3807; Bell 25 Bethel1 and Gold J. 1958 1905. and Young Ghem. Rev. 1956 56 753. E. R. H. Jones and Whiting J. 1957 2597. 182 QUARTERLY REVIEWS relative rates of reaction with a series of reagents of varying nucleophilic reactivity span a wider range than the corresponding relative rates for less stable (and more reactive) carbonium ions.26 Solvation.-Ions-or more generally electric charges-in solution are stabilised by virtue of their interaction with solvent molecules. The phenomenon is perfectly general 27 and its detailed consideration here would be out of place. The existence of solvation implies that many solvent molecules are concerned in every ionisation and for this reason some workers prefer to regard &"I reactions as multimolecular rather than as unimolecular.It is relevant to our discussion but rather a vexed problem- to which a definite answer is not available at the present time-whether some of the solvating molecules are attached in a more intimate fashion than others and whether the forces of attraction involved have directional properties. The indications are that for simple inorganic ions (like Na+ or Br-) solvation forces have no covalent component i.e. the attraction is purely electrostatic and therefore non-directional." This does not of course imply that solvation need be equally intense around the periphery of every ion as the nature of the charge distribution or the shape of the ion inay cause the ion-solvent interaction to be stronger in some directions than in others.Solvation numbers derived from various experimental observations arc statistical averages (the method of averaging being that appropriate t o thc phenomenon studied) and are now generally thought not to indicate a permanent association of the ion with a particular number of solvent mole- cules except in " aquated " ions such as Cr(H20)63+. One might expect these general principles to apply to carbonium ions and hence that solvation has no structural significance in the usual meaning of that word. However the opposite view has recently been taken by Doering and Zeiss 28 who consider that there is a partly or wholly covalent ahtachment of two solvent molecules to a carbonium ion so that the car- bonium carbon atom is " pentacovalent ".This involves overlap of both lobes of the vacant p-orbital of the carbonium ion with orbitals of solvent molecules (S) resulting in two C-S bonds which are weaker than the other bonds of the ion (S). Grunwald Heller and KleinZ9 have made the less specific suggestion that the molecules of the R (a) solvation sheath can be thought to occupy a definite number of sites around the ion. Both these proposals were put forward to account for the preponderance of inver- sion accompanying XN1 reactions and will be further discussed in that con- text. The oriented interaction between ions and polar molecules which has R R 2sC. G. Swain Scott and Lohmann J . Arner. Chem. SOC. 1953 75 136. 27 Gurney " Ionic Processes in Solution " McGraw-Hill New York 1953.28 Doering and Zeiss J . Amer. Chein. SOC. 1953 75 4733. 29 Grunwald Heller and Klein J. 1957 2604. * Some writers (cf. ref. l b ) use the term " solvation " to include ordinary covalent bonding between an ion and the solvent but this is not general practice. BETHELL AND GOLD THE STRUCTURE OF CARBONIUM IONS 183 been postulated to exist in benzene solution 30 will be considered in connection with interionic forces in that solvent (p. 185). A review of methods of estimating the variation of solvation energy with the structure of the carbonium ion and with the nature and properties of the solvent lies outside the scope of this article (cf. ref. 2c). Generally speaking the relatively large size of most carbonium ions will cause their solvation energies to be much smaller than those of metal cations.Interionic Forces.-Closely related to stabilisation by solvation is the external charge dissipation and stabilisation of ions by interionic attraction. In every ionic solution some stabilisation results from the ordering which is a consequence of electrostatic interaction of opposite and like charges. For dilute solutions of large ions in solvents of high dielectric constant (such as water or sulphuric acid) this phenomenon is adequately described in terms of the Debye-Huckel theory as formation of " ionic atmospheres ". At higher concentrations or what is more relevant to the present discussion in solvents of low dielectric constant the situation is less well understood. Electrostatic attraction or repulsion varies inversely with the dielectric constant (D) and in a solvent of lower dielectric constant the ordering effect of electrostatic forces gains in importance relatively to the disordering effect of thermal agitation.The model which can generally be invoked to give a reasonably satisfactory description of the physical behaviour (conductivities indicator ionisation etc.) in such systems is that of aggregation of ions into pairs triple ions and higher cl~sters.~l No structural significance has ever been attached to the formation of ionic atmospheres i.e. to the consequences of interionic attraction in the range of validity of the Debye-Huckel theory. On the other hand the kinetic and stereochemical characteristics of reactions involving carbonium ions especially in media of low dielectric constant have been held to require that the phenomena of ion-pair and cluster formation have spatial properties in the sense that the anion and the cation formed by ionisation of an organic compound tend to retain their relative orientations.This implies for example that the leaving anion tends to '' shield " the carbonium ion from attack by another reagent from the same direction. We shall return to the stereochemical aspects on p. 186. XNl reactions in acetic acid (D = 6.2) and benzene (D = 2-25) show great sensitivity to added electrolyte or polar molecules and the phenomena observed have given rise to two further specific proposals about ion associa- tion. Winstein and his collaborators made an extensive study32 of &',I acetolysis and observed for example that the effect of added lithium perchlorate on the reaction velocity was linear over the approximate con- centration range 0-03-0.1~ (called by the authors the " normal " salt 30 (a) Ingold Proc.Chem. SOC. 1957 279 ; ( b ) Hughes Ingold Mok Patai and Pocker J. 1957 1265. 31 Robinson and Stokes " Electrolyte Solutions " Butterworths London 1955 Chapter 14. 32 Fainberg and Winstein J . Arner. Chem. SOC. 1956 78 2763 2767 2780 ; Fain- berg Robinson and Winstein ibid. p. 2777 ; Winstein and Clippinger ibid. p. 2784 ; see also ref. 2a. 184 QUARTERLY REVIEWS effect) but that a stronger accelerating effect (or “ special ’’ salt effect) obeying a different dependence on salt concentration was operative below this range. These observations were interpreted in terms of the postulated existence of two distinguishable and structurally different types of ion pair.The initially formed ‘‘ intimate ” (or ‘‘ internal ”) ion pairs are thought to differ from the secondary ‘‘ loose ” (or (( external ”) ion pairs in that no solvent molecules separate the ions of the pair in the former case whereas they do so in the latter. Ionisation dissociation and solvolysis may thus involve the following steps Ionisation Dissociation _____ + 1 2 3 -1 - 2 - 3 RX -+ rc-tx- -+ R+ IIX- .- R+ -1- x- Intimate Solvent- Dissociated ion pair separated ions ion pair \ r J Solvolysis product The last form i e . free ions is the only one which is sensit,ive to a common-ion effect and may be of negligible importance in the solvent considered. Regeneration of RX from an ion pair is not aided by the addition of the common ion X- and has been called “ internal return ”.On this model the ‘( special ” accelerating salt effect is attributed to a ‘( scavenging ” reaction between the added salt and the solvent-separated ion pair This reaction competes effectively with the ( ( ion-pair return ” reaction (-2) and leads to a rapidly solvolysed ion pair. I n their series of papers 32 Winstejn et aZ. elaborate this interpretation in particular by con- sidering how the stability of the carbonium ion governs the nature of the salt effect. Thus “ special ” salt effects are considered to be absent if the reactivity of the ion is so high that it is completely destroyed by solvent attack on the (‘intimate” ion pair. The “normal ” salt effect is con- sidered to influence the ionisation rate (1). As an alternative and much simpler explanation it has been suggested 33 that the ‘‘ normal ” and the r c special ” salt effect may be caused by ion pairs and free ions of the added salt respectively and by implication that the assumption of two types of ion pair is unnecessary.In order to assess the adequacy of the simpler model a detailed consideration of Winstein’s results from this point of view would be welcome. Studies of a rearrangement reaction in which a toluene-p-sulphonate group labelled with 1 8 0 in the alcohol-oxygen migrates from one point of attachment in a molecule to another show incomplete equilibration of 1 8 0 among the three oxygen atoms of the sulphonate group. This result has been interpreted as indicating definite structural attachment of the two halves of an ion pair,34 but it might also arise from the existence of non- ionic intramolecular paths for the rearrangement.33 Ref. 30b footnote on p. 1278. 3 4 Denney and Goldstein J . Anzer. Clbevt. Xoc. 1957 ‘79 4948. BETHELL AND QOLD TJIE STRUCTURE OF CARBONIUAT IONS 188 The importance of the state of ionic aggregation of added salts (reagent or otherwise) has been clearly pointed out in connection with a study of X N l reactions in benzene.30 Here too electrostatic catalysis by ion pairs was observed but the outstanding feature of the experiments was the dis- covery of unimolecular chloride exchange azide replacement and methan- olysis of triphenylmethyl chloride the velocity of the reactions being independent of the concentration but not of the nature of the substituting reagent. The interpretation of this result is based on the hypothesis that the unimolecular generation of carbonium ions is followed by an instan- taneous association of the ion pair formed with an ion pair (or molecule in the case of a non-electrolyte) of the reagent.This step is in turn followed by a rearrangement of the cluster to a configuration favourable to reaction e.g. 4- M+X- TC1 T+C1- -5 T+Cl-M+X- -5 Slow Rapid Slow JLapid Rapid Slow Ion Non-reacting pair cluster Fbpid T+X-M’-Cl- _I TX + M+C1- Reacting cluster The postulated rapidity of the second step is made plausible by the magnitude of electrostatic forces in benzene. As Ingold points two univalent counter-ions attract each other in benzene with an energy equal to the mean kinetic energy along a line a t a separation of 500 A. The mean separation of solute neighbours in these solutions was about 50 A and is quite likely to correspond to a distance a t which electrostatic attraction between two ion pairs exceeds their kinetic energy of translation.The application of this model to unimolecular eliminations in the gas phase has also been proposed. The structural concept introduced in this theory is the notion of what might be called “ cluster isomerism ” i.e. different arrangements of four ions in a quadrupole cluster can be distinguished. We are not here con- cerned with other kinetic repercussions of Winstein’s or Ingold’s proposals. If the structural ideas in either scheme have physical reality it should be possible to find support for them outside the immediate phenomena which caused their postulation. From this point of view Ingold’s model appears to be more interesting.The discovery of such confirmatory evidence might open a new vista in structural chemistry. III. The Geometry of Carbonium Ions The Xteric Course of SN1 Reactions.-On a simple view the change in the hybridisation state of the central carbon atom from sp3 to sp2 during the formation of a carbonium ion involves the replacement of the tetrahedral arrangement of bonds in the original molecule by a planar configuration in the ion and hence loss of optical activity if the central atom was the only asymmetric one in the molecule. Recent experiments have shown the need to modify this straightforward statement. Evidence which indicated car- bonium-ion formation has sometimes been accompanied by other observa- tions apparently inconsistent with a planar configuration for the ion.186 QUARTERLY REVIEWS The relation between reaction mechanism and the steric course of replacement reactions was pointed out in the 1930’s by Hughes Ingold and their co-workers.20 Briefly their views were as follows. The unimolecular mechanism ( SNl) of reaction involves intermediate formation of a carbonium ion. If the asymmetric carbon atom in an originally optically active com- pound is the seat of reaction the ionisation will yield an ion which by virtue of its plana’rity will react to give racemic products. Such behaviour has for example been observed in the reactions of certain diarylmethyl compounds.35 I n the single-step (bimolecular XN2) replacement the geometry of the transition state is such that each molecular act of substitu- tion inverts the configuration of groups about the central (asymmetric) carbon atom.However in many cases of reactions known to proceed by the unimole- cular mechanism the product has been found t o be optically active indi- cating that some inversion of configuration had accompanied a predominant racemisation. For example acetolysis of optically active l-phenylethyl chloride gave a product 15% inverted and 85% ra~emised.~~ Degrees of inversion up to 54% have been observed 28 in methanolysis. An interesting case is the oxygen-isotope exchange of alcohols in aqueous acid this is thought to involve carbonium ions generated by loss of water from the protonated alcohol but is accompanied by complete in~ersion.~’ The observations can be reasonably explained in terms of the concepts of shielding and lifetime of the carbonium ion.An inherently stable ion will have a comparatively long life surviving a number of collisions before finally reacting with a nucleophilic reagent to yield the product. During this life i t will be able to free itself completely from the anion (the leaving group) and hence the “ front ” and the ‘‘ back ” of the ion will become equivalent An inherently less stable carbonium ion on the other hand will tend to react very soon after its formation and may do so before it has completely freed itself from the leaving group e.g. while it is still associated with its original ionic partner in an ion pair. The leaving group will thus tend to shield one side of the carbonium ion favouring reaction at the opposite side with consequent partial inversion. This theory implies definite spatial orienta- tion of the ions in a freshly formed ion pair which-in Winstein’s terminology -would probably have to be described as an “ intimate ” ion pair.A more quantitative elaboration of the theory has been put forward by Grunwald Heller and Klein.2B They consider that in a reaction with solvent molecules (which fill the dual rBle of first solvating and then destroy- ing the cation) the departing group will occupy the site of one of the solvating molecules and thereby reduce in an asymmetric manner the number of sites from which attack on the cation can take place. The experimentally deter- mined excess of inversion can then be used to calculate the number of solvation sites. Small and reasonable values are obtained. The reaction product will then of necessity be racemic.35 Davies and Kenyon Quart. Rev. 1955 9 203. 3 6 Steigman and Hammett J . Arner. Chern. SOC. 1937 59 2536. Bunton Konasiewicz and Llewellyn J. 1955 604. BETHELL AND GOLD THE STRUCTURE OF CARBONIUM IONS 187 Doering and Zeiss's model 2 8 s 2c has been expressed in a different termin- ology. It is effectively based on the same picture in that the leaving group is supposed to occupy rz solvation site on one side of the plane of the car- bonium ion. With a stable carbonium ion the anion is replaced by a solvent molecule during the life of the ion producing a symmetrical carbonium ion (and racemisation) ; an unstable carbonium ion is attacked by solvent before the anion has separated. The distinctive feature of the scheme is the hypothesis that only two sites (one on each side of the plane containing the carbonium valencies) are considered to be involved in the solvation of the carbonium ion.Furthermore it is speculated that the two solvent mole- cules (or one solvent molecule and one leaving group) are held covalently (see formula 8). In all discussions of these phenomena it is possible to replace the concept of '' life-time '' of the carbonium ion by the equivalent one of rate constants for the replacement of the shielding ion by a solvent molecule and for the attack of solvent on the carbonium ion as is done by Doering and Zeiss. Xteric AcceZeration.-The formation of a planar carbonium ion from a molecule of tetrahedral configuration implies an opening-out of the remain- ing bonds from an initial (tetrahedral) angle of around 109" to one of 120".Thus any congestion existing between the groups attached to the central atom will be alleviated during ionisation. An effect of this kind is likely to be observed in the values of the equilibrium constants for the formation of triarylcarbonium ions with bulky ortho-substituents.16S l8 It is also to be expected that the solvolysis of organic halides having a congested structure would be faster than that of similarly constituted but non-congested structures since a t the transition state of ionisation the bonds will have opened out to somewhere between 109" and 120". Such behaviour has been reported in solvolyses of a number of bulky tertiary alkyl compounds e.g. tri-ter.t.-butyl halides,38 but it must be borne in mind that alternative explanations of this steric acceleration are sometimes pos- sible e.g.hyperconjugation 39 or the formation of bridged carbonium ions .4O 1 7 Bridgehead Carbonium Ions.-Just as the steric compressions between groups attached to the central carbon atom in certain tertiary halides can be relieved during ionisation and lead to faster solvolysis so structures which prevent the increase in bond angles which is the concomitant of ionisation should show diminished reactivity in unimolecular reactions. Such be- haviour has been observed in bicyclic halides in which the halogen is attached to a bridgehead carbon atom. Thus refluxing l-apocamphanyl chloride (9) with aqueous-ethanolic potassium hydroxide or silver nitrate failed to remove the hal0gen.~1 The broinotriptycene (10) showed a similar lack of reactivity despite the presence of three phenyl groups.In a favourable 38 E.g. Bartlett and M. S. Swain J . Amer. Ciiem. SOC. 1055 '77 2801 ; H. C. 38 Hughes Ingold and Shiner J. 1953 3827. *O Ingold ref. 20 p. 417 ; cf. H. C. Brown and Kornblum J . Amer. Ch,em. SOC. hown J. 1956 1248. 1954 76 4510. *lBartlett and Knos ibid. 1939 61 3184. 188 QUARTERLY REVIEWS orientation (but not in the forced arrangement of formula 10) the phenyl groups would stabilise the corresponding carbonium ion (cf. the triphenyl- carbonium ion). Unlike the triphenylmethyl halides this bromotriptycene dissolves in liquid sulphur dioxide to yield a colourless non- conducting solution .42 The three o-phenylene rings in bromotriptycene (10) are disposed symmetrically about the Hr-C-C-H axis. Kekul6 bonds have been omitted from the forward-protruding ring for c' (9) @$ (10) clarity.In both cases the rigid cyclic structure will not admit coplanarity of the bonds to the bridgehead carbon atom thereby preventing ionisation. It has been pointed out l7 that the cyclic structure by hindering the rearward approach of nucleophilic reagents also prevents reaction by the SN2 mechanism. Direct evidence of the difficulty of forming bridgehead carbonium ions has been provided by electron-impact studies.43 The appearance potentials of the ions (11) and (12) were found to be appreciably greater than that of an analogous planar ion the trimethylcarbonium ion. Significantly the ion (12) has a lower appearance potential than its analogue (ll) in keeping with its more flexible structure which will permit a closer approach to coplanarity of the three bonds to the carbonium-carbon atom.Further a qualitative parallelism exists between the appearance potentials of the ions and the reactivities of the corresponding bromides in unimolecular solvolyses .44 IV. Bridged Caxbonium Ions Origin of the Concept.-In recent years evidence has accumulated that the structures of certain carbonium ions cannot adequately be represented by the normal (or " classical ") valency formulae considered in Sections I1 and 111. Ions which have to be described in terms of the more recent and unconventional formu18 have come to be known as " non-classical " ions. Some of t b structural notions have been outlined in the Introduction. The idea of bridged non-classical ions arises from an explanation offered by I. Roberts and Kimball twenty years ago 45 to account for the observation that polar addition of bromine to a double bond gives exclusively the " truns "-product.They suggested that the ion (13) is an intermediate in 42Bartlett and E. S. Lewis J. Amer. Chem. Soc. 1950 '72 1005. 43Franklin and Field J. Chem. Phys. 1953 21 550. 41 Doering Levitz Sayigh Sprecher and Whelan J. Amer. Ghem. SOC. 1953 75 4 5 I. Roberts and Kimball ibid. 1937 59 947. 1008. BETHELL AND GOLD THE STRUCTURE OF CARBONIUM IONS 189 the reaction the vacant p-orbital of the carbonium-carbon atom (in the classical ion) overlapping with a lone pair orbital of the bromine thus creating a bridged structure. Rotation about the carbon-carbon bond (which could occur in the classical ion 14) would thereby be prevented and subsequent attack by bromide ion would be from the direction opposite to the bromine bridge.Evidence for similar bridged ‘‘ bromonium ” ions has also come from studies of sub~titution.~~ The configuration of the di- bromides formed by reaction of hydrogen bromide with threo- and erythro- bromohydrins is consistent with the hypothesis that the fwst bromine serves to (‘ hold ” the configuration of the central bond in the ionic intermediate. A summary of the substitutions studied is contained in the reaction scheme below in which stereochemical formulz have been drawn according to New- man’s c~nvention.~’ (The molecule is drawn as a substituted ethane viewed end-on. The bonds in front of the circle represent bonds to groups attached to the ethane-carbon atom nearer the reader and the bonds emerging from behind the circle are those of the ethane-carbon further away.) The bonding of the first bromine to both carbons in the intermediate ion is essential to the explanation of the products formed.The formation of the three-membered ring in the first step occurs with a Walden inversion a t the rear carbon atom (from which OH is lost) and so does the attachment of H H&::- OH {-)- threo (Br attacks front or rear carbon) (+)-erythro H WMe Br. enan tiorners Br H Br Br Br iden ti ca I Br Me@Ce H Br 413 Winstein and Lucas J. Amer. Chem. Svc. 1939 61 1576 1581 2845 47 Newman J . Chem. Edtic. 1955 32 344. N 190 QUARTERLY REVIEWS bromide and breaking of the ring in the second step. The reactions have been given in detail since their underlying stereochemical principles also form the basis of Cram's elegant demonstration of bridging by phenyl groups which will be mentioned below although the general consideration of organic cations containing elements other than carbon and hydrogen falls outside t'he scope of this Review.Nevell de Salas and Wilson 48 extended the concept of bridged structures to ions containing only carbon and hydrogen. They suggested that the cationic intermediate in the acid-catalysed rearrangement of camphene hydrochloride (15) to isobornyl chloride (17) had the structure (16). In classical formulze of this ion either C(l) or C(2) would be written as tervalent and carrying the positive charge (corresponding respectively to the initial and the final structure in the rearrangement). In the non-classical formula the carbonium character is divided between these carbon atoms both of which are now joined to C(s) which thus forms a bridge between these atoms.* The last few years have brought to light many instances of chemica behaviour that seems to require this type of structure.The evidence has come in the main from kinetic stereochemical and tracer investigations of XN1 solvolyses. There does not appear to be any evidence for bridged structures in stable carbonium ions. Kinetic Eflects of Neighbouring-group Participation.-Unimolecular sol- volyses (SN1 and E l ) have as common rate-determining first step the generation of a carbonium ion through heterolysis of the starting specics. In some reactions of this type the overall change is attended by a skeletal rearrangement (Wagner-Meerwein change) and it has been found that these particular solvolyses are often (but not always) unexpectedly fast.Simple examples are neopentyl chloride (Me,C*CH,Cl) and 2 2 2-triphenylethyl chloride (Ph,C*CH,Cl) both of which react with wet formic acid a t 95" by a unimolecular mechanism to give rearranged products (tert.-amyl com- pounds and triphenylethylene respectively). The reaction velocity for the first compound has a value of the order expected for the ionisation of a primary halide of this type but the velocity for the second compound is 60,000 times greater,49 too large to be accounted for in terms of the rela.tive inductive effects of the methyl and the phenyl groups. The accelerated unimolecular reaction of the triphenylethyl chloride must be a manifestation of an additional effect facilitating ionisation.This can be explained as 48 Nevell de Salas and Wilson J. 1939 1188. 4 9 F . Brown Hughes Ingold and Smith Nature 1951 168 65. * Ingold has coined the word " synartesis " to describe this kind of phenomenon The name is intended to suggest that " a split single bond ' fastens together ' the loca tions of a split ionic charge ''.20 BETITELL AND GOLD THE STRUCTURE OF CARBONIUM IONS 191 pa'rticipation by a neighbouring phenyl group in the ionisation with forma- tion of a non-classical ion (18) in which there is bridging by the phenyl group." w The various possibilities for ionisation rearrangement and product formation in systems of this general type are summarised in the attached scheme which also indicates the assumed stereochemical course. In the two classical ions rotation about the bond along which we are looking would be possible.The formation of the bridged ion may be likened to 51a an Sx2 displacement on the nearer carbon atom and its rupture to an SN2 displacement on either the rear or the front atom depending on whether or not a rearranged product is formed. Y Unrear ra nqed Bridged ion p- product [&I+ + Y ' 2nd ciassial ion Rearranged product The generation of the bridged ion in preference to the classical ion in any particular reaction is intelligible if the non-classical ion is the more stable. In such a case the transition state for the formation of the bridged ion would also be stabilised in a corresponding manner by the incipient formation of the bridge i.e. by neighbouring-group participation. The rate would accordingly be increased and the primary step of the reaction would be the production of the bridged ion.The occurrence of rearrangement is not itself evidence about the structure 50 Winstein Lindegren Marshall and Ingraham J. Amer. Chem. Xoc. 1953 75 147. 511ngold ref. 20 pp. ( a ) 511 ( b ) 523. * Winstein has proposed the adjective " anchimeric " to describe the effect of neighbouring-group participation as a result of which a rate-determining imisation may be accelerated. The term applies to neighbouring carbon hydrogen and also func- tional groups.50 192 QUARTERLY REVIEWS of the ion. Even when the classical ion is more stable than the bridged one rearrangement may take place. It then involves the activated transforma- tion of one classical ion into the other. The transition state of this re- arrangement resembles the stable bridged ions in the disposition of the groups but differs from them in being more (and not less) energised than the classical structures.The case of the neopentyl cation cited above could be an instance of such a rearrangement.* Abnormally high ionisation rates are important evidence for the forma- tion of bridged ions but this interpretation is not always the only possible one. In particular they may be caused by steric acceleration i.e. a non- bonding effect of neighbouring groups which causes the ionisation to be att'ended by a release of steric strain. I n many cases this possibility can be excluded. For example the S,l etlianolysis of isobornyl chloride (17) is 105 times faster 499 53 than that of bornyl chloride (19). Since the departing chlorine is less crowded in the isobornyl compound than in the bornyl compound the rate difference cannot arise from steric acceleration but it is thought to indicate the formation of the ion (16) as a reaction intermediate.I n isobornyl chloride the conformational dispositions of neighbouring-group and departing chlorine are favourable to the formation of the collinear transition state of the internal displacement reaction whereas they are unfavourable to fhis formation of the bridged ion from bornyl chloride. I n certain instances quite small rate increases (10-fold or less) have been attributed to neighbouring-group participation. Such explanations must be regarded with caution unless they are supported by other evidence. A small acceleration may well be a consequence of the ordinary inductive effect of the neighbouring group.Also it is perhaps sometimes assumed too readily that the solvolyses studied follow the SNl mechanism. Stabilisation of Bridged Structures.-The cause of the peculiar stability ascribed to some bridged carbonium ions is by no means fully understood. In general terms one may seek to classify the stabilising factors under the headings of mesomerism inductive charge spreading and steric (including conformational) effects. Only the first of these appears to be capable of providing extra stabilisstion of the required order of magnitude. Meso- merism is of course implied in formulz such as (16) with the aid of partial bonds. This procedure- structures 0 R + ,c-c - as is sometimes done. is equivalent to writing + ,+ R\ \ / R \ / ,c-c( - ,c-c It is not clear why this a number of resonance R+ / 4-b' - ,c- c resonance should be so 5 2 Winstein and Marshall J .Amer. Chem. Soc. 1952 74 1120. 53 Winstein Morse Grunwald H. W. Jones Corse Trifan and Marshall ibid. * Detailed consideration of the reaction velocities for neopentyl compounds has led to the suggestion that even in this case there is neighbouring-group assistance through methyl bridging.62 p. 1127. BETHELL AND GOLD THE STRUCTURE OF CARBONIUM IONS 193 important here when in other cases the delocalisation of C-C a-bonds (or C-C hyperconjugation) is energetically rather trivial. I n broad qualitative terms the effect may be ascribed to the electron-deficiency (electronic sextet) of the carbonium-carbon atom. As in the case of boron this somehow causes unusual electronic effects to come into play so that the vacant levels may achieve at least partial occupation.51b In bridging by a phenyl group this resonance stabilisation looks perhaps more convincing than in the case of alkyl groups. Resonance in these ions -which Cram calls " phenonium " ions-can now be considered to involve structures such as (20) and (al) in addition to those written above for + bridging by an alliyl group although against this we must set the loss of the normal benzene resonance within the bridging phenyl group. Simonetta and Winstein 54 have calculated the n-electron energy of such a system on the basis of a simple molecular-orbital procedure and have come to the con- clusion that phenyl bridging causes a considerable increase in resonance energy. It is a less satisfactory feature of the quantitative aspect of this interpretation that several solvolyses exist for which stereochemical evidence indicates phenyl-group participation but which are not accelerated (seep.194). It is worth noting that the predicted disposition of the bonds a t the bridging carbon atom of the phenyl group is the same as the tetrahedral arrangement proposed for the intermediate in certain aromatic substitution reactions.55 I n extension of the idea of participation of phenyl groups at a neighbour- ing atom the possibility of aryl participation a t atoms further along an aliphatic chain has been On the basis of detailed com- parisons of reaction velocities Heck and Winstein 56 consider that there is evidence for the following carbonium-ion formation during acetolysis (OBs = p-bromobenzenesulphonate) 56 OMe - OMe * O O M e (22) They could find no evidence for participation a t positions other than 6 along a side chain or by unsubstituted phenyl groups.Xtereochemicul Evidence.-An example of the importance of other types 54 Simonetta and Winstein J . Amer. Chem. SOC. 1954 76 18. 5 5 ( a ) Melander Acta Chem. Xcand. 1949 3 95 ; (b) Corey and Sauers J . Amer. 56Heck and Winstein ibid. p. 3105. Chem. Soc. 1957 79 248. 1 94 QUARTERLY REVIEWS of evidence for bridging is provided by the acetolysis of t,he toluene-p- sulphonates of the isomeric 3-phenylb~tan-Z-ols.~~ In this case there is little or no acceleration but the stereochemical results are sufficient to establish neighbouring-group participation. 3-Phenylbutan-8-01 exists in threo- and erythro-forms each of which is an enantiomeric pair.Acetolysis of the optically active erythro-toluene-p-sulphonate (23) gave the erythro- acetate (24) of about 94% optical purity whereas the threo-sulphonate (25) gave racemic threo-acetate (26). Thus both reactions proceed with retention of configuration but only in the erythro- case is optical activity preserved. These results were interpreted in terms of intermediate formation of Me Me&; OAc Me *fiY4 OTs L-crythro (23) -z + Med$J H c Me OTs (AcOH attacks front or rear carbon ) L - f hreo (25) Ph H Me ident ica I H OAc OAc J ‘‘ phenonium ” ions as illustrated. The threo-sulphonate yields a phen- onium ion possessing a plane of symmetry (in the plane of the paper) and reaction a t either of the central carbon atoms then gives the threo-acetate but with loss of optical activity.No such plane of symmetry exists in the case of the erythro-ion so attack a t either carbon atom yields the same optical enantiomer. It has been suggested 5* that the stable form of the phenonium ion may be represented as a dynamic equilibrium between (27) and (28) (the threo- case being illustrated here). The symmetrical form essential to the explana- tion of the stereochemical course is the transition state for this intercon- version. It is to be noted that some configuration-holding interaction ~ __ 57 Cram J . Amer. Chem. SOC. 1949 71 3863. 58 Winstein Brown Schreiber and Schlesinger ibid. 1952 74 1140 ; Winstein and Schreiber ibid. 1). 2165. BETHELL AND GOLD THE STRUCTURE OF CARBONIUM IONS 195 between the central carbon atoms and the neighbouring groups must occur throughout the reaction; otherwise both carbonium ions would assume the most favourable conformation by rotation of the central bond and inter- conversion of threo- and erythro-forms would occur.Alternatively it must be assumed that the destruction of the carbonium ion is rapid compared with the speed of rotation about a single bond (see also p. 198). A large number of related systems have been examined by Cram and his co-workers with the general conclusion that both open (classical) and phenyl-bridged cations can be intermediates and that the successive forma- tion of both these types may also be 2b The simultaneous possibility of migration of (and bridging by) different groups in the same molecule also brings in another " non-classical " type of isomerism.Redistribution of an Isotopic Label.-As has already been indicated under the preceding heading the non-classical structures for carbonium ions may have different symmetry from that of the corresponding classical structures. As a consequence positions which are non-equivalent in the starting molecule or in the classical ion may become equivalent in the non-classical ion so that isotopic labelling of one such position should result in random distribu- tion of the isotope amongst all positions which become equivalent in the ion I n the hands of J. D. Roberts and of other workers this technique has proved one of the most searching tools in the determination of carbonium-ion structure. It has suggested the inadequacy of classical structures in certain cases confirmed their adequacy in others and revealed unsuspected com- plexities that may require modification of the simplest views of bridging in carbonium ions.An interesting example is the XNl acetolysis of exo-norbornyl p-bromo- benzenesulphonate (29) which because of its high velocity was considered to proceed by formation of a non-classical ion to which the structure (30) was assigned.60 As the projection formula shows this equivalence of positions 1 and 2 and of positions 3 and structure implies 7. Labelling the molecule (29) with 14C equally a t positions 2 and 3 should thus result in equal distribution of the isotope over positions 1 2 3 and 7 in the product. The isotope was in fact found in all these positions but in addition 15% of the total radioactivity appeared a t positions 5 and 6 a result 61 which is not immediately explained by the structure considered.The observations are consistent with the hypothesis that the ion (30) rearranges by a hydrogen shift from position 2 t o position 6 to the extent of about 45%. Hence positions 1 2 and 6 become shuffled to a certain extent and there a,lso is the same amount of interchange among positions 3 5 and 7. It may now be 69Cram and Allinger J . Amer. Chem. Soc. 1957 79 2859. e0 Winstein and Trifan ibid. 1952 74 1154. 61Roberts Lee and Saunders ibid. 1954 76 4501. 196 QUARTERLY REVIEWS asked whether this hydrogen shift indicates some extent of hydrogen par- ticipation in the ionisation step i.e. whether some of the ions formed have a structure involving both carbon and hydrogen bridging (31) or (32).It has been pointed out Pa that the extent of additional carbon shuffling (in- volving the 5- and the 6-position) depends on the nucleophilic activity of the solvent and is zero for very reactive solvents in which the life of the car- bonium ion is short. This would indicate that the hydrogen transfer com- petes with the solvent attack and is therefore subsequent to the formation of the carbonium ion. (30 (32) There is also other evidence for hydrogen shifts across cyclic carboniuin ions. The deamination (with nitrous acid) of cyclodecylamine isotopically labelled a t the l-position yields products with the isotopic label not only in the I- but also in the 5- and the &position. The reaction is believed to involve carbonium ions and the result can therefore be interpreted as a transannular hydrogen shift in the cyczodecyl cation.62 Roberts and Yancey 63a have also looked for evidence of hydrogen bridging in the ethyl cation by examining the distribution of 14C in ethanol formed by deamination of CH,-l4CH2*NH2. The isotopic label stays almost completely in the original position. The two ends of the molecule do not therefore become equivalent a t any stage during the reaction and if the reaction proceeds by way of carboniuin ions the ions cannot have a sym- metrical structure such as the bridged formula (33). Corresponding studies with labelled 2-arylethylamines 63b (to detect aryl- bridging) and with n-propylaniine 63c (to detect methyl- bridging) showed the occurrence of some rearrangement but in no case was the isotope found to be equally distributed between the 1- and the 2-position.The reactions cannot therefore proceed entirely by formation of bridged ions. They may do so in part ; but alternatively the isotope shuffling could be caused by rearrangement of classical ions. Another interesting case is presented by the cyclopropylcarbonium ion. The rapid S,l solvolysis of cyclopropylmethyl derivatives (34) suggests a non-classical structure for the ion. 64y 6 5 Acetolysis of cyclopropylmethyl chloride and of cyclobutyl toluene-p-sulphonate and deamination of cyclo- propylmethylaminey yield mixtures of products containing amongst others similar proportions of cyclopropylmethyl and cyclobutyl derivatives. These 6 2 Prelog Urech Bothner-By and Wursch Helv. Chim. Acta 1955 38 1095 ; see also Prelog and Kiing ibid. 1956 39 1394 ; Urech and Prelog ibid.1957 40 477 ; Prelog Experientia 1957 Suppl. VII p. 261. 63 ( a ) Roberts and Yancey J . Amer. Chem. SOC. 1952 74 5943 ; ( b ) Roberts and Regan ibid. 1953 7'5 2069 ; (c) Roberts and Halmann ibid. p. 5559. 64Roberts and Mazur ibid. 1951 73 2509. 6 5 Bergstrom and Siegel ibid. 1952 74 145. BETHELL AND GOLD THE STRUCTURE OF CARBONIUM IONS 197 results have been explained in terms of rapid interconversion of cyclo- propylmethyl and cyclobutyl cations. As a result of such interconversions each of the original carbon atoms 1 3 and 4 may appear in the side-chain of a cyclopropylmethyl compound in the product. If position 1 is isotopically labelled with 14C the radioactivity should become equally distributed among carbon atoms 1 3 and 4. Such shuffling of carbon is in fact observed although the amounts of radioactivity in the three positions are not quite equal.I n view of the fact that there is rate enhancement for the formation of the ion an attractive (though not the only possible) interpretation of the results is the suggestion that there is intermediate formation of some car- bonium ions of the structure (35) which has a three-fold axis of symmetry. IycH*cH,x $!i$.\ i CH2- /6? - -‘-CH I+ (35) [ lH\\ 1’ ’CH H *c’- CH (33) (34) *CH The structure of the spirocarbonium ion (22) mentioned earlier implies that the cc- and the &position of the starting compound become equivalent (and similarly the ,6- and the y-position). It should therefore be possible to test the correctness of structure (22) by isotopic labelling a t any one of these positions and examination of its distribution in the product.Isotope-labelling experiments have also amplified our knowledge of re- actions for which phenonium-ion formation is a possibility. Experiments have been carried out on the SNl solvolyses of 1 2 2-triphenylethyl compounds such as the acetate (36) labelled on C(l) in the l-phenyl group or in the leaving acetate group.66 The rates a t which the chain and the ring label became distributed over the two halves of the molecule (to appear on oxidation in either the benzoic acid or the benzophenone formed) are a measure of the rate of phenyl migration from position 2 to position 1. Rearrangement by formation of a phenonium ion (37) by an internal SN2 reaction would in view of the symmetry of this structure lead to equal distribution of the phenyl label between the two sides.In fact the phenyl label is in some cases distributed randomly among all three phenyl groups. Since the acetate-labelling experiments indicate that the rearrangement is not accompanied by recombination (“ internal return ”) of the ion pair Ph + [H’ - ‘H ] (37) P h y.’ ‘ ’c / P h* ph ,Ph Ph -5-C-OAc (36) H’ “H formed the result is most easily rationalised in terms of open-chain (“ classi- cal ”) carbonium ions which can undergo equilibration by phenyl migration as follows In view of the great stability of phenyl- and diphenyl-carbonium ions classical structures may be particularly favoumble in this system and these Ph,CH-+CKPh” + +CHPh*CHPhPh* 6 6 Collins and Bonner J. Amer. Chem. SOC. 1955 ‘77 92 99. 198 QUARTERLY REVIEWS results are not irreconcilable with Cram’s interpretation of his stereochemical observations (see p.194). The importance of open carbonium ions in some rearrangements is in fact well recognised by Cram.59 2b In an important continuation 07 of this series of investigations Collins and his co-workers examined the deamination by nitrous acid of 1 1- diphenyl-2-aminopropan- 1-01 (38) stereospeci$caZZy labelled in one of the phenyl groups to form a-phenylpropiophenone (39). In this reaction stereochemically different end products (i.e. configurations a t the migration terminus) were obtained according to whether the labelled or the unlabelled phenyl group migrates as shown below. (The two phenyl groups are not equivalent because the stereospecific labelling produces asymmetry a t C(ll.) The result implies that bridge-formation between C(l! and by the migrating phenyl group cannot occur synchronously with the rupture of the C-N bond (which results in the formation of the carbonium ion) since in one case the phenyl group must attack Cp) from the same side as that occupied by the amino-group and? to be able to do so the C-N bond must have been severed before this attack.This observation therefore also seems to require a classical structure for the carbonium ion (40) and this in a system considered favourable to the formation of bridged ions. To explain the precise product composition it is however still necessary to assume that rotation about the bond between C(l) and C(z) in the ion (40) [to interconvert the enantiomeric conformations] is not very fast compared with the product- forming phenyl migration.This conclusion amounts to the postulation of a rotation-hindering force in carbonium ions other than through the bridge bonding which characterises the “ non- classical ” structures considered up till now. This postulate could perhaps be rationalised in terms of a novel kind of hyperconjugation between the vacant p-orbital on the carbonium- carbon atom and the three splayed a-bonds on the adjacent carbon atom(s). Unless there is an alternative explanation of Collins’s observations drastic reconsideration of the significance of the stereochemical evidence for bridged ions going right back to Roberts and Kimball’s hypothesis (p. lSS) may be required. 67Benjamin Schaeffer and Collins J . Amer. Chem. SOC. 1957 79 6160. BETHELL AND GOLD THE STRUCTURE OF CARBONIUM IONS 199 V.Homoallylic and cycZoPropeny1 Cations The Efect of Neighbouring Double Bonds.-The acetolysis of anti-norborn- 7-enyl toluene-p-sulphonate (41) is 1011 times more rapid than that of the 7-norbornyl atnalogue 6* (42). The magnitude of the acceleration indicates that this is not an inductive effect of the double bond. The phenomenon is attributed to stabilisation of the incipient positive charge on by the n-orbital of the double bond and formation of the ion (43). The bonding in this ion has been likened to that in the cyclopropenyl cation.* Simple molecular-orbital calculations 71 analogous to those applied to phenonium Ts 0 2 6 6 (44) (45) OH (46) ions,j4 support t,he increased stability of such a structure. For syn- norborn-7-enyl ester (44) the rate is 107 times slower than that for the anti- compound since the leaving group now prevents the interaction between the double bond and The fact that the solvolysis of ester (44) is still 1 0 4 times faster than that for (42) has been attributed to formation of the intermediate (45) by methylene participation.Hydrolysis of the acetolysis product gives the alcohol (46) in accordance with this scheme. The interaction involved in ion (43) can also be expressed in terms of the resonance structures for the " homoallylic " representation of the car- bonium ion I' ' + t H - H C-CH or [H2C;-7C<. "&/ \ 2 2 \ / \ CH CH2 CH CH2+ CH CH and stabilisation is found in other molecules containing this arrangement of bonds. This interaction of an cc-carbon and a y9-double bond was in fact recognised by Shoppee 72 in 1946 for cholesteryl chloride (47).I n agree- ment with this suggestion a rate-enhancement has been observed for the solvolysis of the toluene-p-s~lphonate.7~ The reaction of the chloride (47) 68 Winstein Shatavsky Norton and Woodward J . Amer. Chem. SOC. 1955 7'7 4183; Winstein and Xtafford ibid. 1957 79 506. Roberts Streitwieser and Regan ibid. 1952 '74 4579. 70 Breslow ibid. 1957 '79 5318. Woods Carboni and Roberts ibid. 1956 '78 5653. 72Shoppee J. 1946 1147. 73 Winstein and Adams J . Arner. Chem. SOC. 1948 '70 838. * Considerable stabilisa.tion has been predicted for the (symmetrical) cyclopropenyl cation and has been confirmed by the isolation of a salt of the stable 1 2 3-triphenyl- cyclopropenyl cation.70 200 QUARTERLY REVIEWS (47) - + OMe is formulated as shown.Like allylic resonance this type of interaction does not affect the spatial disposition of the bonds in the ion though it will affect bond lengths. VI. Protonated Unsaturated Systems Stable Carbonium Ions formed from Olefins and Acids.-Cationic species of the same empirical formuh can be generated either by loss of an anionic group from a neutral molecule or by addition of a proton to a related unsaturated compound. A simple example is the stable methyldiphenyl- carbonium ion (48) the two modes of formation of which are easily realised by dissolution of the alcohol or the olefin in sulphuric acid the same species being formed in both processes.74 These stable carbonium ions are better represented by classical formulze 74 rather than by structures (" z-complexes ") in which the proton occupies a bridging position or is loosely attached to the n-electron cloud in a less localised fashion.75 This view is based on the similarity of the ultraviolet absorption spectra 74 of the different species obtained by dissolving 1 1 -diphenylethylene triphenylethylene or anthra- cene in sulphuric acid.This similarity is easily understood if the species obtained in the three cases are protonated a t those positions (cf. formulE) which simple molecular-orbital calculations l2 predict would yield the most stable classical ion. In these ions the system of electrons extends over the nuclear frame enclosed within the broken lines and is in each case a diphenyl- carbonium system. The ions should thus have very similar electronic spectra. On the other hand the close similarity experimentally observed would be out of place on other views of the structure of these ions.Support 7 4 Gold and Tye J . 1952 2172. 7 5 Dewar " The Electronic Theory of Organic Chemistry " Oxford Univ. Press London 1949. BETHELL AND GOLD THE STRUCTURE OF CARBONIUM IONS 201 for this kind of structure in ions derived from polycyclic aromatic hydro- carbons has been obtained from the study of hydrocarbon basicities.76 Ionic Reaction Intermediates formed from Unsaturated Compounds in Acidic Media.-The study of the kinetics of reactions initiated by a proton transfer to an unsaturated or conjugated molecule has suggested that another structure of the carbonium ion is also possible. Taft measured the rate of hydration of isobutene in aqueous acid as a function of acidity and found the reaction velocity to be proportional to Hammett's acidity function h ( H = - log h,) rather than to the concentration of hydrogen ion." On the basis of the Zucker-Hammett hypothesis 78 this result was interpreted as implying that the slow step of the reaction involved only a protonated olefin species formed in a rapid proton-transfer pre-equilibrium from olefin and acid.The specific picture proposed by Taft is the slow isomerisation of a '' n-complex " to the classical carbonium ion followed by completion of the reaction through attachment of OH in a sequence of rapid steps which do not affect the kinetics. It is fairly clear that if the first phase of the reaction is a rapid protonation equilibrium the protonated species formed cannot be a classical carbonium ion since the hydration of the isomeric 2-methylbutenes (49) and (50) yields the same product but is not accom- panied by isomerisation of unchanged olefin.The classical ion formed from each isomer would have the same structure (51) and on loss of a proton would obviously regenerate the same olefin mixture. 79 Several other less direct lines of evidence support the conclusion outlined but with all this evidence in mind Long and Paul 81 recently concluded that it " does not seem easy to exclude the possibility that olefin hydration involves a slow proton transfer as the rate-determining step " and by implication that the experiments on olefin hydration have no bearing on the structure of the intermediate carbonium ion. I n connection with this problem Cannell and Taft 82 examined the deamination of isobutylainine (CH,) ,CH*CH,*NH by nitrous acid.This reaction produces trimethylcarbonium i0ns,~3 and involves a t some stage during or after the formation of the carboniurn ion the transfer of hydrogen 'eMackor Mofstra and van der Wads Trans. Paruday SOC. 1958 54 66. 77Taft J . Amer. Chem. SOC. 1952 74 5372. '8Zucker and Hammett ibid. 1939 61 2791. Levy Taft and Hammett ibid. 1953 75 1253. Taft Purlee Riesz and DeFazio ibid. 1955 77 1584 ; Purlee and Taft ibid. 1956 78 5807. 81Long and Paul Chem. Rev. 1957 57 935. 82Cannell and Taft J . Amer. Chem. SOC. 1956 78 5812. s3 Idem Abs. Papers 129th Amer. Chem. Soc. Meeting April 1956 4 6 ~ . 202 QUARTERLY REVIEWS from position 2 to position 1. If this intermediate is identica,l with the protonated olefin formed in the hydration of isobutene then the migrating hydrogen should be easily exchanged with protons from the solvent.How- ever it is found that when the deainination is carried out in deuterium oxide solution the tert.-butyl alcohol formed contains no C-D bonds. This indicates that the migrating hydrogen is a t all times firmly held and the intermediate is different from that postulated for the olefin hydration. Cannell and Taft accordingly propose a new type of isomerism between two kinds of non- classical ions the forms being (1) n-complex ions formed in olefin hydration in which the proton is weakly embedded in the n-orbital of the double bond above the plane of the other olefin valencies and (2) bridged " protonium " ions formed in the deamination in which bonding by a-electrons is involved throughout the reaction.Some of these suggestions are closely related to conclusions drawn con- cerning the interaction of acids with aromatic rings. The rate of hydrogen- isotope exchange (deuterium loss) in acidic solvents has been found to follow Hammett's acidity function over a wide range of acidities and application of the Zucker-Hammett hypothesis 78 again leads to the conclusion that the first step in the reaction is the formation in low concentration of a protonated species which is in rapid proton-transfer equilibrium with the solvent.84 This species cannot have the structure (52) in which protium and deuterium occupy equivalent positions since such a structure would break down to yield either the deuterio- or the protio-compound and there- fore the initial rate of protonation could not be very fast compared with the exchange velocity.It has been suggested that the slow step is a re- arrangement of the first carbonium ion formed. To avoid the structural implications of other names Gold and Satchell refer to the first ion formed as an " out,er complex )' and to the ion (52) as an " inner complex ',. Since hydrogen-exchange reactions appear to be subject to steric " outer complexes " do not readily rearrange over the whole system of conjugation but that the initial site of protonation- if there is to be exchange-must be sufficiently close to the seat of reaction for steric hindrance by ortho-substituents to be perceptible. All these conclusions are again subject to the limitations of the Zucker-Hammett hypothesis and as for Taft's mechanism Long and Paul have speculated about alternative hypotheses which would destroy the structural signi- ficance of the proposals made.The equilibria observed in solutions of aromatic hydrocarbons in strong acids have frequently been interpreted in terms of two kinds of interaction one looser and one more intimate termed respectively n- and o-complexes. The former are generally pictured as involving attachment of the proton to the electron cloud and the second as being like structure (52). Valuable though these studies are perhaps most of the results relating to such equi- libria are not really relevant as evidence of structure. 8 4 Gold and Satchell Nature 1955 176 602; J . 1955 3609 3619. 85 Tiers J . Arner. Chem. Soc. 1956 78 4165. +OH ,-- hindrance 85 in the ortho-position one must conclude that these L-- (52, BETHELL AND GOLD THE STRUCTURE OF CARBONIUM IONS 203 VII.Non-classical Ions in the Gas Phase Cutionuted cycloPropane Rings.-An indication of other structural possi- bilities which have not so far received prominence in chemical speculation is provided by some mass-spectral studies.86 tert. -Butylbenzene in which the a-carbon of the side-chain had been labelled with 13C yielded on electron impact benzyl cations two out of three of which contained no 13C. The reactions involved are formulated as shown. In the cationated cyclopropane CH3-’ Q + e d Q %- CH +,p CH-CH -0 CH2’ \*/ +C,H I / \ CH3 H,C CH C”2 t 3e +CH (53) ring (53) the three side-chain carbon atoms become equivalent thus leading t o t,he statistical distribution of 13C in the benzyl cations finally produced.Evidence also exists for structures similar to (53) in which methyl cations or protons replace the phenyl cation. However it may be that the occurrence of rapid rearrangements is characteristic of high-energy carbonium ions formed on electron impact and that the results need not imply non-classical structures. Thus “ benzyl ” cations C7H7+ formed from labelled alkyl- benzenes by electron impact have been found to decompose in such a manner that all seven positions contribute equally to the break-down pr0ducts.~7 Taken in conjunction with values of the appearance potentials the result has been interpreted as indicating the tropylium structure (in which all seven positions are equivalent) for the ion C7H +. Conclusion It will be apparent from the selection of material presented in this Review that the structure of carbonium ions is a lively field of chemical research and speculation.It is unlikely that all the notions now current will stand the test of time. It may even be that the variety of ‘Cnon- classical ” phenomena observed indicates a serious weakness of the under- lying structural concepts. The Reviewers would like to state that they have made no attempt to assess questions of priority of ideas. The presentation of suitable illustrative material was the only guiding principle in the selection of particular pieces of research. They thank Drs. P. B. D. de la Mare and C. W. Rees for com- ments and Mr. J. S. Coe M.Sc. for checking references. S6Rylander and Meyerson J. Arner. Chem. SOC. 1956 7S 5799. 87 Rylander Meyerson and Grubb ibid. 1957 79 842.

ISSN:0009-2681    

DOI:10.1039/QR9581200173

出版商:RSC    

年代:1958

数据来源: RSC