摘要:

QUARTERLY REVIEWS BIOSYNTHESIS OF STEROLS STEROIDS AND TERPENOIDS. PART 11. PHYTOSTEROLS TERPENES AND THE PHYSIOLOGICALLY ACTIVE STEROIDS By R. B. CLAYTON (DEPARTMENT OF PSYCHIATRY STANFORD UNIVERSITY SCHOOL OF MEDICINE PALO ALTO CALIFORNIA) The Biosynthesis of Phytosterols and Tetracyclic Triterpenes THE steps involved in the generation of the isoprene units of terpenoid compounds and in the biosynthesis of cholesterol have been described in Part I.* There is good experimental evidence that the pathways of bio- synthesis of ergosterol (LXXVII)157-159 and of the CSl tetracyclic triter- pene eburicoic acid (LXXVIII)lso-lsl involve the same intermediates that are involved in cholesterol biosynthesis at least as far as the first tetracyclic cyclisation product of squalene. Indeed the isolation of cholesterol from red algaels2 and more recently from the potato plant163 demonstrates the presence of the whole enzymic mechanism in certain plant tissues for the synthesis of this characteristically “animal” sterol.The origin of the “extra” carbon atoms substituted at C(24) in side chains of plant and fungal sterols and triterpenes and the structure of the steroid or triterpene acceptor molecule in the alkylation are problems of particular interest. The origin of the C(,,)-methyl carbon of ergosterol in the one-carbon pool (labelling from formate) was demonstrated by Danielsson and Blochls4 who interpreted their results to exclude the methylation of squalene before * Part I Clayton Quart. Rev. 1965,19,168 (preceding Review). References in Part I1 which appeared in Part I are as follows 12Saniuels “Metabolic Pathways” ed.Green- berg Academic Press New York 1960 p. 471. l4BergstrOm Danielsson and Samuels- son “Lipid Metabolism” ed. Bloch Wiley Sons New York 1960 p. 291. 15Daniel~~~n “Advances in Lipid Research,” ed. Paoletti and Kritchevsky Vol. 1 Academic Press New York 1963 p. 335. 43Bernfeld “The Biogenisis of Natural Compounds” Per- gamon Press New York 1963. 75Popj8k Goodman Cornforth Cornforth and Ryhage J. Biol. Chem. 1961 236 1934. lo7Arigoni CIBA Foundation Symposium Biosynthesis of Terpenes and Sterols ed. Wolstenholme and O’Connor Little Brown and Co. Boston 1959. loSBirch and Smith ref. 107 p. 245. lo9Birch Kocor Sheppard and Winter J. 1962 1502. lK7 Hanahan and Wakil J. Amer. Chem. SOC. 1963,75 273. 15* Dauben and Hutton J. Amer. Chem. SOC.1956 78 2647. lK9 Dauben Hutton and Boswell J. Amer. Chem. SOC. 1959 81,403. Dauben Ban and Richards J. Arner. Chem. SOC. 1957 79 968. Ia1 Dauben and Richards J. Amer. Chem. Sac. 1956,78 5329. la2 Tsuda Agaki Kishida Hayatsu and Saka Chem. Phann. Bull. Tokyo 1958 6 la3 Johnson Bennet and Heftmann Science 1963 140 198. la4 Danielsson and Bloch J. Arner. Chem. SUC. 1957 79 500. 724. 201 1 202 QUARTERLY REVIEWS its cyclisation. The methylene group of eburicoic acid has also been shown by Dauben and his co-workers to be derived from a one-carbon source and not from a ~ e t a t e . l ~ ~ J ~ ~ More detailed studies by Alexander et af.lSS of the origin of c(28) in ergosterol have implicated methionine as the methyl donor since in- corporation of the labelled methyl group from this source was more efficient than from formate and took place without change of the 14C/3H ratio from [14C,3H]methyl-labelled methionine.Subsequently S-adeno- sylmethionine was shown to be a more effective methyl donor than methionine16'. The conclusion was drawn by Alexander et aZ.,16$ on the basis of their double-labelling experiment that no loss of hydrogen from the methyl group of methionine occurred during its incorporation as C(2a) of ergo- sterol. This conclusion is open to question on the grounds that the isotope effect would favour selective loss of hydrogen rather than tritium from the migrating methyl group. Lederer and his co-workers168 have recently re-examined the methylation process in this light using [2H3]methyl- labelled methionine as the methyl donor for ergosterol synthesis in a methionine-less stain of Neurospora crassa.Mass spectrometry of the ergosterol isolated from the organism indicated the incorporation of only two deuterium atoms per molecule. The introduction of the Q Z s ) carbon atom of ergosterol by transfer from methionine was the first reported example of C-methylation from this source but similar C-methylations have since been shown to occur in the formation of the cyclopropane ring in cyclopropane fatty a ~ i d ~ ~ ~ and the branching methyl group of tuberculostearic acid (10-methyl- stearic acid)173 and in other microbiological p r o c e s ~ e s . ~ ~ ~ J ~ * Experiments with CD,-methi~ninel~~ have shown that in tuberculostearic acid as in ergosterol the incorporation of the methyl group entails an exchange of one hydrogen atom.The acceptor molecule in this reaction is an un- saturated fatty acid derivative in which the ethylenic linkage is the acceptor site. There is so far little biochemical evidence to indicate the structure of the biosynthetic intermediate which undergoes the alkylation reaction at C(J4). There is evidence176 for the conversion of lanosterol but not zymo- sterol into ergosterol in yeast and the structure of citrostadienol 165 Dauben Fonken and Boswell J. Amer. Chem. SOC. 1957 79 1000. 168 Alexander Gold and Schwenk J. Amer. Chem. SOC. 1957 79 2967. 167 Parks J . Amer. Chem. SOC. 1958. 80 2023. 16s Lederer Experientia 1964 20 473. 160 O'Eeary J. Bacteriol. 1959 78 709. 170 Liu and Hofmann Biochemistry 1962 I 189. 171 Zalkin Law and Goldfine J . Biol. Chem. 1963 235 1242. 172 Lennarz Scheuerbrandt and Bloch J.Biol. Chem. 1962,237,664. 179 Schwenk Alexander Gold and Stevens J. Biol. Cheni. 1958 233 1211. Birch English Massey-Westrop Slaytor and Smith J. 1958 365. 175 Jaureguiberry Law McCloskey and Lederer Compt. rend. 1964 258 3587. 176 Schwenk and Alexander Arch. Biochem. Biophys. 1958 76 65. CLAYTON STEROLS STEROIDS AND TERPENOIDS 203 (LXXIX),1i7J78 isolated from citrus fruits suggests that in these higher plant tissues also the acceptor molecule is a ~ l ~ ~ - s t e r o l preceding zymo- sterol in the biosynthetic pathway. The small degree of conversion of cholesterol into 24-methylenecholesterol in could be attributable to the desaturation of cholesterol possibly to desmosterol by these organisms.180 The occurrence of such compounds as the polyporenic acids (LXXVIIIa),1s1~182 eburicoic acid (LXXVIII),183 and cyclolaudenol (LXXX),184 which contain some variant of the lanosterol structure with addition of a carbon atom at C(24) suggests that lanosterol itself may be the acceptor in these cases.The structure of cyclolaudenol with an un- saturation in the 25,26-position suggests that the methylation reaction involves the formation of a carboiiium ion intermediate for which several possible routes of stabilisation may be available depending upon the characteristics of the enzyme involved. Hypothetical schemes are shown in Fig. 19 to account for the formation of sterol and triterpene side chains of the 24-methylene (LXXXI) 24-methyl (LXXXII) and 25-methylene (LXXXIII) types and the cyclopropane fatty acids (LXXXIV). It is now known that methionine provides both carbon atoms of the ethyl group 17' Mazur Weizmann and Sondheimer J.Attier. Chem. SOC. 1958 80 1007 6293. 178 Mazur and Sondheimer J. Amer. Chem. SOC. 1958 80 6296. 179 Fagerlund and Idler Canad. J. Biochem. Physiol. 1961 39 1347. la0 Fagerlund and Idler Canad. J. Biochem. Physiol. 1961 39 505. lS1 Roth Saucy Anliker Jeger and Heusser Helv. Chim. Acta 1953 36 1908. lE2 Halsall and Hodges J. 1954 2385. x83 Holker Powell Robertson Simes Wright and Gascoigne J. 1953 2422. lS4 Henry Irvine and Spring J. 1955 1607. 204 QUARTERLY REVIEWS substituted at C(24) in the C,,-sterols /3-sitosterol (LXXXVII),185J8s and spinasterol (LXXXVIII)lS7 and of the ethylidene group in fucosterol (LXXXIX).168 It seems that a 24-methylene substituent (LXXXI) may either remain as such (e.g.24-methylenecholesterol)18*J89 be reduced to 24-methyl (ergosterol type) or be remethylated giving a 24-ethylidene- sterol (LXXXV) (fucosterol type) which in turn may be reduced to the 24-ethyl derivative (LXXXVI) (sitosterol type). A similar scheme has been suggested by Castle et aZ.lB5 and several possible mechanisms of these reactions are discussed by Lederer.168 The recent findings concerning the origin of the 24-ethyl group in the CZ9 sterols offer a new possibility of studying the mechanism of these methyl group-transfer reactions since the simple mechanisms shown in -7=c; \ H Adenosy' I+ 'i' H2 CH3- S - CH2€Hi CH C q H (UXXIV) ,CH3 -c-c / cH3 c p 3 ,CH3 ,CH3 YH2 ,CH3 - -CH-CF HLy /cH3 -H+ tH ,CH3 -c -c\H - -c-qi ( 3 3 a 3 CH3 (uxxv) (UXXVl) FIG. 19.Possible mechanisms of methylation of ethylenic bonds leading to various Fig. 19 should result in transfer of CD without loss of deuterium in the case of the C, sterols and also in cyclolaudenol (LXXX). sterol side-chains and cyclopropanes. lE5 Castle Blondin and Nes J. Amer. Chem. SOC. 1963 85 3306. la' Bader Gulielmetti and Arigoni Proc. Chem. Soc. 1964 16. 180 Idler and Fagerlund J. Amer. Chem. Soc. 1955 77 4142. Nicholas and Moriarty Fed. Proc. 1963 22 529. Barbier Hugel and Lederer Bull. SOC. Chem. Biol. 1960 42 91. CLAYTON STEROLS STEROIDS AND TERPENOIDS 205 The mechanism of introduction of the d22-bond in such compounds as ergosterol stigmasterol (XC) and brassicasterol (XCI) has not been investigated but studies of the time course of incorporation of rnevalonic H 1 *; acid into p-sistosterol and stigmasterol in the potato plant suggest that the former sterol may be a precursor of the latter.lgO On the other hand the possibility that d22~24(28) compound is a companion of ergosterol in yeastlo' suggests that desaturation at C(22) may occur before reduction of the ~ l ~ ~ ( ~ ~ ) - b o n d .It seems likely that the biochemistry of the side-chain transformations in plants varies not only among species but in different tissues of a given species and in the same tissues at different phases of growth. Biogenesis of the Physiologically Active Steroids FoIIowing the early experiments of Bloch and his co-workers in which deuterated cholesterol was shown to be converted in vivo into cholic acid1** and urinary pregnanediol,lQ3 the main outlines of the biosynthesis of the steroid hormones and bile acids have been clarified.It seems unlikely on loo Johnson Heftmann and Houghland Arch. Biuchem. Biuphys. 1964 104 102. lol Breivik Owades and Light J. Org. Chem. 1954 19 1734. lop Bloch Berg and Rittenberg J. BioZ. Chem. 1943 149 511. Bloch J B i d . Chem. 1945 157 661. 206 QUARTERLY REVIEWS the basis of evidence now a ~ a i l a b l e ~ ~ * J ~ ~ that there is any significant contribution to their formation by pathways not involving cholesterol though the possibility of contributions from precursors such as desmo- sterol appears to exist.lQ6 Some aspects of steroid hormone and bile acid formation that have been the subject of recent iiivestigations will be dis- cussed below. The Steroid Hormones Cleavage of the Side Chain of Cholesterol.- Since evidence began to accumulate that this was probably a key point of action of pituitary adrenocorticotrophic hormone (ACTH) in stimulating adrenal steroid hormone output,lQ7 J~~ considerable interest has centred on the exact mechanism of conversion of the iso-octyl side chain of chole- sterol into the two carbon 20-0x0-structure found in the steroids of the pregnane group.Although the mechanism of action of ACTH is still not clarified the steps leading from cholesterol to pregnenolone the parent Czo steroid are now known in some detail. EasIier reports of the fission of the cholesterol side chain by adrenal HO s-\q c FIG. 20. Biosynthesis of pregnenolene. Demonstrated conversions possible conversions ---* no conversion:.&*. lQC Caspi Dorfman Khan Rosenfeld and Schmid J.Biol. Chern. 1962 237 2085. lQ5 Werbin and Chaikoff Arch. Biochem. Biophys. 1961 93 476. lQ6 Goodman Avigan and Wilson J. Clin. Invest. 1962 41 2135. lS7 Hayano Saba Dorfman and Hechter Recent Progr. Hormone Res. 1956,12,79. IB8 Tchen 6th International Congress of Biochemistry New York 1964 (Abstracts) p. 543. CLAYTON STEROLS STEROIDS AND TERPENOIDS 207 enzymes to yield pregnenolone and isocaproic acid,lg9 t2O0 directed atten- tion to the metabolism of cholesterol derivatives oxygentated at either or both C(20) and C(22) (Fig. 20). The initial cleavage product representing the side-chain carbon atoms is now known to be isocaproic aldehyde201 which is only subsequently converted into the acid. The initial oxidation product appears to be the 20a-hydroxy-compound (XCII).202 This is further converted into the 20a,22&diol (XCIII) which undergoes oxidative cleavage to pregnenolone (XCIV) and isocaproic aldehyde (XCV).203 This may not be the exclusive pathway however since it has also been shown204 that 22-hydroxycholesterol (XCVI) serves as a better substrate for the cleavage enzyme system than cholesterol itself.It is not clear whether the 20a,22&dihydroxy-derivative is oxidised to the 22-ketone before fission of the 20-22 bond but the 20a-hydroxy-22-ketone (XCVII) is an effective substrate for the cleavage enzyme system though 22-0x0- cholesterol (XCVITI) unlike 22-hydroxycholestero1 (XCVI) is not met abolised. These oxidative steps have been studied in soluble adrenal enzyme systems; they require oxygen and utilise either NADH or NADPH though optimally the latter is ~ t i l i s e d .~ ~ ~ ~ ~ ~ ~ A requirement for ferrous ion is also reported.lgs Their mechanism is no doubt similar to that of other mixed function oxidase reactions that are responsible for the various stereospecific hydroxylations of the steroid hormones and bile acids.206 The same pathway of cleavage of the cholesterol side chain has been established207 in the testis in which the production of the androgenic C1 steroids takes place predominantly by removal of the side chain from a C, 20-keto-ster0id.~~~ A proposed alternative pathway of formation of C19 steroids via 3p 1 7a,20a-trihydroxycholest-5-ene208 could not be demonstrated.209 The conversion of 20cc-hydroxycholesterol into preg- nenolone and progesterone in enzyme preparations from corpus luteum has also been observed.210 The derivation of the estrogens from these Czl precursors by way of the C1 androgenic steroids has been reviewed in detail by Breuer.211 A recent development in the study of cholesterol catabolism in the adrenal has been the recognition that cholesteryl sulphate may serve as the substrate in the cleavage reaction.The finding that dehydroepiandrosterone lQ9 Lynn Staple and Gurin J. Amer. Chem. Soc. 1954 76 4048. Staple Lynn and Gurin J . Biol. Chem. 1956 219 845. 201 Constantopoulos and Tchen J. Biol. Chem. 1961,236 65. 202 Shimizu Hayano Gut and Dorfman J . Biol. Chem. 1961 236 695. 203 Shimizu Gut and Dorfman J. Biol. Chem. 1962 237 699. 204 Chaudhari Harada Shimizu Gut and Dorfman J. Biol. Chem. 1962,237,703. 205 Halkerston Eichorn and Hechter J.Biol. Chem. 1961 236 374. 208 Hayano in Hayaishi “Oxygenases,” Academic Press New York 1962 p. 225. 207 Toren Menon Forchielli and Dorfman Steroids 1964 3 381. 208 Dorfman Forchielli and Gut Recent Progr. Hormone Res. 1963 19 251. 2oa Shimizu Biochemistry Tokyo 1964 56 201. z10 Hall and Koritz Biochemistry 1964 3 129. ‘11 Breuer Vitamins and Hormones 1962 20 285. 208 QUARTERLY REVIEWS -0 FIG. 21. Conversion of cholesteryl sulphate to dehydroisoandrosterone sulphate. sulphate (Fig. 21 CII) may be a primary secretory product of the adrena1212 prompted a study of the metabolism of cholesteryl sulphate (IC) doubly labelled with 14C and 32S. When perfused through an adrenal tumour this compound yielded dehydroepiandrosterone sulphate containing the labelled atoms in essentially unchanged ratios.213 Pregnenolone sulphate (C) has also been shown to yield 17a-hydroxypregnenolone sulphate (CI)214 and dehydroepiandrosterone sulphate (CII)216 in adrenal prepara- tions in vitro.Cholesteryl sulphate has recently been isolated from normal bovine adrenal tissue,216 though apparently it is present in very low con- centrations (ca. 1.5 mg. per kg.). Until data as to its rate of turnover and the relative efficiencies of metabolism of free cholesterol and cholesteryl sulphate are available the significance of these observations remains un- certain. Further Metabolism of Pregnenolone in the Adrenal.-Largely as a result of early studies using adrenal perfusion or in vitro incubation in which various substrates were tested for their capacity to undergo hydroxylation to more highly oxygenated corticosteroids it was concluded that pregne- nolone (Fig.22) (XCIV) was oxidised to progesterone (CIII) before further hydr~xylation.~~~ The preferred order of hydroxylation of progesterone according to these earlier experiments was C(17) C(21) C(ll) with a branch- ing of pathways occurring depending on whether the initial attack was at C(17) (CIV) [leading to cortisol (CV)] or at C(zl) [leading to corticosterone (CVI)]. Recent experiments may call for some modification of this scheme. Vande Wiele MacDonald Gurpide and Lieberman Recent Progr. Hormone Res. 1963 19 275. '18 Roberts Bandi Calvin Drucker and Lieberman J. Amer. Chern. Soc. 1964 86,958. a14 Calvin and Lieberman Biochemistry 1964 3 259. 'l* Drayer Roberts Bandi and Lieberman J. Biol. Chem. 1964 239 PC 31 12.*17 Hechter and Pincus Physiol. Rev. 1954 34 459. Calvin Vande Wiele and Lieberman Biochemistry 1963 2 648. CLAYTON STEROLS STEROIDS AND TERPENOIDS 209 FIG. 22. Relationships of pregnenolene and progesterone in the biosynthesis of some corticosteroids. Berliner et showed that adrenal enzymes could catalyse the oxidation of the 3fi-hydroxy-d5-moeity to the d4-3-ketone in steroids that had under- gone prior hydi ..xylation at C(l,) and C(21). Thus 3/3,21-dihydroxy-d6- pregnen-20-one (CVII) yielded deoxycorticosterone (CVIII) and 3~,17a,21-trihydroxy-d5-pregnen-20-one (CIX) gave 1 1-deoxycortisol (CX). Weliky and Enge1219 further found that when [4-14C]progesterone (CIII) and [7w3H] 17a-hydroxypregnenolone (CXI) were incubated to- gether in an adrenal tumour homogenate the latter steroid was converted into cortisol more efficiently (63 %) than the former (17 %).These authors citing other work in support suggest that the major route to cortisol may lie through 17a-hydroxypregnenolone rather than through progesterone. 218 Berliner Cazes and Nabors J . Biol. Chem. 1962 237 2478. 219 Weliky and Engel J. Biol. Chem. 1962 237 2089. 210 QUARTERLY REVIEWS Other evidence consistent with this view has been adduced together with the demonstration that pregnenolone may be converted in the adrenal into 21-hydro~ypregnenolone.~~~ Hence 17a-hydroxylation of pregneno- lone is not a prerequisite for 21-hydroxylation. An evaluation of the relative importance of the “new” and “old” metabolic pathways in vivo or even in normal tissue in vitro remains to be made.The pathway via 17a-hydroxy- pregnenolone would account for the formation of urinary 3/l 17a,20a- trihydroxy-d 5-pregnene.221 Levy et aZ.222 have reported the isolation of 17a-hydroxypregnenolone as a metabolite of cholesterol perfused through the bovine adrenal under conditions in which 1 1 a-hydroxylation is pharmacologically inhibited. These workers also record the remarkable observation that in this system d4-androstene-3 17-dione (CXII) was converted in part into 1 l-deoxy- cortisol i.e. a C19 steroid was converted into a C, steroid. This suggests that under certain conditions the conversion of a 17a-hydroxy-20- ketone into the 17-ketone may be reversed and recalls an earlier though unconfirmed report by H e ~ h t e r ~ ~ ~ that 14C labelled dehydroepian- drosterone (CXIII) appeared to be converted into cortisol or some similar steroid.Another recent instance of the reversal of a steroid metabolic transformation previously considered to be irreversible is the demonstra- tion by Ward and Enge1224 of the reversibility of the enzymic conversion of d5-3/?-hydroxy-steroids into their A4-3 keto-derivatives. In the experiments reported (Fig. 23) d4-androstene-3 17-dione (CXII) was converted by a FIG. 23. Reversibility of d5-3,!-OH-d4-3-ket*steroid transformation. microsomal enzyme preparation from the sheep adrenal into both 3/3-hydroxy-d5-androsten- 17-one (CXIII) and 3/?-hydroxy-d 4-androsten- 17-one (CXIV). In recent years a number of genetically determined defects in adrenal steroid hydroxylation mechanism have been These defects may involve any of the three positions lip 1701 or 21 and give rise to characteristic clinical symptoms in some cases fatal.When C(ll) or 220 Sharma and Dorfman Biochemistry 1964,3 1094. 221 Fotherby Biochem. J. 1958 69 596. 222 Levy Cha and Carlo 6th International Congress of Biochemistry New York 223 Hechter CIBA Foundation Colloquia in Endocrinology London Churchill 224 Ward and Engel J. Biol. Chem. 1964 239 PC 3604. 226 Bongiovanni and Root New Engl. J. Med. 1963,268 1283. 228 Wilkins Amer. J. Clin. Nutr. 1961 9 661. 1964 (Abstracts) p. 584. 1953 vol. 7 p. 161. CLAYTON STEROLS STEROIDS AND TERPENOIDS 21 1 C(21) hydroxylations are defective cortisol output is seriously impaired and feed-back control of ACTH production fails with consequent hyperplasia of the adrenal gland. The major route of metabolism of 17a-hydroxy-20- keto-steroids then becomes their direct conversion into C19 androgenic steroids by cleavage of the side chain.Hence the prevalance of severe virilising symptoms in these conditions. An interesting contribution to the understanding of these diseases is the finding227p228 that several C, steroids whose output from the adrenal is thus enhanced are themselves inhibitors of hydroxylation at both C(ll) and C(21). A curious situation is therefore indicated in which the end product of the primary defect serves pro- gressively to exacerbate the defect. A more physiologically appropriate instance of this type of “feed-back” control is the recently reported inhibition of the conversion of cholesterol into 20~-hydroxycholesterol by pregnen01one.~~~ This effect may well be significant in the normal homeostatic control of the conv6rsion of chole- sterol to the adrenal steroid hormones.Bile Acid Formation.-The biochemistry of bile acid formation has recently been authoritatively reviewed in all its aspects by Danielsson.15 This field owes its somewhat confusing complexity to several factors. The composition of the bile acid mixture of normal bile depends not only upon the action of liver enzyme systems which yield a “primary” secretory pro- duct but also upon the action of intestinal micro-organisms which convert a proportion of the secreted bile acids into derivatives some of which are resorbed from the intestine to be “re-secreted” in the bile either in the same or in modified form. Further no satisfactory in vitro system has so far been devised which would permit a systematic analysis of the enzymic steps leading from cholesterol to cholic acid and there is evidence that just as in some other aspects of sterol and steroid biosynthesis that have been discussed pathways of biosynthesis of bile acids are to some extent flexible.Finally there are marked inter-species differences with respect to the types and proportions of different steroid constituents of the bile which however are of considerable interest from the evolutionary point of view. We shall here be concerned in detail with only a few aspects of the formation and metabolism of chol ic chenodeoxycholic and deoxycholic acids. Sequence of Events in the Formation of Cholic Acid.-Intraperitoneal administration of labelled hypothetical intermediates between cholesterol and cholic acid to rats with cannulated bile ducts thereby obviating most of the problems due to activities of intestinal micro-organisms has been one of the most fruitful techniques to be applied in this field.Much of the evidence for the pathways outlined in Fig. 24 comes from experiments 227 Sharma Forchielli and Dorfman J. Biol. Chem. 1963 238 572. 228 Sharma and Dorfman Biochemistry 1964 3 1093. 229 Koritz and Hall Biochemistry 1964 3 1298. 212 QUARTERLY REVIEWS / \ HO ‘ “‘OH (CXXIII) ’.OH (cxvr) I “1 (CxxVII1) [&$-] (cxx I x) (Continued on facing page). CLAYTON STEROLS STEROIDS AND TERPENOIDS 213 n . (cxxx) Livtr vria p C O H $ & HO" (CXXIV) ' /l J B * c t t r / ~ e ~ o * - \dco2H -OH (CXIX) &%H Hofl \ Ho-.' H (CXXV) (cxx I) FIG. 24. Pathways of bile acid formation.of this type. Work before 1957 indicated that the nuclear transformations preceded scission of the side chain14J5 and the inversion of the hydroxyl group and saturation of the d5-bond took place after hydroxyla- tion at C(,)2309231 and possibly after hydroxylation at C(12).232 In a recent coprostane-3a,7a-diol (CXVIIa) was not hydroxylated at C(12) in a mouse liver preparation in vitro but was converted into the correspond- ing 3a,7a,26-triol (CXVII). Hydroxylation at C(zs) probably constitutes the first step in degradation of the side chain,234 and since 26-hydroxy- cholesterol (CXV) cholest-5-ene-3/3,7a,26-triol (CXVI) and coprostane- 3~~,7a,26-triol (CXVII) were good precursors of chenodeoxycholic acid 230 Lindstedt Acta Chem. Scand. 1957 11 417. 231 Danielsson and Einarsson Acta Chem.Scand. 1964 18 83 I . 232 Danielsson Acta Chem. Scand. 1962 16 1534. 233 Berseus and Danielsson Acta Chem. Scand. 1963 17 1293. 234 Frederickson J . Biol. Chem. 1956 222 109. 214 QUARTERLY REVIEWS ___t (CXXVI) (CXXVI I) (XCVIII) but not of cholic acid (CXIX) which however was formed from coprostane-3a,7a 12a-triol (CXX),235 it seems clear that 12-hydroxylation must occur before any attack on the side chain. Further support for this conclusion comes from the recent demonstration of the enzymic conversion of cholesterol into coprostane-3a,7a 12a-trio1 (CXX) in rat liver.236 Since in the rabbit cholic acid is formed only in poor yield from 12~hydroxy- cholesterol (CXXI) but in good yield from 7a-hydroxycholes terol (CXXJJ),232 hydroxylation at C(12) probably follows that at C(7).The fore- going observations together with others to be discussed below strongly suggest that the major route to cholic acid is as depicted on the left-hand side of Fig. 24. If attack on the side chain is initiated before 12a-hydroxylation the latter is evidently excluded and clearly the specificity of the enzyme systems responsible for side-chain oxidation and hydroxylation at C(,) must be much lower than that of the 12a-hydroxylase. Thus litliocholic acid (CXXIV) is metabolised in the rat to several p r o d ~ c t s ~ ~ ~ including chenodeoxycholic acid and compounds hydroxylated at both Q6) and C(,) but 12a-hydroxy-products are absent. On the other hand deoxycholic acid (CXXV) is converted into cholic It is interesting in view of this that a bile acid formed from cholestan-3/3-01 (CXXVI) in the rabbit is now identified239 as 3a,12a-dihydroxy-5a-cholanic acid (CXXVII) suggesting that the specificity of the 12~-hydroxylase is influenced considerably by the stereochemistry of the junction of rings A and B.The specificity of this enzyme is also subject to important species differences since chenodeoxy- cholic acid is in part converted into cholic acid in the python24o and in the chicken.241 Inversion of the 3IS-Hydroxyl Group and Saturation of the A5-Bond.- The precise stage at which the characteristic structural feature of the bile acid series a 3a-hydroxyl group with a cis-fused A/B ring junction is introduced into the molecule has not yet been determined though it seems likely that it follows the completion of the nuclear hydroxylation 235 Danielsson and Kazuno Acta Chem.Scand. 1964 18 1157. 238 Mendelsohn and Staple Biochemistry 1963 2 577. 237 Thomas Hsia Matschiner Doisy Elliot Thayer and Doisy J. Biol. Chem. 238 Bergstrom Rottenberg and Sjoval Z. physiol. Chem. 1953 295 278. 23D Hofmann and Mosbach J. B i d . Chern. 1964 239 2813. 240 Bergstrom Danielsson and Kazuno J. Biol. Chem. 1960 235 983. 241 Ahlberg Ziboh Sonders and Hsia Fed. Proc. 1961 20 283. 1964 239 102. CLAYTON STEROLS STEROIDS AND TERPENOIDS 215 steps.15 Several aspects of the changes involving the 3-hydroxyl group and the d5-bond of cholesterol are however now clarified as a result of the efforts of Samuelsson and his co-workers. It was shown that the saturation of the d5-bond involved stereospecific addition of hydrogen in the 5/3- and 6 p - p o ~ i t i o n s ~ ~ ~ and further details of the reaction have now been elucidated by means of experiments with [3o1-~H 4-14C]- and [4p-3H 14C]-chole- Loss of 3a-3H occurred during conversion into bile acids but a significant fraction of the 4/3-3H was retained.In chenodeoxycholic acid isolated from the bile in these experiments about 25% of the 4/3-3H was still present and by administration of this material to another bile-cannu- lated rat so that it was metabolised further to a 6/3-hydroxylated derivative it was shown that most of the 3H that had been retained was located in the 6P-position. The conclusion was drawn that the inversion of the 3- hydroxyl group takes place via an intermediate 3-ketone and that reduction of the d5-bond is preceded by its migration to the A4-position (CXXVIII and CXXIX).In the course of this migration an appreciable part of the /I-hydrogen at C(4) migrates to the axial 6/3-position. As the authors point out these observations are similar to those of Wang et aZ.244 with a bac- terial d5-d4 isornerase but conflict with those of Werbin and C h a i k ~ f f ~ ~ ~ who reported complete loss of 4/3-3H from cholesterol during cortisol biosynthesis. While it seems likely that the efficient enzymic isomerisation and reduction of the double bond are facilitated by the formation of a keto- group at C(3) the epimerisation of the 3/3-hydroxyl group does not depend upon the concomitant changes involving the double bond as is evident from the formation of 3or 12a-dihydroxycholanic acid (CXXVII) from cholestanol (CXXVI) and the conversion of coprostanol into various bile acids of usual structure all having 3a-hydroxyl groups.246 Degradation of the Side-Chain-Staple and his c o - w o r k e r ~ ~ ~ ~ demon- strated the conversion of 3a,7au 1 2a-trihydroxy-5/3-cholestan-26-oic acid into cholic acid and have since studied the fate of the three terminal carbon atoms of [26-I4C]3a,7a 12a-trihydroxy-5/3-cholestane (Fig.25 CXX) in A preparation from rat liver converted this compound in vitro into the corresponding 3a,7a 12a,26-tetraol (CXXX) and the 3a,7a7 12a,26-carboxylic acid (CXXXI) both of which were isolated and identified. Carbon atoms 25 26 and 27 were found to be split off as propionic acid labelled in C(l) and C(3) and evidence for the formation of the coenzyme A derivative (CXXXVII) of this fragment was obtained.The authors suggested the pathway shown involving the coenzyme-A 242 Samuelsson J . Biol. Chem. 1959 234 2852. 243 Green and Samuelsson J . Biol. Chem. 1964 239 2804. 244 Wang Kawahara and Talalay J . Biol. Chem. 1963 238 576. 24s Werbin and Chaikoff Biochim. Biophys. Acta 1963 71 471. 246 Bell Elliott and Doisy 6th International Congress of Biochemistry New York 247 Briggs Whitehouse and Staple J. Biol. Chem. 1961 236 688. 248 Suld Staple and Gurin J. Biol. Chem. 1962 237 338. 1964 (Abstracts) p. 564. 216 QUARTERLY REVIEWS FIG. 25. Formation of the bile acid side-chain. derivatives (CXXXII-CXXXV) by analogy with the well established pathway of fatty acid degradation. They pointed out that the final thiolytic cleavage reaction would leave the cholic acid residue in the form of its CoA ester and hence ready for conjugation with glycine or ta~rine.~~O This suggested route of formation of cholic acid is consistent with the views of H a ~ l e ~ o o d ~ ~ ~ ~ ~ ~ ~ concerning the evolutionary significance of the bile acids of different species.According to this author the primary cholesterol metabolites of bile have evolved from C(,,)-hydroxylated products more closely related to cholesterol to the “modern” C(,,)-acidic products whose present-day pathway of formation still recapit dates certain features of the evolutionary trend. Thus 3a,7a 12a-trihydroxycoprostanic acid (CXXXI) the predominant bile acid of certain reptiles and amphibians was sug- ge~ted,~ as a likely precursor of cholic acid in higher organisms. Apart from the evidence cited above which supports this concept it is further strengthened by the recent isolation of 3a,7a 12a-trihydroxycoprostanic acid from human bile.253*254 On the other hand there is also good evidence that several C, and C26 polyhydroxy-steroids found in lower vertebrates 249 Elliott Biochem.J. 1956 62 427 433. 250 Haslewood in “Comparative Biochemistry,” eds. Florkin and Mason Academic 251 Haslewood 6th International Congress of Biochemistry New York 1964 (Ab- 262 Haslewood Physiol. Rev. 1955 35 178. 263 Staple and Rabinowitz Biochim. Biophys. Acta 1962 59 735. *64 Carrey and Haslewood J. Biol. Chem. 1963 238 PC 855. Press New York 1962 vol. 111 p. 205. stracts) p. 539. CLAYTON STEROLS STEROIDS AND TERPENOIDS 217 should be regarded as deviants from the major biosynthetic (and evolu- tionary) pathway to the bile acids of the mammals.235 Some experimental observations that are not consistent with the pro- posed scheme of side-chain degradation are discussed in detail by Daniels- It will suffice here to note briefly that some evidence has been reported for the cleavage of the three terminal carbon atoms in the form of ace- tone.a55*25s 25-Hydroxycholesterol was tentatively identified as a metabolite of cholesterol in ~ i t t - 0 ~ ~ ~ though this was not c ~ n f i r m e d .~ ~ ~ ~ ~ ~ These obser- vations suggest a cleavage mechanism involving a 24,25-glycol analogous to that by which pregnenolone is formed but so far little evidence has been forthcoming to indicate that this is more than a minor pathway. The accumulation of desmosterol in the liver as a result of inhibition of the 24,25-reductase by triparanol has been discussed in a previous section.It seems most probable that under these conditions desmosterol is oxi- disedlQ6 to bile acid mixtures similar to those formed normally from cholesterol .258 Evidence for the conversion of desmosterol into its 26- hydroxylated derivative in mouse liver homogenates has recently been described.259 On further oxidation this compound could yield the ap- unsaturated carboxylic acid which is presumed to be an intermediate in the normal pathway from cholesterol. Formation of Deoxycholic Acid.-The foregoing discussion of the sequence of hydroxylations implies that any normally formed bile acid having a 12~x-hydroxyl group should also have a 7whydroxyl group yet deoxycholic acid (CXXV) and lithocholic acid (CXXIV) are well known constituents of the bile of many species.The presence of these constituents in the bile is due to the action of intestinal bacteria in removing the 7a- hydroxyl group from cholic and chenodeoxycholic acids. The mechanism of this dehydroxylation has been studied by Saniuelsson260 and is discussed in full by Bergstrom et al.14 Although no intermediate in the process has been isolated ingenious experiments with 3H-labelled substrates indicate that the 7a-hydroxyl group is initially eliminated together with the trans- 6/3-hydrogen and that the resulting d6-bond is then reduced by trans- addition of hydrogen in the 6a- and 7P-positions. The micro-organism(s) responsible for these changes remains unidentified but in a recent study261 of the metabolism of cholic acid by the zrobic soil bacterium Coryne- bacterium simplex several transformation products were isolated including 7a 1 2a-di hydroxy-3-oxochol-4-enic acid 12~-hydroxy-3-oxochola-4,6- dienoic acid and 12a-hydroxy-3-oxochol-4-enoic acid.These results 256 Whitehouse Rabinowitz Staple and Gurin Biochim. Biophys. Acta 1960 37 256 Whitehouse Staple and Gurin J. Biol. Chem. 1961 236 68. 257 Danielsson Arkiv Kemi 1961 17 373. 268 Goodman Avigan and Wilson J. Clin. Invest. 1962 41 962. 259 Danielsson and Johansson Acta Chem. Scand. 1964 18 788. 260 Samuelsson J. Biol. Chem. 1960 235 361. 261 Hayakawa and Samuelsson J. Biol. Chem. 1964 239 94. 382. 218 QUARTERLY REVIEWS suggest that in C. simplex the elimination of the 7a-hydroxyl group may also involve the formation and subsequent saturation of a d6-bond.Biosynthesis of Other Classes of Terpenoid The discovery of mevalonic acid and its recognition as a specific pre- cursor of the isoprene unit have enormously accelerated the study of the biogenesis of most of the major classes of terpenoid compound. These in- clude the carotenoids rubber certain alkaloids the ubiquinones the mono- sesqui- di- and tri-terpenes of plants as well as various fungal metabolites of partial terpenoid structure. An adequate review of these extensive areas in an article of this type is clearly impossible. A useful comprehensive survey43 gives good coverage of most of them up to 1962. All that will be attempted here will be to indicate some features of interest in the development of these various fields. Cyclic Terpenes.-Ruzicka262 has recently reviewed the field of bio- genesis of cyclic terpenes of all the major classes with emphasis on testing the hypothetical mechanisms of their formation by the use of labelled biological precursors.Many of the topics mentioned below are fully dis- cussed in that review. Pentacyclic Triterpenes.-Experiments proving the retention of identity by the isopropylidene carbon atoms of squalene in ring A of soyasapogenol have already been noted.lo7 Ruzicka262 has now discussed an experiment designed to demonstrate that in lupeol (cf. Fig. 14 Part I) the terminal methyl groups of squalene distal to the point of initiation of cyclisation similarly retain their stereospecific individuality. The experiment also serves as a test of the postulated stereoelectronic mechanism of squalene cyclisa- tion in the formation of this type of compound.Lupeol biosynthesised from [2-14C]mevalonate was predicted to have the labelling distribution shown (CXXXVIII) and in complete accord with the 3 0 v P 2 HO Lupol (uootvi I I) Ruzicka Pure Appl. Chem. 1963 6 493. CLAYTON STEROLS STEROIDS AND TERPENOIDS 219 theory the methylene group (C,,,)) on oxidation with osmium tetroxide yielded formaldehyde that was devoid of radioactivity while the terminal methyl group (C,,,)) of the resulting ketone contained one-sixth of the total radioactivity of the starting material. Studies with more biological objectives of the incorporation of mevalonate into /3-amyrin (CXXXIX)2s3 and oleanolic acid (CXL)264 have also been reported and have indicated interesting differences between the efficiences of incorporation of the precursor in to sterol and terpenoid products.Diterpenes.-As discussed above rosenonolactone (LIV) and gib- berellic acid (LIII) also retain the steric individuality of the gem.-sl!bsti- tuted carbon C(15) and C(16).1079108 Gibberellic acid biosynthesised from Rosenonolac tone Gibberellic ccid FIG. 26. Biosynthesis of gibberellic acid labelling distribution from [ 2-14C]mevalonic acid. [2-14C]me~alonic acid265 (Fig. 26) was labelled as in (CXLIII) in accord with its possible derivation from geranyl linalool (CXLI) via kaurene (CXLII) with contraction of ring B as shown. If as has recently been im- plied,266 kaurene is converted intact into gibberellic acid the hydroxyla- tion that occurs at C(3! in the latter compound is unusual in that it must be unrelated to the initiation of cyclisation.Results suggestive of the formation of gibberellic froni all-trans-geranyl geranyl pyrophosphate have been Geranyl linaloyl pyrophosphate could of course be an inter- mediate in this transformation. The diterpene mutilin biosynthesised from [2-14C]mevalonate has the labelling distribution shown (CXLIV) which is also rationalised on the basis of its derivation from geranyl linalool (Arigoni quoted in ref. 262). The failure of sclareol (CXLV) to incorporate the label of [2-14C]meval- 263 Baisted and Nes J . Biol. Chem. 1963 238 1947. 264 Nicholas J. Biol. Chem. 1962 237 1481. 265 Birch Richards and Smith Proc. Chem. Soc. 1958 192. 2e6 Cross Galt and Hanson J . 1963 2944. 267 West Dennis Upper and Lew 6th International Congress of Biochemistry New York 1964 (Abstracts) p.601. 220 QUARTERLY REVIEWS onic acid in Salvia officinali~~~~~~~~ as in other similar cases may have been due to the failure of the precursor to reach the appropriate enzymic site. It is noteworthy that labelled C02 was readily incorporated into the diterpene. 288 Sesquiterpenes.-Jones and Lowe269 have used [2-14C]mevalonic acid as a precursor of trichoihecolone (CXLVI) and have rationalised its forma- tion from farnesyl pyrophosphate on the basis of the resultant labelling pattern. The stereochemical aspects of this transformation and its theo- retical relation to the formation of bisabolene (CXLVII) are discussed in detail by Ruzicka.262 The distribution of label from [1-l4C]acetate in carotol (CXLVIII) a sesquiterpene of carrot seed permitted a discrimina- tion between two possible modes of cyclisation of the farnesyl residue OH w (CXLIV) 8; I4 (CXLV) p / p t>H (CXLVI I) (CXLVI I I) mo &I (CXLVI) in the course of its bio~ynthesis.~~~ No such distinction could be made however in a of the labelling of the symmetrically united ses- quiterpenoid fragments of gossypol (CIL).Monoterpenes.-[2J4C] Mevalonic acid has been used in the (Fig. '27) of the biosynthesis of the monoterpenes /I-pinene (CL) and thujone (CLI). Oxidation of /3-pinene to norpinic acid (CLII) and of thujone to the diketone (CLIII) resulted in each case in retention of 14C and thus supported the derivation of both terpenes from geranyl pyro- phosphate via the respective intermediates (CLIV and CLV) with similar skeletal structures but different charge distributions.An ingenious demon- stration of the retention of stereochemical individuality by the superficially identical termjnal gem-dimethyl groups in an open-chain terpenoid structure has been reported by Birch et aL109 (Fig. 28). The structure in 268 Nicholas Biochim. Biophys. Acta 1964 84 80. 269 Jones and Lowe J. 1960 3959. 270 SouEek Cull Czech. Chem. Comm. 1962,27 2929. 271 Heinstein Smith and Tove. J. Biol. Chem.. 1962. 237 2643. 272 Sandermann and Schweer Tetrahedron Letters 1962 257 259. CLAYTON STEROLS STEROIDS AND TERPENOIDS -g (CLII) Co2H (CLIII) 22 1 FIG. 27. Biosynthesis of /?-pinene and thujone. (CLVI) OH k- NH CH3 cH,.CHo. QH3 Robbit_ Q02H (unlabel led) CO,H (CLVI I) (CLVI I I) FIG. 28. Retention of steric individuality in isopropylidene group methyl carbons of mycelianamide.question is the monoterpenoid chain of mycelianamide. This compound was biologically labelled as shown (CLVI) from [2-14C]mevalo.rlic acid and the terpenoid chain (CLVII) cleaved by treatment with sodiuiii in liquid ammonia. The resulting hydrocarbon was administered to a rabbit and recovered from the urine in the form of its dicarboxylic acid meta- bolite (CLVIII). Ozonolysis of this material yielded acetaldehyde that was free from radioactivity indicating that enzymic attack on one of the gem.-dimethyl groups had been confined to that which was labelled. In contrast to this and the other examples that have been cited of the retention of steric individuality by the gem.-dimethyl groups of terpenoid isopropylidene structures Yeowell and S ~ h m i d ~ ' ~ (Fig.29) have reported evidence for a randomisation of the isopropylidene methyl carbons of citronella1 (CLIX) in the course of its presumed conversion via iridodial 473 Yeowell and Schmid Experientia 1964 15 251. 222 QUARTERLY REVIEWS I I -. .o - (CLIX) o$-CH 0 (cu I I) 7% FIG. 29. Biogenesis of plumieride aglycone. = label from [ 2-14C]mevalonate * =- label from [ 1-I4C]acetate GI = glucose. (CLX) into the plant glucoside plumieride (CLXII). Comparative studies of the distribution of label from [ I-14C]acetate and [2-14C]mevalonate respectively led the authors to postulate the union of acetoacetate with an intermediate of the type (CLXI) in which the isopropylidene carbons becomes indistinguishable from each other. Carotenoid~.~~~.~~~-The biogenesis of the carbon skeleton of lyco- persene (CLXV) and related C, acyclic caroteiioids (Fig.30) clearly presents a problem analogous to that of squalene biosynthesis but in- volving tail-to-tail condensation of two C, (geranylgeranyl) units rather than the two C1 (farnesyl) units as in squalene. The terminal structure of the open-chain carotenoid hydrocarbons is appropriate for the formation of the mono- and di-cyclic carotenes and zanthins by what is apparently a proton-initiated mechanism. The utilisation of mevalonic acid in carotenoid synthesis in a variety of organisms has been demonstrated by several and there is evidence for the utilisation of mevalonic pyrophosphate,280 farnesyl pyrophosphate and isopentenyl pyrophosphate.2811282 The enzymic synthesis of geranylgeranyl pyrophosphate (CLXIII) has been reported to occur in yeast,283 and has recently been studied in detail in carrot root and pig liver enzyme preparations284 and in the yellow bacterium Micro- coccus Zys~deiktictis.~~~ These various enzyme systems which apparently 274 Olson J.Lipid Res. 1964 5 281. 27s Chichester and Nakayama ref. 43 p. 475. 276 Purcell Thompson and Bonner J. Biol. Chem. 1959 234 1081. 277 Stele and Gurin J. Biol. Chem. 1960 235 2778. 278 Braithwaite and Goodwin Biochem. J. 1960 76 194. 279 Yokoyama Nakayama and Chichester J. Biol. Chem. 1962 237 681. 280 Suzue Bull. Chem. Soc. Japan 1964 37 613. 281 Yamamoto Yokoyama Simpson Nakayama and Chichester Nature 1961 282 Beeler Anderson and Porter Arch. Biochem. Biophys. 1963 103 26. 283 Grob Kirschner and Lynen Chimia (Switz.) 1961 15 308.284 Nandi and Porter Arch. Biochem. Biophys. 1964 105 7. 285 Kandutsch. Paulus Levin and Bloch J . Biol. Chem. 1964 239 2507. 191 1299. CLAYTON STEROLS STEROIDS AND TERPENOIDS 223 1 - 2 H (3 -Zeacarotene d- Carotene Viol axon t h in FIG. 30. Biogenesis of carotenoids. 224 QUARTERLY REVIEWS have quite similar characteristics synthesise geranylgeranyl pyrophosphate from isopentenyl pyrophosphate and C5 Cl0 and Cls ally1 pyrophos- phates. The mechanism of condensation of geranylgeranyl pyrophosphate to form the CqO structure is at present unknown. In the enzymic synthesis of geranylgeranyl pyrophosphate the formation of small amounts of geranyllinalool (shown as pyrophosphate CLXIV) was n0ted.~~~#~*5 This material is the CzO analogue of nerolidol which in the form of the pyrophosphate (XL Fig.12) is a hypothetical participant in the farnesyl pyrophosphate coupling reaction according to one possible mechanism76 (cf. Fig. 12 Part I) but since this type of structure readily arises by acid- catalysed isomerisation from the farnesol type its appearance in these experiments may not be significant. In any case the analogy between the biosynthesis of carotenoids and of squalene cannot be pressed too far on the basis of evidence available at present since conflicting results have been obtained in the search for the first-formed C40 compound in carotenoid biosynthesis. Grob and his c o - w o r k e r ~ ~ ~ ~ ~ ~ ~ ~ presented results which implicate lycopersene as the immediate product of coupling of two geranylgeranyl units in a reaction which like the formation of squalene requires NADPH.O t h e r ~ ~ ~ ~ t ~ ~ ~ however have obtained evidence for phytoene (CLXVI) as the primary C40 product and in keeping with this less reduced structure no NADPH requirement could be demonstrated. Despite the uncertainty as to the coupling mechanism there seems no doubt that either lycopersene or phytoene is the “parent” carotenoid from which all others arise by progressive desaturation cyclisation and oxida- tion reactions. Evidence for the pathways shown in Fig. 30 comes from a wide variety of biochemical kinetic and comparative structural studies that are reviewed by Olson.274 Some characteristics of the bacterial geranylgeranyl pyrophosphate synthetase which has been obtained in a relatively high state of suggest that it may be a single protein which catalyses the stepwise elonga- tion of an enzyme-bound terpenoid pyrophosphate chain by addition of isopentenyl pyrophosphate units.This type of enzyme system might be expected to participate in the biosynthesis of a variety of compounds having extended terpenoid chains such as rubber the ubiquinones and the vitamins of the E and K groups. Rubber.-In rubber biosynthesiszag acetate is incorporated via mevalonic acid and an enzyme system from Heveu brasizienis converts isopentenyl pyrophosphate into rubber with high efficiency.29o There is also evidence that dimethylallyl pyrophosphate is a “starter” for the polymerisati~n.~~~ The cis-ethylenic bonds are a distinctive feature of natural rubber and it 286 Grob ref. 107 p. 267. 28’ Mercer Davies and Goodwin Biochem. J. 1963 87 31 7.288 Anderson and Porter Arch. Biochem. Biophys. 1962 97 509. 290 Archer Ayrey Cockbain and McSweeney Nature 1961 189 663. 2g1 Archer Audley Cockbain and McSweeney Biochem. J. 1963 89 565. Bonner ref. 43 p. 727. CLAYTON STEROLS STEROIDS AND TERPENOIDS 225 has been shown by experiments with stereospecifically labelled [4-3H]- mevalonate that they arise as a result of steric specificity of the polyiso- prenoid synthetase enzyme. Rubber incorporated 3H from 4 4 - [4-3H I]- mevalonate but not from 4R- [4-3H]mevalonate which on the other hand yielded 3H-labelled trans-trans-farnesyl pyrophosphate in the same enzyme preparation.292 The stereochemical course of rubber biosynthesis can therefore be represented as in Fig. 31. The extension of the rubber polymer chain takes place in a particulate fraction of latex.291 Ribonuclease is reported to destroy the enzymic activity suggesting that ribonucleic acid (RNA) contained in these particles plays some r61e in the maintenance of enzymic /c** H3C CH FIG.31. Absolute stereochemistry of polymerisation in rubber formation from 4-S-[ 4-3H Imevalonate. * indicates hypothetical label distribution from Ubiquinones.-These compounds of general structure (Fig. 32 CLXVII) all contain polyisoprenoid chains of varying lengths bound to a substituted p-quinone nucleus. Their biosynthesis is still far from fully understood. A major problem is the low level of incorporation of labelled precursors and the evident difficulty of obtaining in vitro systems suitable for syste- matic enzymic analysis. Important aspects of the biochemistry of the ubiquinones including both physiological function and biosynthesis are covered in a CIBA Foundation Symposium.294 The most reasonable hypothesis to account for the formation of these compounds visualises the attachment of the preformed isoprenoid chain (presumably reacting in the form of the pyro- phosphate) to the quinone ring or its p r e c u r s ~ r .~ ~ ~ - ~ ~ ~ In keeping with this 282 Cockbain and PopjAk quoted in Samuelsson and Goodman Biochem. Biophys. Chem. Comm. 1963 11 125. 293 McMullen 6th International Congress of Biochemistry New York 1964 (Ab- stracts) p. 586. ss4 CIBA Foundation Symposium on Quinones in Electron Transport eds. Wolsten- holme and O’Connor Churchill London 1961. 286 Martius ref. 294 p. 312. Wiss Gloor and Weber ref. 294 p. 264. 297 Birch ref.294 p. 233. 288 Lynen ref. 294 p. 244. 299 Olson Dialameh and Bentley ref. 294 p. 284. [ P 4 C Imevalonate. 226 QUARTERLY REVIEWS ij (CLXVII) k OH -1 (C u 1 x a) + co + PP (CUXI I) FIG. 32. Ubiquinones and some possible biogenetic relationships. concept M a r t i u ~ ~ ~ described the conversion of tritium-labelled 2,3- dimethoxy-5-methylbenzoquinone (CoQ,) into CoQa (CLXVII n = 4) by rat liver homogenates to which geranylgeranyl pyrophosphate was added. Some synthesis of CoQ9 and CoQ, was also observed simultane- ously. CoQ (CLXVII n = 9) which apparently is the predominant form of coenzyme Q in the rat,29g-301 was not formed from CoQs (CLXVII n = 6)302 but results consistent with the shortening of the isoprenoid chain have been r e p ~ r t e d . ~ * ~ ~ ~ ~ Although Wiss et al.296 were able to demonstrate incorporation of labelled mevalonate into the ubiquinone polyisoprenoid chain in vivo with a labelling pattern consistent with that found in other isoprenoid systems they could not confirm the utilisation of the preformed quinone nucleus as described by M a r t i ~ s ~ ~ ~ nor was Gloor and Wiss Biochem.Biophvs. Res. Comm. 1960 2 222. Lawson Mercer Glover and Morton Biochem. J. 1960 74 38P. 302 Rudney and Sugimura ref. 294 p. 21 1. 303 Threlfall and Glover Biochem. J. 1962 82 14P. CLAYTON STEROLS STEROIDS AND TERPENOIDS 227 mevalonate utilised in a rat liver h~mogenate.~~~ More recently however the incorporation of mevalonate into ubiquinone in a rat liver homogenate has been The structure of the acceptor molecule to which the isoprenoid chain becomes attached is still uncertain.Olson et aL306 confirm the finding of Wiss et ~ 1 1 . ~ ~ ~ that the preformed quinone nucleus is apparently not utilized. The absence of known biosynthetic pathways for the formation of aromatic compounds (other than estrogens) in mammals has directed attention to the possible rale of phenylalanine and tyrosine as precursors of the quinone moiety. While earlier attempts to test this possibility gave negative it now seems clear that these essential amino-acids can provide the quinone moiety of the ubiquinones in the rat.307 When un- labelled p-hydroxybenzoate was administered simultaneously with the labelled amino-acids the labelling of the quinone ring was virtually abolished suggesting that p-hydroxybenzoate may be an intermediate.Benzoate labelled in the ring was poorly incorporated and the label from carboxyl-labelled benzoate was entirely lost. Further studies by Olson et ~ 1 . ~ ~ ~ indicate that 2,4-dihydroxybenzoate and 3,4-dihydroxybenzoate may be precursors of the quinone moiety. The C-methyl group of the qui- none ring and also the 0-methyl groups were derived from the 1-carbon PO 01.307 The above results of Olson and his co-workers in the rat are similar in many respects to those reported by Rudney and Parson for the bio- synthesis of CoQl0 by Rhodospirillum rubrum. These authors observed the incorporation of p-hydroxybenzaldehyde into the quinone moiety with loss of the aldehydic carbon.308 p-Hydroxybenzoic acid was also uti1ised3O9 and the C-methyl group of the CoQ ring was derived from methi~nine.~~~ Rhodoquinone which differs from ubiquinone only in that one phenolic hydroxyl group is unmethylated is probably not a precursor of ubi- q~inone.~O~ Birch and his co-workers have suggested an analogy between the bio- synthesis of the ubiquinones and the mould metabolite aurantiogliocladin (CLXX).This compound is apparently synthesised from four acetate molecules with loss of the carboxyl carbon of one of them and introduc- tion of one C-methyl and two 0-methyl carbons from the 1-carbon poo1.3109297 The intact incorporation of 6-methylsalicylic acid (CLXIX) was reported but orsellinic acid (CLXIXa) originally suggested as an intermediate,310 was not u t i l i ~ e d . ~ ~ ~ The significance of these results for 304 Wiss Gloor and Weber Amer. J. Clitz. Nrttr. 1961 9 27. 306 Green Diplock Bunyan and McHale Biochim.Biophys. Acta 1963 78 739. 306 Olson Dialameh Aiyar Ramsey Riegl and Bentley Proc. 6th International congress of Biochemistry New York 1964 (Abstracts) p. 434. 307 Olson Bentley Aiyar Dialameh Gold Ramsey and Springer J. Biol. Chem. 1963,238 PC 3146. 308 Rudney and Parsons J. Biol. Chem. 1963 238 PC 3137. Parson and Rudney Proc. 6th International Congress of Biochemistry New York 1964 (Abstracts) p. 434. 310 Birch Fryer and Smith Proc. Chem. SOC. 1958 343. 228 QUARTERLY REVIEWS ubiquinone synthesis is unclear. However the observed introduction of the C-methyl group from the l-carbon pool in both the mammaI3O7 and a mould308 together with other evidence306 makes it unlikely that orsellinic acid is a precursor in these systems. The decarboxylation indicated by these various studies could be coupled with the attachment of the polyisoprenoid chain in an enzymic SN2 reaction (CLXXI-CLXXII).An analogous coupled decarboxylation- methylation process may be involved in the introduction of one of the C-methyl groups of aurantiogliacladin. Considerable interest centres around the inter-relationship of the ubiquinones and cyclised derivatives of the ubichromenol type (CLXVIII). Although various lines of e~idence~ll-~l~ have suggested that ubichro- menol may be an artifact of isolation or storage evidence for its bio- synthesis from ubiquinone or from some common precursor has been The possible participation of ubichromenol derivatives in the generation of high-energy phosphate in the process of oxidative phosphorylation has been discussed by several authors.16* 9314-317 Terpenoid Alkaloids.-Several classes of alkaloid have carbon skeletons that are totally or partially isoprenoid.Excellent sources of references up to 1962 are the reviews by Leete318 and Batter~by.~'~ It is now clear that the ergot alkaloids related to lysergic acid (CLXXIII) originate as suggested by Mothes et al.320 by condensation between (pre- sumably) dimethylallyl pyrophosphate and tryptophan. Thus in agro- clavine (CLXXVI R = H) and elmyoclavine (CLXXVI R = OH) the bulk of the label of [2-14C]mevalonic acid is in the branch carbon C(l,) as shown though some labelling was present in C(7). Controversy now surrounds the question of how the isoprene and tryptophan fragments become united in particular whether the dimethylallyl substituent becomes attached initially to C(4) of the tryptophan nucleus (CLXXIV) or to the 2-carbon of the alanyl side chain (CLXXV).Weygand et al.322 have recently discussed a variety of suggested mechanisms and have described experiments to compare the utilisation of their suggested intermediate (CLXXV) with that of the 4-dimethylallyl intermediate (CLXXIV) suggested by Pleininger et aZ.323 who have also reported similar experi- 311 Draper and Csallany Biochem. Biophys. Res. Comm. 1960 2 307. 312 Links Biochim. Biophys. Acta 1960 38 193. 313 Stevenson Henning and Morton Biochem. J. 1964 89 58P. 314 Folkers Shunk Linn Trenner Wolf Hoffman Page and Koniuszy ref. 294 316 Clark and Todd ref. 294 p. 190. 316 Slater Colpa-Boonstra and Links ref. 294 p. 161. '17 Moore and Folkers J. Amer.Chem. Soc. 1964 86 3393. 319 Battersby Quart. Rev. 1961 15 259. 320 Mothes Weygand Groger and Grisebach Z . Naturforsch. 1958 13b 41. 321 Battacharji Birch Rrack Hofmann Kobel Smith Smith and Winter J. 1962 322 Weygand Floss Mothes Groger and Mothes 2. Nuturforsch. 1964 19b 202. 323 Pleininger Fischer and hide Angew. Chem. 1962 74 430. p. loo. Leete ref. 43 p. 739. 421. CLAYTON STEROLS STEROIDS AND TERPENOIDS 229 H02C 8 \ H H H3c cH3 &jH2 H H,C CH Id \ H )( (CLXXVII) OCH ...- f C t - P (CLxXVlIl) " O W (CLXXIX) r n e n t ~ . ~ ~ ~ The suggested precursors were supplied to growing Claviceps cultures either singly or simultaneously with the two compounds differ- ently labelled with 14C and 3H. Elymoclavin (CLXXVI R = OH) was isolated and the relative efficiency of incorporation of the test substances was assayed.Both groups of workers agreed that the 4-dimethylallyl derivative (CLXXIV) was the better precursor a finding which supports the proposals of Battacharji et aZ.,321 but doubts were expressed as to the real significance of the results and the r81e that degradative reactions might have played in making the labelled materials available for alkaloid synthe- sis. The testing of doubly-labelled precursors with (for example) 14C in the dimethylallyl chain and 3H in the alanine chain with comparison of labelling ratios in the starting material and in the product might answer such doubts. The biological substitution of a diinethylallyl residue in the aromatic ring of tryptophan presumably has an analogy in the biosynthesis of the mould metabolite echinulin (CLXXVII) for which tryptophan is a precursor.325 The uncertainties that may arise in assessing the utilisation of a precursor by intact plant tissues is illustrated by the recent reports of non-utilisa- tion32s and utilisation (with low efficiency)327 of labelled mevalonate in the biosynthesis of the Delphinium alkaloid lycoctinine (CLXXVIII). 324 Pleininger Fischer and Leide Annalen 1964 672 223. 326 Birch and Farrar J. 1963 4277. Herbert and Kirby Tetrahedron Letters 1963 23 1505. s27 B ~ M and May Experientia 1964 20 252. 230 QUARTERLY REVIEWS The incorporation of mevalonate into Solariurn alkaloids such as solanidine (CLXXIX) in which the carbon skeleton is identical with that of cholesterol has also been reported.328 That cholesterol itself may indeed be a precursor of these alkaloids should now be seriously considered in view of its proven occurrence in the same plant An intermediary r61e of cholesterol in the biogenesis of sapogenins has been suggested on the basis of studies with 14C-labelled precursors in Dioscorea plants.329 Conclusion It happens that while these paragraphs were being written the 1964 Nobel prizes for medicine were being awarded to Konrad Bloch and Fodor Lynen for their independent contributions to the fundamental under- standing of the biosynthesis of terpenoids and other lipids.In attempting to assemble material for this Review it has become all too apparent how vast are the implications of this new knowledge of terpene biosynthesis for so many different areas of biology and biochemistry. It would be im- possible except in an extensive monograph to cover in detail all of the topics that have been touched upon and even then an author with the requisite chemical and biochemical expertise in all of these fields would be hard if not impossible to find.Moreover the rate at which fundamental un- derstandingof terpenoid biosynthesis is now being utilised in further explora- tions would outdate such a monograph even before it left the printing press. One of the most striking features of the area reviewed is the extensive interaction that has occurred between the “chemical” and “biochemical” disciplines. In the context of historical development many examples have been cited and in terms of recent contributions the extraordinary success of the joint work of Cornforth and PopjAk is pre-eminent. Such interdisciplin- ary contributions will no doubt become more frequent since much work especially in relation to the biogenesis of alkaloids and mould products poses interesting problems calling for study at an enzymic level.Moreover in the light of rapidly increasing knowledge of protein structure compara- tive studies of some of the enzymes responsible for the formation of closely related cyclic terpenoids may provide new insight into the detailed struc- tures of the active sites of these enzymes and consequently into their mechanisms of action. Finally the unerring accuracy with which organic chemical theory has been able to predict the course of many biological transformations in the terpene series poses fascinating but at present unanswerable questions con- cerning biochemical evolution for in no area of biochemistry is it more obvi- ous that enzymes exploit the inherent potential reactivities of their substrates.This Review was written during tenure of an Established Investigatorship of the American Heart Association. 32R Coseva and Paseschnichenko Proc. 5th International Congress of Biochemistry MOSCOW 1961 vol. 7 287-293 Pergamon 1963. 329 Bennett and Heftmann Proc. 6th International Congress of Biochemistry New York 1964 (Abstracts) p. 565.

ISSN:0009-2681    

DOI:10.1039/QR9651900201

出版商:RSC    

年代:1965

数据来源: RSC

摘要:

THEORY AND APPLICATIONS OF VACUUM MICROBALANCE TECHNIQUES By JOHN M. THOMAS and BRIAN R. WILLIAMS* (DEPARTMENT OF CHEMISTRY UNIVERSITY COLLEGE OF NORTH WALES BANGOR) 1. Introduction EVERY chemist is familiar with the straightforward technique of weighing ; and the advantages of using microbalances in routine operations ranging from chemical analysis to thermogravimetry in which the course of a chemical reaction is followed by recording changes in weight are well known. But the versatility of vacuum microbalance techniques in which measurements of small forces are carried out in situ does not seem to be generally recognised although two outstanding examples-the determina- ationl of atomic weights from the measurement of gaseous densities and the direct experimental verification2 of the theories of long-range forces of molecular attraction-have been discussed in previous Quarterly Reviews.This Review is intended (i) to survey the various types of thermo- dynamic kinetic and structural information which have been established using vacuum microgravimetry ; (ii) to indicate other problems where similar techniques may profitably be applied; and (iii) to outline the theoretical principles which govern the operation of vacuum microbalances. Some attention will be given to the limitations and errors associated with the various types of microbalances now extant; but no attempt will be made either to trace the evolution of vacuum microbalances or to provide precise details concerning construction manipulation etc. since reposi- tories of such information became available in 1953 when the uses of micro- balances in the study of solid surfaces were surveyed by Rhodin3 and G~lbransen.~ Since that time the scope and applications of vacuum micro- gravimetry have multiplied as evidenced by the fact that international conferences dealing with the corpus of information on this topic have been held annually since 1960.2. Definition of Terms The term vacuum microbalance itself merits comment. It refers to balances capable of detecting mass changes of down to a few microgrammes (pg) and which are either to be operated under vacuum conditions gener- ally better than torr. mm. Hg.) or have been subjected to similar vacuum conditions during the course of sample pretreatment. The term is by no means precise and is used as much to describe an ultramicro- balance capable of detecting pg in an ultra high vacuum system * Present address National Research Council Ottawa Canada.R. Whytlaw-Gray Quart. Rev. 1950,4. 153. B. V. Derjaguin I. I. Abrikosova and E. M. Lifshitz Quart. Rev. 1956 10 295. T. N. Rhodin Adv. Catalysis 1953,5 39. E. A. Gulbransen Adv. Catalysis 1953 5 119. 23 1 232 QUARTERLY REVIEWS (< torr.) as one that detects 10 pg in a soft vacuum of about lo-’ torr. The designation is also used to describe sensitive devices for measur- ing small forces (less than dyn) not due to gravity. The sensitivity of a balance is the ratio of the reversible response to the change of mass which produces that response. The latter is usually measured especially for beam balances by the de- flexion of the beam-end and is expressed either in radians or in distance moved.Confusion may be avoided when reading microbalance literature if it is borne in mind that the inverse ratio of the sensitivity is frequently used to convey the precision of a balance (sic.). Moreover in some articles the sensitivity is quoted simply as a mass (e.g. 1 pg). even though such an expression is meaningless. Almost invariably however a closer enquiry will reveal that elsewhere in the articles a specified beam deflexion (usually 0.001 cm for beam-type vacuum microbalances) is quoted. The term sensibility often a more useful quantity stands for the minimum variation in mass that may be reproducibly measured with a given ac- curacy. Two other factors are used to describe the performance of a micro- balance the range which is the maximum variation in mass that the balance can take for a given load and the capacity which is simply the maximum permissible load.Sensitivity and Sensibility. 3. Types of Vacuum Microbalances The ideal vacuum microbalance should have large capacity and range be of high sensitivity (the magnitude of which should not vary with load) possess small temperature and pressure coefficients have great zero point stability be free from vibrational and electrostatic disturbances and be capable of straightforward calibration. In addition the material of the balance and other essential accoutrements including the balance frame suspensory fibres and cements must be capable of withstanding baking in vacuo at temperatures high enough to remove chemisorbed and physically absorbed impurities and be chemically inert to the gases used in subsequent studies.In view of these stringent requirements it can be appreciated that most commercial balances cannot with impunity be used as vacuum microbalances. (There are a few notable exceptions see e.g. refs. 5 and 6 where gold-plated and aluminium balances respectively are described). A very large number of special vacuum microbalances have been and are continually being evolved. Following a time-honoured practice the vast majority of balances have been fashioned out of silica. This material certainly meets most requirements; but that there is nothing sacrosanct in this choice and that a material such as duralumin is just as SOC. 1959 55 2166. K. H. Behrndt Plenum Press New York. 1963 p. 29. ri J. J. Scholten P. Zwietering J. A. Konvalinka and J. H. de Boer Trans.Faraday L. Cahn and H. R Schultz in “Vacuum Microbalance Techniques” vol. 3 ed. THOMAS AND WILLIAMS VACUUM MICROBALANCE TECHNIQUES 233 useable has recently been demonstrated theoretically’ and experimentally*. As a general rule vacuum microbalances have capacities that lie between 0.02 to 2 g. and sensibilities that are often as little as pg but more customarily in the region of 1 pg. It can therefore be seen that the ratio of the capacity to the sensibility is not very different (ca lo6) for conventional and vacuum microbalances. What is significant however is that vacuum microbalances have (i) the ability to cope with minute variations in the mass of small samples such as single crystals and thin films and (ii) a very considerable range. It is instructive to consider the operation of vacuum microbalances by using the following arbitrary classification.(a) Cantilever Microba1ances.-The principle underlying the mode of operation of this type of balance is the bending of a thin beam when a weight is suspended from one end the other end being fixed. As the movement of the beam-end is proportional to the total load the ratio of the capacity to the sensibility will be limited. The sensitivity of such a balance is greatest for thin beams which can take but small loads. Although this type of balance can be used to detect mass changes of 10-1 pg it suffers from the disadvantage of being prone to errors arising from temperature inhomogeneities (the balance being unsymmetrical) and of requiring buoyancy corrections during operation. (b) Spring Balances.-Helical springs made of silica have been extensively especially by M ~ B a i n ~ ~ for a variety of vacuum studies.The extension of the spring is proportional to the total load so that again high sensitivity is attainable only at the expense of reduced capacity. The sensitivity is itself dependent on the diameters of the fibre and the coil the temperature of the helix and the number of coils per unit length. The necessity for buoyancy corrections and the notorious fragility of silica springs have tended to diminish the popularity of helical balances but this trend has been offset by recent advances in the construction of springs made of tungstenlO or copper-berylliumll alloy. Such springs have greater internal friction so that oscillations are readily damped in vacuo but they possess the undesirable property of metallurgical creep and re- quire more stringent thermostatting.Spring balances though not capable of the sensitivity of beam-type balances have been used12 for the study of gas-solid interaction at high temperature and are amenable to electro- ’ C. H. Massen J. A. Poulis and J. M. Thomas J . Sci. Instr. 1964,41 302. * J. M. Thomas and B. R. Williams in “Vacuum Microbalance Techniques” vol. 4 (a) F. Emich Monarsh 1915,36 436; (b) J. McBain and A. Bakr J. Arner. Chem. lo S. L. Madorsky in “Vacuum Microbalance Techniques” vol. 2 ed. R. F. Walker l1 P. J. Anderson and R. F. Horlock Trans. Furaduy Soc. 1962 58 1993. l2 P. Connor J. B. Lewis and W. J. Thomas “Proceedings of Fifth Conference on ed. F. A. Brassart Plenum Press New York 1965. Soc. 1926 48 690.Plenum Press New York 1962 p. 47. Carbon” Pergamon Press Oxford 1962 p. 120. 2 234 QUARTERLY REVIEWS magnetic compensation,13 which serves to increase their range for a given sensitivity. (c) Beam-type Balances.-Although vacuum microbalances have been made1* which are similar in design to conventional microbalances both silica and agate knifc-edges being utilised the majority of those now constructed are beam-type gravity balances that are variants either of the classical Nernst-Donau apparatus later developed by Gulbransen* and R h ~ d i n ~ or pivotal balances of the kind made by Gregg15 and others.16 Torsional balances which utilise the torsional moment of the wire consti- tuting the primary fulcrum to restore the balance to an original setting are also used.17 In this type of balance the torsional con- tribution of the suspension wire that acts as primary fulcrum (Fig.1) is ( i ) Gravity microbalances. FIG. 1 . Schematic illustration of beam-type microbalance ( a ) elevation (b) plan. made so small by using a very thin wire (e.g. lop diameter) that it is neglig- ible compared with the “gravity” moment exerted by the sample and counterweight which are suspended from the two secondary fulcrums. To render the balance stable and to produce the desired sensitivity the beam is offset by a calculated amount at a computed distance either side of the primary fulcrum [Fig. 1 (b)]. Comparatively little manipulative skill is required to construct a balance of this type with a sensitivity such that a deflexion of 0.001 cm. of the beam-end corresponds to a mass change of 1 pg.Greater care in securing parity of arm length and uniformity of cross- section enables a sensibility of 10-1 pg to be reached and a sensitivity that remains essentially independent of load (see Section 4).lS By utilising thin silica beams which sagged when loaded Wolsky and Zdanuklg were able to arrange for the sensitivity to increase commensurately with increase l 3 J. Hooley Canad. J. Chem. 1957 35 374. l4 B. D. Steele and K. Grant Proc. Roy. SOC. 1909 A 82 580; S. Machin Ph.D. l6 S. J. Gregg J. 1946 561; 1955 1438. l6 A. W. Czanderna and J. M. Honig Analyt. Cliem. 1957,29 1206. l7 J. Strange Ph.D. Thesis Pennsylvania State University 1964. l9 S. P. Wolsky and E. J. Zdanuk in “Vacuum Microbalance Techniques” vol. 2 Thesis Rennselaer Polytech. 1961. B. R. Williams Ph.D.Thesis University of Wales 1964. ed. R. F. Walker Plenum Press New York 1962 p. 37. THOMAS AND WILLIAMS VACUUM MICROBALANCE TECHNIQUES 235 of load. The limiting value for the sensibility of this type of balance appears to be ca. pg; although it does seem possible to increase the capacity without sacrificing sensitivity.20 The prime advantage of this type of balance which can readily be con- verted to a null-point instrument using electromagnetic adjustment,21 is that owing to its symmetry disturbances arising from temperature and pressure gradients are minimised (but see Section 4) and buoyancy corrections need not be applied if the counterweight is so chosen that its density matches that of the sample. (ii) PivotaZ balances. In this modification two fine points are employed instead of a knife-edge for suspending the balance beam at the primary fulcrum.Gregg15 built a sturdy (capacity 20 g reciprocal sensitivity 3 x 1 0-4 g./div.) pivotal balance using gramophone needles for the primary and secondary fulcrums. More sensitive* variants of this design have been constructed :Is thus Czanderna22 has recently developed an automatically operated pivotal balance that can take a load of 20 g. and has a sensibility (a) Quartz Crystal Microba1ances.-Attempts to circumvent the dis- advantages of the types of microbalances so far discussed e.g. difficulties associated with their usage in ultra high vacua (< torr.) and their tediousness in operation have resulted in the successful d e v e l ~ p m e n t ~ ~ ~ ~ ~ of a new principle of weighing which utilises the fact that the frequency of a piezoelectric crystal is dependent on the total vibrating mass.Commerci- ally available quartz crystals which can be fitted with a gold plate to act as “sample holder” are made part of an oscillating circuit and the change in frequency of the resonating crystal as a result of addition or loss of mass to or from the holder (by condensation or evaporation of adsorbed species) is monitored with an electronic counter. This kind of balance can be calibrated by evaporating a metal on to the holder and determining the mass of the deposit by subsequent spectrochemical analysis.24 Frequency changes equivalent to 1 x pg./cm2 can now be measured such precision implies that less than a thousandth of a monolayer of oxygen may be detected. * Some caution is required in interpreting the meaning of increased sensitivity.For convenience the sensitivity is assessed by the smallness of the mass change corres- ponding to a fixed beam deflexion (Section 2). If ever-decreasing values of the deflexion can be detected (e.g. electronically) then for a fixed balance performance the mass change decreases proportionally. Hence it appears as if the balance densitivity is en- hanced purely by the improvement in measuring deflexion. 2o R. F. Hampson and R. J. Walker J . Res. Nat. Bur. Stand. Sect. A. 1961,65 289; J. M. Thomas E. L. Evans and B. R. Williams to be published. 21 T. Gast Chem.-Ing.-Tech. 1957 29 262. 22 A. W. Czanderna in “Vacuum Microbalance Techniques” vol. 4 ed. F. A. Brassart Plenum Press New York 1965. 23 (a) G. Sauerbrey 2. Plzysik.1959 155 206; (b) A. W. Warner and C. D. Stock- bridge in “Vacuum Microbalance Techniques” vol. 2 ed. R. F. Walker Plenum Press New York 1962 p. 71 ; p. 93. 24 I. Haller and P. White Rev. Sci. Instr. 1963,34,677. of 2 x 18-1pg. 236 QUARTERLY REVIEWS (e) Vertically Suspended Balances.-In measuring small forces there is much to be gained by using an instrument that rotates about a vertical rather than a horizontal axis. The recording of very small forces exerted on samples placed in inhomogeneous magnetic fields can be acc~mplished~~ with great precision using what is effectively a vertically suspended torsion balance advantage being taken either of modern electromagnetic means of operating the balance as a null-point instrument2j or of recent refinements in the use of the Poggendorf optical lever26 to measure the precise amount of deflexion.It will emerge later that one of the most efficacious means of measuring the vapour pressures of solids involves what is nowadays designated the torsion-effusion technique [see Section 5(a)]. Introduced by Vo1mer2' in 1931 this uses a vertically suspended torsion fibre attached to which there is a small concave mirror and at the free end a small cell with two orifices situated eccentrically on opposite faces [Fig. 2(a)]. The top view of a FIG. 2. A selection of cells used with vertically suspended balances (a) and (6) torsion- efusion cells (c) arrangement for measuring growth or evaporation of crystals and ( d ) arrangement for study of moIecular beants. typical2* sample holder used in a torsion-effusion balance is shown in Fig.2(b) the axis of rotation passing perpendicularly to the plane of the paper through X. The type of assembly employed to measure the rate of growth or e v a p ~ r a t i o n ~ ~ of a crystal is illustrated by Fig. 2(c) and to record the intensity of a molecular beam3* by Fig. 2(d). It is possible to devise systems that can simultaneously be operated both as vertically suspended torsion balances and horizontally mounted beam-type balances [see Section 5(a)]. 25 J. A. Poulis C. H. Massen and P. van der Leeden Appl. Sci. Res. B. 1961,9 133. 26 R. V. Jones J. Sci. Instr. 1961,38 37. *' M. Volmer,Z. Physik Chem. Bodenstein Festband 1931 836. 28 R. S. Bradley and T. G. Cleasby J. 1953 1681. 2 9 (a) S. A. Kitchener and R. F. Stickland-Constable Proc. Roy. Soc. 1958 A 245 30 D.W. Bassett and A. J. B. Robertson Brit. J . Appl. Phys. 1959,10 534. 93; (b) E. K. Rideal and P. M. Wiggins Proc. Roy. SOC. 1951 A 210 291. THOMAS AND WILLIAMS VACUUM MICROBALANCE TECHNIQUES 237 4. Theory of Microbalances We shall for heuristic purposes briefly discuss the quantitative ex- pression for the sensitivity of a beam-type gravity microbalance and the deflexion of a vertically suspended balance. Then we shall consider the theoretical limitations of microbalances. (a) Beam-type Gravity Balances.-If (see Fig. 3) C denotes the primary fulcrum (i.e. the suspension axis) and A and B the secondary fulcrums then if the length of the beam is 21 the masses of the sample counter- weight and beam are respectively (M + m) M and G the distance of the centre of mass below the primary fulcrum is s and a is the distance of + M+m FIG.3. Illustration of theory of beam-type microbalances. the primary fulcrum above the centre of the beam we have by taking moments about C 9- 1 - dm G s + 2 M a + K d+/dm being the sensitivity. K is the term that takes into account the torsional contribution of the main suspension wire it is negligible for thin wires so that equation (1) becomes identical with that which expresses the sensitivity of an ordinary knife-edge balance.31 If the sensitivity is to be independent of load a must be zero; in other words all the fulcrums must be coplanar and this is the main reason for offsetting the beams. Maximum sensitivity is reached when I is large and both G and s are small. Long thin beams are obviously desirable but a compromise has to be made as sagging will be more pronounced with such beams.It is more prudent to construct a light sturdy balance and to arrange for some means of preselecting a value of s while still maintaining stability through having D lower than C so as to permit the use of a range of sensitivities.* The sensitivity of a beam balance is improved? marginally if the beam has a rectangular rather than a circular ~ r ~ ~ ~ - ~ e ~ t i o n . * ~ * To this end the procedure pioneered by Lambert and Phillips3a of adding silica weights above the centre of the beam has often been adopted. t The weight of the beam for a given moment of inertia is decreased by using a rectangular cross-section. 31 Notes on Applied Science No. 7 (National Physical Laboratory) H.M.S.O. Lon- don 1954. 32 B. Lambert and G.S. C. Phillips Phil. Trans. A 1950,242,415. s3 E. A. Gulbransen and K. F. Andrew in “Vacuum Microbalance Techniques” vol. 1 ed. M. J. Katz Plenum Press New York p. 1. 238 QUARTERLY REVIEWS (b) Vertically Suspended Torsion Balances.-If a vessel with two effusion holes is suspended in a vacuum on a fibre of torsion constant T the pressure p in the vessel in the neighbourhood of the orifices is related to the angle of twist 8 by an equation which takes the form p = k 8 where k is an apparatus constant. The precise value of k can be evaluated by considering the recoil force arising from effusion through the holes of area a and a at distances b and b from the suspension axis X [Fig. 2(b)]. It transpires that p = 2 T B/(a b + a2 b2) However since the effusion holes in any real cell must be of finite length it has been found3.‘ necessary to incorporate a factorfinto the expression on the right hand side of equation (2).In effect,fis the ratio of the force resulting from the effusion of molecules through an orifice of finite length to that expected for effusion through an infinitely thin orifice and the value off depends on the ratio of the length of the hole to its radius. In general if a force I acts normal to the surface of a vane suspended on a fibre b being the distance from the suspension axis to the point at which the force acts then Bassett and showed how the force produced by a molecular beam of non-condensible gas could be calculated by assuming that the impinging molecules were first adsorbed and subsequently evaporated according to the cosine law.35 By measuring the angle 8 through which the vane was turned and using equation (3) they could calculate the momen- tum carried by the molecules in the molecular beam.(c) Theoretical Limitations.-We have now to consider two distinct types of limitations imposed on the performance of vacuum microbalances. One of these is totally unavoidable; the effect of the other can however be meliorated. Since present-day improvements in the design and construction of microbalances are rendering them pro- gressively more sensitive it is pertinent to enquire where the theoretical limit to the sensitivity lies. A limit is imposed by the random fluctuations in the motions of the atoms of which the balance is composed and of the gas molecules which surround the balance when in operation.These fluctuations disturb the balance so that any mass variation that is smaller than the background (Brownian motion) “noise” cannot be detected. Quantitative estimates of the magnitude of the spurious couple or force caused by Brownian motion have been made by several investigat01-s.~~ Barnes and S i l ~ e r m a n ~ ~ ~ in 1934 were under the false impression that the F b = r B (3) (i) Injluence of Brownian motion. 34 A. W. Searcy and R. D. Freeman J. Chern. Phys. 1955,23,88. 35 M. Knudsen Ann. Physilc. 1910,31,205,633. 36 (a) R. Bowling Barnes and S. Silverman Rev. Mod. Phys. 1934 6 162; (b) J. A. Poulis and J. M. Thomas in ref. 6 p. 1. THOMAS AND WILLIAMS VACUUM MICROBALANCE TECHNIQUES 239 limit of detection of microbalances was about 10 pg* and hence regarded the magnitude of the “spurious mass” estimated by them pg) as of academic interest only.Poulis and Thomas,36b in a recent assessment of the relevance of fluctuation theory38 to the performance of beam-type and spring vacuum microbalances concluded that for the balance designs currently in vogue the “spurious mass” arising from Brownian motion is likely to lie in the range pg. It appears that we are not far from the state already reached with galvanometers where endeavouring to increase sensitivity is pointless because by so doing the fluctuations are simultaneously increased. (In “electronic” weighing e.g. with quartz resonating crystals it is possible to cope to a certain extent with the Brownian noise by integrating the overall response in such a way as to neutralise the noise and so extract a meaningful signal).(ii) Disturbances arising from thermomolecular flow. If the sample or counterweight or even the fibres on which these are suspended experience small temperature gradients when the ambient pressure is low it is possible that spurious couples or forces several orders of magnitude as great as those to be measured may be registered. For example Williams,18 who used a beam-type gravity balance found that when one limb in which was suspended a sample of single crystals of graphite was cooled in liquid air (the graphite counterweight suspended in the other limb being at room temperature) the apparent mass change versus pressure curve shown in Figure 4 was obtained. The apparent mass change is very considerable to SOL ’“i io+ I o-2 I 0-‘ 10 log. press. (torr) FIG.4. Apparent mass change arising from thermo-molecularflow. Sample at - 193”~. counterweight at room temperature. (Reprinted by permission from B. K. Williams Ph.D. Thesis University of Wales 1964). * In 1914 Petterson3’ had weighed 0.2 g. with an accuracy of 1 part in lo9. .s7 H. Petterson Dissertation University of Goteborg 1914. 38 C. W. McCombie Reports Prog. Phys. 1953,16,266. 240 QUARTERLY REVIEWS and passes through a maximum in the Knudsen pressure range to 10-1 torr.) where the mean free path of the gas molecules is comparable with the dimensions of the sample and annular space between sample and the balance limb. Spurious mass changes such as these can seriously impair the precision of vacuum microgravimetry" and can be minimised but perhaps never completely eliminated by diminishing the smallest of temperature gradients along sample counterweight or suspension fibres.Thomas and Poulis4* have formulated a theory which ascribes the occur- rence of these spurious mass changes to thermomolecular flow (thermal transpiration). They use the well-known Knudsen equation where p1 and p2 are the pressures associated with the temperatures Tl and T2 to evaluate the magnitude of the mass changes together with the pressures at which the changes are maximal for various gases. From their theory it emerges40 that microbalances when operating in the Knudsen pressure range will as has been borne out e~perimentally,~~ be prone to relatively large errors unless unconscionably elaborate thermostatting is incorporated. 5. Applications of Vacuum Microbalance Techniques to Problems of Chemical Interest Our discussion of the problems that can be tackled by vacuum gravi- metry will be deliberately eclectic and our arrangement and sub-division of the topics unavoidably somewhat arbitrary.We exclude many notable applications such as the use of quartz microbalances to the compressibilities and virial coefficients of gases and adduce several that are not widely known and speculate upon others as yet untried. In this section we shall concern ourselves with the measurement and use of vapour pressures of substances in the condensed phases. (a) Vapour Pressure and Related Thermodynamic Data.-We recall that the enthalpy and entropy of vaporisation of solids and liquids may be obtained using the second law of thermodynamics from the variation of vapour pressure with temperature.The Clausius-Clapeyron equation relates the vapour pressurep to the absolute temperature T via the standard enthalpy of vaporisation AH" But the enthalpy in general varies with temperature and its value at a * It has been reportedsQ that with a sample held ostensibly at 1O0OoC. the spurious mass change arising from a largely unknown temperature difference was lo3 pg in a pressure of 10-l torr. of helium. sQ A. W. Czanderna ref. 33 p. 129. 40 (a) J. M. Thomas and J. A. Poulis ref. 6 p. 15; (b) J. A. Poulis and J. M. Thomas J. Sci. Instr. 1963 40 95; (c) J. A. Poulis B. Pelupessy C. H. Massen and J. M. Thomas J. Sci. Instr. 1964,41,295. 41 S. P. Wolsky ref. 33 p. 143. d In p/dT = AHo/RT2 ( 5 ) THOMAS AND WILLIAMS VACUUM MICROBALANCE TECHNIQUES 241 particular temperature AH; can be written Ab AH; = AH," + An T + - T2 + AC T-' + .. . . . . 2 (6) AH; being the enthalpy of the hypothetical vaporisation process at 0°K. All terms beyond the first on the right hand side of equation (6) arise from the change in heat capacity ACp for the vaporisation since for both the gaseous and condensed species the heat capacity may be written Cp = a + bT + C T - ~ + . . . . . . . . . If ACp is zero equations (5) and (6) combined yield an equation of the form (7) l n p = - AT-l + B where A is clearly equal to - A H ~ / R and B to AS"/R AS" being the standard entropy of vaporisation. If ACp is finite but constant which is tantamount to saying that we may write Cp = a for both the gaseous and condensed species then AH; = AH," + Aa T l n p = -AT-l + C l n T + D (8) (9) so that substitution in equation (5) leads to from which we may again extract the enthalpy and entropy of vaporisation.The enthalpy of vaporisation at 298"K. often required in thermodynamic calculations (see later) can be obtained from AH," using equations (6) or (8). Note that whether ACp is zero or finite and whatever the range of temperature employed with the second law approach to the thermo- dynamics of vaporisation we obtain one value (an average) of AH,". Using the third law approach we may take advantage of tabulations of free energy functions and hence arrive at an independent value of AH," (or AH;) for every temperature at which vapour pressure is measured. This can be gleaned from the following equations which relate the Gibbs standard free energy of vaporisation AG; to the vapour pressure and to the free energy function change A(G; -H,")/T and Values of (G; --H,")/T also written as f e f may be readily computed for the species in the vapour phase using statistical mechanical methods provided the appropriate values of bond length vibration frequency etc.are known. For the condensed phase,fefvalues are obtainable from heat capacity measurements by using the following equation I T T fef = T 1 CpdT - 1 Cpd In T (12) 0 242 QUARTERLY REVIEWS Having seen how the enthalpy and entropy of vaporisation may be evaluated from vapour pressure data we shall now outline the various ways in which vacuum microbalances can be used actually to determine the vapour pressure. The great advantage in using microbalances is that they enable pressures as low as lo-’ torr.to be determined so that the tech- nique can be used for metals alloys and a vast range of organic and in- organic compounds at temperatures close to 298°K. The rate of loss of mass dw/dt by effusion of vapour through a small hole (area a,) and infinitely small thickness when the (low) pressure inside the effusion vessel is p and that outside is zero is given by:35 ( i ) The Knudsen efusion method. p = (-dw/dt)(2~RT/M)~/a (13) where M is the molecular weight. The value of dwldt is readily determined using a beam-type or spring microbalance the direction of effusion from the suspended cell being so arranged that the recoil force does not con- tribute to that recorded by the balance. The value of a can be obtained either directly by measurement or by calibration using substances such as benzophenone or sulphur of known vapour pressure.Such a calibration will remove uncertainty concerning the effective as opposed to the geo- metric area of the hole. (No effusion hole is infinitely thin [see Section 4(b)] consequently the effusion area has to be multiplied by a so-called Clausing factor,42 closely similar to the factor f previously mentioned ). This method has been used to determine the vapour pressures of paraffins,43 aromatic hydrocarbon^,^^ hydrogen-bonded organic crystals,4j and a large number of metals alloys and non-metallic elements.46 However when used for the determination of the enthalpy of vaporisation of ferro- cene misleading results were [probably because the coefficient of condensation as defined in Section 5(a) (ii) was less than unity].In a recent variant4* of the method the rate of effusion of Ag and Pd from a Knudsen cell has been measured by recording the mass increase of a collector plate which was suspended on a microbalance directly above the effusion hole.* The rate of loss of mass by vaporisation from a sample of area a2 in a vacuum when none of the (ii) Langrnuir free-evaporation * This method is however not as sensitive as the analogous one developed by Carson 42 P. Clausing Ann. Physik. 1932 12 961. 43 A. R. Ubbelohde Trans. Furuduy Soc. 1938 34 282. 44 R. S. Bradley and T. G. Cleasby J. 1953 1690. 4s R. S. Bradley and A. D. Care J. 1953 1688. 46 J. L. Margrave in “Physicochemical Measurements at High Temperatures” ed. J. O’M. Bockris J. L. White and J. D. Mackenzie Butterworths 1959 p.225. 47 J. W. Edwards and G. L. Kington Trans. Faraday SOC. 1962 58 1323. 48 P. D. Zavitsanos Rev. Sci. Znstr. 1964,35 1061. 4B A. S. Carson R. Cooper and D. R. Stranks Trans. Faraduy SOC. 1962,§8,2125. 6o I. Langmuir,Phys. Rev. 1913,2 329. et U Z . ( ~ using isotopically labelled solids. THOMAS AND WILLIAMS VACUUM MICROBALANCE TECHNIQUES 243 gaseous molecules returns to the surface is related to the equilibrium vapour pressure by the equation The factor a known either as the coefficient of condensation or the coeffi- cient of evaporation is the fraction of the gaseous molecules which would condense if they were to impinge on the vaporising surface under equili- brium condition^.^^ In measuring the vapour pressure of metals by this method it is often assumed that ct is unity although there are indisputable indications that ct is fractional and tends to vary with temperature.Mar- grave and his ~ o - w o r k e r s ~ ~ ? ~ ~ have shown that for many metals a = l and the vaporising species are monatomic. A difficulty with this method is that the area a2 is not easy to evaluate. For metals and alloys the roughness factor (true area divided by geometric area) is about two; for other substances it can be several hundreds and its true value ought to be determined as outlined in Section 6(a). Beam-type microbalances have been used to study the vaporisation of metals and refractory substances up to 20OO0C. and the reproducibility of the results can be judged from the following typical data for platinum. Using the second law approach adumbrated in the previous section Dreger and Margrave51b found the slope of the In p versus 1/T plot together with tabulated heat capacity data to yield = 135 & 2 kcal.mole-l the vapour pressure having been determined at eleven temperatures from 1638” to 1786°K. Using the third law [equation (ll)] the same workers found dH0298 = 135.2 & 0.85 kcal. mole-l. An independent determinationz0 carried out using a similar rnicrobalance gave = 134.9 Zt 1-0 kcal. mole-l. From these two “third law” values of dHoZg8 the normal boiling point of platinum is estimated to be 4100 & 100°K. (iii) The torsion-eflusion or torque-Knudsen method. We have already cited [equation (2)] the relationship between the vapour pressure within a cell which has two effusion holes arranged as in Fig. 2(b) and the angle of twist of the fibre on which the cell is suspended in vacuum.Equation (3) shows that the vapour pressure determined this way does not require a knowledge of the molecular complexity of the vaporising species a point that was emphasised when the technique was first used2’ to measure the vaporisation of benzophenone Hg and K at temperatures up to 200°C. The technique has been used up to 1900°C. to measure the enthalpy of vaporisation of several e l e m e n t ~ ~ ~ t ~ ~ and c o r n p o ~ n d s ~ ~ + ~ ~ ~ ~ and in asso- ciation with the results from the Knudsen and Langmuir methods to 61 (a) Ke-Chin Wang L. H. Dreger V. V. Dadape and J. L. Margrave J. Amer. Ceram. SOC. 1960,43,509; (6) L. H. Dreger and J. L. Margrave J. Phys. Chem. 1960 64 1323; (c) R. C. Paule and J. L. Margrave J. Phys.Chem. 1963 67 1896. 62 ( a ) K. Niwa and 2. Sibata J. Chem. SOC. Japan 1940 61 667; (b) R. F. Barrow P. G. Dodsworth A. R. Downie E. A. N. S. Jeffries A. C. P. Pugh F. J. Smith and J. M. Swinstead Trans. Faraday SOC. 1955,51 1354; (c) A. N. Nesmeyanov “Vapour Pressure of the Chemical Elements” Elsevier Amsterdam 1963 p. 47. 63 R. F. Barrow E. A. N. S. Jeffries and J. M. Swinstead Trans. Farahy Soc. 1955 51 1650; 1657. p = ( -dw/dt)(2nRT/M)*/aza (14) 244 QUARTERLY REVIEWS determine the atomicities of Se and other elements in the vapour phase. It has recently been employed for the investigation of the vapour pressure of liquid alloys (such as Ag-Bi) and solid alloys (such as Ag-Mg).54 WesseP and Bradley and Cleasby28 ingeniously arranged for both the Knudsen effusion and the torsion-effusion methods to be employed simultaneously.A torsion effusion cell was suspended on a fibre from one end of a beam balance so that both mass loss due to ordinary Knudsen effusion and the angle of twist arising from torsion-effusion could be measured. The operative equation for such an arrangement incorporates equation (13) into one of the form p = k 0 (see Section 4(b)) M = -(dw/dt)2 2 ~ R T / a ~ ~ ( k 0 ) ~ (1 5 ) Bradley et aZ.28866 used equation (1 5) to determine the molecular weight of the species produced during the vaporisation of a range of organic compounds. As an illustration the molecular weight of oxamide was determined to be 87-1 (actual value 88-05) proving that no decomposition or polymerisation occurs in the sublimate. (b) A Summary of the Information Derivable from Vapour Pressure Measurements.Quite apart from the fact that enthalpies and entropies of vaporisation are required for the compilation of trustworthy thermodynamic data for elements and compounds much valuable information can be extracted from the direct use of the functions themselves.* (i) Bond energies. The mean dissociation energy of the metal-carbon bonds in an organo-metallic compound MR, containing the quatervalent metal M and a monovalent organic radical R is given by D(M-C) = B{4AH,”(R,g) + AH,”(M,g) - AH,”(MR,,g)) (16) where AH,” ( ,g) is the symbol representing the standard enthalpy of formation from the elements of the compound within the brackets. This equation includes the standard enthalpy of formation of the organo- metallic compound MR, in the gaseous state.Calorimetry yields data which are applicable to that state of the compound which is stable at 25°C.; this is usually solid or liquid. The desirability of having reliable data on the enthalpy of vaporisation of the organo-metallic compound (and of the metals) is therefore evident. The enthalpy of vaporisation of the dihalides of iron determined by * An excellent example of the importance of precise measurement of calorimetric entropy is to be found in recent work concerning the entropies of the gaseous dihalides of Be Mg and Zn. The discrepancies between the spectroscopic and calorimetric en- tropies of these compounds has now been tracked down to an error in the vibrational partition fun~tion.~’ 64 J. N. Pratt and A. T. Aldred J. Sci. Instr. 1959,36,465. 5s G. Wessel Z.Physik. 1951 130 539. 66 R. S. Bradley and S. Cotson J. 1953 1684. 57 D. L. Hildenbrand J. Chem. Pliys. 1964 40 3438. THOMAS AND WILLIAMS VACUUM MICROBALANCE TECHNIQUES 245 Sime and dissociation energies of the Fe-X bonds to be evaluated using the torsion-effusion method enabled the mean (17) It is interesting to note that in principle all the quantities on the right hand side of the last equation can be obtained from Knudsen effusion measure- ments. [Wise59 determined the enthalpy of dissociation of F,-hence AH,”(F,g)-by measuring the rate of effusion of the gas at different temperatures and thereby determining the degree of dissociation at each temperature use of the van? Hoff isochore thence yielded the required enthalpy 1. A far-reaching consequence of the availability of enthalpies of vaporisa- tion is the demonstration by Eley,so that the bond formed by the chemi- sorption of H2 on metals is predominantly covalent.Eley arrived at a value of D(M-M) the strength of a metal-metal bond required for his calculation of the strength of the chemisorbed link by writing for face- centred cubic metals (which have twelve-fold co-ordination) D(Fe-X) = ${ 24H,”(X,g) + dH,”(Fe,g) -AH,”(FeX,,g)) D(M-M) = AH0/6 (18) (ii) Lattice energies. When the vapour pressure is low the vapour behaves essentially as an ideal gas and so the enthalpy of vaporisation of a solid may be equated to the lattice energy. The lattice energies of a large number of paraffins hydrogen-bonded crystals and aromatic ring com- pounds with or without functional groups have been tetermined by vacuum microgravimetry.Davies et a1.61a have also determined the lattice energies of numerous hydrogen-bonded solids by precise weighing (but not involving vacuum microbalances). The magnitude of the lattice energies of crystals of carboxylic acids amides etc. is determined28 by contributions from electrostatic interaction energy energy of repulsion and quantum-mechanical dispersion energy. But part of the lattice energy may be assigned to hydrogen-bonding be- tween nearest neighbours. The following values of the hydrogen-bond energy in kcal. mole-l in crvstals of the specified compounds have been computed by Davies and Kybett61b from systematic measurements of the enthalpies of vaporisation straight-chain monohydric alcohols straight-chain even-carbon monocarboxylic acids 6.8 & 0.6 8-9 & 0.4 Bradley and C ~ t s o n ~ ~ found that the difference in lattice energy between the ct-form of anhydrous oxalic acid which crystallises in molecular sheets and the /3-form which has a chain-like arrangement of the mole- cules was small.But the difference in entropy of vaporisation was rather s8 R. J. Sime and N. W. Gregory J. Phys. Chem. 1960,64 86. 5s H. Wise J. Phys. Chem. 1954 58 389. 6o D. D. Eley Discuss. Furuduy SOC. 1950,8 34. (a) M. Davies and J. I. Jones Trans. Furuduy SOC. 1954 50 1042; (b) M. Davies and B. Kybett Nature 1963,200,776. 246 QUARTERLY REVIEWS larger than might be expected from the packing in the two forms and may be associated with the restricted rotation about the C-C axis of the eight- membered ring in the /3-form which has the higher entropy (lower entropy of vaporisation) whereas in the or-form this rotation is suppressed.P Little or no work appears to have been done using vacuum micro- balances to determine the vapour pressure of essentially ionic solids knowledge of the lattice energy of such solids aids quantitative intrepreta- tion of absorption spectra and point-defect concentrations.62 (iii) Phase diagram investigation and thermodynamic activities of alloys. The torsion-effusion method offers a ready means of determining the vapour pressure of one component of an alloy system as a function both of t e m p e r a t ~ r e ~ ~ and composition.63+ The composition is determined by using the equation (19) where R and R are the atom ratios of volatile to non-volatile component initially and at any particular stage when the total weight loss is A w ; MI and M are the atomic weights of the volatile and non-volatile com- ponents respectively and w2 is the weight of the non-valotile component in the effusion cell.This approach is being used to supplement (in some instances to supplant) the traditional methods based on X-ray diffraction and metallography of phase-diagram investigation. B l a ~ k b u r n ~ ~ who was the first to use microbalances for this purpose explored the oxygen dissociation pressures over uranium oxides an established that there are three stable oxides U 0 2 U40 and U,O,,. Many alloy systems have been investigated by means of this t e c h n i q ~ e . ~ ~ * ~ ~ Progress is being made,64 using microbalances in the study of alloys from the standpoint of classical thermodynamics of non-electrolyte solutions.In an alloy AB the component A alone being volatile the thermodynamic activity a A of component A taking pure A as the standard state is given by R = Ro - (Mz Aw/M,w~) aA = P A / P o A (20) where p o A is the pressure of pure A and p A that of component A for a 62 P. Gray Quart. Rev. 1963 17 441. 63 P. E. Blackburn J . Phys. Chem. 1958,62 897. 64 (a) E. Veleckis C. L. Rosen and H. M. Feder J. Phys. Chem. 1961 65 2127; G6 K. M. Myles Acta Metallurgica 1965,13 109. 1962,66,362; (6) K. M. Myles and A. T. Aldred J. Phys. Chem. 1964,68,64. THOMAS AND WILLIAMS VACUUM MICROBALANCE TECHNIQUES 247 particular alloy composition. The activity aB of the non-volatile component can be computedaa using the Gibbs-Duhem equation where NA and N are the respective atomic fractions of the components.Pratt and Aldred,54 upon computing the activities of bismuth in Ag-Bi alloys found the systems to behave non-ideally (activities not synonymous with corresponding atomic fractions) ; and Myles and Aldred64b have shown that in V-Fe alloys iron exhibits fairly large negative deviations from Raoult’s Law throughout the entire compositional range. It has already been implied that the vapour pressure measured using the methods outlined above may differ from the true equilibrium value unless the coefficient of condensation a is unity. Values of cc may be obtained either directly [Section 8(b)] or by taking the ratio of the pressures froin Langmuir and Knudsen effusion experiments when every precaution has been taken (e.g. operating with a large ratio of sample area to orifice area) to ensure that the effusion method yields equilibrium pressures.It is p o s ~ i b l e ~ ~ ~ ~ ~ to draw conclusions about the mechanisms of the vaporisation and condensation process from the magnitude of a; and a plausible theory of the evaporation of metal crystals involving the migration of surface species to and from monatomic ledges and the regions of emergence of dislocations has been formu- lated.a7b However in view of some of the discrepancies reported by in- dependent on identical systems it seems that theory tends to be ahead of experiment in this particular branch. Many systems notably the evaporation of droplets of certain esters and hydrocarbons are on the other hand well ~ n d e r s t o o d . ~ ~ (iv) Coefficients of condensation.6. Applications in Surface Chemistry Irrespective of whether one is following the course of chemisorption of a gas on a solid surface in the sub-monolayer region or the progress of physical adsorption in the multilayer region or the embryonic stages of the oxidation of metals and alloys there is much to commend the gravi- metric over the manometric method of study. When the surface area of the solid in question is small (say a few hundred cm.2) the advantages are overwhelming because unlike the manometric method which becomes progressively more insensitive as the total pressure of adsorbate or oxidant increases the gravimetric method is potentially capable of being just as sensitive at 1 atm. pressure as at 10-lo torr. Moreover when the dead- 66 L. S. Darken and R. W. Gurry “Physical Chemistry of Metals” McGraw-Hill Book Co.Inc. New York 1953. 67 (a) 0. Knacke and I. N . Stranski Progr. MetaZ Phys. 1956 6 181 ; (b) J. P. Hirth and G. M. Pound “Condensation and Evaporation” Pergamon Press Oxford 1963. 68 R. S. Bradley Proc. Roy. Soc. 1951 A 205 553. 8B (a) R. S. Bradley M. G. Evans and R. Whytlaw-Gray Proc. Roy. Soc. 1946 A 186 368; (b) R. S. Bradley and J. Binks ibid. 1949 A 198 226,239. 248 QUARTERLY REVIEWS space volume of the apparatus is large and the surface area of the solid very small (a few cm.2) the manometric method is extremely inaccurate if not impracticable. Bearing in mind that if an area 1 cm.2 is covered by a chemisorbed layer of 0 molecules the mass increase is about 6 x lo- pg we may appreciate that vacuum microgravimetry is ideally suited for the study of both the thermodynamics and kinetics of adsorption.Many applications of this type have already been cited by Rl~odin,~ Gulbransen4 and Gregg.’O The estimation from physical adsorption isotherms of the surface area and pore volume of solids using the B.E.T. and Kelvin equations along with the determination of the heats and entropies of physical adsorption of gases and vapours using the Clausius-Clapeyron equation will not be considered here (see refs. 70 and 71). (a) Study of Chemisorption.-In the last decade it has become in- creasingly evident that extremely high precision may be attained in the study of the chemisorption of various gases on single crystal thin film and polycrystalline adsorbents by using microbalances under ultra-high vacuum conditions.Thus Wolsky and Z d a n ~ k ~ ~ were able to study the cleaning of Ge surface by ion-bombardment record the sputtering rate as a function of ion energy follow the subsequent chemisorption of gases on the “clean” surface all in one microbalance assembly which could be evacuated to 10-lo torr. Stockbridge and Warner,23b using the resonating quartz microbalance were able to follow the evolution and re-adsorption of gases during the fracture of quartz in vacua close to 10-l2 torr. Further significant advances are represented (i) by the work of Bassett and who studied the efficiency of exchange of translational energy in collision between cold gas molecules and a hot surface under conditions where the time of adsorption was small; (ii) by the study carried out by Haller and White,73 of the nucleation of molecular beams on surfaces prepared by evaporation in ultra-high vacuum; and (iii) by the kinetic using the ultra-sensitive pivotal type beam balance of the adsorption of O2 on Ag in which activation energies were obtained for the dissociative adsorption molecular adsorption and surface migration of oxygen.(b) Distinguishing Adsorption and Absorption.-Although it is in principle easy to distinguish between adsorption on the one hand from the various forms of absorption on the other the distinction may in practice present difficulties. When in gas-solid systems the specific area of the solid is small the difficulties are exacerbated and vacuum micrograv- imetry constitutes one of the most satisfactory techniques to employ. ‘O S. J. Gregg “The Surface Chemistry of Solids” Chapman and Hall London 1961.71 J. M. Thomas Sci. Prog. 1962 50,46. 72 S. P. Wolsky and E. J. Zdanuk ref. 33 p. 46. 73 I. Haller and P. White in “Proceedings of lnternatiorial Symposium on the Con- densation and Evaporation of Solids” Dayton Ohio 1962. A. W. Czanderna J. Phys. Chem. 1964,68 27. THOMAS AND WILLIAMS VACUUM MICROBALANCE TECHNIQUES 249 Suppose that a small equilibrium uptake of oxygen by a single crystal of nickel at a fixed temperature had been recorded gravimetrically. If after cleavage in situ the amount taken up by the nickel is increased by about 100 % it can be concluded that adsorption predominates. It has been demonstrated1 gravimetrically that in contrast to the Br,-graphite system where intercalation leads to the formation of lamellar the I,-graphite system involves chemisorption only.When H and U or H and Nb interact the initial chemisorption is followed by extensive absorption often with the formation of separate phases. Katz and G~lbransen~~ have studied the thermodynamics of such systems and the general phenomenon of occlusion of gases in transition metals with admirable precision. In their technique phase-diagrams are constructed by noting amounts of gas uptake-the very reverse of the method pioneered by Blackb~rn~~ and described in Section 5(b) (iii). (c) Oxidation of Metals and Alloys and the Combustion of Solid Fuels.- The systematic study of the kinetics of metal and alloy oxidation at high temperatures and the influence of trace impurities on the pattern of behaviour has been pertinaciously executed over the last two decades by Gulbransen and his Apart from producing a wealth of tech- nically useful data such studies have added to our understanding of the mechanism of the oxidation process and it is now possible to appreciate why in one set of circumstances the rate of oxidation is linear and in another parabolic.The microbalance technique is equally applicable to other gas-solid reactions that undergo mass changes and the oxidation of solid fuels notably the various forms of carbon and graphite by a variety of oxidising gases typifies the niceties of the technique. Thorough outgassing of the sample in high vacuum subsequent oxidation in either static or dynamic conditions at low pressures (to avoid troublesome diffusional effects) and the physical adsorption of an inert gas at low temperature (to yield surface areas and pore volumes) can all be carried out in situ.78 7.The Study of Decomposition Reactions Since standard text~~~v* on thermogravimetric analyses describe a wide range of chemical decompositions that can be followed by weighing only * The “Thermal Analysis Reviews” ed. J. P. Redfern published by the Chemistry Department Battersea College of Technology in collaboration with Stanton Instru- ments Ltd. London constitute excellent surveys of current literature on thermogravi- metric analysis. 75 R. C. Croft Quart. Rev. 1960 14 1. ‘13 0. M. Katz and E. A. Gulbransen in “Non-Stoichiometric Compounds” ed. L. Mandelcorn Wiley New York 1964 ch. 3. 77 E. A. Gulbransen and K. F. Andrew J. Electrochem. SOC. 1963 110 476 and references therein. E. A. Gulbransen K.F. Andrew and F. A. Brassart Westinghouse Research Lab. 7 9 C. Duval “Inorganic Thermogravimetric Analysis” Elsevier Press Amsterdam Report 1963 63-139-120-PI. 1963. 250 QUARTERLY REVIEWS a brief account will be given here of some reactions that are particularly well suited for study by vacuum gravimetry. There are two types of study which merit consideration. The first of these embraces the decomposition of surface phases that extend to no more than a monolayer (i.e. chemisorbed phase) or to a thin film of a distinct phase consisting of several monolayers. Decompositions of such phases are conveniently studied by the analogue of the Langmuir free-evaporation method [Section 5(a) (ii)]. Both beam-type and helical balances have been used to (i) detect the presence and follow the break- down of an integument of molybdic oxide on the surface of MOS,;~O and (ii) prove that a tenaciously held film of water is not removed from TiO surfaces unless a temperature of 800°C.(at torr.) is attained and that point defects consisting of TiO++ ions and quasi-free electrons begin to be produced according to the reaction (22) at about 875"C.81 The formation and breakdown of surface oxides on graphite powders during oxidation and the photodesorptioii of chemisorbed phases on non-stoichiometric oxide catalystss4 have also been elucidated by means of microgravimetry. The second mode of studying decomposition reactions utilises the effu- sion cell technique either straightforwardly in the manner of Knudsen or by employing the principle of torsion-effusion (see Section 5).The ordinary Knudsen method has been extensively used to accumulate thermodynamic data on the decomposition of metal halides carbonates hydrates organic complexes etc. which conform to the general reaction A(s) -+ B(s) and C(g) the solid phases possessing low volatility. Gregory and his collaboratorss5 have produced reliable enthalpy and entropy data for a diversity of thermal decompositions some of which are quite complex reactions owing to the occurrence of simultaneous disproportionation or dimerisation of one of the participants. Thus a successful study has been mades5 of the vaporisation of Fe,Cl and the concomitant release of Cl from solid FeCl between 120" and 150". Farber and Darnells6 were able to study the decomposition of Tic& (between 220" and 480°C.) according to TiO,(s) + TiO++(s) + 4 O,(g) + 2e the pyrolysis of carbonaceous materials in general 2 TiCl,(s) + TiCl,(g) + TiCl,(s) (23) and simultaneously measure the vapour pressure of solid TiCl,.However the decomposition pressures measured by effusion in the systems7; P. Cannon Nature 1959,183,1612. A. W. Czanderna and J. M. Honig J. Phys. Chem. 1959,63,620. R. Jongepier and G. C . A. Schuit J . Catalysis 1964,3,464. Hammer and N. W. Gregory ibid. 1962 66 1705; 1964 68 314. M. Farber and A. J. Darnell J. Phys. Chem. 1955,59 156. 82 E. A. Gulbransen and K. F. Andrew Ind. Eng. Chem. 1952,44 1039. 83 P. L. Waters Analyt. Chern. 1960,32,852. 85 (a) J. 13. Stern and N. W. Gregory J. Phys. Chern. 1957 61 1226; (b) R. R. THOMAS AND WILLIAMS VACUUM MICROBALANCE TECHNIQUES 25 1 Mg(OH),(s) + Mg 0 6) + H 2 W (24) were found to be seriously in error a fact which can be a s ~ r i b e d ~ ~ ~ * ~ [cf.Section 5(a)] to an abnormally small coefficient of condensation. In the kinetic study of thermal decompositions the use of the tortion- effusion technique is advantageous. For example the decomposition of polytetrafluoroethylene has been found to be first order the activation energy being ca. 76 kcal. mole-' and the molecular weight of the vaporising species 100 & 6.89 8. Combined Study of Two or More Phenomena Although combined measurements-such as the use of a mass spectro- meter in Knudsen effusion experiments-have extended the range of in- formation derivable from vapour pressure determinations it is in the study of the surface chemistry of solids that combined or concurrent measure- ments utilising vacuum microbalances have proved most practicable.A selection of the information gained in this way will now be given. (a) Magnetic Susceptibility Measurements.-In the study of solid catalysts apart from measuringg0 the mass changes resulting from ad- sorption and desorption of intermediates formed on the surface micro- balances can be used to record changes in the magnetic susceptibility either of the solid itself or of the adsorbed species. Czandernagl has recently constructed a sensitive vacuum microbalance for determining the magnetic susceptibility of adsorbents and catalysts using the Faraday method. Extremely minute changes in susceptibility accompanying the processes of adsorption desorption oxidation and reduction of the specimens can be detected and the apparatus has so far been used to (i) establish the valency of metallic ions buried substitutionally in alumina powders ; (ii) study the magnetic properties of reduced cupric oxide; and (iii) measure the diamagnetic anisotropy of single crystals of graphite.(b) The Study of Thin Films.-Fundamental studies of thin films of metals and alloys require a knowledge of their mass and thickness optical transmis~ion,~~ ease of deposition and other properties most of which can be obtained using vacuum microbalance techniques. The pre- cision of mass measurements will be enhanced once the resonating quartz devices replace the beam-type balances although the latter have proved remarkably successful in measuring the rates of deposition of thin films. Mayer et al. 93 have completed a particularly elegant study of the formation of thin films during the impingement of a beam of Hg atoms on a silica 87 E.Kay and N. W. Gregory J. Phys. Chem. 1958,62 1079. 88 K. Motzfeldt J. Phys. Chem. 1955 59 139. 8 s C. L. Rosen and A. J. Melveger J. Phys. Chem. 1964 68 1079. C. B. McCarty Diss. A h . 1962,22,2323. A. W. Czanderna see ref. 8. A. W. Czanderna and H. Wieder J. Clrem. Phys. 1961,35,2259. O3 H. Mayer R. Niedermayer W. Schroen D. Stiinkel and H. Gore ref. 6 pp. 75 and 87. 252 QUARTERLY REVIEWS substrate maintained at low temperatures. At - 133 “C. the coefficient of condensation [Section S(b) (iv)] was unity but at higher temperatures 01 was no longer constant at any one temperature and was found to be dependent on the mass of deposit previously formed the time of deposi- tion and the thermal history of the substrate.Although some of the finer details are still obscure the mechanism of thin film formation which has long defied explanation is now reasonably well understood and the dependence of the rate of nucleation on surface temperature and beam intensity as well as the r6le of surface diffusion and time of adsorption are appreciated. (c) Micrography and Micr0gravimetry.-Now that our knowledge of the nature of lattice imperfections has been considerably enlarged 94 and the availability of single crystals of ultra-pure materials has increased efforts can be made to investigate what relation if any exists between the chemical reactivity and the defective character of a solid. By combining the techniques of optical and electron microscopy on the one hand with microgravimetry on the other Boggs and his coworkersg5 have’ established aprima facie case for believing that in the oxidation of metals and alloys oxide nuclei are formed preferentially at the regions of emergence of dis- locations.The precise r61e of dislocations and point defects in thermal decomposition reactions has yet to be clarified. But it appears possible that vacuum microbalances may contribute in a special way to the under- standing of the importance of such defects in e.g. the thermal decomposi- tion of carbonates azides and hydroxides. The precision of microbalances is so great that it enables a very small percentage of decomposition to be detected thus permitting a micrographic study to be made of the distribu- tion of dislocations and point defects at the onset of decomposition or at the culmination of the induction period preceding decomposition.9. Future Possibilities On the basis of previous work and current trends it is possible to enumerate several future uses of vacuum microbalances. (i) Study of polymorphic changes. By obtaining the relevant vapour pressure data Bradley and C o t ~ o n ~ ~ showed which of the two anhydrous forms of oxalic acid was stable at room temperature; they also were able to compute the temperature at which the 01 and /3 forms were in equilib- rium. It should be possible to locate transition temperatures by utilising the fact that a phase-transition is usually accompanied by a density change large enough to cause a readily measurable change in buoyancy when the sample (and “physically stable” counterweight) are suspended in D4 S.Amelinckx “The Direct Observation of Imperfections” Academic Press London 1964. 86 W. E. Boggs R. H. Kachik and G. E. Pellissier J. Electrochem. Soc. 1961 108 6; 1963 110,4. 96 (a) J. M. Thomas and G. D. Renshaw Trans. Faraday Soc. 1965 61 791; (b) G. D. Renshaw and J. M. Thomas in the press. THOMAS AND WlLLIAMS VACUUM MICROBALANCE TECHNIQUES 253 a fixed pressure of an inert gas. (This principle has been employed to follow the crystallisation and melting behaviour of various polymers using a microdensity balance. 97) Margrave46 has made the interesting suggestion that it ought to be possible to measure both the pressure and average molecular weight of a species by recording the apparent mass increase caused by the effusion of that species from a Knudsen cell in a direction which contributes to the force recorded by the microbalance.This procedure is quite distinct from the torsion-effusion method and it offers just as many rewards. (iii) Anisotropy of heats of chemisorption. The resonating quartz balance and the more refined forms of beam-type microbalances offer the attractive possibility of determining the heats of chemisorption of certain gases on well-defined crystallographic planes of single crystal adsorbents. The quartz balance can be directly as a calorimeter to measure the heat liberated; but in order to use the other balances the chemisorptions must be reversible. Such experiments would decide whether as suspected the bewildering range of different chemisorption bonds formed in a given system* are attributable solely to crystallographic factors.(iv) Stud' of adhesion. Just as evaporation kinetics-as the reverse of crystal growth-can be studied using torsion-effusion techniques so also could adhesion-as the reverse of condensation from molecular beams-be studied using either torsion-effusion platesg9 or resonating quartz crystals. In this way the strength of bonding between condensates (e.g. thin films) and substrates could be ascertained. Esoteric use of a vacuum microbalance has been made by Ohmann.lo0 By utilising the thermo-magnetic properties of gadolinium metal which has a Curie point close to 20°C. he measured the intensity of electromagnetic radiation using a microbalance. Calcula- tions showlol that the sensibility of present-day vacuum microbalances is already sufficient for them to be used in the precise measurement of the gravitational constant.And in view of their remarkable sensitivity to temperature gradients at low pressures [Section 4(c) (ii)] they may find use in some specialised aspects of thermometry. We wish to thank Drs. J. A. Poulis C. H. Massen S. J. Gregg Manse1 Davies 0. M. Katz F. A. Brassart and P. L. Waters for stimulating discussions. We acknowledge also the helpful correspondence of Drs. E. A. Gulbransen K. F. Walker J. L. Margrave A. W. Czanderna and W. E. Boggs and the encourage- ment of Professor s. Peat. * For a system as simple as the W + CO one it has been establishedas by the technique of flash-desorption that there are four distinct states of binding a at 20 b1 at 53 pa at 75 and P3 at about 100 kcal. mole-]- (heat of adsorption). n8 P. A. Redhead Trans. Faraday Suc. 1961 57 641. @@ L. M. Fitzgerald private communication. loo Y. ohmann and B. Rydgreen ref. 6 p. 193. lol J. A. Poulis and C. H. Massen private communication. (ii) Molecular complexity of species in the vapour phase. (v) Miscellaneous studies. F. T. Simon and J. M. Rutherford J. Appl. Phys. 1964,35,82.

ISSN:0009-2681    

DOI:10.1039/QR9651900231

出版商:RSC    

年代:1965

数据来源: RSC

摘要:

OSMIUM AND ITS COMPOUNDS By W. P. GRIFFITH (DEPARTMENT OF INORGANX CHEMISTRY IMPERIAL COLLEGE LONDON S.W.7) Discovery and Natural Occurrence.-The element was discovered in 1 802 by Smithson Tennant (1 76 1 -I 8 I 5 ) who prepared the volatile tetroxide by the acid distillation of the black residues remaining from digestion of native platinum with aqua regia. He decided to call the new element ptkne (from ptenos volatile) but wiser counsels prevailed and he subsequently wrote that “as the smell is one of its most distinguishing characters I should on that account incline to call the metal Osmium.”l Like ruthenium which it in many ways resembles osmium is a rare and consequently expensive element (the current price for the pure metal is &60 an ounce). Its abundance in the upper part of the earth’s crust is probably no greater than 0.001 part per million although the abundance in meteor- ites exceeds the terrestrial figure by a factor of about 1500.In nature it is always found with one or more of the other platinum metals and occasionally with gold. The most important mineral sources are osmirid- ium which is an osmium-iridium alloy containing varying amounts of other heavy metals found mainly in Alaska and on the Witwatersrand gold reef; and laurite an osmium-ruthenium disulphide found chiefly in Borneo and the Transvaal. Small quantities of osmium-bearing minerals have also been discovered in the Urals Ethiopia and Colombia. Extraction.-The modern method for extracting osmium in essence duplicates Tennant’s original procedure the ores are treated with aqua regia and the insoluble residues extracted with an oxidising flux such as sodium peroxide and the resulting mass is dissolved in water filtered and the filtrate distilled with nitric acid to give osmium tetroxide which can then be reduced directly to the metal.Properties of the Metal.-Osmium has atomic number 76 and atomic weight 190-2. Seven naturally occurring isotopes have been identified with mass numbers (percentage abundances in parentheses) 184 (0.018) 186 (1.-59) 187 (1.64) 188 (13-3) 189 (16-1) 190 (26-4) 192 (41.0) and of these only two have nuclear spins (187 I = Q and 189 I = 3/2). The following artificial radioisotopes have been made 18 1 182 183 185 187 189 190 191 193 194 and 195 and of these Ig4Os is the longest-lived with a half-life of about seven hundred days.2 The physical properties of the six platinum metals have recently been reviewed :3 annea!ed osmium metal is the hardest of the six and has the highest melting and boiling points Smithson Tennant Phil.Trans. 1804 94 411; Nicholson’s Journal of Natural D. Strominger J. H. Hollander and G. T. Seaborg Rev. Mod. Phys. 1958,30,585; Platinum Metals Review 1963 7 147. Philosophy 1805 10 24. K. Rankama “Progress in Isotope Geology,” Interscience New York 1963 458. 254 GRIFFITH OSMIUM AND ITS COMPOUNDS 255 (estimated as 3050" and 5500"~). It was once thought to be the densest of all elements but the latest figures3 indicate that it has the same specific gravity as iridium (22.61 for osmium and 22-65 for iridium at 2O"c). Like ruthenium but unlike the other platinum metals it has a close-packed hexagonal crystal lattice structure (a = 2-7341 A c/a = 1.5800).Chemically it is the most reactive of the platinum metals. Finely powdered osmium is very slowly oxidised even at room temperatures by the air and so always carries a perceptible odour of the tetroxide but in more condensed forms the metal is not affected by air or oxygen below 150"~. The product of oxidation is the tetroxide though there is evidence of surface film formation of the dioxide below 400"c.* It is attacked by aqua regia and oxidising acids only over a long period of time and is barely affected by hydrochloric and sulphuric acids. Attack by fluorine and chlorine takes place above 1OO"c to give a mixture of halides. It is dissolved by molten alkalis and by oxidising fluxes while the reaction with hydrogen peroxide can be of al- most explosive violence.Catalytic properties. The price of the metal limits its commercial application as a catalyst. Like all the platinum metals it does possess considerable catalytic powers though there is little evidence that these are superior to those of palladium or platinum. It will catalyse the oxidation of carbon monoxide carboxylic acids aldehydes alcohols and amines and the hydrogenation of ketones olefins acetylenes and heterocyclic systems. Little detailed study has been made of such reactions apart from an investigation of the high-temperature synthesis of ammonia at atmos- pheric pressures over an osmium-silica catalyst5 and of the hydrogena- tion of olefins and acetylenes over osmium-alumina.6 Applications of Osmium and its Compounds.-These are few owing largely to the price of the materials their nature and the difficulty of working osmium metal.The latter was once used in alloy for electric light filaments and for fountain pen nibs and it is still used for the latter purpose and for watch and clock bearings. Osmium tetroxide remains a valuable reagent for organic hydroxylations (see below) and is extensively used in microscopy for staining tissues. Other occasional uses for osmium include passivation of iron production of artificial diamonds detection of speeds of explosion in gelignite and photographic processing. Compounds of Osmium.-No less than nine oxidation states (VIII to 0) have been established for the compounds (or complexes; in this Review the terms will be used synonymously) of osmium. The tendency of certain ligands to stabilise higher or lower oxidation states is well brought out by the compounds (Table 1).Thus all the "high" oxidation state complexes (VIII to v) contain either fluoride oxide or nitride ligands all of which are J. C. Chaston Platinum Metals Review 1965 9 51. M. Temkin and S. L. Kiperman Zhur.fiz. Khim. 1947,21,927; S . L. Kiperman and G. Webb and P. B. Wells Traits. Faraday SOC. 1965 61 1232. V. S. Granovskaya ibid. 1951 25 557. 256 QUARTERLY REVIEWS TABLE 1. Stabilisation of oxidation states in osmium complexes. Oxidation Donor group state VII oxyhalides VI V IV VIII VII VI V IV I11 I1 I 0 (OSO,F~, OSO (OSO,F&~- OSOF (oQp- (0So5)3- OSOCI [OSO,(OH),]~ (oSo3F3)- [OSO~(OH)~]~- - (osF6) OsOF, OsO,? ( OsO,X,)" (X = C1 Br) - - (OSF,) (OsF,j- OSF, OsCl, (OS~OCI,~)~- OSO, OSX OsBr., (X = S Se Te) .. (OsX,)2- (X = F C1 Br I) OsCI, Os13 - Os,O,? (oSx,)3- (0s acaqJ (X = C1 Br I) OsCl, OsBr, - oso OSI (os(soaa4- [Os(NH&,I3+ Carbonyl ( 0 s phena3+ halides ( 0 s phenJ2+ Carbonyl (0s bipya2+ halides (0s b i ~ ~ a 3 + [acw813+ [wc~,14- [os(co),]~+ ? - Carbonyl halides - [0S(CO),l [O~,(CO~,,I good r-donors and small in size while good r-acceptor ligands such as CO CN- NO+ 2,2'-bipyridyl phosphines arsines and stibines stabilise the "low" (11 to 0) states; the intermediate (IV and 111) states are stabilised by ligands which are good 0-donors but poor r-donors or acceptors (NH, C1- B r I- ethylenediamine,* etc.). Osmium most closely re- sembles rhenium particularly in its oxide and fluoride chemistry and ruthenium particularly in its halogen and nitrogen chemistry and in its ability to form polynuclear complexes with nitrogen and oxygen donors.Almost all osmium complexes are octahedrally co-ordinated. The chief exceptions are the tetroxide and the osmiamate (tetrahedral) and the pentacarbonyl (probably trigonal bipyramidal) ; pentaco-ordination may also occur in the gaseous pentafluoride and in OsO,,NH, while hepta- and octa-co-ordination may be found in (OSOC~,)~- and [O~en,(en-H),]~+ respectively. The general principle that third-row transition elements may often show higher co-ordination numbers than the corresponding second- or first-row elements is exemplified by the oxy-complexes of ruthenium- *Ethylenediamine = en. Ethylenediamine minus a N-hydrogen is shown as en-H. GRIFFITH OSMIUM AND ITS COMPOUNDS 257 (vI) -(vII) and -(VIII) all of which are tetrahedral while for osmium- (vI) -(wI) and -(VIII) the only tetrahedral oxy-species is OsO, the others being octahedral.Wherever possible mention has been made of work relating to structure and bonding in the complexes. There is little direct structural data (Table 2 summarises the available information) but there is a large body of work Oxidation Complex Bond lengths (A) Ref. TABLE 2. Structural data on osmium complexes. state VIII Y VI ? * 9 ) 9 7 9 V IV I1 9 0 Tetrahedral; 0s-0 = 1.717 Tetragonal bisphenoid; 0s-0 = Octahedral; 0s-F = 1.831 0s-N = 1.56 D4h; 0s-0 = 1.77; Os-(OH) = 2.03 D4h; 0s-0 = 1-75; 0s-C1 = 2.38 C ~ V ; 0s-N = 1.61 ; 0s-C1 (trans) =z 2.16 ; 0s-C1 = 2.40 C ~ V ; 0s-N = 1.61 Os-(OHJ = 2.07 Octahedral; 0s-F = 1.82 Octahedral; 0s-Cl = 2-36 0s-Br = 2.60; 0s-P = 2.56 (1) and 2.34 (2) “Octahedral”; 0s-ring = 1.86; Octahedral; 0s-C = 1.95; C-0 = C-C = 1.42 ; 0s-C = 2.22 1.14; 0 s - 0 s = 2.88 14 25 33 39 40 42 42 48 53 102 99 105 * Electron diffraction determination; all others X-ray.t Infrared and Raman studies also reported (see text). # Some distortion from full octahedral symmetry. on electronic spectra of the halogeno-complexes and in particular there has been much careful and systematic work on magnetic properties (Table 3). There is however room for very much more physicochemical research on the element particularly on structures and on reaction kinetics. Osmium (vIII).-Osmium octafluoride was long thought to exist and in fact a cubic structure was assigned toit as a result of early electron- diffraction measurements but recent work has made it clear that the compound is the hexafluoride,’ and it seems likely that old reports of an octachloride* may also be discounted.The oxyfluorides (OsO,F,) and the derived salts (Os03F,)- are however well establishede as well as salts of the (OSO,F,)~- anion.l0 The greater tendency of fluorine to stabilise high-oxidation states than chlorine is illustrated by the fact that the highest oxychloride is of osmium(v~). ’ B. Weinstock and J. G. Malm J. Amer. Chem. Soc. 1958 80 4466; G. B. Har- greaves and R. D. Peacock Proc. Chem. Soc. 1959 85. H. Moraht and C. Wischin Z. anorg. Chem. 1893 3 153 166. M. A. Hepworth and P. L. Robinson J. Znorg. Nuclear Chem. 1957 4 24. lo F. Krauss and D. Wilken 2. anorg. Chem. 1925 145 166. 258 QUARTERLY REVIEWS TABLE 3.Magnetic properties of osmium complexes. d” Complex Temperature range (OK) R.T. 81-297 1 0 1 -295 80-300 90-300 90-300 90-301 80-300 4 K,(OsT,) 80-300 5 [OsC1,(AsPh2Me),] 80-295 P eff (R.T.) 1 -44 1.50 2.06 3.34 1.31 1.51 2.1 3 1.21 1*5F5 1.96 1-38 1 -80 P eff Also (Pre- studied dicted)a 1.3 3- 1 1.4 Cs salt Cs salt Cs,NH,,Bun,N Cs salt Cs NH, Me,N salts Cs salt 1.75 Ref. 31 7 44 46 b b 52 b 52 b a * Measured at “infinite dilution” (see text). R.T. = Room temperature a B. N. Figgis and J. Lewis Progr. Itzorg. Chem. 1964 6 162. A. Earnshaw B. N. Figgis J. Lewis and R. D. Peacock J. 1961 3132. Osmium tetroxide OsO, is probably the commonest compound of osmium although it is one of the most dangerous to handle on account of its high vapour pressure (the solid has a v.p.of 11 mm. at 25”c). It can be made by oxidation of the metal or of any of its compounds with nitric acid (ruthenium tetroxide reacts with nitric acid to give non-volatile products so nitric acid distjllation is a convenient and widely used method of separating the two elements). Osmium tetroxide fonns yellow crystals (m.p. 40.6” b.p. 131.2”) which exist in only one modification not two as was once claimed. In the vapour phase and in organic solvents it is monomeric and has no dipole moment. Michael Faraday was the first to demonstrate its diamagnetism. A considerable amount of thermodynamic data has been accumulated on the compound.ll Both X-ray12 and infrared with Raman13 measurements show the molecule to be tetrahedral and electron-diffraction determinations confirm this1 (earlier electron-diffraction studies were in- correctly interpreted to indicate a distorted structure).The electronic absorption spectrum of the vapour has been measured15 and band assign- ments proposed,16 and oxygen17 nuclear resonance measurements made.17 l1 E. Ogawa Bull. Chem. SOC. Japan 1931 6 314. l3 L. A. Woodward and H. L. Roberts Trans. Faraday. SOC. 1956,52,615; 1960,56 l4 A. F. Wells “Structural Inorganic Chemistry” 3rd edn. Oxford 1962 pp. 456 l5 A. Langseth and B. Qviller Z. phys. Clzem. 1934 27 B 79. l8 A. Carrington and C. K. Jlzrrgensen Mol. Phys. 1961 4 395; S. P. McGlynn and l7 B. N. Figgis R. G. Kidd and R. S. Nyholm Proc. Roy. SOC. 1962 A 269 469. A. Zalkin and D. H. Templeton Acta Cryst. 1953 6 106. 1267. 920. M. Kasha J . Chem. Phys. 1956 24 481. GRIFFITH OSMIUM AND ITS COMPOUNDS 259 Chemically the compound is reactive and a powerful oxidising agent though less so than ruthenium tetroxide.With alkalis it gives rather unstable octahedral osmium(v1Ir) complexes of the form trans- [OSO,(OH),]~- and [OSO,(OH)H~O]-,~~~~~ while with such donor groups as ammonia and phophorus trihalides 1 :1 and 2:l adducts of unknown structures are obtained.lg Polarographic studies on solutions of the compound in various media and at various pH are reported,20 and there is polarographic evidence for the existence of a peroxyosrnate(vr~r).~~ Osmium tetroxide is extensively used in organic chemistry as an oxidis- ing agent and as a catalyst for oxidation. When dissolved in an inert organic solvent it will react smoothly at room temperatures with olefinic double bonds to give cis diols and in the presence of bases such as pyridine the reagent is sufficiently reactive to attack the n-system of aromatic hydrocarbons.Criegee showed in a classic piece of work that the reaction involved an osmium(vr) intermediate which can in many cases be iso- lated :22 I HO-CH I 0 I CH 11 + oso >os(,-;. - I 3- oso3 CH 0 0-CH HO-CH I I I Olefinic hydroxylations can be carried out catalytically with a mixture of osmium tetroxide with barium or silver chromate (the mechanism is obscure; intermediates such as oSo,,clo3 and cyclic osmium esters have been proposedz3) or with Milas’s reagent which consists of an organic solution of hydrogen peroxide activated with a little osmium tetroxide. A recent review on alkene hydroxylations has been published.24 There is some rather unconvincing evidence for the existence of an osmium(vrrr) tetrasulphide.The osmiamates. Addition of aqueous ammonia to an alkaline solution of osmium tetroxide yields the very stable yellow ammonium osmiamate (NH,)(Os03N). X-Ray studies on the potassium salt show that the anion has a slightly distorted tetrahedral configuration (tetragonal bisphenoidal a structure also shown by potassium per-ruthenate KRuO,) in which the osmium-oxygen and osmium-nitrogen distances are both given as l8 W. P. Griffith J. 1964 245. 2o L. Meites J. Amer. Chem. SOC. 1957 79 4631 ; R. E. Cover and L. Meites ibid. 1961 83,4706; J. Perichon S. Palous and R. Buvet Bull. SOC. chim. France 1963 982. 21 K. Fulop and L. J. Csanyi Acta Chem. Acad. Sci. Hung. 1963 38 193. 22 R. Criegee Annalen 1936,522 75; R.Criegee B. Marchand and H. Wannowius ibid. 1942 550 99. 23 L. J. Csanyi Acta Chem. Acad. Sci. Hung. 1959 21 35. 24 F. D. Gunstone “Advances in Organic Chemistry” Interscience New York 1960 vol. 1 110. M. L. Hair and P..L. Robinson J. 1960 2775; 1958 106. 260 QUARTERLY REVIEWS 1~5681;~~ and a similar conclusion was reached from studies on the infra- red and Raman spectra of the compound.26 The elucidation of the struc- ture by purely chemical methods was one of the later triumphs of Werner who showed in 1901 that the previously accepted formulation as a nitrosyl derivative [Os(NO)O 2]- was unlikely since thermal decomposition of the compound gave nitrogen rather than oxides of nitrogen and treatment with hydrochloric acid gave the oxygen-free species K,(OsNCl,) indicating the presence of a direct metal-nitrogen bond.27 Free osmiamic acid is one of the products of the rather complicated reaction between liquid ammonia and osmium tetroxide;lg infrared studies on this and its deuterated form show that the proton in the anhydrous acid is attached to the oxygen rather than to the nitrogen to give [OSN(OH)O~].~* OS~~U~(VII).-T~~S is a rare oxidation state both for osmium and ruthenium.An oxyfluoride (OsOF,) has been made by the fluorination of osmium dioxide. It is paramagnetic and shows an unresolved fluorine magnetic resonance spectrum.29 Recently some heptavalent osmates have been prepared these include alkali metal salts of (0~0,)~ and ( O S ~ ) ~ - made by heating potassium osmate and potassium superoxide with alkali- metal The magnetic moment of Ba,Li(OsO,) is consistent with the presence of osmium(v~~).~~ It is possible that the (0~0,)~- ion should in fact be formulated as [OSO,(OH),]~- the osmium(vr1) analogue of the “perosmate” ion [OsO,(OH) *I2-.Osmium(vr).-The hexafluoride can be made by direct union of the elements. It forms yellow crystals (m.p. 32.1 b.p. 45*9”~).~ A considerable amount of thermodynamic information is available on the X-Ray studies show that the molecule is octahedral7 and the osmium- fluorine bond length of 1.831 A is found from electron-diffra~tion.~~ The magnetic susceptibility has been measured over a temperature range7 and has been interpreted on the basis of the orbital reduction factor being about 0.5 [implying that the two magnetic 5d electrons of osmium (VI) spend about an equal amount of time on the metal and on the ligand~].~ The electronic absorption spectrum has been measured and assignments ~uggested.~ The compound is of particular interest because its 26 F.M. Jaeger and J. E. Zanstra Proc. A c d . Amsterdam 1932 35 610. 26 L. A. Woodward J. A. Creighton and K. A. TayIor Trans. Faraday. SOC. 1960 27 A. Werner and K. Dinklage Ber. 1901,34,2698; 1906,39,500. 28 W. P. Griffith J. 1965 3694. 29 N. Bartlett S. Beaton L. W. Reeves and E. J. Wells Canad. J. Chern. 1964 so R. Scholder and G. Schatz Angew. Chem. 1963 2 264; R. Scholder ibid. 1958 31 A. W. Sleight J. Longo and R. Ward Znorg. Chem. 1962 1 245. 32 G. H. Cady and G. B. Hargreaves J. 1961 1563. 33 B. Weinstock and J. G. Malm Proc. U.N. Con$ on Peaceful Uses of Atomic Energy 34 J. C. Eisenstein J.Chem. Phys. 1961 34 310. 56 1267; J. Lewis and G. Wilkinson J. Inorg. Nuclear Chem. 1958 6 12. 42 2531; N. Bartlett N. K. Jha and J. Trotter Proc. Chem. SOC. 1962 277. 70 591. Geneva 1958 28 125. GRIFFITH OSMIUM AND ITS COMPOUNDS 26 1 t282 ground state suggests that Jahn-Teller effects may be observed although the electrons are in the “non-bonding” tzs level distortion effects are likely to be small but may become apparent as a form of vibronic coupling to give a “dynamic” Jahn-Teller in which case this should perhaps be observed in the vibrational spectra of the compound. Studies on the infrared and Raman spectra of the gaseous hexafluoride do reveal the expected anomalies the v2 symmetric stretching mode is not observed in the Raman spectrum in conditions in which it should normally be easily apparent and all combination bands involving v2 are abnormally broad [somewhat similar effects are found for (osc16)2-].The vibrational spectra confirm the octahedral syrnmetry of the molecule.36 An oxyfluoride OsOF is formed in small amounts during the preparation of the hexa- fluoride and OsOC1 can be made by treating osmium metal with an oxygen-chlorine mixture (no hexahalides other than the fluoride are known for osmium). Both the complexes are diamagnetic owing no doubt to the low molecular symmetries. An apparently heptaco-ordinated complex c ~ ~ ( o ~ o c 1 ~ ) can be made from caesium chloride and osmium oxychloride. There is some doubt about the existence of osmium(v1) oxides the trioxide has not been obtained pure but the product of decomposition of the cyclic osmium(v1) esters mentioned above may be a hydrated trioxide.It is curious that the parent acid of the osmates H,[OsO,(OH),]aq has not been made. Osmium shares with rhenium ruthenium and the actinide elements the property of forming very stable trans dioxo-complexes. The “osmyl” complexes trans-(O~O~X,>~- ( X = C1- Br- +ox2-) can be made by the action of the appropriate acid on the tetroxide and other complexes of the same type ( X = CN- OH- OCH,- SO?-) can be made from potas- sium osmate and the appropriate alkali-metal The “oxy-osmyl” series originally formulated as (OSO~X~)~- (X = C1- Br- NO2- and ;ox2-) are made by similar methods,37 and have recently been shown to be of the forni trans- [Os02(OH)2X2]2-.1s~38 The trans arrangement of the oxy-groups in K2[0sO2(OH),] 39 and in K2(Os02C1,)40 has been confinned by X-ray studies.All the osmyl complexes are diamagnetic and it has been suggested that this arises from the axial compression of the octahedron (0 to D4h) brought about by the two short trans 0 s = 0 bonds the com- pression being sufficient to split the t g s lower triplet into a lower singlet [into which the two 5d electrons of osmium(v1) are paired] and an upper 35 H. A. Jahn and E. Teller Proc. Roy. Suc. 1937 A 161 220; J. H. van Vleck J. Chem. Phys. 1939 7 61 12. 36 B. Weinstock H. H. Claassen and J. G. Malm J. Chem. Phys. 1960 32 181; H. H. Claassen ibid. 1959 30 968. 37 L. Wintrebert Ann. Chim. phys. 1903 28 54 102 134. 38 W. P. Griffith J. 1962 3248. 39 M. A. Porai-Koshits L. 0. Atovyman and V. G. Adrianov J. Struct. Chem.40 F. Kruse Acta Cryst. 1961 14 1035. U.S.S.R. 1960 2 686. 262 QUARTERLY REVIEWS doublet an explanation which receives support from the electronic spectra of the cornplexe~.~~ Infrared spectra have been measured on a wide range of these complexes.18 It is worth noting that potassium osmate was long formulated as K2(Os04) by analogy with the tetrahedral potassium ruthenate K,(Ru04) but is of course octahedral K,[OsO,(OY),]. A series of nitrido-osmium(v1) complexes of the form (OsNX,),- (X = C1- BI-),,~ trans-[OsN(H,O)X,]- ( X = Br- CN- OX^-)^^^^^ and trans-[OsN(H,O)(OH),X,]- (X = F- can be made by reaction of the appropriate acid with potassium osmiamate; it is noteworthy that in this reaction it is the three osmium-oxygen bonds which are attacked while the osmium-nitrogen triple bond remains intact.X-Ray studies on K,(OsNCl,) show that the chlorine atom trans to the nitrogen is 0.3 A closer to the metal than the four equatorial groups,42 a situation very difficult to understand since it might be expected that the strongly r- donating nitrido-group should labilise the ligand opposite to it. Infrared studies have been carried out on these nitride complexes.28 Osmium(v).-This somewhat rare oxidation state for the element is represented only by the fluoride and its derived anionic complexes. Re- duction of osmium hexafluoride with tungsten carbonyl yields a mixture of the tetrafluoride and the blue pentafluoride; the latter melts at 70"c to a green liquid and boils at 225-9" to a colourless vapour. These colour changes on heating suggest that the compound in the solid and liquid forms has a polymeric form (probably like RuF which has recently been shown to be tetrameric with fluorine bridges43) and monomeric in the gas phase; this is supported by the unexpectedly low magnetic moment.44 The mixed compound OsIF has also been reported.44 A series of salts containing the (OsF,)- ion can be made from osmium tetrabromide bromine tetrafluoride and an alkali-metal bromide.45 The susceptibilities of these (Table 3) give values much closer to those expected for octahedral osmium(v) than does the pentafl~oride.,~ The fluorine magnetic resonance of the solid potassium salt has been measured,47 and X-ray studies made on a wide range of the Osmium(iv).-The tetrafluoride can be made in the same way as the pentafluoride; it is a yellow solid (m.p.230"c) probably polymeric.44 A tetra~hloride~~ and a tetrabromide are both known,50 being made 41 K.Lott and M. C. R. Symons J. 1960,973. 42 E. A. Atovyman and G. B. Bokii J. Struct. Chem. U.S.S.R. 1960 1 501. 43 J. H. Holloway R. D. Peacock and R. W. H. Small J. 1964 644. 44 G. B. Hargreaves and R. D. Peacock J. 1960 2618. 45 M. D. Hepworth P. L. Robinson and G. J. Westland J. 1954,4269. 46 B. N. Figgis J. Lewis and F. E. Mabbs J. 1961 3138. 47 D. Elwell Proc. Phys. Soc. 1964 48 409. 48 M. A. Hepworth K. H. Jack and K. J. Westland J . Inorg. Nuclear Chem. 1956 48 Vauquelin Ann. Chim. phys. 1841 (I) 89 248. 5 0 I. N. Semenov and N. I. Kolbin Russ. J . Znorg. Chem. 1962 7 111; S. A. 2,79; R. D. W. Kemmitt D. R. Russell and D. W. A. Sharp J. 1963,4408. Schukarev N. I. Kolbin and I.N. Semenov ibid. 1961 6 638. GRIFFITH OSMIUM AND ITS COMPOUNDS 263 directly from their elements but the long-claimed tetraiodide has been shown not to exist.51 The complexes (OSX,)~- (X = C1- Br- I- F-) are all known and are very stable. The fluorides are made by reduction of (OsF,)- and the others by prolonged action of the appropriate halogen acid on osmium tetroxide. An extensive study has been made of the magnetic properties of these complexes (Table 3) but interpretation of the results is not a simple matter. It has been demonstrated that there is magnetic exchange in salts of (OsC1,)2- and of (OsBr,)” since the moments of these ions when dispersed in a suitable diamagnetic host lattice are much higher than the bulk values and it is only at “infinite dilution” that the Kotani theory predictions for tZg4 configurations are obeyed.It has been sug- gested that this exchange operated through drs-prr (metal-ligand) and pi~n-prr (ligand-ligand) bonding Finally to add to the complexities of the problem recent observations on the solid-state fluorine magnetic resonance of K,(OsF,) have been interpreted as indicating that there is a considerable departure from simple occupancy of the t2s4 ground state by the four osmium(1v) 5d electrons although confirmation of these results must await single-crystal experiments.*’ The X-ray crystal structure of K,(OsCI,) has been determined and the octahedral symmetry confirmed.53 Infrared and Raman studies of (OSCI,)~- indicate that as with the hexafluoride dynamic Jahn-Teller effects may be operative.54 Mixed ~hloro-bromo-~~ ar,d chloro-iodo-osmium(~v)~~~ species have been detected in solution and in the former case separated by high-voltage paper electrophoresis.In this way all five complexes55 of the (OSC~,B~,-,)~- series were isolated and it was found that variations in absorption spectra maxima and migration velocities formed a consistent pattern.55 Jarrgensen has made an extensive survey of the electronic absorption spectra of platinum metal hexahalogeno-complexes including (OSX,)~- and (OSX,)~- (X = C1 Br I) over the range 180 to 1000 mp and he has assigned the ligand field bands and also the many electron- transfer transition^,^^ a subject also dealt with in a recent paper by Englman.57 The colours of some of these osmium(1v) hexahalides vary remarkably with the nature of the cation; thus while (0sC1J2- ion in aqueous solution is yellow the potassium salt is deep red the caesium salt orange and the silver and thallous salts are respectively deep brown 51 J.E. Fergusson B. H. Robinson and W. R. Roper J. 1962 2113. 62 A. D. Westland and N. C. Bhiwandker Canad. J. Chem. 1961 39 1284 2353; A. D. Westland ibid. 1963 41 2692; R. B. Johannesen and G. A. Candela Znorg. Chem. 1963 2 67 63 J. D. McCullough 2. Krist 1936 A 94 143. 64 L. A. Woodward and M. J. Ware Spectrochim. Acta 1964 20 71 1. 66 E. Blasius and W. Preetz 2. anorg. Chem. 1965 335 16. 61a E. Fenn R. S. Nyholm P. G. Buxton and A. Turco J. Znorg. Nuclear Chem. 1961,17,387; C. K. Jrargensen Acta Chem. Scand. 1963,17 1043; F. Rallo J. Chroma- tog. 1962 8 132. 56 C. K. Jrargensen and J. S. Brinen Mol. Phys. 1962 5 535; C.K. Jrargensen Acta Chem. Scand. 1962,16,793; 1963,17 1043; Mol. Phys. 1959,2 309. 57 R. Englman Mol. Phys. 1963 6 345. 264 QUARTERLY REVIEWS and orange-green. Studies of the reflectance spectra of these and other hexahalide salts suggests that these effects arise from shifting and broaden- ing in electron-transfer bands brought about by varying lattice para- meters caused by the change in cation.58 Shifts and broadening effects in these bands are also produced when high pressures are applied to solid solutions of osmium(1v) he~ahalides,~~ and the dielectric constant of the solvent also affects the position of the bands in liquid solutions.58 A recent study of the rate of exchange of labelled bromide ion with (OsBrJ2- is one of the very few kinetic investigations on osmium com- plexes; another is the study of aquation of (OSCI~)~-.~O Group VI donors.The dioxide is normally made by heating the metal in a stream of nitric oxide or in the vapour of the tetroxide. It is said to exist in a black and also in a brown form the latter (which is rather less reactive) having the rutile structure. Osmium disulphide diselenide and ditelluride can all be made from their elements at red heat. No aquo- complex is reported although there is some polarographic evidence for the existence of aquo- or hydroxy-complexes formed by reduction of buffered solutions of the tetroxide [by comparison with the chemistry of ruthenium one might expect the existence of such complexes of both osmium-(Irr) and -(Iv)]. Infrared evidence suggests that the complex originally formulated as (NH4)2(0sC150H)61 is in fact binuclear (N€Q4 (Os,OCl,,) with a linear O=Os=O skeleton similar to the analogous ruthenium compound.Nitrogen and carbon donor complexes are somewhat rare for this oxidation state of the element. The reaction between ethylenediamine (en) and osmium(rv) hexahalides is a complicated one yielding [Os It’(en -H) en,]Br, [Os1v(en-H)2en]Br2 [OsIv(en -H)2en2]12 together with some rather obscure products believed to contain quinque- and sexi-valent osmium.62 [en-H is an ethylenediamine molecule with one proton re- moved i.e. H2NCH2CH2.NH-; it will have a lone pair on the depro- tonated end of the molecule suitably placed for n-donation to the metal and might therefore be expected to stabilise osmium(1v); it is noteworthy that (0s en3)4+ is unknown].The diamagnetism of these quadrivalent osmium complexes may well arise from the asymmetry of the ligand field brought about by the non-equivalent ligands. It was suggested that [Os~v(en-H),en,]T contained eight co-ordinate osmium(Iv),62 but a more reasonable explanation of the unusual formula would be that two of the ligands probably (en -H) were unidentate leading to the conventional octahedral co-ordination for the metal atom. A few osmium(1v) complexes 68 C. K. Jnrrgensen J. Inorg. Nuclear Chem. 1962,24,1587; Actu Chern. Scand. 1963 6@ A. S. Balchan and H. G. Drickamer J. Chem. Phys. 1961 35 356. 6o G. Schmidt 2. Nuturforsch. 1961,16a 748; G . A. Rechnitz and H. A. Catherino 61 F. P. Dwyer and J. W. Hogarth J. Proc. Roy. SOC. New South Wales 1950,84,194. 62 F. P. Dwyer and J. W.Hogarth J. Amer. Chem. SOC. 1955,77 6152. 17 1034. Inorg. Chem. 1965,4 1 12. GRIFFITH OSMIUM AND ITS COMPOUNDS 265 of o-phenanthroline (phen) and 2,2'-bipyridyl (bipy) are known ,63 these being ligands which by virtue of their conjugated ring systems are capable of stabiiising a fairly wide range of oxidation and in fact osmium- (111) and -(II) complexes with phen and bipy are also known. A single arsine complex [OsBr4(AsPh3),] is reported with a magnetic moment of 1.5 B.M. at room temperature^.,^ Although no unsubstituted ammines of quadri- valent osmium are known the binuclear species [OS 2N(NHJ,X2]X3 (X = C1- Br-) can be made by treating the hexahalogeno-complexes with ammonia under pressure., Infrared spectra indicate that the bromide contains a linear N-0s-N skeleton analogous to (OS,OC~,,,)~- and (Ru,OCI~,,)~- mentioned above ; the n-donor abilities of the bridging nitrido-group no doubt help to stabilise the high oxidation state.28 Carbon donor complexes.The only one reported for this oxidation state is a cyclopentadiene complex probably [T-C~H~),OS(OH)~] made by the oxidation of osmocene in solution (chronopotentiometric studies show that ruthenocene but not ferrocene can also be oxidised to their respective quadrivaleiit oxidation states).,' Recently amine salts of the form (amine-H),[Osl"(CN),] and (amine H),[RUI"(CN),] (amine = ani- line 1,2,3-benzotriazole etc.) have been claimed made by addition of the amhe to K [Os(CN),] and K,[Ru(CN),].~' Since osmium(1v) and ruthen- ium(1v) cyano-complexes are not elsewhere reported in the literature and as it seems unlikely that the cyanide group would stabilise quadrivalent states alone the Reviewer wrote to the firm who supplied the cyano- complexes to the original workers and they confirm that their products were the normal K4[Os(CN),] and Kd[RU(CN),] but had been incorrectly labelled.It is clear therefore that the amiiie salts should be reformulated as (amineH),H [OS~~(CN),] and (arnineH),H,[R~~~(CN),] similar to the salts (amineH),H [FeI1(CN),] which have long been known69 for ferro- cyanides. Osmium(IrI).-While both ruthenium trifluoride and (RuF,J3- salts are well known the osmium analogues do not exist exemplifying once again the tendency to stabilise higher oxidation states in the third row of the Periodic Table; thus for osmium the quadrivalent is the commonest oxidation state the tervalent for ruthenium and the bivalent for iron.Osmium trichloride tribromide and tri-iodide are known the chloride made from the elements or by heating ammonium hexachloro-osmate(rv) 63 D. A. Buckingham F. P. Dwyer H. A. Goodwin and A. M. Sargeson Austral. J . Chem. 1964,17 315 325. O4 J. Chatt J. Inorg. Nuclear Chem. 1958 8 515. 65 L. Vaska Chem. and Znd. 1961 1402. F. P. Dwyer and J. W. Hogarth J . Proc. Roy. SOC. New South Wales 1950,84,117; V. I. Belova and Y. K. Syrkin Russ. J . Znorg. Chem. 1958 3 2016; G. Watt and L. Vaska ibid. 1958 6 246. 67 T. Kuwana D. E. Bublitz and G. Hoh J . Amer. Chem. Soc. 1960,82,5811. 68 R. F. Wilson and P. Marchant J . Znorg. Nuclear Chem. 1964 26 1057; R . F. Wilson and J. James 2. Anorg. Chem. 1963,321 180; 1962 315 235. 6 g W.M. Cumming J. 1922 1287. 266 QUARTERLY REVIEWS in a current of chlorine70 (data on the dissociation of both the tetra- and tri-chlorides of osmium have recently been presented71) and the iodide also from its In addition to the tribromide,*O Os,Br (made by the prolonged action of hydrobromic acid on the tetroxide)* and Os,Br (formed as a by-product during preparation of the tetrabromide50) have been claimed. Salts of (OSX,)~- (X = C1- Br- I-) are known but are unstable with respect to oxidation to the quinquevalent hexahalogeno- species from which they are usually prepared by electrolytic or chemical reduction processes. The electronic absorption spectra of all three ions in solution have been measured and the electron-transfer and ligand-field bands assigned.56 It is likely that complexes of the form [OsIIIX (HzO)6-n](3-n)f should exist since such systems are known for osmium(Iv) chloro-species60 and have been extensively investigated for both ruthe- nium(1n) and rhodium(m).Although the sesquioxide Os,O, is reported in the early literature its existence is highly questionable and neither is there convincing evidence for the existence of the sesqui-sulphide -selenide or -telluride. There are few Group VI donor complexes of osmiuni(II1); none with oxy- hydroxy- or aquo-ligands has been isolated although as with the quadrivalent state there is some polarographic evidence for their existence in reduced solutions of osmium tetroxide.,O Neither are there any sulphato- sulphito- ni trat o- nitri t o- nor (surprisingly) oxalat o-complexes. The species normally used for the colorimetric analysis of the element was long believed to be [OS1"(thi0Urea),]C1,(OH) [made from (OSC~,)~- and thiourea] but it is in fact [Os(thiourea),]CI and its instability constant has been measured.72 It can be reduced electrolytically to [Os(thio~rea),]~+.An analogous osmium(m) selenourea complex allegedly with a 1 :8 metal ligand ratio has been made.73 With acetylacetone (acac) osmium(rv) hexahalogeno-complexes give (0s acac,) together with a number of unidentified halogen-containing species of higher oxidation A number of mixed acac-phen and acac-bipy complexes have also been prepared., The most important donors for this oxidation state are of Group V. The hexammines [Os(NH3)6]& (X = C1- Br- I-) are difficult to prepare the method normally used being the high-pressure reaction between ammonia and the hexahalogeno-osmates(1v) at 280"~,~~ a method which also gives [Os"'(NH,),](Os'"Br,).The latter complex has recently been 70 0. Ruff and E. Bornemann 2. anorg. Chem. 1910 65 450 501. N. I. Kolbin and I. N. Semenov Russ. J. Inorg. Chem. 1964,9 108; N. I. Kolbin 72 R. D. Sauerbrann and E. B. Sandell J. Amer. Chem. SOC. 1953 75 3554; 73 A. T. Pilipenko and I. P. Sereda Russ. J. Znorg. Chem. 1961,6,209. 74 F. P. Dwyer and A. M. Sargeson J . Amer. Chem. SOC. 1955,77 1285. 76 F. P. Dwyer and J. W. Hogarth J . Proc. Roy. SOC. New South Wales 1951 84 117; 1952 85 113; G. W. Watt and L. Vaska J . Inorg. Nuclear Chem. 1958 6 246; 1958 5 308. I. N. Semenov and Y. M. Shutov ibid. 1964 9 563. Analyt. Chim. Acta 1953,9 86. GRIFFITH OSMIUM AND ITS COMPOUNDS 267 examined by infrared spectroscopy with normal deuterated and nitrogen- 15-substituted ammine groups and the metal-nitrogen fundamental stretching vibration found near 450 CM.-~.~ The room-temperature magnetic moments of the hexammines are close to 1.8 B.M.75 With a solution of potassium in liquid ammonia products which analyse as [Os(NH,),]Br and [os(NH&] are obtained; it is possible that these may be hydrido-complexes of osmium(~r).~~ Measurements of the rates of exchange of hydrogen with a number of metal ammines [M111(NH3)sJ3+ and ammines (MI11 (M = Co Rh Ir Cr Ru 0 s ) have been made and conclusions drawn as to the relative strengths of metal-nitrogen bonding between the The halogenopentammines are some- what more stable than the hexammines and are prepared by similar methods;76 it is likely that a wide range of substituted osmium(r1r) am- mines could be made.The reaction between (OSB~,)~- and ethylenediamine in the presence of a reducing agent yields ( 0 s [curiously no tris-ethylenediamine ruthenium(rn) complexes are reported].62 With the aromatic bases o- phenanthroline (phen) and 2,2'-bipyridyl (bipy) the complexes (0s hen,)^+ and ( 0 s bi~y,)~+ are formed and some salts of these have been while mixed phen and bipy complexes with other ligands have also been made.63 A considerable amount of work has been done on the electron- transfer reactions between these and the corresponding bivalent tris- phenanthroline and -bipyridyl complexes (in acid solution oxidation potentials are +046 and +0.73 v respectively) and in fact these systems have been used as redox indicat01-s.~~ The rates of electron exchange between the bi- and ter-valent tris-phenanthroline and -bipyridyl complexes are extremely high suggesting that little energy is needed to reorganise the co-ordination shells of reactants and products in the reaction and the observed rates are in excellent agreement with the predictions of the Marcus theory of electron exchange.s0 Such agreement is also found for the reac- tion of ( 0 s bi~y,)~+ with ferrocyanide ions.80,s1 Although pyridine is a stronger base (Le.a better a-donor) than either phenanthroline or bipy- ridyl it is a less effective n-acceptor owing to its lower degree of conjuga- tion and the only known pyridine-osmium complexes are (OsIVCI,py,) (OsIVBr,py,) and ( O ~ ~ ~ ~ C l p y ) . ~ ~ A number of phosphine and arsine complexes of osmium-(Iv) -(HI) and -(I[) are reported.Tertiary phosphines arsines and stibines react65,ez with hexahalogeno-osmates(1v) to give 76 W. P. Griffith unpublished work. 77 G. W. Watt E. M. Potrafke and D. S. Klett Znorg. Chem. 1962 2 868. 78 J. W. Palmer and F. Basolo J. Inorg. Nuclear Chem. 1960 15,279. 7 s F. P. Dwyer and E. C. Gyarfas J . Amer. Chem. SOC. 1952,74,4699; F . P. Dwyer E. C. Gyarfas and N. A. Gibson J. Proc. Roy. SOC. New South Wales 1950,54 80 83. E. Eichler and A. C. Wahl J. Amer. Chem. SOC. 1958 80 4145; A. Wahl 2. Elektrochem. 1960,64,90; R. Campion N. Purdie and N. Sutin J . Amer. Chem. SOC. 1963 85 3528. D. H. Irvine J. 1959 2977; 1957 1841. 82 F. P. Dwyer R. S. Nyholm and B. T. Tyson J . Proc. Roy. Soc. New South Wales 1947 81 272.268 QUARTERLY REVIEWS [OsX,(MR,),] (X = C1- Br-; M = P As Sb) while the versatile ligand o-phenylenebisdimethylarsine (diars) gives complexes of the form ( 0 s diar~,X,)+;~~ these can be reduced to (0s diar~,X,),~~ and the far infrared spectra of both series have been measured.83 Carbon donor complexes of osmium(m) are rare. There is polaro- graphic evidence for the existence of [os(cN)6]3- in and the oxidation potential of the [os(cN),]3-/ [Os(CN)6]4- couple is -0.99~ (for the corresponding ruthenium system the potential is + 0 ~ 8 6 v ) . ~ * Osmium(II).-The dichloride is said to exist in both a green and brown form but the latter is probably the only pure form of the compound and it is made by thermal decomposition of the trichloride in a stream of chlorine.70 It is insoluble in water but soluble in alkalis possibly to give hydroxy-complexes and surprisingly it is claimed to be paramagnetic.86 There seem to be no reports of a dibromide but the di-iodide can be made from the elements.51 The reduction products of acid solutions of (o~C1,)~- are vi~let-blue,~~ possibly owing to the presence of (OSC~,)~- or an aquated form of this.Oxygen donor complexes. The oxide OsO is said to be formed when K6[Os(H,0)(S0,),] is heated,87 but the evidence for the products being the monoxide is inconclusive. The monosulphide may exists8 but neither the selenide or telluride are known. Osmium(I1) complexes with Group VI donors are rare mixed acetylacetone complexes with 2,2'-bipyridyl and with o-phenanthroline have recently been reported63 and [Os(thiourea),l2f has been made in ~olution.'~ Stable sulphito-complexes are however known although the nature of the bonding (whether via oxygen sulphur or both) is not clear.Treatment of hexahalogeno-osmates(rv) with sul- p h i t e ~ ~ ~ gives compounds claimed to be [OS~~(SO~)~(€€SO,)~]~- and [OSIV(H,O)(SO~),]~- but the fact that they are colourless suggests that they may in fact be derivatives of bivalent osmium. Treatment of the species formed by the tetroxide with glycols and sodium sulphite yields Na [OS(SO~)~],~H,O,~~ which may contain bridging or bidentate sulphito- groups or could alternatively be formulated as [OS(SO~)~(H,O),]~- with unidentate ligands. Claus reported in 1863 a blue sulphite OsSO, as a volatile product of the complicated reaction between osmium tetroxide and sulphur dioxide.Group V donors are important for bivalent osmium. Although [Os (NH3),I2+ has not been isolated there is evidence for its intermediate formation in the reaction between potassium in liquid ammonia and the 83 J. Lewis R. S. Nyholm and G. A. Rodley J . 1965 1483. 84 W. P. Griffith Quart. Rev. 1962 16 188. m5 W. Hieber and H. Stallmann 2. Elektrochem. 1943 49 288; Ber. 1942 75 B 8e B. Cabrera and A. Duperier Compt. rend. 1927 185 415. 8 8 J. E. McDonald and J. W. Cobble J. Phys. Chem. 1962 66 791. 1472; W. Manchot and J. Konig Ber. 1925 58 229. C. Claus J. prakt. Chem. 1863 90 80. A. Rosenheim and E. A. Sasserath 2. anorg. Chem. 1889 21 143; A. Rosenheim ibid. 1900 24,422; A. Sachs Z. Krist. 1901 34 166. GRIFFITH OSMIUM AND ITS COMPOUNDS 269 tervalent hexamine~,~7 and it is possible that the products of this reaction may be hydrido-ammines of osmium(r~).Complexes with aromatic bases are very stable; these include ( 0 s ~hen,)~+ ( 0 s b i ~ y ) ~ + ~ ~ ( 0 s tripy,),+ (terpyridyl a tridentate ligand),gO and ( O ~ E r ~ p y ~ ) ~ ~ while mixed bipy and phen complexes have also been prepared.63 Electron-exchange reac- tions between the tris-bipy and -phen complexes with their tervalent analogues have already been and to this may be added the extensive work carried out on the kinetics of the fast electron exchange between (0s bi~y,)~+ with a large number of other spin-paired oxi- dants*“ and on the kinetics of reaction of the same complex with thallous and persulphate ions.*l A wide range of phosphine and arsine complexes with bivalent osmium have recently been prepared.These ligands are good 7r-acceptors and behave like phenanthroline and bipyridyl in stabilising low oxidation states. A review on such complexes has recently been publishedg1 so only a very brief summary is given here. With hexahalogeno-osmates(Iv) tertiary phosphines and arsines give [OsX,(MR,),] ande5~a2 [0s2C1 (PR3),]CI (the latter are dimeric with halogen bridges),92 while with diphosphines cis- and trans- [OsX,(diphosphine),] (X = C1- I-) are formed.92 The latter complexes can be reduced with lithium aluminium hydride to trans- [OsIIHX(diphosphine),] in which there is a direct metal- hydrogen bond.93 Infrared studies on these hydrides show that there is a relationship between the metal-hydrogen stretching frequency and the trans-effect of the group X; the frequency decreases along the seriesg3 I- > Br- > C1- > SCN- > NO2- > CN- > H- With o-phenylenebisdimethylarsine (diars) the complexes (0s diars,X,) (X = C1- B r ) are Carbonyl and alkyl phosphines are dis- cussed below.A number of osmium(r~) nitrosyl complexes are known although their range is not comparable with that found for ruthenium(11). The complex originally formulated by Wintrebert37 as K2 [Os(NO,),] [prepared by the prolonged action of potassium nitrite on K,(OsC&)] is diamagnetic and is in all probability a nitrosyl derivative K,[OsII(NO)(OH)(N0,)4].76 Action of halogen acids on this gives [OS(NO)X,]~- (X = Cl- B r I-).37 In all these compounds the nitric oxide is likely to be functioning as a three-electron donor as in most of its cornple~es.~~ Although no penta- cyanonitrosyl is reported it is probable that the “nitr~cyanide”~~ obtained by the reaction between nitric acid and K,[OS(CN),] is or contains O 0 G.Morgan and F. H. Burstall J. 1937 1649. Dl G. Booth Adv. in Inorg. Chem. and Radiochem. 1964 6 1. Be J. Chatt and R. G. Hayter J. 1961 896. O3 J. Chatt and R. G. Hayter J. 1961 2605. 94 J. Lewis Science Progress 1959 47 506; C. C. Addison and J. Lewis Quart. O6 C. A. Martius Annalen 1861 117 362. Rev. 1955 9 115. 270 QUARTERLY REVIEWS K,[Os(NO)(CN),] since this is precisely the method normally used to make the pentacyanonitrosyls of ruthenium(I1) and iron(1r). Carbon donor complexes are also of importance for bivalent osmium chiefly as might be expected with good n-acceptor ligands such as CO CN- and -C5H5-.The cyanide complex is prepared by fusion of osmium compounds with potassium cyanide and has long been known;Q5 its electronic and vibrational spectra have been reported.gs The free acid H4[Os(CN)6] can easily be made and it has recently been shown by infra- red and far infrared spectroscopy that the anhydrous solid acid contains unsymmetrical N - H - - * N hydrogen The “sandwich” com- pound osmocene [(T-C,H,),OS] is made from the tetrachloride and sodium cyclopentadienide 98 and an X-ray crystal-structure determination on the compound shows that the two rings are in the eclipsed position as in r~thenocene.~~ The only olefin complex so far reported is the cyclo-octa- 1,5-diene (C8H12) species [OsC~,(C8Hl,)(PEtPh,),] prepared from the olefin and [ O S C ~ ~ ( P E ~ P ~ ) ~ ] ; ~ ~ since this is a very stable substance it seems likely that an extensive range of osmium olefin complexes could be made.Alkyl and aryl phosphines in which the alkyl and aryl groups are attached to the osmium by a metal-carbon a-bond have been made from the appropriate lithium salts together with [OsCl,(diphosphine),][ di- phosphine = [C,H,(PPh,),]}; these are cis- and trans-[OsR,(diphos- phine),] (R = Me Et Ph). Reduction of either the cis or trans isomer yields the hydride [OsRH(diphosphine),] (R = Me Et) and similar products are obtained from the ruthenium analogues. Infrared nuclear magnetic resonance and dipole moment studies on these are reported.QQa The carbonyl complex [ o ~ ( c o ) ~ ] c ~ ~ has been very briefly reported ;lo4 its existence is somewhat surprising since this is a very high oxidation state for carbon monoxide ligands alone to stabilise.There are three types of carbonyl halides of osmium(rI) all made by the action of carbon mon- oxide under pressure on osmium dihalide~.*~ The monomeric [Os(CO),X,] and dimeric [Os(CO),X,] (X = Cl- Br- I-) species are all quite stable while the action of heat on [Os(CO),l,] yields the inert and highly poly- meric [Os(CO) ,I 2]n.85 The reaction of hexahalogeno-osmates(1v) with tertiary phosphines and arsines in the presence of alcohols or ethanolic potassium hydroxide gives the remarkable series of carbonyl complexes of the general form [Osr1HX(CO)(MR3),] (X = C1- Br-; M = P As);loo these were earlier wrongly formulated as square planar derivatives of g6 K. Masumo and S. Waku Nippon Kagalcu Zasshi 1962 83 116; I.Nakagawa and T. Shimanouchi Spectruchim. Acta 1962 18 89; J.-P. Mathieu and H. Poulet Compt. rend. 1959 248 2325. g7 D. F. Evans D. Jones and G. Wilkinson J. 1964 3164; W. Beck and H. S. Smedal 2. Naturforsch 1965 20b 109. g8 E. 0. Fischer and H. Grubert Ber. 1959,92,2301. gg F. Jellinek 2. Naturforsch. 1959 14b 737. J. Chatt and G. Hayter. J. 1963 6017. loo L. Vaska and J. W. diluzio J . Amer. Chem. Suc. 1961 83 1262; L. Vaska ibid. 1964 86 1943; J. Chatt and B. L. Shaw Chem. and Ind. 1960,931. GRIFFITH OSMIUM AND ITS COMPOUNDS 27 1 univalent osmium [OsX(MR,),]. The mechanism of their formation is not clear but a suggestion recently put forward to explain the correspond- ing reaction with ruthenium complexes (and presumably the osmium reaction takes the same path) is that first an ethoxide complex is formed followed by the transfer of a hydride ion (H-) from the a-carbon atom of the ethoxide group to the metal giving an acetaldehyde complex which finally breaks down to methane and the carbonyl halide;lol the overall reaction is then [Ru,Cl,(PEt,Ph),]Cl+ 2KOH + 2C2H50H -+ 2[RuHC1(C0)(PEt2Ph),] + 2CH4 + 2KC1 + 2H20 Methane has been detected amongst the reaction products,lol and tracer experiments with carbon-14 show that the carbonyl group comes from the alcohol.loO The X-ray crystal structure of [OsBrH(CO)(PPh,),] shows that the phosphine group trans to the supposed position of the hydrogen atom is 0.2 A further from the metal than the other two (equatorial) phosphines which may perhaps be taken as an illustration of the trans effect of the hydride group.lo2 Osmium(r).-This is the rarest oxidation state for the element.The iodide OsI can be made from the elements and has a room-temperature magnetic moment of 0.5 B.M.51 No other halides or oxides are known. The carbonyl halides [Os(CO),X] (X = Br- I-) are extremely stable and probably have halogen bridges and metal-metal interaction ; they are made by the action of carbon monoxide and the halides at high temperatures in a high-pressure apparatus.85 As mentioned earlier [Os(NH,),]+ may be a hydride of bivalent osmium while the "univalent" phosphines [OsX (PPh,),] are in fact the carbonyl hydrides [OSXH(CO)(PP~,),].~~~ Osmium(O).-The only established examples of this oxidation state are the carbonyls [Os(CO),] and [Os,(CO),,] although it is to be expected that the element should form zero-valent phosphine complexes of the type recently reported for ruthenium.lo3 The pentacarbonyl is a colourless monomeric liquid (m.p.- 1 5 " ~ ) which can be made by the action of carbon monoxide under pressure on the trihalides together with silver or copper powder to take up the halogen. Although its structure has not yet been determined its monomeric nature suggests that it may have the trigonal bipyramidal structure of iron p e n t a c a r b ~ n y l . ~ ~ ~ ~ ~ ~ During its preparation yellow crystals (m.p. 224"c) of another carbonyl originally formulated as the ennea-compound Os,(CO), are formed (the same product results from the high pressur lol J. Chatt B. L. Shaw and A. E. Field J. 1964 3466. loe P. L. Orioli and L. Vaska Proc. Chem. SOC. 1962 333. lo3 J. Chatt and J.M. Davidson J. 1965 843. H. Hieber and H. Fuchs 2. anorg. Chem. 1941 248 256; W. Hieber and T. Kriick Angew. Chem. 1961 73 580. 272 QUARTERLY REVIEWS reaction of osmium tetroxide with carbon m o n ~ x i d e ) . ~ ~ ~ ~ ~ A recent X-ray determination however shows that the correct formulation is as a dodeca- carbonyl O S ~ ( C O ) ~ ~ . The molecule has approximately D3h symmetry the three metal atoms forming an equilateral triangle with four terminal carbonyl groups each two approximately perpendicular to and two parallel to the triangular plane. The three tetracarbonyl osmium units are joined by metal-metal bonds (2.88 A) and it was suggested that the bonding could best be understood in terms of octahcdrally disposed metal orbitals which instead of pointing directly at each other form “bent” metal-metal bonds.lo5 The infrared spectrum of [Os,(CO),,] has been recorded and is consistent with the above structure.lo6 Hieber and Stallmann report that small quantities of a highly volatile substance probably the carbonyl hydride [Os(CO),H,] [strictly an osmium(I1) derivative] were formed during preparations of the car- bon yls.85 Analysis of Osmium.-Almost all analytical methods for osmium are The commonest and probably the best provided great accuracy is not demanded makes use of the complex (0s ruthenium interferes and should not be present to an extent greater than 10 % of the osmium concentration.lo7 For determination of small quantities of the element (down to 0.01~) the accelerating effect of the tetroxide on the reaction between arsenic(rI1) and cerium(rv) in acid solution has been used.lo8 In the presence of large amounts of ruthenium osmium can be estimated colorirnetrically as (Ph4A~)2(0sC1,),10g but in general it is preferable to separate the metals first by acid distillation.Toxicology of Osmium.-Osmium tetroxide is a violent poison but fortunately has a characteristic and powerful odour. There is only one fatality (in 1874) recorded in the literature from tetroxide poisoning death having been caused by capillary bronchitis and confluent pneu- monia. The vapour attacks the eyes and may cause temporary blindness lasting in some cases for a number of days conjunctivitis and corneal ulceration and it also attacks the nose throat and bronchial passages. The best treatment for chest symptoms is to administer penicillin or sulphapyridine drugs and a propamidine lotion for any conjunctivitis which may develop.l1° The traditional treatment with hydrogen sulphide gas dates from 1847 and is not recommended.A method for the quantita- tive estimation of tetroxide in the atmosphere has been developed and loS E. R. Corey and L. F. Dahl Inorg. Chem. 1962 1 521 lo6 E. R. Corey and L. F. Dahl J. Amer. Chem. SOC. 1961 83 2203; W. Beck and K. Lottes Chem. Ber. 1961 94 2578; D. K. Huggins N. Flitcorft and H. D. Kaesz Inorg. Chem. 1965 4 166. E. B. Sandell “Colorimetric Methods for Traces of Metals” Interscience New York 1959 702. lo8 R. D. Sauerbrann and E. B. Sandell Microchim. Acta 1953 22. log R. Neeb Z. analyt. Chim. 1957 54 23. A. I. G. McLaughlin R. Milton and K. M. A. Parry Brit. J. Ind. Medicine 1946 3 183.GRIFFITH OSMIUM AND ITS COMPOUNDS 273 there are reviews on the toxicology of the element.llo~lll Almost all osmium compounds are very easily oxidised to the tetroxide particularly in acid solution and consequently all work involving the element should be carried out in a well-ventilated laboratory. The author is indebted to Dr. D. F. Evans and Dr. L. Pratt for their help- ful and constructive criticism of the manuscript. ll1 F. R. Brunot J. Industr. Hygiene 1933 15 136; D. Hunter J. Pharm. Pharmacol. 1953 5 149; Brit. Med. Bull. 1950 7 11.

ISSN:0009-2681    

DOI:10.1039/QR9651900254

出版商:RSC    

年代:1965

数据来源: RSC

摘要:

THE CHEMISTRY OF THE PHENALENES By D. H. REID (DEPARTMENT OF CHEMISTRY THE UNIVERSITY ST. ANDREWS SCOTLAND) 1. Introduction MOST of the interesting features of the chemistry of the phenalenes have emerged during the past 15 years. Developments have been along two lines. First the phenalene hydrocarbons have only recently become ac- cessible and their availability has been followed by experimental and theoretical studies of the phenalene ions and radical and of novel carbo- cyclic and heterocyclic aromatic systems based on the phenalene unit. Secondly several highly oxygenated derivatives have been isolated as plant and fungal pigmeilts. This Review is of these developments and in- cludes all important earlier work. The nomenclature and numbering adopted are those used in Chemical Abstracts.Compounds in this series are named as derivatives of phenalene (1) or phenalenone (2). Earlier names for the hydrocarbon (1) were perinaphthene perinaphthindene and benzonaphthene and the ketone (2) has been variously referred to as “pyrene ketone” phenalone-9 perinaphthindone 1,8-naphthindenone 9-ketoperinaphthindene and perinaphthenone. 2. Formation and Synthesis Oxidation of the hydrocarbon pyrene with chromic acid gave phenale- none-6,7-dicarboxylic acid the first recorded phenalene derivative,l later decarboxylated to phenalenone.2 All useful syntheses give a phenale- none or 2,3-dihydrophenalenone as an intermediate or final product. 2,l. Phena1enes.-Synthesis via /3- 1 -naphthylpropionic acids. This is the only versatile synthesis of phenalenes 2,3-dihydrophenalenes and 2,3-dihydrophenalenones.1-Halogenomethylnaphthalenes prepared by ch10romethy1ation3~* or indirectly by successive fonnylation (Vilsmeier) or acylation reduction with lithium aluminium hydride,6 and treatment E. Hintz Dissertation (Strasbourg 1878). E. Bamberger and M. Philip Annalen 1887 240 147. G. Darzens and A. Levy Compt. rend. 1935,201 902. L. F. Fieser and M. D. Gates J. Amer. Chem. SOC. 1940 62 2335. V. Boekelheide and C. E. Larrabee J. Amer. Chem. SOC. 1950,72 1240. 274 REID THE CHEMISTRY OF THE PHENALENES 275 C H R ~ HO,C' 'CHR' RH - RCJ4X.R' R.CO.R' - R.CH(OH).R' t & (3) I Reagents 1 LiAIHp; 2 HCl-EtOH. with a phosphorus trihalide are converted into /3- I-naphthylpropionic acids (3) by the malonic ester ~ynthesis.~-ll Cyclisation best with hydrogen f l ~ o r i d e ~ ~ ~ gives 2,3-dihydrophenalen- 1 -ones (4).Reduction of the ketone (4) and dehydration of the carbinol(5) gives the phenalene (6).s-7J* The scope of this synthesis is subject to steric and electronic effects which affect the direction of cyclisation. Thus while the acid (7; R = H) gave a small amount of the benzindanone (9; R = H) along with the major product (8; R = H),4 the 7-alkyl substituted acids (7; R = Me Et Pr') gave exclusively the six-membered ring ketones.13 However the acid (7; R = But) gave a mixture of ketones (8 and 9; R = But) in which the latter pred01ninated.l~ Cyclisation of the series of acids (10; R = Me Et Prn But) also gave mixtures of five- and six-membered ring ketones in which the latter predominated except in the case where R = Again while the V. Boekelheide and C.E. Larrabee J. Amer. Chem. Soc. 1950,72 1245. D. H. Reid and R. G. Sutherland J. 1963 3295. J. von Braun G. Manz and E. Reinsch Annalen 1929 468 277. lo W. Klyne and R. Robinson J. 1938 1991. l1 L. F. Fieser and F. C. Novello J. Amer. Cltern. Soc. 1940 62 1855. l2 L. F. Fieser and L. W. Newton J. Amer. Chem. SOC. 1942 64 917. l3 A. J. M. Wenham and J. S. Whitehurst J. 1956 3857. l4 A. J. M. Wenham and J. S. Whitehurst J. 1957 4037. l6 M. F. Ansell J. 1954 575. * F. Mayer and A. Sieglitz Ber. 1922 55B 1835. 276 QUARTERLY REVIEWS cyclisation of 5- and 7-methoxy-/3-l -naphthylpropionic acid gave ex- clusively the dihydrophenalenones (1 I ) and (1 2) the 6-methoxy-acid gave only the benzindanone.16 Phenalenes from phenalenium salts. Reduction of 1,4,7-trirnethyl- phenalenium perchlorate (1 3) (section 4,3) with lithium aluminium hydride gave 3,6,9-trimethylphenalene ( 14).17 Phenalenes from phenalenones.Wolff-Kishner reduction of phenale- none afforded phenalene in modest yield; this was the first preparation of this hydrocarbon.l* Reduction of phenalenone with lithium aluminium hydride gave phenalene (14 %) 2,3-dihydrophenalenone (8; R = H) (65 %) and phenolic rnateriaL6 Since phenalenones are abnormally polarised the yield of phenalenes from phenalenones could doubtlessly be improved by the use of the lithium aluminium hydride-aluminium chloride system. 2,2. Phena1enones.-Dehydrogenation of 2,3-dihydrophenalenones. 2,3- Dihydrophenalenones are dehydrogenated preparatively to phenalenones by palladi~m-charcoal,~~ by bromination followed by dehydrobromina- tion,13 by quinones,20 and by triphenylmethyl perclilorate.21 2,3-Dihydro- phenalenones are slowly transformed into phenalenones when exposed to air and light.Metal halides of the type used in the Friedel-Crafts reaction bring about rapid dehydrogenation. It has been concluded* that early attempts to prepare 2,3-dihydrophenalenones by the action of aluminium or stannic chloride on /3- 1 -naphthylpropionic acid chlorides invariably gave some of the phenalenone. Cyclisation of 1-P-hydroxyacryloylnaphthalenes. The Claisen ester condensation of alkyl 1-naphthyl ketones gives compounds (1 5) which are cyclodehydrated by 80-90 % sulphuric acid to p h e n a l e n o n e ~ . ~ ~ ~ ~ ~ ~ 1 - or 2-Naphthol when heated with glycerol sulphuric acid and a mild oxidising agent give Condensation of naphthols with acraldehyde.l6 A. L. Green and D. H. Hey J. 1954 4306. l7 D. H. Reid and R. G. Sutherland J. 1963 3295; R. G. Sutherland Ph.D. Thesis la G. Lock and G. Gergely Ber. 1944 77B 461. l9 J. D. Loudon and R. K. Razdan J. 1954,4299. D. H. Reid and R. G. Sutherland unpublished results. W. Bonthrone and D. H. Reid J. 1959 2773. 2z A. Liittringhaus and F. Kaeer Ger.P. 489,571 and 490,358. 23 H. Silberman and S. Silberman Austral. J. Chern. 1956 19 115. University of St. Andrews 1962. REID THE CHEMISTRY OF THE PHENALENES 277 phenalenone (2).12,24 This procedure applied to 2,7-dihydroxynaphthalene gave 6-hydroxyphenalenone ( 16).25 The reaction of acraldehyde with 1- or 2- naphthol in the presence of hydrogen fluoride also gives phenalenone spontaneous dehydrogenation occurring.26 R' R2 0 0 5) OH (16) R' = H a l k y l a r y l R2- a l k y l a r y l CO,R Ring-expansion of acenaphthene derivatives.Aliphatic diazo-compounds bring about ring-expansion of acenaphthenequinone with C-insertion between carbonyl and the aromatic nucleus. 2-Hydroxyphenalenones (1 7) are formed.27 Successive ozonolysis and treatment with aqueous alkali of the sulphonate (1 8) gave 3-hydroxy-2-phenylphenalenone ( 19).28 Rearrangement of oxygen heterocycles. with dimethyl sulphate and alkali gave the The dihydrocoumarin (20) acid (21) whose chloride was cyclised with aluminium chloride to the dihydrophenalenone (22).29 Treatment of the dihydrocoumarin (23) with aluminium chloride at an elevated temperature gave 4-hydroxyphenalenone (24) dehydrogenation accompanying the rearrangement.lg Under similar conditions the chro- 24 M.A. Kunz and G. Kochendorfer Ger.P. 614,940; G. B. Silberman and S. M. Barkov J. Gen. Chem. U.S.S.R. 1937 12 1733; L. F. Fieser and E. B. Herschberg J . Amer. Chem. SOC. 1938 60 1658. 25 R. G. Cooke B. L. Johnstone and W. Segal Austral. J. Chem. 1958 11 230. 26 W. Calcott J. M. Tinker and V. Weinmayr J. Amer. Chem. Soc. 1939 61 949. 27 B. Eistert and A. Schonberg Chem. Ber. 1962 95 2416; B. Eistert and H. Selzer 28 E. Henmo P. de Mayo A. M. B. Abdus Sattar and A. Stoessl Proc. Chem. Soc. 2B C . F. Koelsch J. Amer. Chem. Soc. 1936 58 1326. Chem. Ber. 1963 96 314. 1961 238. 278 QUARTERLY REVIEWS manone (25) gave a nlixture of 9-hydroxyphenalenone (26) and its di- hydroderivative (27).lS The recent report30 that treatment of the tetra- hydrocoumarin (28) with methanolic hydrogen chloridemgives 9-methoxy- phenalene (29) requires further investigation since previous attempts*# 's31 to prepare methoxyphenalenes have given 2,3-dihydrophenalenones (see following section).Miscelkdneous. A large number of 3-hydroxyphenalenones have been prepared (a) by the condensation of naphthalene or its alkyl or alkoxy- derivatives with malonic acid or a substituted malonic acid and hydrogen fluoride,32 or with malonyl chloride and aluminium chloride,33 (b) by the reaction of naphthalic anhydride in the presence of zinc chloride with com- pounds containing a reactive methyl(ene) group for example with diethyl 30 J. Cologne G. Descotes and R. Puthet Bull. SOC. chim. France 1963 553. 31 G. M. Badger W. Carruthers and J.W. Cook J. 1949 1768. 32 H. Greune and G. Langbein Ger. P. 753,210. 33 K. Fleischer and E. Retze Ber. 1922 55B 3280. REID THE CHEMISTRY OF THE PHENALENES 279 rnal~nate,~~ phenylacetic and 2-meth~lpyridine.~~ 5-Hydroxytetra- lone reacts in the presence of sulphuric acid with glucose and other hexoses also with glyceraldehyde to give a compound formulated as (30) but more likely to possess the tautomeric structure (3 1) (see section 6,l .).37 3. Tautomerism of Phenalenes Klyne and Robinson suggested by analogy with indene that phenalenes might exhibit six-fold tautomerism each of the three rings assuming the aromatic and the unsaturated character in turn.38 In the case of phenalene the six tautomers are identical but they become chemically distinguish- able in a 1-substituted phenalene.Consequently oxidation of 1 -methyl- phenalene might yield a mixture of three dicarboxylic acids i.e. naph- thalic acid and its 2- and 4- methyl derivatives. Unfortunately this hypo- thesis could not be tested since the required 1-methylphenalene could not be synthesised. Fieser and Gates subsequently showed that the carbinol (32) obtained by the reaction of o-chlorophenylmagnesium bromide with 2,3-dihydrophenalenone (8; R = H) gives rise to a mixture of two acids (34) and (35) both of which are rearrangement products and interpreted (32) (33) (34) (35) Reagents 1 MeC0,H; 2 H,-Pt; 3 Cu,(CN),; 4 KOH. this as resulting from tautomerisation of the phenalene (33) which how- ever was not i~olated.~ In a related reaction sequence the carbinol (36) upon dehydration gave the ketone (39) by tautomerisation of the primary dehydration product (37) to the isomeric phenalene (38) which as an enol ether hydrolyses and rearranges to the product (39)31 (see also section 4,3.).Reagents 1 HCl-MeOH-C,H,; 2 Tautomerisation. 34 G. Errera Gazzetta 191 1 41 190. 36 M. Cesaris Gazzetta 1912 42 453. 36 A. Taurins J. prukt. Chem. 1939 153 177. 37 T. Momose and Y. Ohkura Chem. Pharm. Bull. (Tokyo) 1958 6 412. W. Klyne and R. Robinson J. 1938 1991. 280 QUARTERLY REVEWS In an elegant series of experiments Boekelheide and his co-workers attempted to prepare four different methylphenalenes (40)-(43) by the acid-catalysed dehydration of the alcohols (44)-(47) respectively in order to determine whether interconversion would occur.6 The same hydrocarbon shown to be (43) or (48) was obtained from all four alcohols.Reagents 1 HCI-EtOH; 2 PhLi MeI; 3 MeMgI. This hydrocarbon was also obtained by methylation of phenalene with phenyl-lithium followed by methyl iodide and from the reaction of phena- lenone with methylmagnesium iodide showing that isomerisation is not accounted for by acid or base catalysis alone. Further the hydrocarbon was not isomerised by treatment with phenyl-lithium followed by hydro- lysis. These results illustrate the tendency of the phenalene nucleus to behave as a unit making it difficult to determine which ring should be designated the peri (incompletely conjugated) ring. Different rings may be involved in different reactions.39 For example catalytic hydrogenation of 2-methyl- phenalene (49) gives the hydrocarbon (50) while oxidation gives the acid ( 5 1).Methylation of 2-methylphenalene (49) gives a dimethylphenalene as a tautomeric mixture. This on further methylation gives the hydrocarbon (52) but on catalytic hydrogenation affords the hydrocarbon (53).39 a9 V. Boekelheide and M. Goldman J. Org. Chern. 1954 19 575. REID THE CHEMISTRY OF THE PI-ENALENES 28 1 Me Me Me A (53) Reagents 1 H2-Pt; 2 KMnO,; 3 PhLi MeT. In a further study the preparation and degradation of a 14C-labelled phenalene was carried The alcohol (54) was synthesised from 1- chloromethylnaphthalene and diethyl [2-14C]malonate (section 2,l.) and dehydrated to the labelled phenalene (55). Permanganate oxidation of the latter in acetone gave naphthalic anhydride (57) having two-thirds of the specific activity of compounds (54) and (55).Oxidation was also carried out stepwise first with sodium dichromate in acetic acid to give labelled phenalenone (56) in order to avoid possible formation of the phenalene anion (section 4,2.) when positions 2,5 and 8 would become equivalent and the ketone (56) was further oxidised with permanganate to naphthalic anhydride. The specific activity of the anhydride in both degradations was the same. It was concluded that the even distribution of 14C between positions 2 5 and 8 had occurred at the dehydration step. Reagents 1 HCI-EtOH; 2 Na2Crz0,-CHSC02H ; 3 KMn0,-acetone. It is clear from the foregoing that although ready isomerisation of phenalenes occurs under the conditions employed in their preparation the nature of the isomerisation has not yet been fully defined.4. Ions and Radicals 4,l. Theoretical Considerations.-Phenalene is remarkable in that it gives rise to a relatively stable anion (58) cation (59) and radical (60). These result both in practice and in theory by the loss of a proton hydride ion or hydrogen atom respectively. 4 n M. Nakazaki U.S. Atomic Energy Commission Reports UCRL-3700 (1957). 282 QUARTERLY REVIEWS All three entities possess three-fold rotational symmetry about an axis which passes through the internal carbon atom perpendicular to the molecular plane also three identical two-fold axes in the molecular plane. It was first pointed out6 that this symmetry might make possible consider- able resonance stabilisation of these entities. Quantum chemical studies later provided a satisfactory explanation of their stability.Huckel molecular orbital (HMO) calculation^^^-^^ predict that the phenalene anion cation and radical should all possess the same v-electron delocal- isation energy namely 5.83/3. Fig. 1 shows the energy levels for the FIG. 1 HMO energy levels for phenalenyl ~ysterns~l-~~ Energy Cation Radical Anion 0 - + 1.6678 + 2.450/3 .f T fl 1 3 . - f I 1.1 I J . - f I I S phenalenyl system which in addition to possessing six bonding molecular orbitals has one of zero energy relative to the energy of an electron in a p z orbital of an isolated sp2 hybridised carbon atom. The phenalenium cation possesses 12 n-electrons which exactly fill pairwise the six bond- ing molecular orbitals. The extra one and two electrons of the phenalene radical and anion occupy the zero-energy orbital. Since the phenalenyl radical is an odd-alternant radical the electron density is unity at each position.The electron density distribution in the phenalenium cation and in the anion however is non-uniform. The charge densities (= l-electron density) in the phenalenium cation are + 0.167 at positions 1,3,4,6,7 and 9 and zero at the remaining positions. The charge distribution in the phenalene anion is the same but with negative sign. '' R. Pettit Chem. undlnd. 1956,1306;J. Amer. Chem. SOC. 1960,82 1972. *2 M. E. Dyatkina and E. M. Shustorovitch Dokludy Akud. Nuuk S.S.S.R. 1957,117 1021. 43 R. Zahradnik J. Michl and J. Koutecky Coll. Czech. Chem. Comm. 1964 29 1932. REID THE CHEMISTRY OF THE PHENALENES 283 4,2. Phenalene Anion.-Treatment of phenalene with phenyl-lithium6 or potassium m e t h ~ x i d e ~ ~ ~ ~ gives solutions containing the red anion (58) which reacts with methyl iodide to give 4(9)-methylphenalene (section 3) and condenses with benzaldehyde to give presumably compound (61).6 Comparative studies showed that phenalene is more acidic than tri- phenylmethane (pK 25) and less so than cyclopentadiene (pK 16).s Using HMO theory Streitwieser calculated the change in n-bond energy in going from phenalene to its anion and on the assumption that this quantity is a direct measure of the acidity calculated the pK of phena- 1ene.46 The value found (pK 22) agrees only moderately satisfactorily with experimental data.The true value must in fact be nearer to that of cyclopentadiene since the anion (58) is liberated by methoxide ion (pK of methanol 17). Aerial oxidation of the anion gives the phenalenyl radical (see section 4,4.).44 Reagents 1 NaOH SOCII NaN,; 2 HCl; 3 HONO; 4 AgC104; 5 0- Chloranil- 70% HC104.4.3. Phenalenium Cation.-The phenalenium cation was first prepared as its perchlorate (64) by the sequence (62)-(64).41 Compound (62) was obtained by the addition of ethyl diazoacetate to acenaphthylene. At- tempts to prepare the alcohol (63; OH in place of C1) were unsuccessful. Formation of the covalent chloride (63) appears to result from the collapse of a diazonium chloride ion pair with internal nucleophilic attack with- out the intervention of a free carbonium ion. A more general synthesis of phenalenium salts involves hydride abstraction from phenalenes by high potential quinones in the presence of perchloric acid7 or by triphenyl- methyl perchlorate.21 Several alkyl- and methoxy-phenalenium per- chlorates have been prepared by this method for example the salt (13) as well as benzo [alphenalenium perchlorate (65).44 D. H. Reid Chem. and Znd. 1956 1504. 45 D. H. Reid Tetrahedron 1958 3 339. 40 A. Streitwieser Tetrahedron Letters 1960 No. 6 23. 284 QUARTERLY REVIEWS Methoxyphenalenium perchlorates result directly from treatment of 1 -hydroxy-2,3-dihydrophenalenes with these reagents. For example dehydration of the alcohol (66) gives 2,3-dihydrophenalenone (68) by isomerisation and demethylation of the primary dehydration product (67) (section 3) but in the presence of a quinone or the triphenylmethyl cation rapid hydride abstraction diverts the intermediate (67) to the salt (69) before isomerisation can occur.The tribenzo-derivative (7 1) results from the oxygenation of the hydrocarbon (70) in acetic acid containing per- chloric Reagents and reactions 1 HCI-EtOH ; 2 Isomerisation-demethylation ; 3 HC1O4-CH,CO2H; 4 Abstraction of H- by o-chloranil. Phenalenium perchlorate is rapidly attacked by moist air. 7p21w41 Alkyl- and alkoxy-phenalenium perchlorates are more stable. All these salts are hydrolysed irreversibly in water. Phenalenium and alkylphenalenium perchlorates give an equimolecular mixture of the corresponding phenalene (or its tautomeric mixture) and phenalenone. The process doubtless proceeds by hydride abstraction possibly as shown or by a related route involving the carbinol(72) or the derived ether (73). Alkoxyphenalenium perchlorates give phenalenones as the only hydro- lysis product.1-Methoxyphenalenium perchlorate (69) gave phenalenone (2) while 1,5-dirnethoxyphenalenium perchlorate (74) afforded a methoxy- phenalenone believed to be (75).20 47 E. Clar and D. G. Stewart J. 1958 23. REID THE CHEMISTRY OF THE PHENALENES 285 + Using the correlation between the constant for the equilibrium ROH + H+ + R+ + H20 and the difference in n-bond energies of ROH and R+ calculated by HMO theory pKR+ for the pair phenalenium-I-hydroxyphenalene was estimated to be around &2.48 Unfortunately this could not be checked owing to the irreversible disproportionation of 1-hydroxyphenalene (72). Good agreement was obtained between the predicted and experimentally determined ultraviolet and visible transitions [A,,, 400 378 (shoulder) and 226 J when the effect of interaction between ground and excited states on the spectrum of the phenalenium cation was taken into Chemical shifts due to ring current effects have been calculated for the phenalenium cation by both the Huckel and the Self Consistent Field (SCF) procedure^.^^ The predicted shifts for H-1 and H-2 expressed as ratios to the corresponding effect in benzene are respectively 0.85 (HMO) 1-02 (SCF) and 0.77 (HMO) 0.92 (SCF).The larger shifts shown by the SCF method arise because the central carbon atom which has a negative charge of 0.024 in the SCF treatment and is neutral in the HMO version increases the ring current by making the periphery more nearly of the (4rz + 2) type. Unfortunately it has not yet been possible to obtain the proton magnetic resonance spectrum of phenalenium perchlorate owing to its great tendency to form radical-containing material and con- sequent line-broadening in the spectrum.The spectrum of the derivative (13) in trifluoroacetic acid shows a methyl singlet (9H) at 8 3.36 and an 48 D. Meuche H. Strauss and E. Heilbronner Helv. Chim. Ada 1958 41 57. N. S . Ham J. Chem. Phys. 1960,32 1445. G. G. Hall A. Hardisson and L. M. Jacban Discuss. Farachy SOC. 1962,34 15. 286 QUARTERLY REVIEWS AB system (6H) with components centred at 8 8-18 (H-2) and 9.30 (H-3) p.p.m. (J = 8.3 c./sec.) consistent with its symmetry.51 There is thus no evidence of abnormally large ring current effects. 4,4. Phenalenyl Radical.-Early attempts to prepare the phenalenyl rad- ical involved the bromination of phenalene with N-bromosuccinimide and dehydrobromination of the dibromide (76).6 It was hoped that 1-bromo- phenalene (77) or phenalenium bromide might result and that these might be transformed into the radical.Green-blue solutions were obtained. In a further attempt the alcohol (78) was dehydrated with the expectation that tautomerisation of the primary dehydration product (79) would give the bromo-compound (77).52 A yellow solid was obtained which rapidly changed into a high-melting substance doubtlessly peropyrene (80). The formation and preparation of several derivatives of the phenalenyl radical has been reported. Treatment of phenalenone with magnesium in ethereal acetyl or benzoyl chloride is reported to give blue solutions of the 1 -acetoxy- and 1-benzoyloxy-phenalenyl radicals (8 1 ; R = MeCO or Br Br B& “a (78) Br (79) Br (76) (77) PhC0).63 Reduction of 3-phenylphenalenone (82) with zinc in acetic acid gave a colourless compound formulated as (83) which on partial re- oxidation with air gave a red substance to which structure (84) was assigned.This substance was also formed upon mixing solutions of equimolecular amounts of compounds (82) and (83) and was claimed to disproportionate 61 W. Bonthrone and D. H. Reid unpublished results. s2 V. Boekelheide and M. Goldman J . Amer. Chem. Soc. 1954,76,604. 63 E. Clar “Aromatische Kohlenwasserstoffe” Springer Verlag Berlin 1952 2nd edn. p. 431. REID THE CHEMISTRY OF THE PHENALENES 287 into these compounds in solution.s4 The properties of compound (83) are not in accord with those of simple phenalenes and further studies of these reactions seem desirable.Preparation of the phenalenyl radical (60) was first carried out by shaking a solution of the anion (58) in an atmosphere of oxygen. Continued uptake of oxygen gives a green “peroxide” of unknown constitution which breaks down thermally to a mixture of phenalenone (2) and peropyrene (80).44145 The phenalenyl radical has not been isolated but exists in solution (Amax. 61 3 mp). Formation of the phenalenyl radical has also been observed in the oxidation of phenalene with quinones or osmium tetroxidee20 In the latter case the diol(85) was also produced. In a different approach the diol(87) obtained by bimolecular reduction of the ketone (86) gave solutions of the radical when treated with The sequence (87) -+ (88) + (89) -+ (60) Reagents 1 Af-EtOH-C,H,; 2 HCI-CH,CO,H; 3 Heat.was suggested to account for this transformation. Solutions of the phena- lenyl radical when boiled give peropyrene (80). Treatment of phenalenium perchlorate with zinc dust also gives peropyrene4’ as does a boiling solu- tion of the diol(87) in acetic acid containing a catalytic amount of mineral acid.45 The course of these reactions is rationalised by postulating the equilibrium (60) $ (89) which is disturbed by the irreversible transforma- tion of the dimer (89) into peropyrene. In the foregoing experiments no attempt was made to exclude oxygen. In the absence of oxidising agents the 54 C. F. Koelsch and J. A. Anthes J. Org. Chem,. 1941 6 558. 288 QUARTERLY REVIEWS phenalenyl radical is stable for indefinitely long period^.^^^^^ Phenalenyl has been reported to be present in the pyrolysis products of petroleum fractions of widely varying boiling range and “aromatics” ~ o n t e n t .~ ~ ~ ~ Although the electron density at each position of an odd-alternant hydrocarbon radical is unity,57 the density distribution of the electron in the non-bonding molecular orbital is non-uniform. Calculation within the framework of the HMO procedure predicts that the unpaired electron density in the phenalenyl radical should be zero on seven carbon atoms and have positive values at the remaining six. This is shown in structure (90) from which a seven-line pattern is predicted in the electron spin 0 Q.l67(\10*167 Oib7@0467 0 0 0467 0467 (90) resonance (e.s.r.) spectrum with a total spread of about 28 gauss. In fact however seven principal lines are present each is further split into a quartet and the total spacing is 49 gauss.58* The relative intensities of the principal lines were 1 6 15 20 1 5 6 1 those of the components of the quartets 1 3 3 1.The hyperfine splittings were 7.3 and 2.2 gauss later r e k i ~ e d ~ ~ to 6.30 and 1.82 gauss. The structure of the e.s.r. spectra of phenalenyl and other odd-alternant hydrocarbon radicals are accounted for by refined molecular orbital theory which introduces the concept of negative spin density.59 This takes into account the disturbing effect of the unpaired electron on the orbitals of the paired electrons. Partial unpairing occurs which induces at the adjacent carbon atom odd-electron density of opposite sign to that of the electron responsible. The spin densities at positions 2 5 and 8 of the phenalenyl radical therefore are negative those at positions 1 3,4,6,7 and 9 positive and correspondingly greater.60 The negative spin density accounts for the fine structure of the e.s.r.spectrum of phenalenyl. Since the total width of the e.s.r. spectrum of a conjugated radical or ion is a measure of the sum of the absolute values of the spin densities the large spread (49 gauss) in the case of phenalenyl is accounted for. It is interesting that simple vaIence bond theory gives directly positive and negative spin densities in good agree- ment with experiment.61 * In these studies a solution of the phenalenyl radical was by allowing a solution of phenalene in carbon tetrachloride sealed in air to stand for several months. 56 J. E. Bennett Proc. Chern.Sac. 1961 144. 56 K. W. Bartz and F. C. Stehling J . Chem. Phys. 1961 34 1076. 57 C. A. Coulson and C. A. Rushbrooke Proc. Cambridge Phil. SOC. 1940 36 193. 58 P. B. Sogo M. Nakazaki and M. Calvin J. Chem. Phys. 1958 28 107. 5 0 H. M. McConnell and D. B. Chesnut J. Chem. Phys. 1958 28 107. 6 o R. Lefebvre H. H. Dearman and H. M. McConnell J. Chem. Phys. 1960,32,176. 61 H. M. McConnell and H. H. Dearman J. Chenz. Phys. 1958 28 51. REID THE CHEMISTRY OF THE PHENALENES 289 5. Aromatic Systems Based on the Phenalene Nucleus Consideration of the fact that phenalene forms both a stable cation and anion (preceding aromatic systems section) led to-the suggestion that stable non-alternant (91) and (92) might result by fusion of these units to the cyclopentadienide and tropyliuni ions r e ~ p e c t i v e l y ~ ~ ~ ~ ~ that is by ortho- fusion of two stable oppositely-charged cyclic n-electron systems.Indeno- [2,l-a]phenalene (93) a derivative of the hydrocarbon (9 l) has been synthesised by the route shown and a study of its properties has substan- tiated the expectation of its aromatic properties.s2 The results of a theo- retical treatment 63 of the hydrocarbons (9 1)-(93) and their heterocyclic CH-OH \ HO 1 / / Reagents 1 1-C,,H,MgBr H,O; 2 HC0,H; 3 HC0,Et-KOMe; 4,93 % H2SOo. analogues are in excellent agreement with those of chemical studies. The hydrocarbon (93) is readily formed by dehydrogenation of a dihydro- derivative (94 or isomer) with palladium-charcoal at room temperature or triphenylmethyl per~hlorate,~~ in agreement with its considerable delocali- sation energy (0.402 P/n-electron).The relatively low level of the frontier bonding orbital (0.37413) explains its general chemical stability. The hydro- carbon (93) reacts under mild conditions with electrophiles for example with trifluoroacetic anhydride triphenylmethyl perchlorate and tetra- nitromethane to give monosubstitution products.20J21s62 These results are consistent with the reactivity indices (charge densities superdelocalisa- bilities and localisation energies) all of which predict position 12 to be the most reactive in electrophilic substitution reactions. Half-protonation of the hydrocarbon (93) occurs in 64% sulphuric acid (H = -4.80) placing it alongside the azulenes and heptalene as one of the most basic 62 I. M. Aitken and D. H. Reid J. 1956 3487. R. Zahradnik and J.Michl Coll. Czech. Chem. Comm. 1965,30,520. 290 QUARTERLY REVIEWS hydrocarbons. The nuclear magnetic resonance (n.m.r.) spectrum in trifluoroacetic-sulphuric acid (4 1 v/v) shows a methylene signal at 8 4-51 and aromatic signals in the range 6 7.4-9-5 p.p.m. (intensity ratio 2 11) supporting the a-protonated structure (95) for the cation.64 The high basicity of the hydrocarbon is consistent with the low electrophilic localisation energy (1.827/3) for position 12. Addition reactions occur easily. Reduction with zinc and acetic acid gives a dihydro-deri~ative.~~ The hydrocarbon (93) is a reactive diene affording with maleic anhydride the compound (96),62 and reacting with benzyne with concomitant dehydrogenation to give the hydrocarbon (97).65 Bicentric localisation energies are lowest for the pairs of positions 7 12 (3.4513) and I, 12 (3.52/3) both of which are low enough to explain the ease of Diels-Alder addition reactions.63 The dihydro-derivative is thus predicted to be (94) if formed under thermodynamically con trolled reaction conditions. Neither of the hydrocarbons (91) and (92) has yet been synthesised. Attempts to dehydrogenate the tetrahydro-derivative (98) were unsuccess- ful. However the conditions employed were severe.45 Energy characteris- tics and reactivity indices of (91) and (92) augur favourably for their synthesis and stability.62 Several derivatives (102) of a heterocyclic analogue (99) of the hydro- carbon (93) are known. These have been prepared by the Fischer indoliza- tion of the phenylhydrazones (100) of 2,3-dihydrophenalenone followed by dehydrogenation of the resulting phenalenes (101).The stability of the cation in the salts (102) is reflected in their ready formation from the phenalenes (101) by disproportionation in boiling acetic acid containing perchloric acid. The heterocycle (99) is a strong base. It is reactive and attempts to isolate it have given insoluble high-melting products of un- known structure. 6. Phenalenones Phenalenones constitute the largest group of phenalene derivatives. They are stable and readily formed by mild oxidation of the phenalene hydrocarbons. Samples of phenalenes become yellow in air within a few hours owing to aerial oxidation to phenalenone~.~~~J~ Most methods of formation of the phenalene nucleus give phenalenones directly. All the 64 D. H. Reid unpublished results.65 I. M. Aitken and D. H. Reid J. 1960 663. REID THE CHEMISTRY OF THE PHENALENES 29 1 w 0-co (96) (99) LJ (97) (101) or isomer R-H,Me,Ph X = I ,ClO Reagents 1 HC1-CH3COOH; 2 I,-MeOH; 3 HCIOo in CH3-C02H boil. known naturally occurring compounds in this class are polyhydroxy- phenalenones. 6,l. Physical Properties.-Phenalenone is an abnormally highly polarised ketone resembling in many of its properties tropone cyclo- propenones and some heterocyclic ketones. The polarisation is attested to by the high dipole moment ( 3 . 8 9 ~ ) ~ ~ and the low infrared carbonyl frequency (1 637 cm.-l). 67 Infrared data demonstrating the effect of sub- stitution on the carbonyl frequency are available for a large number of phenalenones. 68 Phenalenone is abnormally highly basic (p& = 0-4).69 It dissolves reversibly in ccncentrated hydrochloric acid and forms stable salts with strong acids (see section 6,4.). This property is useful in the purification of phenalenones.20$21 A polarographic investigation showed that the reduction of phenalenone proceeds in two steps each involving the uptake of one The first and more positive step is reversible; it gives rise to the 1-hydroxyphenalenyl radical (103). The second stage is irreversible and its exact nature is not entirely clear. Values of E* for the 66 V. A. Kogan 0. A. Osipov 0. E. Shelepin and Y . A. Zhdanov Doklady Akad. Nauk S.S.S.R. 1959 128 719. 13' N. H. Cromwell and G. V. Hudson J . Amer. Chem. SOC. 1953 75 872. H. L. Josien N. Fuson J. M. Jacobs and T. M. Gregory J. Chem. Ph-vs. 1953,21 331 ; N.Fuson and M. L. Josien Bull. SOC. chim. France 1952 389; R. D. Campbell and N. H. Cromwell J. Amer. Chem. SOC. 1957,79,3456; W. I. Awad and 0. M. Aly J . Org. Chern. 1960,25 1872; B. Eistert and A. Schonberg Chem. Ber. 1962,95,2416; B. Eistert and H. Selzer Chem. Ber. 1963 96 314. ' O P. Beckmann Austral. J. Cliem. 1961 14 229. T. Handa Bull. Chem. SOC. Japan 1955 28 483. 292 QUARTERLY REVIEWS reversible step in the polarography of a series of phenafenones and the pKa values of their conjugate acids show a parallel dependence on the nature of the substituents. 71 Irradiation of phenalenone in certain polar solvents notably isopropyl alcohol produces the radical (103) detected by its e.s.r. spectrum and ultimately the ketone (68).72 6,2. Tautomerism of Hydroxyphena1enones.-Tautomerism is possible among the hydroxyphenalenones involving prototropic shifts from hydroxy to carbonyl oxygen.Three cases are distinguishable. (a) 3-,34s 73 6-25 and 9-Hydroxyphena1enonel9 can tautomerise but the tautomers are identical. The first two give rise accordingly to only one methyl ether.25s34s73 9-Hydroxyphenalenone (104) has the properties of a strongly intramole- cular hydrogen bonded h y d r o ~ y - k e t o n e . ~ ~ ~ ~ ~ ~ ~ ~ It shows striking resistance to hydroxyl and carbonyl reagents. Hydrogenation proceeds cleanly to give the dihydro-derivative ( l O y With the boron trifluoride-ether complex it forms the stable complex ( 106).75 Unsymmetrical substitution of 3- 6- and 9-hydroxyphenalenone gives rise to pairs of non-identical tauto- 72 H. Koller G. P. Rabold K.Weiss and T. K. Mukherjee Proc. Chem. Soc. 1964 73 M. Goldman J. Amer. Chem. SOC. 1954 76 4032. 74 R. G. Cooke and W. Segal Austral. J. Chem. 1955 8 413. 75 1. C. Paul and G. A. Sim Proc. Chem. SOC. 1962,352. P. Beckmann Chem. and Ind. 1955 1635. 332. REID THE CHEMISTRY OF THE PHENALENES 293 mers. Accordingly 4-phenyl-6-hydroxyphenalenone (107) affords two methyl ethers.74 (b) 4- (108) and 7-Hydroxyphenalenone (109) are tauto- mers of one another.ls9 74 Methylation gives a mixture of 4- and 7-methoxy- phenalen~ne.~~ The first of these ethers has been synthesised unambigu- ously by alternative r o ~ t e s . ~ ~ ~ ~ ( c ) 2- 5- and 8-Hydroxyphenalenone cannot tautomerise. Only the first (1 10) is known. It is formed by the base- catalysed condensation of the ketone (68) with p-nitrosodimethylaniline followed by acid hydrolysis 76 and by rearrangement of 2,3-epoxyphenale- none (1 1 1) with acid.12 6,3.Reactions of Phena1enones.-The abnormal polarisation of phenalenones modifies the ketonic properties. Phenalenone forms a hydrazone in boiling glycol,l* but not a 2,4-dinitrophenylhydrazone. Hydrogen peroxide with sodium carbonate in ethanol gives the epoxide (1 1 1).12 The 2,3-double bond of phenalenone is unreactive as a dienophilic centre unless further activated for example as in the 3-carboxy-deriva- tive.12 Phenalenones undergo 1,4-addition with organometallic re- a g e n t ~ . ~ ~ 77 Position 9 is involved. Aromatisation frequently follows addition. Thus phenalenone with phenylmagnesium bromide gives the intermediate (1 12) which when distilled dehydrogenates to 9-phenyl- phenalen~ne.~~ 9-Substituted phenalenones also result from the reactions of 3-hydroxy- and 3-ethoxy-phenalenone with organometallic reagents.77-70 1 I R' R'= C H, R'= H I . I (116) N. P. Buu-Hoi and P. Cagniant Revue Sci. 1942 80 271. C. F. Koelsch and R. H. Rosenwald J . Org. Chem. 1938 3 462. 77 C. F. Koelsch and R. H. Rosenwald J. Amer. Chem. SOC. 1938 59 2166. 7 8 E. Calderera Gazzetta 1913 43 632. 294 QUARTERLY REVIEWS Phenalenone condenses with two molecules of 2,4-dimethylpyrrole in ethanolic hydrogen bromide to give a blue salt formulated as (1 13).80 The bromine atom in 2-bromophenalenone is inert to silver acetate,12 but is replaceable by primary or secondary amino-groups in some cases with rearrangement.s1 Treatment with piperidine at 85 O gives 2-piperidino- phenalenone (115) as the major product but at 25" the 3-piperidino- derivative (1 14) is obtained.Parallel behaviour is shown with morpholine. With cyclohexylamine however the azirine (1 16) was isolated ; this with acid gave 2-cyclohexylaminophenalenone. The annexed mechanism has been suggested. Detailed stereochemical information is lacking. 6,4. Hydroxyphenalenium Salts and Related Compounds.-The long known2 basicity of phenalenone depends on the stability of the conjugate acid (1 17) a derivative of the phenalenium cation which is formed when phenalenone dissolves in strong acids. The ultraviolet spectrum of phenal- enone in 60 % sulphuric acid is similar to that of phenalenium per~hlorate.~' Phenalenone with triethyloxonium fluoroborate gives l-ethoxyphenal- enium fluoroborate (1 18)64982 whose ultraviolet and nuclear magnetic resonance spectra correlate well with those of 1 -methoxyphenalenium p e r ~ h l o r a t e .~ ~ ~ ~ The salt (118) reacts with secondary amines to form 1-dialkylaminophenalenium salts (1 19),64 and with the sodium salt of malononitrile or cyanoacetic ester to yield compounds of the type While solutions of phenalenone in acids contain the cation (1 17) the molecular ratio of ketone to acid in the solid salts varies. Hydrogen bromide and sulphuric acid give 1 -hydroxyphenalenium salts (l2l),* but perchloric and tetrachloroferric acid give 2 1 possibly containing the hydrogen-bridged cation (1 22). Phenalenone reacts with inorganic halides (AICI, SbCI, SbCI, HgC12 SnCl, PCI,) to form 1 1 co-ordination compounds ( 123).8435 The phenal- enone moiety in these substances is substantially polarised as shown by the high dipole moment of the complex with SbC1 (8.49~).The acceptance of an electron pair of the oxygen atom by the metal atom disperses the negative charge and promotes the ground-state polarisation of the phenal- enone. Similar compounds are formed with dinitrogen t e t r o ~ i d e ~ ~ ~ ~ and sulphur trioxide.88 Phenalenone is recovered from all these compounds by (120).82 8o E. Herrmann A. Treibs and E. Meissner Annalen 1958 612 229. D. B. Capps N. H. Cromwell and S. E. Palmer J. Amer. Chem. SOC. 1951 73 1226; N. H. Cromwell J. Amer. Chem. SOC. 1959,81,4706. 82 H. Prinzbach and V. Freudenberger Angew. Chem. Internat. Edn. 1965 4 243. 83 A. M. Lukin Bull. Acad. Sci. U.R.S.S. 1941 29 411.84 G. B. Silberman and S. M. Barkov Zhur. obshchei Khim. 1937,7,1733. 85 0. E. Shelepin and Y . A. Zhdanov Zzvest. Vysshikh Ucheb. 1960,3 1036. 86 A. M. Lukin and L. D. Dachevskaya Compt. rend. Acad. Sci. 1947,554 825. A. M. Lukin and L. D. Dachevskaya Zhur. obschei Khim. 1948 18 1703. 8 8 A. M. Lukin and G. B. Zavarikhina Compt. rend. Acad. Sci. 1947 55 617. REID THE CHEMISTRY OF THE PHENALENES 295 hydrolysis. Phenalenone hydrazone forms salts (1 24) with acids and co- ordination compounds with metal halides.89 With bromine in benzene or acetic acid at room temperature phenal- enone forms an orange-yellow complex C13H100,Br2,90 from which it is recovered by treatment with thiosulphate or upon attempted recrystallis- a t i ~ n . ~ ~ ~ ~ ~ In boiling benzene however a colourless compound is obtained formulated as (76) which when treated with ethanolic ammonia or simply on being boiled in acetic acid gives 2-brornophenalen0ne.~~~~~ Similar behaviour is shown when phenalenone is treated with chlorine.93 The nature of the intermediates in these reactions is by no means certain.R2 (1 20) I" R'= R2=CN) b R'=CO,Et RLCN) (121) X = Br HSO (1 22) (1 28) (I 2 9) (1 30) 6,5. Miscellaneous.-The reaction of 3-hydroxyphenalenone with phenylhydrazine does not produce a phenylhydrazone ; instead the amino- ketone (125) and aniline are formed.94 Acid hydrolysis of the amino-ketone (1 25) gives 2,3-di-hydroxyphenalenone (1 26) whose N-acetyl derivative is reported to cyclise to the heterocycle (127).94 Oxidation of the diol (126) 0. E. Shelepin and Y. A. Zhdanov Zzvest.Vysshikh Ucheb. Zavedenii 1959 2 200. K. Brass and E. Clar Ber. 1939 72 1$82. A. M. Lukin Compt. rend. Acad. Sci. U.R.S.S. 1940 28 60. @* A. M. Lukin Bull. Acad. Sci. U.R.S.S. 1941 29 695. u3 A. M. Lukin Bull. Acad. Sci. U.R.S.S. 1941 29 565. 94 G. Errera Gazzetta 1914 44 18; G. Errera ibid. 1913 43 583. 296 QUARTERLY REVIEWS with hypobromous acid gives the yellow trione hydrate (128; R = H) which when heated loses water reversibly to give the red trione (129).94 The hydrate or hemiketal (128; R = H or Et) heated with ethanolic sodium carbonate ring-contracts to a mixture of acenaphthenequinone and the lactone (130).95 The trione (129) when heated with selenium in air also gives acenaphthenequinone. 96 The trione (129) behaves normally with o-phenylenediamine forming a quinoxaline but is reduced to the diol(l26) with hydroxylamine or phenyl- hydrazine.It may be used as an alternative to iiinhydrin in the detection and quantitative estimation of a-amino acids. 97 Carbon dioxide ammonia and an aldehyde or ketone are produced. Primary amines are also degraded to aldehydes.98 The trione (129) has also been employed in the quantita- tive determination of ascorbic acid and other reductones in biological samples. g9 7. Naturally Occurring Pkenalenones The presence of the phenalenone nucleus in a naturally occurring com- pound the plant pigment hzmocorin was first recognised in 1955. Shortly afterwards the constitution of several fungal pigments was found to be based on the phenalenone structure. None of these compounds has yet been syn thesised.7,l. Plant Pigments.-H~mo~orin.~~’ 74~100 This the only plant pigment known to contain the phenalenone nucleus is the red colouring matter of the roots of HRmodorum corymbosum Vahl. one of about seventeen species of the genus Hmnodorum which is found in Australia. The plant is reputed to be toxic to livestock. The roots when roasted have been used as food by the aborigines and the plants medicinally. Hzmocorin is a glycoside which crystallises as the hydrate C3&04 H,O. Acid hydrolysis gives a purple-red aglycone C20H1404 and cello- biose.loO The aglycone (1 3 1) contains one methoxyl group forms a diace- tate and is a weak acid giving a blue colour with alkali. Methylation gave a mixture of monomethyl ethers A (132) and B (133) C21H1604. Each ether on further methylation gave a dimethyl ether A’ (134) and B’ (135) C22H18O4 respectively.The aglycone did not form carbonyl derivatives. Monomethyl ether B showed infrared absorption in Nujol at 1620 cm.-l indicating the presence of a carbonyl group and broad absorption at 96 G. Errera and G. Ajon Guzzetta 1914 44 92. 96 A. Mostafa R. Moubasher and A. Schonberg J. 1946 966. O7 W. I. Awad and R. Moubasher J. Biol. Chem. 1949 179 915; W. I. Awad and R. Moubasher J. 1949 1137; A. Mostafa R. Moubasher and A. Schonberg J. 1948 476. 99 R. Moubasher J. Biol. Chem. 1948 176 529; Z. F. Hassan R. Moubasher and R. Moubasher and A. M. Othman J. Amer. Chem. SOC. 1950 72 2666. M. S . El-Ridi Biochem. J. 1951 49 246. loo R. G. Cooke and W. Segal Austral. J. Chern. 1955 8 107. REID THE CHEMISTRY OF THE PHENALENES 297 3200 cm.-l due to intermolecular hydrogen-bonded hydroxyl.In chloro- form however the hydroxyl absorption at 3519 cm.-l was sharp. Mono- methyl ether A showed evidence of internal hydrogen bonding since the hydroxyl absorption in chloroform (3387 cm.-l) was broad and little different from that in Nujol. These and other physical properties suggested that the aglycone was an extended tautomeric system containing a 1,2- dicarbonyl structure. Reagents 1 Me,S04-NaHCO ; 2 H,SO,-EtOH-Me,CO ; 3 MelS04-K,C0,-Me,C0 ; Oxidation of the dimethyl ether A' gave a neutral compound C,,H,,O containing two methoxyl groups indicated to be a derivative (136) of naphthalic anhydride by its characteristic infrared absorption at 1764 and 1727 cm.-l and by its chemical behaviour. Thus it dissolved in alkali and acidification gave a colourless acid which lost water giving back the anhydride upon attempted recrystallisation but which formed a dimethyl ester with diazomethane.Parallel behaviour was shown by the dimethyl ether B' which gave an isomeric anhydride (137). Both anhydrides and the aglycone showed infrared absorption characteristic of the unsubstituted phenyl group (700 and 750 cm.-l). The foregoing data taken with the composition indicated that these compounds are phenyldimethoxy- naphtlialic anhydrides. Comparison of the ultraviolet spectra of the an- hydrides and the derived dimethyl ethers with those of model compounds 4. KMn0,-Me,CO; 5 KMn0,-NaOH ; 6 HgO-NaOH. 4 298 QUARTERLY REVIEWS supported those conclusions as did further oxidation of the anhydrides (136) and (137) both of which gave the acid (138).The accumulated evidence at this stage suggested that the aglycone is a dihydroxymethoxyphenylphenalenone. Ultraviolet spectral studies of model hydroxyphenalenones showed that different arrangements of hy- droxyl (methoxyl) substituents in the phenalenone nucleus can be clearly distinguished and it was concluded that the aglycone is a derivative of 2,5,6-trihydroxyphenalenone. It remained to determine the position of the phenyl and methoxy-substituents. This was obtained as follows. The anhydrides (1 36) and (1 37) gave upon decarboxylation the dimethoxy- phenylnaphthalenes (1 39) and (140) respectively whose structures fol- lowed from independent syntheses. Thus the phenyl group must occupy position 9 of structure (131). Monomethyl ether A and the dimethyl ether A' were oxidised to the same anhydride (1 36) which must therefore contain the methoxyl group of the original pigment.It follows that the methoxyl group in the aglycone occupies position 5. The structure of the aglycone is thus (1 3 1). The tautomeric structure (1 3 la) is also possible but structure (131) is favoured because it would be stabilised by internal hydrogen bonding (see section 6,2). The position and stereochemistry of the glycoside link in hamocorin have not been established. It seems however that the sugar is not attached to the hydroxyl group at position 6 (structure 131) since the ultraviolet spectrum of the glycoside resembles that of the aglycone monomethyl ether (1 33) rather than (132). 7,2. Fungal Pigments.-Several phenalenone pigments have been iso- lated from two series of the section Biverticillata Symmetrica of the genus Penicilliurn.Two of these compounds h e r q ~ e i n o n e l ~ l - ~ ~ ~ and norher- queinone,lo2 account in large measure for the colour of the mycelium of stationary cultures of Penicilliurn herquei Bainier and Sartory while a third a t r ~ v e n e t i n ~ ~ * s ~ ~ ~ is produced by Penicillium atroveneturn G. Smith. These pigments are chemically related,106~107 but only the structure of atrovenetin has been fully e l ~ c i d a t e d . ~ ~ ~ ~ ~ ~ ~ Information derived from the chemistry of herqueinone and norher- queinone was used in determining the structure of atrovenetin the most important being summarised in Fig. 2. Atrovenetin is stable to acid but lol F. H. Stodola K. B. Raper and D. I.Fennel Nature 1951 167 773. lo* J. A. Galarraga K. G. Neill and H. Raistrick Biochem. J. 1955 61 456. lo* R. E. Harman J. Cason F. H. Stodola and A. L. Adkins J. Org. Chem. 1955 lo4 K. G. Neill and H. Raistrick Chern. and Ind. 1956 551. lo5 K. G. Neill and H. Raistrick Biochern. J. 1957 65 166. lo6 D. H. R. Barton P. de Mayo G. A. Morrison W. H. Schaeppi and H. Raistrick lo' D. H. R. Barton P. de Mayo G. A. Morrison and H. Raistrick Tetrahedron lo* G. A. Morrison I. C. Paul and G. A. Sim Proc. Chem. Soc. 1962 352; I. C. 20 1260. Chem. and Ind. 1956,552. 1959 6 48. Paul and G. A. Sim J. 1965,1097. REID THE CHEMISTRY OF THE PHENALENES 299 FIG. 2. Chemical relationship between atrovenetin herqueinone and norherqueinone Deoxynorherqueinone = Atrovenet in (141) CI,H,,O l- Norxanthoherquein + MeCOCHMe Norherqueinone (1 52 ?) C19H1807 (143) C,,H,oO CIbH12a / Xanthoherquein + MeCOCHMe Herqueinone considered to be a Deoxyherqueinone monomethyl ether of norherqueinone Reagents 1 Zn-CH,C02H ; 2 H,O+.herqueinone and norherqueinone break up into two fragments which together account for the total carbon content of the pigments. Atrovenetin. Atrovenetin (141) C19H1806 is optically active forms two tetramethyl ethers thus showing the presence of potentially four hydroxyl groups and contains an inert hydrogen-bonded conjugated carbonyl groups ( v c = ~ 1620 cm.-l). It also contains (Kuhn-Roth) three C-methyl groups. The tetramethyl ethers are insoluble in alkali indicating the re- maining oxygen function to be etherea1.1°4J05 (153b?) C,OH,OO~ (151a or b) C,,H,oO @; 0- 0 44) OH H o m o Me (148) Me Me Acid hydrolysis of herqueinone and norherqueinone both of which are optically active gave xanthoherquein (C1,H1,07) and norxantho- herquein (CI4Hl0O7) both of which are optically inactive together with isopropyl methyl ketone.lo2J03 This information suggestedlo7 that nor- xanthoherquein represents the atrovenetin nucleus and that the five-carbon hydrolysis product contains both the ethereal oxygen atom and the asymmetric centre responsible for the optical activity of atrovenetin.These conclusions were supported by comparative ultraviolet spectral studies which also suggested that these pigments are derivatives of 9-hydroxy- phenalenone. 300 QUARTERLY REVIEWS In order to establish the presence of the phenalenone nucleus nor- xanthoherquein xanthoherquein and atrovenetin were degraded with concentrated nitric acid.All three gave nitrococussic acid (142). If nor- xanthoherquein is a phenalenone then the acid (142) must represent the ring which carries the methyl group and bears only one hydroxyl group. Since norxanthoherquein has seven oxygen atoms attached to the nucleus every pheripheral carbon atom must be linked to oxygen except the angular ones that one carrying the methyl group and the C-H carbon which is nitrated during the oxidation. Norxanthoherquein must therefore be (143) and this taken with the already established inter-relationship of the pigments indicates that atrovenetin also contains the phenalenone nucleus. Degradative studies of atrovenetin showed that the ether bridge is attached to ring B. Brief oxidation with alkaline hydrogen peroxide gave a compound C18H1606 which was optically active formed a diacetate and was assigned the part structure (144) with hydrogen bonding between hydroxyl and carbonyl on both sides of the anhydride ring.This assign- ment rested on tlie correspondence in position of the carbonyl bands of the anhydride (144) (1 703 and 1660 cm.-l) and 2,7-dihydroxynaphthalic anhydride (1720 and 1685 cm.-l) and of the bands of the diacetate of the anhydride (144) (1750 and 1720 cm.-l) and 2,7-diacetoxynaphthalic anhydride (1760 and 1720 cm.?). The foregoing evidence restricts the mode of fusion of the ether ring of atrovenetin to the two possibilities (145) and (146). The latter was favoured because oxidation of atrovenetin with nitric acid had gi~en,~O*~~O~ together with the acid (142) a phenolic product to which structure (147) was assigned on the basis of chemical and spectral studies.Atrovenetin was therefore assigned structure (1 48). However subsequent X-ray crystallographic studieslo8 showed that the ferrichloride of a trimethyl ether of atrovenetin possesses the phenalenium structure (149) thus establishing conclusively structure (141) for atrovenetin. It is apparent that a skeletal rearrangement must have occurred during the nitric acid de- gradation of atrovenetin through it is suggested,lo8 the intermediate (150) as shown. Deoxyherqueinone has been shown to possess structure (151a or b).1079108 These compounds posssss tlie same skeleton as a t r o ~ e n e t i n l ~ ~ ~ ~ ~ ~ but contain an extra atom of oxygen. They are also unstable to acid as already noted in contrast to atrovenetin which is stable.It has been suggested that the extra oxygen atom in her- queinone and norherqueinone is situated at a tertiary position such that it blocks the aromatic conjugation of the system present in a t r ~ v e n e t i n . ~ ~ ~ ~ Reduction of norherqueinone to atrovenetin (Fig. 2) removes this conjuga- tion barrier. Structure (1 52) has been suggested for norherqueinone; this would account for the reduction to atrovenetin and the acid-catalysed cleavage to norxanthoherquein (143) and isopropyl methyl ketone. Since deoxyherqueinone has structure (151a or b) herqueinone would then be Heryueinone and Norherqueinonc. REID THE CHEMISTRY OF THE! PHENALENES 30 1 (153a or b). Cason and his co-workers arrived at structure (154) for a trimethyl ether of herqueinone on the basis of degradation and nuclear magnetic resonance studies thence deducing structure (1 55) for her- q~einone.~O~ Since the arguments leading to this structure were based on formula (148) for atrovenetin Cason's structure for herqueinone becomes identical to the alternative (1 53b) suggested by Paul and Sim.loB OMe MeOfiOMc 8.Biosynthesis of Naturally Occurring Phenalenones Since the structurally similar atrovenetin herqueinone and nor- herqueinone have their origin in closely related fungal species it seems reasonable to expect their carbon skeletons to arise by the same bio- synthetic mechanism. Barton suggested that the origin of the phenalenone nucleus is a poly-j?-diketone system possibly (156) and the ether ring in atrovenetin is derived from a mevalonic acid precursor.1o7 Alternatively one or more of the three rings might arise from shikimic acid.l1° Barton's scheme was shown to be correct by the results of degradative experiments on 14C-labelled norherqueinone produced by feeding Penicillium herquei with sodium [ l-14C]acetate and DL- [2-14C]mevalonic acid lactone in parallel experiments.l1° The norherqueinone was hydrolysed by acid to norxanthoherquein and isopropyl methyl ketone for radioassay.The norxanthoherquein from both experiments was degraded (Kiihn-Roth) to acetic acid and carbon dioxide. The sodium [ l-14C]acetate-derived loo J. Cason J. S . Correia R. B. Hutchison and R. F. Porter Tetrahedron 1962,839. R. Thomas Biocl:em. J. 1961 78 807. 302 QUARTERLY REVIEWS Ho&$H2 -( ,C@,CO co co / OH (I 59) norherqueinone contained two-ninths of the total activity in the side chain and one ninth in the methyl-substituted carbon atom of the norxantho- herquein nucleus ; the methyl group itself was inactive.The carbon dioxide accounted for the remaining activity. The DL- [2-14C]mevalonic acid lactone- derived norherqueinone on the other hand contained all its activity in the side-chain. Hypohalite degradation of isopropyl methyl ketone from both experiments showed that one-half of the activity resided in the non-gem- methyl group. These results demonstrate the isoprenoid nature of the side- chain the mevalonate precursor arising from acetate probably by way of p-hydroxy-b-methylglutarylcoenzyme A. Acetate is incorporated into the aromatic nucleus and the isoprenoid ether ring to the same extent.Although the foregoing results are consistent with the derivation of the phenalenone nucleus (1 57) from the suitably coiled linear structure (1 56) formation from a branched structure is not ruled out. It is interesting that the presumption of normal head-to-tail linking of acetate units predicted the recently revised structure (141) which moreover contains more 18- orientated oxygen substituents. The biosynthetic mechanism whereby the C-2 hydroxyl group is introduced into these pigments remains to be eluci- dated. The anhydride (144) has been isolated from cultures of Penicillium herquei along with atrovenetin herqueinone and norherqueinone.lll A different biosynthetic mechanism has been suggested for the formation of hiemocorin since this pigment contains two sets of ortho-oxygen substituents which is unusual among acetate-derived pheno1s.ll0 An inter- mediate (1 58) is envisaged as arising from two shikimic acid-derived C,-C3 units namely cinnamic and 3,4-dihydroxyphenylpyruvic acid. These substitute into the methyl group of an acetate unit before decarboxylation and ring-closure to the nucleus (1 59) of the haemocorin aglycone. ll1 N. Narasimhachari and L. C . Vining Canad. J . Chern. 1963 41 641.

ISSN:0009-2681    

DOI:10.1039/QR9651900274

出版商:RSC    

年代:1965

数据来源: RSC

摘要:

THE STRUCTURAL CHEMISTRY OF MERCURY By D. GRDENI~ (LABORATORY OF GENERAL AND INORGANIC CHEMISTRY FACULTY OF SCIENCE UNIVERSITY ZAGREB YUGOSLAVIA) 1. Introduction IN recent years knowledge of the structures of mercury compounds has been greatly expanded mainly by X-ray work. Wells1 has provided a good general summary and there have been recent reviews of mercury-nitrogen,2 halogenomercurate(~~),~ and mercurous compo~nds.~ This review deals with the structural chemistry of mercury in both molecular and macro- molecular compounds. The applications of ligand field theory5-’ to dlO-metal ion^,^^^ though promising are not yet numerous and it is hoped that the present survey will stimulate further work in this field. 2. Radii of Mercury The nearest neighbours of mercury in a crystal structure are those in “contact” with it.In a free molecule (e.g. in the vapour state) these con- tacts are chemical bonds and their number is defined by valency or co- ordination number. With a crystal structure the position is less clear and it is necessary to restrict the consideration of nearest neighbours (the atoms in the co-ordination sphere) of mercury to those within a defined distance. This is difficult for mercury as frequently the distances between the sur- rounding atoms and mercury do not follow the additivity rule of the atomic radii currently accepted yet are less than the sum of the van der Waals radii. Even when only atoms of one element surround mercury markedly different distances may be found. Consequently the adoption of a suit- able set of atomic radii should precede any discussion of the structural chemistry of mercury.(a) The MetuZZic Radius.-The parameters of the crystal structure of solid mercurylO at low temperature have recently been determined very accurately.ll The structure differs from the common metallic close packed Wells “Structural Inorganic Chemistry,” 3rd edn. Clarendon Press Oxford Lipscomb Annals New York Acad. Sci. 1957 427-435. Deacon Rev. Pure and Appl. Chem. 1963 13 189; see also BergerhoE Angew. Tarayan Uspekhi Khim. 1953,22 1002. rj Nyholm “Report of the 10th Solvay Conference in Chemistry,” Brussels 1956 Gillespie and Nyholm Progr. Stereochem. 1958,2 261. ’ Orgel “An Introduction to Transition-Metal Chemistry-Ligand Field Theory,” * Orgel J. 1958 4186. lo McKeehan and Cioffi Phys. Rev. 1922 19,444. l1 Barrett Acta Cryst.1957 10 58. 1962 pp. 890-900. Chem. Internat. Edn. 1964 3 686. p. 225. Methuen and Co. Ltd. London 1960. Dunitz and Orgel Adv. Inorg. Chem. Radiochem. 1960 2 34. 303 304 QUARTERLY REVIEWS structure and may be interpreted as a packing of flattened rotational ellipsoids instead of spheres. There are two groups of six equal interatomic distances 3.000 and 3.466 A.12 Each mercury atom is surrounded by six closer mercury atoms in an octahedron drawn along one of its triad axes and six equatorial more distant mercury atoms in a hexagon (Figure 1). This gives mercury twelve-co-ordination in a flat cuboctahedron. FIG. 1. The environment of mercury atoms in the crystal of metallic mercury each mercury atom has six closzr neighbours at 3.000 8 (dashed lines) in a distorted octa- hedron and six more at 3.466 8 (dotted lines) in a hexagon.The resulting co-ordination polyhedron is a flattened cuboctahedron which relation to the rhombohedral crystal lattice (a = 3.000 A 01 = 70” 32’) is shown. As a single Hg-Hg distance cannot be given for the metal both values have to be used in defining the metallic radius p (Hg) so that 1.50 A < p (Hg) < 1.73 A. In amalgams this means that Hg-Hg distances of 2.90-3-24 8 represent true contacts between mercury atoms while those longer than 3.466 A do not. (b) The vdn der WaaZs Radius.-In a discussion of the stereochemistry of mercury the distinction between bonded and non-bonded atoms surrounding mercury is very important. The van der Waals radius of mercury R(Hg) is expected to be nearly equal to its metallic radius. The experimental results are rather unsatisfactory because there are very few contacts between mercury and non-bonded atoms in crystal structures.The lowest value of R(Hg) (1.58 A from the Hg . . . . C contact13) is only slightly larger than the lower metallic radius of mercury. Thus the value R(Hg) = 1.50 A will henceforth be used for the van der Waals radius. Provided R(Hg) < 1-73 A some form of bonding is considered to occur. (c) The Ionic Radius.-Anhydrous mercuric fluoride is the only mercury Pauling “The Nature of the Chemical Bond” 3rd edn. Cornell University Press Kitaigorodskii Khotsyanova and Struchkhov Zhur. fiz. Khim. 1953 27 780. Ithaca 1960. GRDENIC THE STRUCTURAL CHEMISTRY OF MERCURY 305 compound which is purely ionic. It crystallises in a fluorite type structure14 with a = 5.54 A and Hg-F = 2.40 A.Thus R(Hg2+) = 1-04 A {using Pauling’s value12 for R(F-). } [The value R(Hg2+) = 1-10 A which is usually quoted in tables was probably obtained by using Goldschmidt’s value for (d) The Covalent Radii.-As stated in section (b) each Hg-L (L is any atom ion etc. in a mercury compound) approach of R(Hg) < 1.73 A can be considered as bonding. The larger Hg-L distances indicative of weak interactions are not observed in the vapour or in apolar solvents but only in crystal structures melts and polar solvents. This review is concerned with the environment of mercury in free molecules or crystal structures. Digonal covaZent radius. The most familiar mercury compounds are those of the type L-Hg-L (e.g. mercuric chloride) and GHg-L’ (e.g. methylmercuric chloride).In addition there are also many polymeric mercury compounds with linear L-Hg-L bonding (e.g. amidomercuric chloride which contains the polymeric ion +NH,.Hg.NH,+.Hg). The differing values of r(Hg) derived from Hg-L distances in these compounds are usually explained by the postulation of double bonding or partial ionic bonding for relatively short or long bonds respectively. Because the influence of surrounding atoms on L-Hg-L bonds is sig- nificant (see section 5) the most reliable digonal covalent radius of mercury r[Hg(2)] should be one derived from data obtained for free L-Hg-L molecules in the vapour. The Hg-L bond lengths obtained by electron diffraction or microwave spectroscopy are given in Table 1. W-). 1 TABLE 1. Hg-L bond lengths (in A) observed in the vapour phase by Hg-CI.Mercuric chloride 2.20 (e)” ; 2-34 (e)b; 2.27 (e)C. Methylmercuric Hg-Br. Mercuric bromide 2-40 (e)”; 2-44 (e)b; Methylmercuric bromide Hg-I. Mercuric iodide 2-55 (e)“; 2-61 (e)b. Hg-C. Dimethylmercury 2-20 (e)e ; 2.23 (e)b. Methylmercuric chloride 2.061 (mw)d. Methylmercuric bromide 2-074 (mw)d. =Braune and Knocke 2. phys. Chem. 1933 B23 163. bGregg et al. Trans. Faraday Soc. 1937,33,852. CMaxwell and Mosley Phys. Rev. 1940,57,21. Gordy and Sheridan J. Chem. Phys. 1954 22 92. eBrockway and Jenkins J . Amer. Chem. SOC. 1936 58 2036. electron difraction (e) or microwave (mw) spectroscopy chloride 2-282 (mw)d. 2.406 (mw)d. To derive the bicovalent radius of mercury it is preferable to use data from compounds in which the electronegativity difference x(Hg) - x(L) is as low as possible i.e.in Hg-C and Hg-I derivatives.* Thus from Hg-I (2.61 A)*r[Hg(2)] = 1.28 A which is close to the values of 1.29 and 1-30 A derived from C-Hg data for methylmercuric chloride and * Using electronegativities and radii given by Pauling.la l3 Ebert and Woitinek 2. anorg. Chem. 1933 210 269. 306 QUARTERLY REVIEWS bromide. The corresponding Hg-C1 and Hg-Br data give values of 1.29 and 1.27 A respectively. However the value obtained from the Hg-C distance of dimethylmercury (2.23 A) is significantly higher { r [Hg(2)] = 1.46 A}. Although the Hg-C bond in methylmercuric chloride or bromide must be influenced by the halogen the extent of bond shortening from Me2Hg to MeHgX seems surprising and a re-investigation of the structure of Me2Hg is desirable. Theoretically rwg(2)] could be defined as half the observed Hg-Hg distance in mercurous compounds.However this distance varies with change of L in L*Hg*Hg*L derivatives. For example in mercurous halides15 the Hg-Hg distance increases from 2.43 8 in the fluoride to 2-69 A in iodide. The crystal structure16 of mercurous dihydrate contains @3,0. Hg.Hg.OH,l2+ ions with an Hg-Hg distance of 2.54 A which gives r[Hg(2)] = 1.27 A. As this agrees well with the above r[Hg(2)] values the condition of minimum electronegativity difference x(Hg)-x(L) must be fulfilled. By taking into account the influence of the electronegativity difference12 (Schomaker-Stevenson coefficient = 0*03) the mean value for the digonal covalent radius is 1-30 A. This obeys the additivity rule satisfactorily for the free LzHg or LHgL’ molecules.In crystal structures additivity is not preserved because of the close approach of neighbourjng groups to mer- cury. Support for the value of r[Hg(2)] comes from the fact that it is less than the metallic radius of mercury. The bicovalent radius for sp (or ds- see section 6) hybridisation should be much shorter than the corresponding metallic radius especially as metallic bonding is relatively weak with mercury. In the literature the bicovalent and tetrahedral covalent radii of mercury have often been confused. TABLE 2. Bond lengths (in A) in crystal structures having tetrahedrally co-ordinated mercury Hg-I. Mercuric iodide (red) 2.7P; 2*80b; 2~77~; 24Od. Silver(r) tetraiodo- mercurate 2.74d; Copper(1) tetraidomercurate 2 ~ 6 4 ~ . Hg-S. Mercuric sulphide (black metacinnabarite) 2-53eJ; 2.549; 2~525~ ; Cobalt(i1) tetrathiocyanatomercurate 2-50i; Bisethylenediaminecopper(I1) tetrathiocyanatomercurate 2.56j; Nickel@) tetrathiocyanatomercurate di- hydrate 2-50k; Potassium tetrathiocyanatomercurate 2.54z.Hg-Se. Mercuric selenide 2.62. Hg-Te. Mercuric telluride 2.78. UHavighurst Amer. J . Sci. 1925 10 556. bBijvoet Classen and Karssen Proc. k . ned. Akad. Wetenschap. 1926,29,529. “Huggins and Magill J. Amer. Chem. Soc. 1927 49 2357. dKetelaar 2. Krist. 1931,80 190. Lehmann Z. Krist. 1924 60 379. fKolk- meijer and Bijvoet Proc. k . ned. Akad. Wetenschap. 1924 27 390 847. gBuckley and Vernon Mineralog. Mag. 1925 20 382. hGoldschmidt Geochemische Verteilungs- gesetze der Elemente Skrifter Norske Videnskaps Akad. vol. 8 1926. tJeffery Nature 1947 159,6106.jScouloudi Acta Cryst. 1953,6 1953. kK’uo-Hsiang Chou and Porai- Koshits Kristallografya 1960 5,462. ZZvonkova Zhur. jiz. Kllin?. 1952 26 1952. l5 GrdeniC and DjordjeviC J. 1956 1319. l6 GrdeniC J. 1956 1312. GRDENI~ THE STRUCTURAL CHEMISTRY OF MERCURY 307 The tetruhedral covalent radius. Interatomic Hg-L distances deter- mined by X-ray or neutron-diffraction methods for crystals in which mercury has regular tetrahedral stereochemistry are recorded in Table 2. The HgL group is either a complex ion [as in K,Hg(SCN),] or the struc- tural element of a layer (as in red mercuric iodide) or of a three-dimensional framework of the zinc blende type (as in metacinnabarite). From the Hg-L distances and the covalent radii1 of sulphur (1.04 A) selenium (1.14 A) tellurium (1.32) A and iodine (1.28 A) the tetrahedral covalent radius of mercury is found to be 1-48 A.This radius is not of general use as those structures in which mercury has distorted tetrahedral stereochemistry are actually derived from bicovalently or tricovalently bonded mercury with additional close approaches of other usually more electronegative atoms or ions. To summarise the following mercury radii are adopted and used in this review. Metallic radius p(Hg) 1-50 < p(Hg) < 1.73 A van der Waals radius R(Hg) = 1-50 A Ionic radius R(Hg2+) = 1.04 A Digonal covalent radius r[Hg(2)] = 1-30 A Tetrahedral covalent radius r[Hg(4)] = 1.48 A 3. Valency Bond Angle of Mercury The only compounds of mercury that exist as discrete molecules or give them in vapour or solution are of the type LHgL (e.g.mercuric chloride) and LHg-HgL (e.g. mercurous trichloroacetate). They are not very numerous especially the latter. The presence of linear bonds has been confirmed in free molecules (vapour phase) by electron diffraction and microwave spectroscopy and in some crystalline compounds by X-ray diffraction. A slight departure from linearity has been found for the C-Hg-C bonds in mercury diethylene oxidel' (bond angle 176") and in mercuric cyanide18 (bond angle 171 ") but in all other crystal structures of organomercurials the C-Hg-C or C-Hg-X bond angles are close to 180". However the dipole moments of various diarylmercurials in benzene and decalin solution are unexpectedly not ~ e r o l ~ - ~ l suggesting either a bent C-Hg-C skeleton or at least significant oscillation of the molecules away from linearity in solution.Further investigations as to the configurations of free molecules of the diary1 mercurials seem desirable. In the crystal structures of these compounds because of a combination of inter- and intra-molecular factors linearity of C-Hg-C bonds seems a requirement for minimum energy but this is not necessarily true for free molecules. In other crystalline mercury compounds particularly those with halo- gen oxygen or sulphur atoms nonlinear L-Hg-L skeletons are frequently Grdenic Acta Cryst. 1942 5 367. Hvoslef Acta Chern. Scand. 1958 12 1568. l 9 Hampson Trans. Faraday SOC. 1934 30 877. 2o de Laszlo Trans. Faraday Soc. 1934 30 884. 21 Horning Lautenschlaeger and Wright Canad. J. Chern. 1963 41 1441. 308 QUARTERLY REVIEWS found due to the electronegativity of the ligands or the packing condi- tions in the crystal structure.4. Co-ordination Chemistry of Mercury All atoms surrounding mercury at a distance of less than the sum of the van der Waals radii [D(Hg-L)<R(Hg) + R(L)] are considered to belong to the mercury co-ordination sphere. (The values used for R(L) are those given by Pauling.12) The following classification of the co-ordination be- haviour of mercury is suggested. (a) Characteristic Co-ordination-The mercury atom is considered to have the characteristic co-ordination number (m) in the system HgL when all m Hg-L bonds are of the same length. In free molecules or com- plex ions characteristic co-ordination numbers of two three and four for covalent bonding and eight for ionic bonding are known.Examples are given in Table 3 and are illustrated in Figure 2. Regular octahedral I S 2 26 Cl- Hg -Cl a I 's 12.70 2 31 2 43 Hg 'w2e s/ G3 \s F-Hg-Hg-F I'x\I C d b FIG. 2. Examples of characteristic co-ordination of mercury (a) (b) digonal as in mercuric and mercurous halides,' (c) trigonal as in (Me3S)Hg13,33 and (d) tetrahedral as in tetrathiocyanatomercurates (for references see Table 2). stereochemistry for mercury has not been authenticated by full structure determinations but in the recently prepared HgL,(C104) (L is pyridine 1 -oxide dimethylsulphoxide etc.) complexes,22 this stereochemistry is probably realised. Only one characteristic ionic co-ordination number is known (m = 8). The co-ordination polyhedron is a cube in crystalline mercuric fluoride but other possible polyhedra are not excluded for eight co-ordination.The structure23 of potassium tetranitritomercurate(I1) nitrate has been described as containing the tetrahedral complex ion Hg(N0,),2-. Hou- ever eight oxygens are equidistant from mercury and the Hg-0 distance (2.4 A) indicates almost pure ionic bonds. Thus the co-ordination poly- hedron may be a distorted trigonal dodecahedron. Ionic structures with m = 6 have not yet been established (see however Table 3). Such an arrangement requires R(Hg2+)/R(L-) to be less than 0-73 .These conditions apply when L = C1 but the structures of mercury-chlorine derivatives show a departure from octahedral symmetry two Hg-C1 distances being markedly less than the other four. (b) Eflec tive or Actual Co-o rdina tion .-The effective co-o rdi 11 at io n 23 Carlin Roitman Dankleff and Edwards Inorg.Chem. 1962 1 182. 23 Hall and Holland Proc. Chem. Soc. 1963 204. GRDENIC THE STRUCTURAL CHEMISTRY OF MERCURY 309 number n is an extension of the characteristic co-ordination number to include all ligands fulfilling the condition D(Hg-L) < R(Hg) + R(L). This arises from the close approach of adjacent molecules etc. in a crystal structure. Thus HgL becomes HgL where n>m. The value n = 6 is the highest known (Table 4). Only when m = 2 or 3 can additional ligands be accommodated. With m = 4 n = 4 digonal characteristic co-ordination leads to distorted trigonal square planar tetrahedral pyramidal or octahedral stereochemistry while trigonal co-ordination gives elongated trigonal bipyramidal or very distorted tetrahedral effective co-ordination.Linear bonds in digonal characteristic co-ordination become bent when the molecule acquires distorted trigonal or tetrahedral effective co-ordination. The new ligands lengthen the Hg-L bonds of the characteristically co- ordinated ligands. To summarise in effective co-ordination two groups of D(Hg-L) distances are observed (1) shorter ones which reprzsent charac- teristic co-ordination and (2) longer ones due to close approaches in crystal structures. This criterion is also applicable to systems involving different ligands. Examples of all known effective co-ordination arrange- ments are given in Figures 3-5 the most common being distorted octahedral based on digonal characteristic co-ordination. TABLE 3. Characteristic co-ordillation (m) of mercury in molecules complex ions und crystal structures in relation to the electronegativity (XL) of the ligands Covalent species XL HgI,(red) .1 HgI,- HgBr,- HgC1,- .1 PbHgP,, HgTe HgSe HgS (cubic) Hg(SCN)42- 2.1 2.5 2.5 3.0 HgF, HgCl, HgBr, and HgI in vapour or 2.5 dissolved in apolar solvents; M%Hg MeHgCl; I molecules with -0-Hg-0 -N-Hg-N-.Mer- 1 curous compounds with -0-Hg-Hg-0- as in 4 Hg,(OH,),++ or Hg,(OCOCCI,) 3.5 Ionic Probable but not established yet. Partly realised 3-0 3.5 in Hg(OHgCl) J. 3.5 4.0 5. 310 QUARTERLY REVIEWS TABLE 4. Eflective co-ordination (n) of mercury in crystal structures in relation to the characteristic co-ordination (m) and bond distances Dwg-L) n Species 6 HgO Hg(NH3)2C12 Hg(CN) HgPy2C12 NH,Hg Cl K,HgCl,*H,O HgBre HgI,(yellow) HgS(cinnabar) Hg2X2 5 Distorted square (MeS),Hg pyramid 5 Elongated trigonal Me,SHgI bipyramid Hg,NHBr Characteristic bonds m = 2 m = 3 D(Hg-L) 0-Hg-0 N-Hg-N Cl-Hg-Cl Br-Hg-Br C-Hg-C I-Hg-I - S-Hg-S and mixed nt distances with D(Hg-L 2 rlHg(2)I + r(L) D(Hg-L) Probable X and n - rn dis- tances with but not I observed Hg Yet / \ < R(Hg)+W) X X X = Br or I X I X /""\ X X = C1 Br or I HgSO4 0-Hg-0 C-Hg-0 - (NCHg),O HgCl ,2Hg0 0-Hg-Cl Hg(SCN),,KSCN S-Hg-S ~ ~ _ _ _ ~ 3 Planar or Mercury nearly diethylene C-Hg-C - planar oxide (c) Factors IPtJluencing Co-ordination.-The following are the main features of the crystal chemistry of mercury.(1) Linear covalent bond formation. (2) Tetrahedral co-ordination in zinc blende type structures. (3) Metal-metal bond formation. (4) The formation of ionic structures. All rules valid in crystal chemistry can be applied to any groups of mercury GRDENIC THE STRUCTURAL CHEMISTRY OF MERCURY 31 1 a b C 0 S 0 s d e f FIG.3. Examples of trigonal (a) and tetrahedral efective co-ordination of mercury (a) mercury diethylene oxide,17 (b) mercuric o x y ~ y a n i d e ~ ~ (c) trimercuric oxy~hloride,~~ ( d ) bis(tripheny1arsine 0xide)mercuric chloride,124 (e) anhydrous mercuric ~ ~ I p h a t e ~ ~ (f) mercuric dithizonate.66 Br Br 0 Br Br b S 13 24 C FIG. 4. Examples of tetrahedral (a) bipyramidal (b) and pyramidal (c) efective co-ordination of mercury. The distorted tetrahedron (a) in (Me4N)HgBr3,35 and the trigonal bipyramid (b) in imidonzercuric bromide 76 are based on trigonal characteristic co-ordination while the pyramid in dimethylthio mercury7" originated from digonal characteristic co-ordination.compounds from ionic to molecular and mercury co-ordination could be correlated with all influencing factors e.g. ionic radii polarisability electronegativity etc. It has been found that the electronegativity differ- ence [x(Hg) - x(L)] is most suitable for a rational classification of the crystal structures of mercury compounds. The value x(L) = 2.5 is a critical one because for x(L) > 2-5 the characteristic co-ordination is generally digonal and for x(L) < 2.5 the characteristic and effective co-ordination is generally tetrahedral. Sulphur and iodine occupy a transitional position digonal co-ordination being more stable for sulphur (cinnabar more stable than metacinnabarite) and tetrahedral for iodine 312 Br QUARTERLY REVIEWS 0 Br a N N d 0 ci 0 b 0 O h S S C C C f CL Ct FIG.5. Examples of octahedral efective co-ordination based on digonal characteristic co-ordination of mercury (a) mercuric br~mide,~' (b). orthorhombic mercuric oxide,44 (c) cinnabar,s3 (d) cubic8' and (e) orthorhombicS2 amidomercuric bromide (f) K1,Hg (CN),,lao (g) mercuric o~ycyanide,~~ (h) (HgO)2,NaI,68 (i) mercrrric cyanide,97 (j) chloromercuric thio~yanate,~~ (k) mercuric sulphate monohydrate,68a (1) the addition compound of mercuric chloride with ~oumarin.'~ (red modification of mercuric iodide more stable than yellow one). Thus x(S) > ~ ( 1 ) . Digonal characteristic co-ordination is the most common because the majority of mercury compounds contain ligands with x(L) > 2.5 and in this group octahedral effective co-ordination predominates.Even with very close approaches of the additional ligands the co-ordination polyhedron is irregular. The most symmetrical is a square bipyramid which is flat along its tetrad axis. These octahedra are usefully classified into two GRDENIk THE STRUCTURAL CHEMISTRY OF MERCURY 313 groups depending whether the basic L-Hg-L skeleton is linear or bent. The first group contains the point groups Ci (very frequent) D4h CdV D2h C2,, Cz C, C, and the second the point groups C, CZv C2 and C (Figure 5). Tetrahedral effective co-ordination is very rare for x(L) > 2.5 and may be the result of steric factors in the crystal structure e.g. mercuric ~ u l p h a t e ~ ~ or cyanomercuric oxide (CNHg)20.25 The size of the atoms and packing conditions do influence the final arrangement but x(L) gives a satisfactory correlation for most structures.5. Crystal and Molecular Structures of Mercury Compounds (a) Halogenornercurate(r1) Compounds.-The crystal structures of the mercuric halides are an example of a morphotropic transition dependent on the electronegativity of the halogen (4.0 3-0,2.8 and 2.5 for F C1 Br and I respectively). The structures of the chloride bromide and iodide are given in Figure 6. While mercury has distorted octahedral stereochemistry a CL c1 Cl CL / / I 'Cl FIG. 6. The environment of mercury in the crystal structures of mercuric chloride (a) bromide (b) and iodide (c). in the chloride the structure is essentially because two pairs of Hg * - - - Cl distances (3.34 and 3.63 A) are larger than the sum of the van der Waals radii (3.30 A).Mercuric bromide has a layer structure of a deformed brucite type,27 with six Hg-Br distances less than R(Hg) + R(Br). 24 BonefaEiC Croat. Chem. Acta. 1962 34 119; 1963 35 195. 25 SdavniEar 2. Krist. 1963 118 248. 26 Braekken and Harang 2. Krist. 1928 68 123; Braekken and Scholten Z. Krist. 27 Verweel and Bijvoet 2. Krist. 1931 77 122; Braekken 2. Krist. 1932 81 152. 1934 89,448; GrdeniC Arhiv Kem. 1950 22 14. 3 14 QUARTERLY REVIEWS Thus the effective co-ordination is octahedral (Figures 5 and 6). The crystal structure of the iodide is composed of layers of fully corner-linked HgI tetrahedra.28 The mercuric halides are a good example of allotropy. The yellow unstable modification of the iodide crystallises with the HgBr ~ t r u c t u r e ~ * ~ ~ ~ while the bromide probably has an unstable modification with the HgCI ~ t r u c t u r e .~ ~ In the vapour state mercuric chloride bromide and iodide consist of linear molecules. The behaviour of the fluoride is not known but should be similar in the vapour. Halogenomercurates of the type (Mz+)b-2a/lHgaXb (M is a simple or complex cation X is CI Br I) are well known the most common belonging to the classes (M+),HgX and (M+)HgX,. Recently some pyridinium tetra- fluoromercurate(1r) derivatives have also been made.30 The properties of these compounds in solutions and in the melt have been extensively in- vestigated3 and have been recently reviewed.31 Here the discussion is restricted to their crystal structures. The electronegativity rule is again relevant. The characteristic co-ordina- tion for chloro-complexes is generally digonal though there are excep- tions (see below) while that for iodo-complexes is trigonal or tetrahedral.Relatively little is known about the structures of crystalline bromonercur- ates and the structures of the tetrafluoromerc~rates~~ have not yet been determined. The crystal structure of yellow tetragonal Ag,HgI has cubic close packed iodine atoms with some of the tetrahedral holes filled by silver and mercury atoms in a regular manner. It is ordered below 50.7" but the red cubic modification obtained above this temperature has a dis- ordered zinc blende structure.32 The isomorphous Cu ,Hg14 shows analog- ous thermochromic properties. Tetraiodomercurate ion are found in [Me,S]2HgI,,33 but the complete structure has not yet been reported. The structure of [Me3S]Hg13 (m = 3) has been mentioned (Figure 2 Table 4) while [Me,N]HgI is isomorph~us~~ with [Me,N]HgBr, the structure of which 35 has been considered (Figure 4 Table 4).There is one exception to the electronegativity rule conclusion that in chloromercurate the characteristic co-ordination of mercury should be digonal. The tetrachloromercurate of the alkaloid perloline contains tetrahedral HgCl 42- ions.36 In we,N]HgCl, which is isornorpho~s~~ with the corresponding tribromo-deri~ative,~~ the characteristic co-ordina- tion is trigonal (Table 3). A classification of the other known chloro- 2 * Havighurst Amer J. Sci. 1925 10,556; Bijvoet Claasen and Karssen,Proc. k.ned. 2@ van Nest 2. Krisr. 1909 47 263. 30 Dotzer and Meuwsen 2. anorg. Chem. 1961 301 79. 31 Wait and Janz Quart. Rev.1963 17 225. 32 Ketelaar,Z. Krist. 1931 80 190; 1934 (A) 87 435; Wells ref. I pp. 174 533. 33 Fenn Oldham and Phillips Nature 1963 198 381. 34 Fatuzzo Nitsche Roetschi and Zingg Phys. Rev. 1962 125 514. 35 White Acta Cryst. 1963 16 397. 36 Jeffreys Sim Burnell Taylor Corbett Murray and Sweetman Proc. Chein. SOC. Akad. Wetenschap. 1926,29,529. 1963 171. GRDENI~ THE STRUCTURAL CHEMISTRY OF MERCURY 315 mercurate structures has been given.37 As rn = 2 and n = 6 in all these compounds then the configuration of the anion does not follow from the stoicheiometry. Thus K2HgCI,,H,0 does not contain discrete HgCl q2- ions,38p39 but distorted HgCl octahedra form columns sharing two op- posite edges (Figure 7a). There is a linear CI-Hg-Cl skeleton (Hg-CI 2.29 A) with two pairs of longer bonds (2-29 and 3.13 A).= If two of these ct C l a b c1 C C l Cl Cl C l c1 c1 C l ct .c1 C l C l c1 c1 C t C l c1 c1 c1 FIG. 7. The chloromercurate anions in the crytsal structures of mercury chloro- complexes (idealised) (a) ribbon of (HgC14),2a- in K2HgC14,Hz0,38*39 (b) twofold ribbon of (Hg2C16)a2a- in NaHgC1340 and related corn pound^,^^**^ and (c) layer of (HgCl,).a- in NH4.HgC1,.43 37 Damm and Weiss 2. Naturforsch. 155 lob 535. 38 Zvonkova Samodurova and Vorontsova Doklady Akad. Nauk S.S.S.R. 1955 39 MacGillavry de Wilde and Bijvoet 2. Krist. 1938 A 100 212. 102 1115. 316 QUARTERLY REVIEWS linked octahedral columns are condensed by sharing three edges of an octahedron,the polymeric (Hg2C16),2a- ion is formed. This has been found in the structures of NaHgC13,40 NaHgC1,,2H20,41 and Na2Hg2CI,0H (Figure 7h).42 NH,HgCl is not isomorphous with the sodium salt but has a layer structure43 in which HgCl octahedra share four chlorine atoms (at 2-96 A) and have colinear Hg-CI bonds of 2.34 A (Figure 7c).The corresponding caesium compound has a cubic structure38 in which mercury and czsium occupy perowskite positions the environment of each mercury being two chlorines at 2.29 A and four at 2.70 A. Three acids H2HgCI4,3Hz0 H,Hg2C1,,xH20 and HHg2C1, have been identi- fied in the solid The first is isostructural with K,HgC14,H20 and thus may be written (H,0)zHgC14,Hz0. The detailed structures of the others are not known but are considered to involve HgCl octahedra.,’ (b) Mercuric Oxide Oxyhalides and Related Compounds.-Both modifications orthorhombi~~~ and hexagonal45 of mercuric oxide are based on O.Hg.O.Hg-O.Hg zig-zag chains (Figure S) which are planar in the former and spiral in the latter.In both n = 6 and m = 2 and the characteristic Hg-0 distance is 2.03 A. Four oxygen atoms from adjacent chains are only slightly closer than the s u ~ n of the van der Waals radii viz. at 2.79 A (two) and 2.90 A (two) in the hexagonal form and at 2-82 (four) in the orthorhombic form. The crystal structure parameters have been refined by neutron-diffraction methods and hence are very accurate. a H H H H \ / b \ / N+ \ / N+ N + \ / \ H H I \ H H FIG. 8. The characteristic mercury-oxygen and mercury-nitrogen chain Jtructures (a) mercuric 0xide,4~-~’ (b) the amidomercuric cation as e.g. in H2NHgC1.82 40 Weiss and Damm 2. Naturforsch. 1954 9b 82. 41 NinkoviC Bull.Inst. Sci. “Boris Kidrit” Belgrade 1957 7 81 ; MalEiC Bull Inst. Nuclear Sci. “Boris Kidric‘” Belgrade 1959 9 1 15. 43 Weiss and Damm 2. Naturforsch,. 1955 lob 537. 43 Harmsen Z. Krist. 1938 A 100 208. 44 Aurivilius Acta Chem. Scand. 1944 8 523; 1956 10 852. 45 Aurivilius and Carlson Acta Chem. Scand. 1957 11 1069; 1958 12 1297. GRDENIC THE STRUCTURAL CHEMISTRY OF MERCURY 317 TABLE 5. Structure types of mercuric oxyhalides Hg,CI 30a-l in relatioir to oxygen :mercury atomic ratio a 1 3 4 312 5 5 00 Formula 2HgCl,,HgO HgC12,2Hg0 HgC12,3Hg0 HgCl2 HgC12,4Hg0 HgBr2,4Hg0 HgO 415 415 1/1 Structure Cl-Hg-C1 (ClHg),O+Cl- Related to kleinite a naturally occur- ing oxychloride containing nitrogen. Probably distorted tridymite-like struc- ture Hg(OHgC1)2 (Hg4c14)4+(Hg608)4- (HS‘p4)1+ (Hg60,)4- Chains of Hg-0-Hg-0-Hg-0 Ref.26 49 50 51 52,449 47 2 53 53* 4-47 * Added in Proof.-The structure of HgBr2,4Hg0 as determined recently by Aurivilius (Arkiv Kemi. 1965 23,469) is best defined as Hg(OHg),Br,. Many basic mercuric halides or oxyhalides of the general formula Hg,C120,- (a = 1.5 2 3 4 5) have been des~ribed.~~ Recent investiga- tions have shown that the characteristic co-ordination is digonal [x(CI) = 3.0 x(0) = 3-51 and the effective co-ordination is tetrahedral or octa- hedra1.49-53 The structures are strikingly dependent on the oxygen mercury ratio and each species has its own distinct structure (Table 5). With increasing oxygen content the structures become more differenti- ated into mercury-chlorine and mercury-oxygen structural elements until complete separation into alternating layers of complex halide cations and complex oxide anions occurs53 with an atomic ratio 0 H g = 4 5 .The structure of 2HgCI2,HgO contains the tris(ch1oromercuri)oxonium ion. This should be pyramidal if the Hg-0 links are covalent single bonds. However X-ray diffraction has shown the ion to be approximately planar49$50 (Figure Sc) which might indicate partial double bond character for the Hg-0 links. Neutron-diffraction analysis54 shows a slight devia- tion from planarity. The 0-Hg-Cl bonds are linear (Hg-0 = 2-03 A Hg-Cl = 2.28 A) and mercury has a very distorted octahedral environment. The structures of the analogous tris(methylmercuri)oxonium compounds55 are now being investigated. 46 Aurivilius and Heidenstamm Acta Chem. Scand. 1961 15 1993.47 Aurivilius Acta Cryst. 1956 9 685. 4 8 Mellor “A Comprehensive Treatise on Inorganic and Theoretical Chemistry,” 4 9 Weiss Nagorsen and Weiss 2. Naturforsch. 1953 8b 162; Weiss Nagorsen and 5 0 Grdenic and SCavniEar Nature 1953 172 584; SCavniEar and GrdeniC Acta 51 SCavniEar Acta Cryst. 1955 8 379. 52 Heritsch Tschermaks tnineralog. u. petrolog. Mitt. 1949 1 300. 63 Weiss Nagorsen and Weiss Z. Naturforsclr. 1954 9b 81. 54 Aurivilius Acta Cryst. 1963 16 A25. 55 GrdeniC and Zado Croat. Chem. Ada 1957 29 425; Grdenic and Zado J. 1962 vol. 4. Longmans Green and Co. London 1957. Weiss 2. anorg. Chem. 1953 274 151. Cryst . 1955 8 275. 521. 318 QUARTERLY REVIEWS Mercuric oxide forms compounds with salts other than those of mercury. The crystal structure of (HgO),,NaI consists of layers formed from HgO chains intermingled with iodide ions.56 The effective co-ordination of each mercury is octahedral if the fairly close approach of a mercury atom in an adjacent layer is considered to be significant (Hg .- - - Hg = 3.334 A). (c) Salts ofOxy-acids.-Digonal characteristic co-ordination is expected [x(O) = 3-51. The following crystal structures have been investigated (1) anhydrous salts HgS0424v57 (2) hydrated salts HgS04,H205* (3) basic salts HgS04,2Hg059-61 2HgS04,Hg0,2H20,61 Hg(X03)2,Hg0,H,062 (X is C1 Br) (4) complex or double salts K3[Hg(N02)4]N03.23 Mercuric sulphate and selenate are isomorphous as are dioxotrimercuric sulphate selenate and c h r ~ r n a t e . ~ ~ In anhydrous mercuric sulphate the stereochemistry of mercury is distorted tetrahedral.The 0-Hg-0 bond angles in two mutually normal planes are 159" and 144" while the Hg-0 bond lengths are 2.14 (two) for the former and 2.08 and 2.28 A for the latter OHgO groups. The hydrate has a completely different structure on H20Hg05 octahedra arranged in parallel chain which are crosslinked through sulphate ions. The characteristic H,O-Hg-0 bonds are bent to 169" the H,O-Hg and Hg-0 distances being 2.24 and 2-17 A respectively.58b The Hg-0 distances of the more remote neighbours are 2-50 A (two) and 2.51 A (two) [N.B. R(Hg2+) + R(02-) = 2.44 A]. The structure of the basic mercuric sulphate Hg,0,S04 (turpeth mineral)59 has been solved quite recently.60s61 The structure is formed from layers of polymeric mercurioxonium cations (Figure 9d) with sulphate anions between them. Thus the compound is (Hg302)a2a+(S042-)a.The characteristic Hg-0 distances are 2.03 A and mercury has octahedral effective co-ordination. A polymeric oxonium ion (Figure 9b) has also been found62 in crystals of basic mercuric chlorate and brornate monohydrate Hg(X0,) ,,H,O and the correct formulation of these salts is therefore (HgOH),a+(XO,-),. The occurrence of mercuri- oxonium ions in basic mercuric salts is thus fairly general. Similar be- haviour is found for ammonia derivatives of mercury compounds where mercuriammonium ions occur The compound 2HgS04,Hg0,2H,0 has a complex structure61 with two crystallographically different mercury atoms. 56 Aurivilius Actu Chem. Scund. 1960 14 2196; 1964 18 1305. 57 Aurivilius and Malmros Acta Chem. Scund. 1961 15 1932. 58 (a) BonefaEiC Actu Cryst.1961 14 116; (b) Templeton Templeton and Zalkin Acta Cryst. 1964 17,933 (refinement by the use of three-dimensional X-ray diffraction data). E . ~ PaiC Bull. SOC. chim. France 1930 47 1254; PaiC Ann. Chim. (France) 1933 19 427. 6o Nagorsen Angew. Chem. 1962 74 119. G1 BonefaEiC Thesis University of Zagreb 1963; BonefaEid Actu Cryst. 1963 16 G2 Weiss Lyng and Weiss 2. Nuturforsch. 1960 15b 678. ,430. GRDENIk THE STRUCTURAL CHEMISTRY OF MERCURY 319 H I 2.1s + a H26- Hg5Hg -OH H I ct 1 b j o y 0 3 iO\ Hg i"\ 03 Hg Hg A 28 \ ct \O/Hg '0 Hg Hg / 0 H I H C l I H i C FIG. 9. The oxonium ions in the crystal structures of different mercury compounds (a) the hydrated mercurous ion in Hg2(N03)2,2H20,1E (b) the hydroxomercuric ion in basic mercuric chlorate and bromate,6a (c) the tris(ch1oromercuri)oxonium ion in 2HgC12,Hg0,40*50 and (d) a (Hg302)a2a+ layer in the basic mercuric sulphate.60*E1 Only one of them has Hg-0 bond lengths corresponding to covalent bond formation (Hg-OH = 2-12 A).All other Hg-0 distances are larger than 2.41 A. There is no characteristic mercuri-oxide chain in the structure which must be considered to be essentially ionic. The structurez3 of potassium tetranitritomercurate(r1) nitrate also seems to be essentially ionic. The nitrogen atoms of the nitrite ions surround mercury tetrahedrally but eight oxygen atoms at 2.4 A do not follow necessarily the tetrahedral symmetry. (d) Compounds with Mercury-Sulphur Bonding.-As the electro- negativity of sulphur is 2.5 both tetrahedral and digonal characteristic co-ordination is observed. The stable modification of mercuric sulphide (cinnabar) has octahedral effective co-ordination.63 The compound is isostructural with the hexagonal form of mercuric oxide.45 The character- istic S-Hg-S bonds are bent to an angle of 172.4" and the Hg-S length (2-36 A) is equal to the sum of the digonal covalent radii. Two other pairs of Hg-S distances (3.10 and 3-30 A) are less than the sum of the van der Waals radii. Both black mercuric sulphide obtained by precipitation from solution and metacinnabarite crystallise as the metastable cubic modifica- tion the structure of which is related to those of cadmium and zinc sulphides. 6s Aurivilius Acta Chem. Scand. 1950 4 1413. 320 QUARTERLY REVIEWS Tetrathiocyanatomercurates contain the tetrahedral Hg(SCN),,- ion in which the Hg-S bond lengths are equal to the sum of the tetrahedral covalent radii.The double salt Hg(SCN)2,Ni(SCN)2,2H20 has been shown6 to be nickel tetrathiocyanatomercurate(I1) dihydrate. It contains tetrahedral Hg(SCN),,- ions (Hg-S = 2.50 A) and nickel cations in an octahedral environment of two oxygen and four thiocyanate nitrogen atoms (Ni-0 and Ni-N = 2-08 A). In the compound KHg(SCN), mercury has digonal characteristic and distorted tetrahedral effective co-ordination. The characteristic S-Hg-S linkage is bent ( 1 5 5 O ) and two sulphurs (at 2.78 A) from adjacent thiocyanate ions complete a distorted tetrahedral envir~nment~~ of the type in Figure 3f. The analogous ammonium salt is isostructural and mercuric dithizonate has a similar structure. 66 The crystal chemistry of mercuric thiocyanate has been d i ~ c u s s e d .~ ~ ~ ~ ~ Mixed deriva- tives of the type XHgSCN (X is C1 or Br) have digonally bonded mercury in distorted octahedral environment 69 (Figure 5f). In dimethylthiomercury Hg(SMe,), mercury has two short linear bonds and a unique square pyramidal effective co-ordination. 70 The structures of RSHgC171 and (EtS)2Hg72 seem to need further refinement. The structure of the sulphochloride Hg,S,Cl consists of layers of poly- meric mercurisulphonium cations and halide ions73 and is similar to that of the basic mercuric sulphate shown in Figure 9d. The S-Hg-S bonds are linear (Hg-S = 2.44 A) with the sulphur atom at the apex of a trigonal pyramid and an Hg-S-Hg angle of ca. 95 O. The analogous mercuric seleno- and tellurohalides have similar structures. Mercurisulphonium ions probably occur as commonly as the oxonium analogues.The sulphonium compounds are more stable tris(methy1mercuri)sulphonium ions having exceptional stability. 74 (e) Coinpounds with Mercury-Nitrogen Bonding.-The large number of mercury compounds with ammonia and its derivatives were originally classified into three groups 75 (1) mercuric ammines e.g. “fusible white precipitate” or the various addition products of amines with mercuric salts ; (2) mercuric amides e.g. “infusible white precipitate” and analogous 64 K‘uo-Hsiang Chou and Porai-Koshits Kristallografya 1960 5 462. 65 Zhdanov and Sanadze Zhur. fz. Khim. 1952,26,469. 66 Harding J. 1958 4136. 67 Zhdanov Communications au XIIIe Congr&s International de Chimie Pure et Appliquee Stockholm 1953 Editions de l’Academie des Sciences de l’U.R.S.S.MOSCOU 1953 p. 175. 68 Zhdanov and Zvonkova Trudy Inst. Krist. Akad. Nauk S.S.S.R. 1954 10 30. 6Q Zvonkova and Zhdanov Zhur. fiz. Khim. 1952,26,586. 7 0 Bradley and Kunchur Chem. and Ind. 1962 1240; J. Chem. Phys. 1964 40 71 Johansson Arkiv Kemi Min. Geol. Ser. A 1939 13 1 . 72 Wells Z. Krist. 1937 A 96 432. 73 Puff and Kohlschmidt Naturwiss. 1962 49 299; Puff and Kuster Naturwiss. 74 GrdeniC and MarkuSiC J. 1958 2434. 75 Ley in Abegg und Auerbach “Handbuch der anorganischen Chemie,” vol. 2 2258. 1962,49,299; 49,464. Part 2 S. Hirzel Leipzig 1906 p. 666. GRDENI~ THE STRUCTURAL CHEMISTRY OF MERCURY 32 1 tetrahedra - I -Hg-N-Hg- I Hg a I amidomercuric derivatives ; (3) Millon’s base and its salts. This classifica- tion still remains good in principle.X-Ray analysis has shown that the structure of these compounds are generally based on substituted ammo- nium ions. Brodersen and Rudorff have given the following classification 76 isolated linear groups polymeric zig-zag chains hexagonal puckered layers +H,N-Hg-NH$ -+NH,-Hg-NH,+-Hg- [HgdNH) 2 1 a2‘+ three-dimensional framework I 1 Hg I la+ e.g. “fusible white precipitate e.g. infusible white precipitate e.g. dimercury imidobromide Hg(NH3) ZC1 2 HgNHzCl e.g. Hg,NHBr, viz. Millon’s base and its salts [ Hg3(NH) 2 1 a2’+(Br-) a(HgBr3-1 a There are also some Hg-N compounds in which nitrogen apparently retains an unshared electron pair e.g. mercuric derivatives of acid amides and hydrazides. 77 In many of the known structures the mercury-nitrogen bonds are coli- near mercury having octahedral effective co-ordination.The Hg-N bond length is equal to or less than the sum of the covalent radii. Hence the Hg-N bond is covalent and the positive charge is located on nitrogen. Four adjacent atoms or ions approach closer than the sum of the van der Waals radii to complete the co-ordination octahedron (e.g. Figure 5d e). Diadnemercuric halides (NH,),HgX2 (X is C1 or Br) have a statistical cubic structure built of unit cubes of halide ions.78 In each cube one of the six possible positions at the centre of a face is occupied by a mercury atom so that the linear +H3N-Hg-NH3+ ions are randomly orientated along one of the three axes with nitrogen atoms approximately at the centre of each cube. The structures of the analogous alkylamine derivative^^^ are still unknown.The complex Py,HgCI is less stable losing pyridine at room temperature. The Hg-N distance (2.60 A) is considerably larger than the sum of the covalent radii. There are two short (2.34 A) and two longer (3.25 A) Hg-C1 bonds.*O 76 Brodersen and Riidorff 2. Naturforsch. 1954 9b 164; Brodersen Acta Cryst. 77 Brodersen Chem. Ber. 1957 90 2703; Brodersen und Kunkel Z. anorg. Chem. 78 Bijvoet and MacGillvary Chem. Weeliblad 1935,32,346; MacGillvary and Bijvoet 79 Hofmann and Marburg Annalen 1899 3Q5 196. 1955 8 723. 1959 298 34. Z . Krist. 1936 A 94 231. GrdeniC and Krstanovic Arhiv Kem. 1955 27 143. 322 QUARTERLY REVIEWS Amidomercuric chloride and bromide are actually salts of a polymeric mercuriammonium ion (Figure 8b). The bromide is cubic when pure,*’ otherwise it is orthorhombic as is the chloride.82 Dimercurihydrazonium chloride Hg,(N2H2)CI2 has a layer structure with a trans-conformation and its formation may be interpreted as re- sulting from dehydrogenation of aminomercuric chloride :83 An analogous chain structure has been found in ethylenediamine- mercury(I1) chloride.84 In ‘‘hidomercury bromide” Hg,NHBr, layers of HgBr,- ions alternate with [Hg3(NH)2]a2a+ layers which also contain bromide ions.76 The compound is therefore written wg,(NH),] a2a+ (Br-)a (HgBr3-),.The cation layer has the same structure as (Hg,0,)a2a+ (Figure 9) oxygen atoms being replaced by NH groups. There are two kinds of effective mercury co-ordination in the structure octahedral and trigonal bipyramidal. Steric conditions do not allow the accommodation of an anion other than bromide.Millon’s base and its salts are known in cubics5” and h e x a g ~ n a l ~ ~ b ~ ~ * modifications. In the cubic form only the nitrate is stable whereas the hydroxide and other salts are stable in the hexagonal modification.86 The cubic modification has a cristobalite structure and the hexagonal form a tridimite type of structure in which silicon is replaced by nitrogen and oxygen by mercury. The NHg2 framework so formed is a polymeric cation (NHg,) aa+ having interstitial holes sufficiently large to accom- modate different anions. A regular tetrahedron of mercury atoms sur- rounds each nitrogen (Hg-N = 2.04-2.09 A depending on the anion). Additional close approaches give mercury octahedral effective co-ordina- tion. The structure has been confirmed by infrared spectroscopys7 and anion-exchange experiments.88 The fluoride HgNH,F is a double fluoride of the ammonium ion and Millon’s basesg [(NHg,)F,NHpF].The interstitial space is sufficiently large to accommodate ammonium as well as fluoride ions. The hydrated acid fluorides9 should probably be formulated (NHg,)F,(H,O)F while the nitrate sulphate phosphate and Rudorff and Brodersen Z . anorg. Chem. 1952 270 145; Brodersen and Rudorff Z. anorg. Chem. 1954 275 141. Lipscomb Acta Cryst. 1951 4 266; Nijsen and Lipscomb Acta Cryst. 1952 5 604. 83 Brodersen Z . anorg. Chenz. 1956 285 5; Brodersen Z. anorg. Chem. 1957 290 24. 84 Brodersen 2. anorg. Chem. 1959 298 142. 85 (a) Lipscomb Acta Cryst. 1951 4 156; (b) Nijsen and Lipscomb Acta Cryst. 86 Brodersen and Rudorff Angew. Chem. 1952 64 617; Rudorff and Brodersen Brodersen and Becher Chem.Ber. 1956,89,1487. Arora Lipscomb and Sneed J. Amer. Chem. SOC. 1951 73 1015; Seeger and Brodersen and Rudorff Z . anorg. Chem. 1956 287 24. 1954 7 103. Z. anorg. Chem. 1953,274 323. Pualuan Bol. SOC. Chilena Quim. 1962 12 25. GRDENI~ THE STRUCTURAL CHEMISTRY OF MERCURY 323 arsenate should have a similar structural framework.g0 The structure of the degradation product of Millon’s base has been studied. 91 The structures of a number of phosphine and arsine derivatives of mercuric salts have also been in~estigated.~~ In PbHgP14 mercury is surrounded by four phosphorus atoms in a slightly deformed tetrahedrong3 with Hg-P = 2.52 A. (f) Compounds with Mercury-Carbon Bonding.-Potassium tetracyano- mercurate(rI) which is isomorphous to the analogous zinc was the first mercury compound to be examined by X-ray methods but the exact structure is still not known.Mercuric cyanide is molecular rather than ionic.95 The full structure was determined by Zhdanov and S h ~ g a r n ~ ~ and neutron diffraction has subsequently been used to obtain the light- atom positions acc~rately.~~ The Hg(CN)2 molecules which are linear in solution or vapour are bent in the crystal lattice because of Hg-N inter- actions. The effective co-ordination is distorted octahedral (Figure 9 the C-Hg-C bond angle being 171 O and the Hg-C and Hg-N bond lengths 1.986 and 2-70 A respectively. The former is significantly shorter than the sum of the digonal covalent radii hence the Hg-C link may have some double-bond character. Mercuric oxycyanide HgO,Hg(CN), is actually cyanomercuric oxide (NCHg) ,O.989 99 Mercury has digonal characteristic co-ordination (Hg-C= 1.97 A) but the two mercury atoms in each (NCHg),O unit have different effective co-ordination numbers viz. tetrahedral (C 3 0 ) and octahedral (C 0,4N). The closest mercury-to- oxygen approach between two molecules 2-53 I$ is exceptionally short for an intermolecular contact (cf. Hg-0 bond length of 2.04 A) but this does not influence the Hg-C length. However in the structure of KI,Hg(CN) the close approach of four iodine atoms (3.383 A) causes a slight lengthening of the Hg-C bondsloo (2.079 A). Comparatively little is known about the crystal structures of R2Hg organomercurials. Both phenyl radicals of diphenylmercury are co- planar101,102 in the cLystal structure but this seems less likely in solu- t i ~ n .l ~ - ~ l All cyclic organomercurials so far investigated contain essentially linear C-Hg-C bonds. Thus “o-phenylenemercury” which was considered Airoldi Ann. chim. (Rome) 1958 48 491. Weber Naturwiss. 1957 44 465. 82 Evans Mann Peiser and Purdie J. 1940 1209. 93 Krebs and Ludwig 2. anorg. Chem. 1958,294 257. B4 Dickinson J. Amer. Chem. Soc. 1922 44 774. B5 Hassel 2. Krist. 1926 64 218. B6 Zhdanov and Shugam Zhur. jiz. Khim. 1945 19,433. B7 Hvoslef Acta Chem. Scand. 1958 12 1568. B9 Weiss and Hofmann 2. Naturforsch. 1960 156 679. SCavniEar Acta Cryst. 1960 13 A 57; SCavniEar 2. Krist. 1963 118 248. loo Kruse Acta Cryst. 1963 16 105. lol Kitaigorodskii and GrdeniC Izvest. Akad. Nauk S.S.S.R. Odtel. Khim. Nauk. lo2 Ziolkowska Roczniki Chem. 1962,36,1314; Ziolkowska Myasnikova and Kitai- 1948 262.gorodskii Zhur. strukt. Khim. 1964 5 737. 324 QUARTERLY REVIEWS to be 9,lO-dimercura-anthracene is actually trimeric,lo3 the molecule containing six linear C-Hg-C bonds.lo4 “Pentamethylenemercury” for which a structure having a ring of five carbons and one mercury atom has been suggested,lo5 is actually dimeric,lo6 but is not isostructural with mercury diethylene oxide.lo7 The latter has a twelve-membered ring in which the C-Hg-C bonds are slightly bent (I 76”) due to mercury-oxygen interaction (Hg - 0 = 2-21 A). “2,2-Biphenylenernercury” is not 9- mercurafluorene but is a cyclic tetramer.lO* The crystal structures of mixed organomercurials RHgX (X is C1 Br or I) are characterised by linear C-Hg-X bonds and octahedral effective co-ordination.In alkyl derivatives the structure is dictated by a tendency for parallel alignment of CHgX groups.log There is disagreement between the Hg-Cl distance for MeHgCl as obtained by X-ray methods (solid statc)log and microwave spectroscopy (vapour stafe)llo (2.50 and 2.282 A respectively). Such a difference between the bonding in the two states seems improbable. The X-ray value is doubtful since it was obtained from two-dimensional Fourier synthesis from a small number of reflexions. Thns a three-dimensional Fourier synthesis of the electron density is desirable. For methylmercuric bromide the difference is much smaller [250 A (X-ray) 2-406 A (microwave)]. Chlorovinylmercuric bromide has a similar structurelll with an Hg-Br length of 2-43 A. The available data suggests that the Hg-C bond is longer in aryl- than in alkyl-mercuric derivatives and thus may be more ionic in the former.For crystalline aryl derivatives the dependence of the Hg-C bond length on the nature and position of substituents in the aromatic ring has yet to be determined. A cornmon property of most known organo- mercurials is a low ability to accept further ligands. Thus in crystals of mixed mercurials the additional Hg - - - X contacts are only slightly less than the sum of the van der Waals radii. However perfluoroalkylmer- curials112 and perfluoroarylmer~urials~~~ do give co-ordination derivatives. Thus Hg(CF,) forms the complex anions,l12 Hg(CF3)2X- and Hg(CF,),X;- (X is halogen) while Hg(C6F,) forms bi~yHg(C~F~)~.ll~ (8) Mercurous Compounds.-Derivatives containing mercury bonded to halogens oxygen and nitrogen are known.The compounds are of particu- lar interest because of the presence of metal-metal bonding. Although mercury has a formal oxidation state of one the structures are based on lo3 Wittig and Bickelhaupt Chern. Ber. 1958,91 883. lo4 GrdeniC Chem. Ber. 1959 92 231. lo5 Hilpert and Griittner Ber. 1914 47 186. lo6 GrdeniC and GoriEan unpublished molecular weight determination. lo’ GrdeniC Actu Cryst. 1952 5 367. lo* Wittig and Lehmann Chem. Ber. 1951 90 876. loQ Grdenic and Kitaigorodskii Zhur. fiz. Khim. 1949 23 1162. Gordy and Sheridan J. Chem. Phys. 1954 22 92. 111 Kitaigorodskii Izvest. Akad. Nauk. S.S.S.R. Otdel. khim. Nauk 1947 259. Lagowski Quart. Rev. 1959 13 233 and references therein. I J 3 Chambers Coates Livingstone and Musgrave J.1962 4367. GRDENIk THE STRUCTURAL CHEMISTRY OF MERCURY 325 digonally co-ordinated mercury as in mercuric compoufids. The presence of linear XHg-HgX bonds in mercurous halides has been long estab- li~hed,~* but less is known about the structures of other derivatives. The most important interatomic distances in the structures of the mercurous halides are given in Table 6. The Hg-Hg bond distance increases from fluoride to iodide although the opposite is expected because of the probable higher ionic character of the Hg-F bond. This behaviour can be attributed to the repulsion between the electron clouds of the ligand and the secondary mercury atoms i.e. by a trans-influence which therefore weakens the Hg-Hg bond more in the iodide than in the fluoride. However the Hg-Hg bond distance of 2-90 8 found in mercurous diacetylhydra- zide,l14 Hg2N2(COCH3)2 cannot be explained in this way as the nitrogen electron cloud is unlikely to exert such a strong repulsive influence and this behaviour is not yet understood.TABLE 6. Interatomic distances in crystal structure of mercurous halides Hg2F2 Hg2CI2 Hg2Br2 Hg212 Hg-X 2-31' 2.52'; 2 ~ 4 1 ~ 2-53'; 2.45b 2-68' Hg-Hg 2-43' 2.53O; 2 ~ 4 5 ~ 2.W; 2.5@ 2.69" x . - * x along the (c-axis direction) no contact 3-33"; 3 ~ 7 0 ~ 3.4OU; 3 - S b 3-55" Krisf,. Akad. Nnuk S.S.S. R. 1949,5,57. CGrdeniC and DjordjeviC J. 1956 1316. OHavighurst J. Amer. Chem. Soc. 1926,48 21 13. bBelov and Mokeeva Trudy Inst. Ct 2.53 I Cl a FIG. 10. The environment of mercury (a) in mercrtroxs chloridez8 and (b) in merciirous oxychloride Hg6CI,0 (mineral eglestonite116).The occurrence of tlie oxonium ion [H20.Hg.Hg-OH2]2+ in the struc- ture of mercurous nitrate dihydrate16 shows that the mercurous ion is a strong electron acceptor along the metal-metal bond axis. The Hg-0 distances are close to the SUM of the covalent radii. Further examples of 114 Brodersen and Kunkel Chem. Ber. 1958 91,2698. 326 QUARTERLY REVIEWS the Hg-Hg-to-oxygen relationship are found in the structures of mercurous oxychlorides. The X-ray diffraction pattern of the mineral eglestonite is best interpreted by the formula Hg6C1,0.115 The effective co-ordination is similar to that of mercurous chloride (Figure lo). Two chlorine atoms are replaced by one oxygen atom and the Hg-Hg groups are reorientated. The oxygen is quite loosely accommodated in an octahedral hole at the centre of the cube the Hg-0 distance being 2.80 A which is more than the great- est Hg-0 distance observed at all (2.66 A in the addition compound of mercuric chloride and 1,4-dioxan116).The Hg-Hg distance is 2.41 A. Oxymercuric mercurous chloride,l17 2HgO,Hg2Cl2 consists of slightly de- formed mercurous chloride molecules and mercuric oxide chains with rather close intermolecular approaches. The dipole moment of mercurous trichloroacetate 2.65 D excludes a rigid chelate structure.l18 The nature of “alkylmercurous” compounds (RHg) is not yet fully understood. To explain their high electrical conductivity it has been sug- gested that they are organic Recent results exclude their formula- tion as organomercurous compounds RHgSHgR or as amalgams between mercury and organic free radicals.120 (h) Amalgams.-In a comprehensive discussion of the structural chemistry of mercury the amalgams should not be entirely omitted.As alloys they have to be treated from the standpoint of the theory of metals and their structural inter-relations in terms of the crystal chemistry of intermetallic phases. Most of them have a peculiar structure in which some mercury atoms are grouped more closely than others. Characteristic arrangements are triangles and pentagons in Mn2Hg,,121 isolated squares in Na3Hg2,122 or squares shared in puckered ribbons in NaHg.122 6. General Concepts of the Stereochemistry of Mercury There are two main features in the stereochemistry of mercury which have to be explained (1) reduction of the characteristic co-ordination number from six or four (as with zinc and cadmium) to two and (2) additional ligand approaches leading to distorted effective co-ordination generally distorted octahedral.An explanation of both phenomena has been given by Orgel.8,g In the third transition series d-s mixing is more energetically favourable for d10 ions than in the first or second series. Hybridisation of the CE, and s orbitals gives two new orbitals and depending on which of these is filled lib Hedlik Tschermaks mineralog. u. petrolog. Mitt. 1949 1 378. 11’ SCavniEar Acta Cryst. 1956 9 956. 11* Davidson and Sutton J. 1942 565. 119 Coates Quart. Rev. 1950 4 217. lao Gowenlock and Trotman J. 1957 21 14; Gowenlock Pritchard Jones and lZ1 de Wet Acta Cryst. 1961 14 733. la2 Nielsen and Baenziger Acta Cryst. 1954 7 277; Duwel and Baenziger Acta Hassel and Hvoslef Acta Chem.Scand. 1954 8 1953. Ovenall J. 1958 535. Cryst. 1955 8 705; 1960 13 476. GRDENI~ THE STRUCTURAL CHEMISTRY OF MERCURY 327 mercury can either form two short strong bonds in the z direction and four weak ones in the xy plane or four strong ones in the xy plane and two weak in the z direction. The former arrangement is preferred though no explanation is yet available. The tendency for digonal co-ordination can also be explainedlZ3 by the fact that mercury has a larger s-p separation than cadmium or zinc which makes sp3 or sp2 hybridisation more difficult for mercury hence favouring sp hybridisation. Nyh01ml~~ has shown that while the d10 -+ d9s separation favours the third series when the charge on the metal is high the order is not maintained when the charge is low.However in the latter case the dlOs -+ dl0p separation in the third series is high favouring sp hybridisation. For distorted octahedral mercury complexes based on digonal character- istic co-ordination the residual charge on mercury in the linear L2Hg skeleton must be reasonably high or further co-ordination would not occur. Thus it is valid to use ds hybridisation to explain the stereochemistry. However the low acceptor properties of organomercurials show that in these compounds the charge on mercury is low and sp hydridisation probably predominates. Added in Proof.-Some structural data have been obtained recently confirming the main structural principles outlined above. 5(b). It has been established by X-ray single-crystal and powder methods that the alkali metal oxomercurates(r1) are isostr~ctural.~~~ The Hg02- ion is linear with an Hg-0 bond length of 1-95 8,.There are no other mercury-to-oxygen contacts in the structure and the effective co-ordination of mercury is octahedral if four equidistant mercury atoms at 3.42 8 are included. An interesting oxygen bridge structure has been found for the dimeric 1 1 tri(pheny1arsine oxide) adduct with mercuric chloride.lZs The HgCl molecule is as strongly deformed as in the 2:l adduct.lZ4 In the 1 1 addition product of mercuric chloride with cyclohexane-1 ,4-dionelZ7 the Cl-Hg-Cl bond angle is 173" and the Hg-Cl and Hg * * - - - 0 distances are 2.30 and 2.79 8 respectively. The effective co-ordination of the mercury atom is a deformed octahedron. Remarkable refinements of the structure of trischloromercurioxonium chloride have been made by neutron diffraction12* as well as by three- dimensional X-ray diffraction The oxygen atom is displaced out of the plane of mercury atoms by 0-43&0.04 A so that the oxonium ion has the shape of a very flat pyramid with a vertex angle of 175.5".S(c). In anhydrous mercuric s ~ l p h a t e l ~ ~ the mercury atom is co- ordinated by four oxygen atoms at distances 2-12 (two atoms) 2.29 and lZ3 Nyholm Proc. Chem. SOC. 1961 273. lS4 Branden Acta Chem. Scand. 1963 17 1363. lz6 Hoppe and Rohrborn 2. anorg. Chem. 1964,329 110. lZ6 Branden Arkiv kemi. 1964 37 485. 12' Groth and Hassel Acta Chem. Scand. 1964 18 1327. lZ8 Aurivilius Arkiv kemi 1964 22 517. 12s Aurivilius Arkiv kemi 1964 22 537. 130 Kokkoros and Rentzeperis 2. Krist.1963 119 234. 328 QUARTERLY REVIEWS 2.43 A in a deformed tetrahedron the bond angle corresponding to the two shortest Hg-0 approaches being 157.4". 5(d). Interesting structural differences have been found among the mercury mercaptides. In t-butylmercury mercaptide131 mercury has tetrahedral co-ordination in a polymeric structure with Hg-S bond lengths to 2.59 and 2-66 A. The ethyl derivatives according to the authors' un- published results has a simple molecular structure. The adducts of mercuric chloride with diethyl ~ulphidel~~ and with tetrahydr~thiophenl~~ are sul- phonium salts i.e. (Et ,SHgCI)+Cl- and (C,H,SHgCI)+Cl-. The character- istic mercury co-ordination in both structures is digonal with a departure from linearity shown by angles of 158" in the former and 143 O in the latter compound.The effective co-ordinations are octahedral and trigonal- bipyramidal respectively. The Hg-S bonds (2.41 and 2.40 A) are shortened in comparison with those given in Table 2. In the crystal structure of the clathrate CoHg,(SCN),,C,H mercury again has tetrahedral ~o-ordinationl~~ but it is deformed in such a way that a side of the benzene molecule approaches the mercury atom with C * * * * - Hg distances of 3.52 and 3-66 A. Co(NCS),-octahedra are joined by mercury-sulphur bridges the benzene molecules being accommodated in holes. In the 1,6-dithiacyclodeca-cis-3,cis-8-diene adduct (1 :2) one molecule of mercuric chloride is linear and the second is deformed.135 In the bis(thiourea) coniplex a cation (thiourea),HgCl with mercury in the planar trigonal characteristic co-ordination has been found the effective co-ordination having not been r e ~ 0 r t e d .l ~ ~ The adduct with phen~xathiinl~~ has the mercury atom in an effective octahedral co-ordina- tion with the structural pattern similar to that found in bispyridine adduct.80 The preliminary structure investigation of two mercury thiourea com- plexes has also been r e ~ 0 r t e d . l ~ ~ 5(f) The crystal of di-p-tolylmercury contains planar centrosymmetrical molecilles so that the high dipole moment in solution must be due either to atom polarisation or to s ~ l v a t i o n . ~ ~ ~ The author thanks Professor R. S. Nyholm F.R.S. for suggesting this review 131 Kunchur Nature 1964 204 468. 132 Branden Arkiv kemi 1964 22 83. 133 Branden Arkiv kemi 1964,22,495 and p. 561. 134 Grnrnbaek and Dunitz Helv. Chim. Acta 1964 47 1889. la6 Cheung McEwen and Sim Nature 1965,205 383. 136 Korczynski Roczniki Chern. 1963 37 1645; 1963 37 1647. 13' Kunchur and Mathew Proc. Chem. Soc. 1964 414. and for stimulating discussion.

ISSN:0009-2681    

DOI:10.1039/QR9651900303

出版商:RSC    

年代:1965

数据来源: RSC