摘要:

QUARTERLY REVIEWS INTERRALOGEN COMPOUNDS AND POLYHALIDES By A. G. SHARPE M.A. -.D. (FELLOW OF JESUS COLLEGE CAMBRIDGE AND UNIVERSITY DEMONSTRATOR IN CHEMISTRY) THE existence of compounds formed by union of the halogens with one another has been recognised for more than a hundred years and eleven such substances are now known. No ternary compound is yet kn0wn.l These are tabulated below. M.p. . B.p. . M.p. . B.p. . M.p. . B.p. . M.p. . B.p. . M.p. . B.p. . Type AX. CIF - 156" - 100 BrF - 33 + 20 BrCl - 66 + 5 ICI tc + 27.2 + 13.9 + 97-4 IBr + 36 + 116 Type AX1. CLF - 83" + 12 BrF + 9 + 127 ICl + 101 (decomp.) Type AX&. BrF - 61" + 40 IF + 8 + 97 Type AX,. IF + 5" (2 stm.) (v.P. = 1 atm.) + 4 All the interhalogen compounds may be (and almost invariably are) obtained by direct combination of the elements.The chlorides of iodine were discovered by Gay Lussac and by Davy in 1814. Balard first made iodine bromide in 1826 ; he also noticed that when bromine is mixed with chlorine the intensity of the colour is much diminished but it was not until a century later that spectroscopic evidence established beyond doubt the existence of bromine chloride.2 Iodine pentafluoride may have been obtained by G. Gore in 1871 but its preparation from iodine and fluorine and its properties were first described by H. Moissan 4 in 1902. and E. B. R. Prideaux * independently discopered bromine trifluoride in 1905 and observed its great reactivity. The remaining compounds were first prepared by 0. Ruff and his collaborators at Breslau between 1925 1 A. N. Campbell and L. W. Shemilt Trans.Roy. SOC. Canada 1946 40 111 17. 2 For the history of bromine chloride see ref. (12). 3 Phil. Mag. 1871 41 309. P. Lebeau 4 Compt. rend. 1902 135 563. Ilbid. 1905 141 1015. J. 1906 89 316. 115 116 QUARTERLY REVIEWS and 1933. Chlorine monofluoride and trifluoride were made from chlorine and fluorine at 250" ; the unstable bromine monofluoride 9 was prepared from the elements a t 10" ; and bromine pentafluoride lo and iodine hepta- fluoride l1 were obtained by fluorination of bromine trifluoride a t 200" and iodine pentafluoride a t 280" respectively. The most recent general review of the interhalogen compounds is by N. V. Sidgwick; l2 the halogen fluorides however have been the subject of a separate review by H. S. Booth and J. T. Pinkston,l3 who have assembled nearly all the relevant information published before 1947.This consists mainly of methods of preparation and descriptions of physical properties such as melting and boiling points vapour pressures and den- sities ; 13a the sections of the review which deal with chemical properties emphasise how much is known about the reactivity of these compounds and how little about their reactions. Recent work on the chlorine fluorides bromine trifluoride and iodine pentafluoride has done something to remedy this defect and it is with these compounds that this Review is most concerned. Investigation of the chemistry of bromine trifluoride has shown however how closely the chemistry of the interhalogen compounds is related to that of the poly- halides. These substances hsve not been reviewed for many years; and this Review therefore deals in turn with aspects of current interest of the chemistry of iodine chlorides and bromide polyhalides (up to 1947) recent work on halogen fluorides and related substances the thermochemistry of the interhalogen compounds the uses of these substances in organic chemistry and finally the stereochemistry of the interhalogen ions and molecules.The Iodine Chlorides and Iodine Bromide.-Relatively little work on the older interhalogen compounds has been reported during the last twenty years. Convenient methods for the preparation of iodine monochloride and trichloride have been given by J. Cornog and R. A. Karges l* and by H. S. Booth and VV. C. Morris,15 respectively. The monochloride has a Trouton constant of 2 7 ~ 5 . l ~ Its polymorphism 17 l8 remains unexplained no investigation of the structure of the solid substance ever having been made.Reactions of iodine monochloride with a number of salts have been studied l9 the chlorides of potassium ammonium rubidium and caesium 7 0. Ruff and E. Ascher 2. u w g . Chem. 1928 176 258. 8 0. Ruff and H. Krug J. 1930 190 270. 9 0. Ruff and A. Braida ibid. 1933 214 81. 10 0. Ruff and W. Menzel ibid. 1931 202 49. 11 0. Ruff and R. Keim ibid. 1930 193 176. 12 Ann. Reports 1933 30 128. 134 A. A. Banks and A. J. Rudge ( J 1950 191) have given further physical proper- 1* " Inorganic Syntheses " Vol. I 1939 p. 165. 16 J. Cornog and R. A. Karges J . Arner. Chem. SOC. 1932 54 1886. 17 W. Stortenbecker 2. physikal. Ghrn. 1889 3 11. 18 J. R. Partington " General and Inorganic Chemistry " 1946 p. 816.19 J. Cornog H. W. Horrabin and R. A. Karges J . Amer. Chem. Soc. 1938 60 l3 Chem. Reviews 1947 41 421. ties of ClF,. l5 Ibid. p. 167. 429. SHARPE INTERHALOGEN COMPOUNDS AND POLYHALIDES 117 are soluble in the molten substance forming polyhalides whilst those of lithium sodium silver and barium are only very slightly soluble ; cyanates and thiocyanates react to form iodine tricyanate I(CNO), and iodine tri- thiocyanate I( CNS),. The chemistry of such halogen-pseudohalogen com- pounds and indeed that of the pseudohalogens themselves has been little investigated and deserves further attention. 2O Fused iodine monochloride was found by Faraday and many later inves- tigators to conduct electricity and it is now certain that the conductivity which approaches that of some fused salts is not due to impurities.The most recent values for the specific conductivity are 4.58 x ohm-l cm.-l and 4.52 x 10-30hm-lcm.-1 (both at 35") by Cornog and Karges,lG and by Y. A. Fialkov and 0. I. Shor.21 Fialkov and Shor also studied the variation of specific conductivity with temperature ; they found an extended maximum in the range 50-70° and above that temperature a steady decrease which they attributed to thermal dissociation. H. J. Emeldus and N. N. Greenwood 22 have confirmed their results and noted in addition that there is no decomposition potential. The electrolysis of the fused substance however still awaits detailed investigation. On electrolysis of iodine monochloride in nitrobenzene 23 or acetic acid 24 both iodine and chlorine appear a t the anode; their determination by the use of silver electrodes indicates that about 1.5 equivalents of chlorine and less than one equivalent of iodine per Faraday are liberated a t the anode and seems to indicate dissociation according to the equation 2IC1 = I+ + IC1,-.Electrolytic studies 23 on the system IC1-AlC13-Ph*N02 have indicated the existence of a compound IAICI,. F. Fairbrother 25 found the dipole moments of iodine monochloride in carbon tetrachloride and cyclohexane solution (1.49 and 1.47 D. respectively) to be about twice the dipole moment in the vapour phase and suggested that in solvents of higher dielectric constant free ions should be formed; his suggestion thus appears to be correct though the chloride ion is probably solvated yielding (IC12)-. Few data for the conductivity of iodine trichloride are available.Solu- tions of the compound in nitrobenzene,26 acetic and bromine 28 have been shown to conduct and fused iodine trichloride in a closed vessel is a good conductor,22 the specific conductivity a t the melting point (101") being 8.4 x 10-3 ohm-1 cm.-l-nearly twice the value for iodine mono- chloride at 35". No decom- position potential has been found but for iodine trichloride (unlike the monochloride and the bromide) the conductivity of the solid a t the melting point is of the same order of magnitude as that of the liquid. The specific conductivity of iodine bromide 29 is 6.4 x ohm-l cm.-l 2o For a review see P. Walden and L. F. Audrieth Chem. Reviews 1928 5 339. 21 J . Gen. Chem. Russia 1948 18 14. 23Y. A. Fialkov and K. Y. Kaganskaya J . Gen. Chem. Russia 1948 18 289.2 p C. Sandonnini and N. Borghello Atti R. Acead. Lincei 1937 25 46. 26 J. 1936 847. 27 R. P. Bruns ibid. 1925 118 89. 28 W. A. Plotnikov Chem. and Ind. 1923 42 750. 29 Y. A. Fialkov and N. I. Goldman J . Qen. Chem. Russia 1941 11 910. It increases up to 111" and then decreases. 22 J . 1950 987. 26 V. Finkelstein 2. physikal. Chem. 1925 115 303. 118 QUARTERLY REWElWS at 45" and has a maximum value a t 65" No recent measurements on the conductivity of iodine bromide solutions have been reported though solu- tions in nitrobenzene liquid sulphur dioxide and arsenic trichloride are known to be conductors.26 has attracted some atten- tion. The thermal reaction at temperatures between 205" and 240" was followed iodometrically by W. D. Bonner W. L. Gore and D. M. Yost,30 who obtained a value of 33.9 kcals.for the activation energy of the slower reaction in a suggested mechanism Hz + IC1 = HI + HCl (slow) . * (1) HI + IC1 = HC1 + I (fast) . * (2) Activation energies for these reactions were calculated by A. Sherman and N. Li s1 to be 39 and 41 kcals. respectively. This would make (2) the slower reaction and Sherman and Li therefore suggested an alternative mechanism of lower overall activation energy to replace reaction (2). The hydrogen-iodine monochloride photochemical reaction has also been studied ; 32 the primary process is dissociation of the interhalogen com- pound into iodine and chlorine atoms. Polyhalides.-The solubility of iodine in potassium iodide solution and the formation of compounds between the alkali-metal halides and the chlorides and bromide of iodine attracted early attention to the poly- halides and there is %I considerable nineteenth and early twentieth century literature on them.H. W. Cremer and D. R. Duncan made a careful and thorough re-examination of the methods of preparation,33 physical proper- ties,34 and reactions in solution 35 and in the solid state,36 of the poly- halides; and their systematic survey summarises most of the work done before 1931 and adds a considerable amount of fresh material. The general method for the preparation of polyhalides is direct combina- tion between metallic halide and interhalogen compound (or for polyiodides iodine). This may be brought about by mixing solutions of the reactants in a suitable solvent by exposing the halide to the vapour of the inter- halogen compound or by producing the interhalogen compound and effect- ing combination in situ.Cremer and Duncan described the preparation of compounds of the types MIBr, MIClBr MTCI, and MIC14 and confirmed the existence of the c d u m polyhalides CsBr, &I3 CsBr2C1 CsBrCI, and Cs1,Br. Indications of the existence of polyhalides of sodium were obtained but none was prepared pure and anhydrous. Methods for the determina- tion of the dissociation pressures of the polyhalides (into the halide con- taining the most electronegative halogen present and an interhalogen com- pound) were reviewed critically and the general order of magnitude of the dissociation pressures for series of polyhalides containing the ions Is- I&,- IBrCl- and IC12- was established as Na > K > NH4 > Rb > Cs.A casium polyhalide containing fluorine CsIBrF was prepared. 34 Poly- The reaction + 2IC4,) = 2HCl, + 30 J . Amer. Chem. Soc. 1935 57 2723. 32 L. J. E. Hofer and E. 0. Wiig ibid. 1945 67 1441. 34 Ibid. p. 2243. 31 Ibid. 1936 58 690. 33 J. 1931 1857. 36 J. 1933 181. 35 J. 1932 2031. SHARPE INTERZZALOQEN COMPOUNDS AND POLYHALIDES 119 halides of large organic cations were described and those of symmetrical cations were found to be the most stable.3* Polyhalides are like typical ionic compounds insoluble in liquids of low dielectric constant (such as carbon tetrachloride) but are decomposed by them to an extent which depends on the dissociation pressure of the compound under consideration. An example of the importance of the dis- sociation pressure in influencing reactivity is provided by the reactions of the dibromoiodides (dibromohypoiodites) with ammonia those which have a dissociation pressure greater than about 0.005 mm.react (via the inter- halogen compound) to yield nitrogen tri-iodide ; those of lower dissociation pressure merely form addition products.3B Potassium rubidium and caesium compounds of the type MIC13F have been made from the fluorides and iodine trichloride; they exhibit the usual mode of thermal decomposition and order of stability.37 The exis- tence of a compound of empirical formula CsI was established in a study s8 of the system CsI-1,-H,O ; this substance was later found 3& to be dia- magnetic and was therefore allotted the formula Cs,I,. The isomorphism of the thallium compound TlI and the tri-iodides of rubidium and caesium 4O suggests that TlI is a thallous compound though from a study of its reactions in solution the formulation as thallic iodide is said to be sup- ported.Its preparation from thallous iodide and iodine however shows it to be a thallous compound since it is clear from consideration of the appropriate standard redox potentials that iodine will not oxidise thallous thallium to the tervalent state. The relation between thallous tri-iodide and the iodothallates such as ]KITH, which it forms with solutions of iodides requires further study. The anhydrous acids from which the polyhalides are derived have not been prepared but HIC1,,4H20 has been obtained by passing chlorine through a suspension of iodine in concentrated hydrochloric acid,42 and conductivity and transference experiments 43 on solutions of iodine mono- chloride in hydrochloric acid prove that the interhalogen compound is present as (H30)+(1C12)-.This accords with the well-known fact that a t the end of an Andrews titration (i.e. by iodate in presence of ca. 5~-hydro- chloric acid) the iodine monochloride is in the aqueous layer. The use of an organic liquid as indicator in this titration may be avoided incidentally by employing certain organic dyestuffs (such as Brilliant Ponceau 5R) as irreversible redox indicators oxidised by the (10,)- but not by the (IC12)- The structures of the polyhalide anions caused much speculation until 37 H. S. Booth C. F. Swinehart and VV. C . Morris J . Arner. Chern. Soc. 1932 38 T. R. Briggs and S. S. Hubard J . Physical Chern. 1941 45 806. 39 S. S. Hubard ibid.1942 46 227. 40 H. L. Wells and S . L. Penfield 2. anorg. Chern. 1894 6 312. *lA. J. Berry T. M. Lowry and R. R. Goldstein J . 1928 1748. 4 2 V. Caglioti Atti R. Accad. Limei 1929 9 563. 43 J. H. Fad1 and S. Baeckstrom J . Amer. Chem. Soc. 1932 54 620. 4 4 G. F. Smith and C. S. Wilcox Id. Eng. Chem. (Anal.) 1942 14 49. ion. 44 54 2561. 120 QUARTERLY REVIEWS X-ray studies of CsICI by R. W. G. W y ~ k o f f ~ ~ of CsI and CsIBr by R. M. Bozorth and L. P a ~ l i n g ~ ~ and especially of (NH4)13,47 (NH4)IC1Br,4* KICl,,49 (NMe,)IC12,50 (NMe4)I,,51 and other p0lyiodides,~2 by R. C. L Mooney led to the general conclusion that in such anions the heaviest halogen is multivalent and the others are distributed symmetrically around it. The ions containing three atoms are linear and the (IC1,)- ion is planar ; in all of them the interatomic distances are very nearly equal to the sums of the single-bond covalent radii of the appropriate halogens.In a series of polyiodides of organic bases the crystal symmetry varies according to the size of the cation and the stability increases with increas- ing cation size.52 The relation between the electron configuration and the stereochemistry of these ions is discussed below. The Halogen Fluorides.-The preparation of chlorine trifluoride on a semi-technical scale has been described by H. R. W. K~asnik,~4 and R. W. Porter.55 Leech's description of a German process using nickel reaction vessels may be taken as typical chlorine monofluoride is first formed (at 200') and is converted into the trifluoride at 280" ; this is con- densed and stored in steel vessels.Kwasnik 54 has also described the preparation of bromine trifluoride from bromine and fluorine at 80-100" in iron apparatus whilst A. G. Sharpe and H. J. EmelBus 56 have reported a modification of the laboratory method used by 0. Ruff and A. Braida.57 A convenient apparatus for the laboratory preparation of iodine penta- fluoride has also been de~cribed.~8 This compound has been shown 59 to be a product of the reaction between fluorine and iodine pentoxide at 250". Some examples of the uses of halogen fluorides prepared in Germany during the war as fluorinating agents in inorganic chemistry have been given. Thionyl chlorofluoride SOCIF is obtained from thionyl chloride and bromine trifluoride or iodine pentafluoride ; 6o sulphuryl bromofluoride SO,BrF from sulphur dioxide bromine trifluoride and bromine in an autoclave at ordinary temperatures.61 The preparations of carbonyl chloro- fluoride COCIF from carbon monoxide and chlorine monofluoride of carbonyl bromofluoride COBrF from carbon monoxide and bromine tri- fluoride and of carbonyl iodofluoride COIF from carbon monoxide and iodine pentafluoride have also been reported. Sharpe and EmelBus investigated the action of liquid bromine trifluoride on the chlorides bromides and iodides of a number of metals ; the general reaction was found to be fluorination to the highest known fluoride of the 45 J . Amer. Chem. SOC. 1920 42 1100. 47 2. Krist. 1935 90 143. 6o Ibid. 1939 100 519. 5 2 lbid. 1943 64 315. 54 F.I.A.T. Review of German Science (1939-1946) Inorganic Chemistry Pt.I 55 Chem. Eng. 1948 55 No. 4 102. 57 2. anorg. Ch.em, 1932 206 62. 59 G. H. Rohrback and G. H. Cady J . Amer. Chem. Soc. 1948 70 2603. 6o J. Soll ref. (54) p. 192. 61 TT. Kwasnik and co-workers ibid. p. 193. 46 Ibid. 1925 47 1561. 48 Ibid. 1938 98 324. 4g Ibid. p. 377. 51 Physical Rev. 1938 53 851. 53 Quart. Reviews 1949 3 22. 1948 p. 168. 56 J. 1948 2135. R. N. Haszeldine J. 1949 2856. 6 2 Idem ibid. p. 243. SHARPE INTERHALOGEN COMPOUNDS AND POLYHaLIDES 121 element (silver provided an exception) though the formation of reagent- insoluble fluoride films on the surface often prevented the reaction from going to completion. Plumbous thallous and cobaltous halides for example yielded mixtures of lower and higher fluorides. When the pro- duct was volatile (e.g. uranium hexafluoride) or soluble in the reagent (e.g.the fluorides of the alkali metals) conversion into the fluoride was quantitative. Sodium potassium rubidium casium silver(I) and barium fluorides were found to be soluble in bromine trifluoride and removal of the solvent by distillation in vacuo a t room temperature yielded solid polyhalides of a new type. Four such compounds the bromotetrafluorides (tetrafluorobromites) of sodium potassium silver and barium were isolated ; those of rubidium and czesium were found to be relatively unstable. Potassium bromotetrafluoride KBrF, the most closely studied of the new compounds is a white crystalline solid very much less reactive than bromine trifluoride but immediately decomposed by water. It neither reacts with nor dissolves in any of the common organic solvents.Thermal decomposition which is slow at temperatures below 200° yields potassium fluoride and bromine trifluoride. Lack of reactivity thermal stability analogy with KICl, and the combination of two (BrF,) units with the invariably bivalent barium ion indicate the existence of the (BrF4)- ion in these new polyhalides. X-Ray powder photography of KBrF shows it to possess a cubic or pseudocubic unit cell,63 but it is not yet possible to give a full description of the structure. The small dissociation pressures of the alkali-metal bromofluorides have not been measured but their stabilities in wacuo show that amongst these compounds the order of stability is K > Na > Rb > Cs. In this respect therefore and also in the formation of a stable silver polyhalide the bromo- tetrafluorides differ from all other known series of polyhalides ; this is attributed to the relatively small size of the (BrF4)- ion.A structure similar to that of (ICl,)- being assumed the (BrF4)- ion should be a square of side 4.84 A. compared with 6.66 A. for the former ion. A fuller dis- cussion of the relation between ionic size and stability must however await the determination of detailed structures. The preparation of a polyhalide KIF (potassium iodohexafluoride or hexafluoroiodate) from potassium fluoride and iodine pentafluoride has also been described; 64 in its ready hydrolysis by water and its method of thermal decomposition this resembles the bromotetrafluoride and must be formulated K+(IF,)-. Chlorine trifluoride does not yield similar poly- halides,56 and the validity of G.Beck’s claim 65 to have made a compound K,ClF or KF,KClF has been disputed.64 Bromine trifluoride reacts with many uranium compounds converting them quantitatively into the hexafl~oride.~~~ 66 This led to an investiga- tion of its action on other naturally occurring radio-elements and their 6 3 F. J. T. Harris unpublished. 64 H. J. Ernelhus and A. G. Sharpe J. 1949 2206. 6 5 2. anorg. Chem. 1937 235 77. A. A. Banks Thesis Cambridge 1948. 122 QUARTERLY REVIEWS compound^.^^ It was found that only radon accompanies uranium on treatment of radioactive mixtures with bromine trifluoride and this pro- cess therefore provides a new and rapid method for the removal of uranium from U-X (thorium) and U-X (protoactinium) preparations. The electrical conductivities of chlorine trifluoride bromine trifluoride and iodine pentafluoride have been measured by A.A. Banks H. J. Emelhus and A. A. Woolf.68 Values obtained for the specific conductivities (in ohm-1 cm.-l) of these substances are CIF, < a t 0"; BrF, 8.0 x 10-5 a t 25' ; IF, 2 x at 25". The temperature coeBcient of conductivity is negative for bromine trifluoride in the range 15-60' and positive for iodine pentafluoride in the range 1040°". Ohm's law is obeyed by bromine trifluoride but not by iodine pentafluoride. Direct-current electrolysis of the former compound produces no evolution of gas but the liquid round the cathode becomes brown whilst the colour of that round the anode is unchanged. The suggested explanation of this is bromine trifluoride is partly dissociated into (BrF,)+ and (BrF,)- ions ; (BrFz)+ is discharged a t the cathode and disproportionates to pale yellow BrF and brown BrF ; (BrF,)- discharged a t the anode disproportionates giving BrF and colour- less BrF,.The negative temperature coefficient is attributed to thermal instability of the ions this factor outweighing in effect the decreased vis- cosity of the medium. In view of the very great reactivity of bromine trifluoride the precautions taken in making such measurements are especially important. These included the use of all-quartz apparatus the direct dis- tillation of the liquid into the conductivity cell in vucuo and the investi- gation of the possible effects of reaction with the quartz or the presence of bromine on the conductivity. Consistent values were obtained and it may therefore be taken that the observed results were not due to impurities.The reactions of bromine trifluoride with oxides and oxy-salts have been examined by H. J. Emelkus and A. A. W00lf.6~ The degree of completion of the reactions (determined by measuring the oxygen evolution with a Topler pump) varied in the manner described for the halides. From com- pounds of the alkali metals however even when all the oxygen present was liberated bromotetrafluorides were not always formed. Metaphos- phates for example yielded hexafluorophosphates whilst orthophosphates gave mixtures of hexafluorophosphates and brornofluorides in molecular proportions 1 2. In the latter instance the formation of a mixture clearly originates in the 3 1 atomic ratio for metal phosphorus in an ortho- phosphate. Investigation of the action of bromine trifluoride on sodium and potassium salts of the oxy-acids of sulphur showed that the perdisul- phates and pyrosulphates yield only fluorosulphonates that the pyrosul- phites thiosulphates and sulphates yield 1 1 mixtures of fluorosulphonate and bromotetrafluoride and that the hydrosulphites (dithionites) and sul- phites yield only bromofluorides.Here the amount of oxygen available for the formation of (SO,F)- ions is as important as the metal sulphur ratio. 67 H. J. Emelh A. G. Maddock C. A. Miles and A. G. S h q e J. 1948 1991. 68 J. 1949 2861. The mechanism of such reactions is discussed below. gg J. 1950 164. SHARPE INTERHALOGEN COMPOUNDS AND POLYHALIDES 123 The most interesting result of the study of the action of bromine tri- fluoride on oxides however was the isolation of a stable compound of antimony pentafluoride and bromine trifluoride.This substance antimony bromo-octafluoride SbBrF, gave conducting solutions in bromine tri- fluoride ; conductometric titration of such solutions with silver bromotetra- fluoride in bromine trifluoride indicated a minimum conductivity when the SbBrF AgBrF ratio was 1 1 and on removal of the solvent silver fluoroantimonate Ag SbF, remained .7O This experiment provides strong evidence for the existence of an ionic equilibrium ZBrF = (BrF,)+ + (BrF4)- in liquid bromine trifluoride. Adopting the definitions of acid and base given by H. P. Cady and H. M. Elsey,?1 a substance which gives rise to (BrF,)+ ions (the cation characteristic of the solvent) is an acid and a substance which gives rise to (BrF,)- ions a base.The reaction which takes place during the titration is then represented by the equation [BrF2]+[SbF,]- + Ag+[BrF,]- = Ag[SbF,] f 2BrF There is also evidence for the existence of a compound between one mole- cule of stannic fluoride and two molecules of bromine trifluoride. A solution of this substance in bromine trifluoride has been titrated conductometrically with potassium bromotetrafluoride ; the minimum conductivity corresponds to one molecule of the tin compound and two of the bromotetrafluoride and the product of the reaction is potassium fluorostannate K,SnF,. The tin compound therefore reacts in solution as (BrF2),snF6 difluorobromonium fluorostannate. Gold dissolves in warm bromine trifluoride and evaporation of the solution yields a compound of empirical formula AuBrF,; a solution of this sub- stance in bromine trifluoride gives with silver bromotetrafluoride in the Same solvent an immediate precipitate of silver fluoroaurate AgAuF,.Instantaneous reactions in solution are usually reactions between oppositely charged ions for which the activation energies are negligible and this reaction must therefore be represented by the equation [BrF2]+[AuF4]- + Ag+[BrF,]- = Ag+[AuF,]- + 2BrF Decomposition of [BrF,][AuF,] at 180" produces the hitherto unknown auric fluoride the properties of which are described elsewhere.? Further recent work by H. J. Emelbus and V. Gutmann 73 has shown that two more acids of the bromine trifluoride system may be isolated in the solid state. These are compounds formed by bromine trifluoride with niobium and tantalum pentafluorides [BrF,][NbF,] and [BrF,][TaF,] respectively.With bromotetrafluorides they yield hexafluoroniobates and hexafluorotantalates. A bismuth acid [BrF,][BiF,] also exists. For the preparation of complex fluorides by neutralisation reactions in &4 third example of a neutralisation reaction has been described.72 70 A. A. Woolf and H. J. Eme16usy J. 1949 2865. 71 J . Cfiem. Educ. 1928 1425. 7 2 Sharpe J. 1949 2901. 73 J. 1950 1046. 124 QUARTERLY REVIEWS bromine trifluoride it is not however necessary to prepare the acid and the base separately. The action of an excess of bromine trifluoride on a mix- ture of equivalent quantities of gold and silver for example followed by removal of the excess yields nearly pure silver fluoroaurate.72 It therefore seems likely that as suggested by Emelbus and Woolf an ionic reaction mechanism involving the unstable acid [BrF,]+[PF6]- is involved in the conversion of potassium metaphosphate into the hexafluorophosphate.A similar mechanism has been suggested for the preparation of potassium hexafluorovanadate KVF, by treating potassium chloride and vanadium trichloride with bromine trifluoridee7* Some of the new complex ffuorides obtained by the use of bromine trifluoride are unstable even in 40% hydro- ffuoric acid and a useful method for their preparation in a non-aqueous system has therefore been devel0ped.~4~ It should be mentioned however that they are not always obtained in a pure condition and that retention of the elements of bromine trifluoride often occurs to some extent.70172 This has been attributed 78 to solvolysis by bromine trifluoride but experi- mental confirmation of this suggestion is still required.Other solvents for which ionic equilibria not involving the proton have been suggested include dinitrogen tetr0xide,~5 sulphur dioxide,76 and In view of the relationship between the bromotetrafluorides and bromine trifluoride it seems very likely that some a t least of the other polyhalides may be bases in as yet unknown acid-base systems in other interhalogen compounds ; and the extension of the work described here to these substances appears to be a promising field for future investigation. Thennochemistry of the Interhalogen Compounds.-The thermodynamic properties of bromine chl0ride,7~ iodine mono~hloride,7~ iodine trichloride,80 and iodine bromide 81 have been measured by Yost and his collaborators who have also given data for iodine monochloride iodine bromide and bromine chloride in carbon tetrachloride solution.82 For iodine mono- chloride and -bromide their results are in agreement with those obtained spectroscopically by other 84 The thermochemistry of the chlorine fluorides has recently received considerabfe attention and the heat of formation of the monofluoride is 74Eme16us and Gutmann J.1949 2979. 74a Emelbus and Woolf (J. 1950 1050) have prepared nitronium complex fluorides such as (NO,)(BF,) (NO,)(PF,) and (NO,)(AuF,) from dinitrogen tetroxide ; Woolf (ibid. p. 1053) has mctde similar nitrosyl salts from nitrosyl chloride. 76C. C. Addison and R. Thompson J. 1949 5211. 76 G. Jander and K. Wickert 2. physikal. Chem.1936 A 178 67 ; see also however L. C. Bateman E. D. Hughes and C. K. Ingold J. 1944 243. 77 G. Jander and K. H. Bandlow 8. physikal. Chem. 1943 A 191 321 78C. M. Beeson and D. M. Yost J . Amer. Chem. SOC. 1939 61 1432. 79 J. McMorris and D. M. Yost ibid. 1932 54 2247. N. P. Nies and D. M . Yost ibid. 1935 5'7 306. 81 McMorris and Yost ibid. 1931 53 2625. 82 C. M Blair and D. M . Yost ibid. 1933 55 4489. 8 3 G. E. Gibson and R. C. Ramsperger Physical Rev. 1927 30 598. 84 W. G. Brown ibid. 1932 42 355. SHBRPE TNTERIIALOC4EN COMPOUNDS AND POLYHALIDES 125 now important in fixing the heat of dissociation of fluorine. 0. Ruff and F. Laass 85 measured the heat of the reaction CIF + H = HC1+ HE' and from their results 0. Ruff and W. Menzel,S6 taking the heat of formation of hydrogen fluoride as 64.0 kcals.calculated that of chlorine monofluoride to be 27.4 kcals. H. Schmitz and H. J. Schumacher 87 determined the heats of the smooth and rapid reactions NaCl + ClF = Cl + NaF . (1) NaCl + HF = *GI + NaF (2) as 24.5 & 0.1 and 39.5 & 0-5 kcals. respectively and from them found the heat of formation of chlorine monofluoride to be 15.0 & 0.5 kcals. at 18O. The reaction 3NaCl + ClF = 3NaF + 2C1 . (3) was found to be exothermic to the extent of 76.5 kcals. leading to values of 27.0 1.5 kcals. for the heat of the reaction ClF + F2 = ClF 9 (4) and 42 2 kcals. for the heat of formation of chlorine trifluoride from chlorine and fluorine. The heat of reaction (4) at 300" was determined by investigating the variation of Kp with temperature in a coated nickel or magnesium vessel using however a quartz manometer.The results were moderately con- sistent and indicated a value of 25 & 2 kcals. This agreement with the calorimetric determination and the fact that the result for reaction (2) was in very close agreement with a previous determinationsS suggest that Schmitz and Schumacher's value for the heat of formation of chlorine monofluoride is correct at least to within 8 few kilocalories. Another recent determination of the heat of formation by E. Wi~ke,~Q who used a direct calorimetric method led to a value of 11.6 & 0.4 kcals. a t Z O O which considering the difficulties created by the reactivity of the substance under examination is in fair agreement with Schmitz a,nd Schumacher's result. A. L. Wahrhaftig 90 and Schmitz and Schumacher O1 have independently examined the absorption spectrum of chlorine monofluoride in the visible and the ultra-violet region.Some of the experimental details given by the latter authors are interesting the sample was contained a t a pressure of 1.5 atm. in a 380-cm. long iron tube fitted with fluorspar windows; and two methods for the preparation of the chlorine monofluoride from chlorine and fluorine and from chlorine trifluoride and chlorine were used. A weak band system in the visible region was found and from the extrapolated series limit the heat of dissociation of chlorine monofluoride was calculated. Wahrhaftig assuming the products of dissociation to be excited chlorine and normal fluorine atoms evaluated the energy required for dissociation Z. anorg. Chem. 1929 183 214.2. Naturforsch. 1947,. 2a 362. 86 Ibid. 1931 198 375. r8H. v. Wartenberg and 0. Fitzner 2. anorg. Chem. 1926 151 313. SfJ Nachr. Akad. Wiss. GGttingen Math.-physik. Klasse 1946 89. QOJ Chem. Physics 1942 10 248. 2. Naturforsch. 1947 2a 359. K 126 QUARTERLY REVIEWS into two normal atoms as 60.3 kcals. and taking the dissociation energy for chlorine to be 57.2 kcals. deduced that Dg,) + 2Qf(ClF) = 63.4 kcals. where is the dissociation energy of fluorine and &t(ClF) the heat of formation (thermochemical) of chlorine monofluoride. Schmitz and Schumacher obtained for the dissociation energy of chlorine monofluoride into normal atoms values of 58.9 kcals. (assuming production of excited chlorine and normal fluorine) and 60.3 kcals. (assuming normal chlorine and excited fluorine).These values lead to 60.6 and 63.8 kcals. respec- tively for D(F4) + 2Qf(C1F). If therefore the heat of formation of chlorine monofluoride lies between 12 and 15 kcals. the dissociation energy of fluorine should be 35-40 kcals. Such a value is much lower than that suggested by H. von Wartenberg G. Sprenger and J. TaylorPg2 vix. 63-3 kcals. This was obtained from the absorption spectrum of fluorine and involved an extrapolation from the properties of the other halogens for which there was no theoretical justi- fication.93 A lower value for the dissociation energy of fluorine has not been universally accepted however and Wicke 9* has criticised the spectro- scopic work on chlorine fluoride and has produced fresh evidence in favour of a high dissociation energy. Meanwhile A. D.Caunt and R. F. from the ultra-violet absorption spectra of rubidium and cmsium fluorides and thermochemical data for these compounds have suggested a value of 50 & 6 kcals. ; and whether the source of the disagreement between dif- ferent results lies in the interpretation of the spectrum of chlorine mono- fluoride or the value for the heat of formation of chlorine monofluoride or the thermochemical data (especially the heats of sublimation) for the alkali-metal fluorides is not yet known. Examination of the absorption spectrum of bromine monofluoride 96 led to values for the dissociation energy of 59.9 and 50.3 kcals. fluorine and bromine respectively being taken to be the excited atom. The ultra-violet absorption spectra of chlorine trifl~oride,~' bromine triflu~ride,~~ and iodine pentafluoride 98 have also been studied but no band structures have been found.The heats of formation of the bromine and iodine fluorides are all unknown though methods for the determination of some of them have been suggested.69 The deviation of chlorine trifluoride from the perfect-gas laws at pressures of 300-800 mm. was studied tensimetrically by Schmitz and Schumacher,gg who showed that the results obtained could be explained by dimerisation the heat of the reaction 2C1F3 = (ClF,) being 3.3 0.5 kcals. in the tem- 92 2. physikal. Chem. 1931 Bodenstein Festband 61. R. S. Mulliken J. Chem. Physics 1934 2 792. 94 2. Elektrochem. 1949 53 212. 95 Nature 1949 164 753. 96 P. Brodersen and H. J. Schumacher 2. Naturforsch. 1947 2a 358. 97 Schmitz and Schumacher ibid. p. 363. 88 C.F. White and C. F. Goodeve Trans. Faruday Soc. 1934 30 1049. 9* Z. Naturforsch. 1947 2a 363. SHARPE ZNTERHALOCIEN CONPOUNDS AND POLYHALIDES 127 peratarc range 9-24'. The degree of association a t 20' calculated from their results is 0.054. Interhalogen Compounds in Organic Chemistry.-Iodine monochloride in acetic acid (Wijs's solution) has long been used for the estimation of the " iodine value " of unsaturated compounds,100 and this reagent is now also important for the iodination of aromatic compounds. That the attacking entity in this reaction is the I+ ion is suggested by the conversion of com- pounds containing groups which activate the aromatic nucleus into o- and p-iodo-compounds acetanilide for example yields p-iodoacetanilide,lol and salicylic acid gives 2- hydroxy-3 5-di-iodobenzoic acid.lo2 Under similar conditions however iodine bromide effects bromination phenol in carbon tetrachloride forms p-bromophenol and a-naphthol in acetic acid gives 4-brom0-l-naphthol.~~~ The bromide is dissociated more than the chloride into its elements and since bromination by bromine is a much faster reaction than iodination by the I+ ion the bromide is a brominating agent even in nitrobenzene ; the chloride chlorinates phenol only in the absence of a s0lvent.103~ The kinetics of the reactions of iodine monochloride with acetanilide and anisole in acetic acid solution have been studied by L. J. Lambourne and P. W. Robertson,104 who found the reaction to be of the first order with respect to the concentration of the aromatic compound and of the second order with respect to that of the iodinating agent.The hydrogen chloride formed in the reaction competes with the aromatic com- pound for the remaining iodine monochloride (with which it forms HICI,) and the rate falls off rapidly as the reaction proceeds. The first published account of a controlled fluorination with bromine trifluoride was the conversion of carbon tetrachloride into a mixture of chlorofluoromethanes.105 This work was later amplified by A. A. Banks H. J. Emeleus R. N. Haszeldine and V. Kerrigan,lO6 who also showed that bromine trifluoride reacts with carbon tetraiodide forming carbon tetrafluoride and bromofluoromethanes. By the action of bromine tri- fluoride on hexachlorobenzene E. T. McBee V. V. Lindgren and W. B. Ligett lo7 obtained products corresponding in composition to approximately C6Br,C14F6 ; and the same authors,los by successive treatment of hexa- chlorobenzene with bromine trifluoride antimony pentafluoride and zinc dust and alcohol prepared perfluorocycZohexadiene.Iodine pentafluoride is only moderately reactive and it appears to be a mild fluorinating agent which will replace iodine but not hydrogen by fluorine. From the slow reaction between iodine pentafluoride and carbon tetrachloride 0. Ruff and R. Keim loo obtained trichlorofluoromethane and a little dichlorodifluoromethane. J. H. Simons R. L. Bond and R. E. 1935 p. 411. loo A. D. Mitchell " Sutton's Volumetric Analysis " J. and A. Churchill London lo2 G. H. Woollett and W. W. Johnson Org. Synth. Coll. Vol. 11 1943 p. 343. lo3 W. Militzer J. Amer. Chem. Soc. 1938 60 256. 1030F.W. Bennett and R. G. Sharpe J. 1950 1383. lo* J. 1947 1167. lo6 H. 8. Nutting and P. S. Petrie U.S.P. 1,961,622 (1934). 106 J. 1948 2188. lo' Ind. Eng. Chern. 1947 39 378. lo* U.S.P. 2,432,997 (1947). loo 2. anorg. Chern. 1931 201 246. lol F. D. Chattaway and A. B. Constable J. 1914 105 124. 128 QUARTERLY REVIEWS McArthur 110 identified fluoroform as the principal product of the reaction between iodine pentafluoride and iodoform ; this reaction was carried out in a copper apparatus and the authors reported that under similar con- ditions carbon tetraiodide yielded hexafluoroethane. Banks Emelhis Haszeldine and Kerrigan however showed that in glass carbon tetra- iodide and iodine pentafluoride yield iodotrifluoromethane ; lo8 and it has been found that even in copper apparatus using the technique employed by Simons Bond and McArthur iodotrifluoromethane is still the main product of the reaction.lll Iodine pentafluoride has also been found to react with carbon tetrabromide forming bromofluoromethanes and with tetraiodoethylene forming iodopentafluoroethane.log Further properties of iodotrifluoromethane and iodopent afluor oethane have recently been described.6s 112 An addition compound of iodine pentafluoride and dioxan C4Hs02,1F5 has been made; 113 this is analogous to similar compounds obtained from iodine iodine monochloride and iodine bromide.114 Information concerning the reactions of the other halogen fluorides with organic compounds is scanty and no controlled reactions of chlorine mono- fluoride bromine monofluoride and pentafluoride and iodine heptafluoride with organic substances have yet been described.R. W. Porter 55 and R. le G. Burnett and A. A. Banks lloa have reported that the controlled reaction of chlorine trifluoride with organic compounds results in the intro- duction of both chlorine and fluorine; and R. N. Haszeldine,lll by the vapour-phase fluorination of benzene and toluene has obtained evidence for the formation of chlorofluoro-addition and -substitution products. In view of the relatively low cost of chlorine it seems that it is in this field that discoveries of future industrial importance are most likely to be made. The Structural Chemistry of Interhalogen Molecules and Ions.-Reference has already been made to the electrical conductivities of certain interhalogen compounds and to X-ray data for ions.Other structural information is scanty. H. Braune and P. Pinnow 116 studied iodine pentafluoride by the electron-diffraction method ; they concluded that the 1-F distances are all equal but when a trigonal bipyramidal shape was assumed the bond length was calculated to be 2-58 A. The sum of the covalent radii for iodine and fluorine is only 1-97 A. and Braune and Pinnow’s result did not meet with general acceptance. Investigation of the infra-red absorption spectrum of the vapour and the Raman spectrum of the liquid led to the conclusion that the molecule has the shape of a tetragonal pyramid.l18 Similar inves- tigations on iodine heptafluoride the only known AB molecule suggested J . Amer. Chem. Soc. 1940 62 3477. 1100 Communicated at the Chemical Society Symposium on Fluorine Chemistry 111 R.N. Haszeldine private communication. 1la H. J. Emel6us and R. N. Haszeldine J. 1949 2948. llsA. F. Scott and J. F. Bunnett J . Amer. Chem. Soc. 1942 64 2727. llaH. Rheinboldt and R. Boy J . pr. Chem. 1931 129 273. 116 2. physikal. Chem. 1937 B 35 239. Nov. 30 1949. R. C. Lord M. A. Lynch jun. W. C. Schumb and E. J. Slowinski jun. J . Amer. Chem. SOC. 1950 72 522. SHARPE INTERHALOGEN COMPODNDS AND POLYHALIDES 129 that this compound is pentagonal bipyramidal in configuration a structure previously unknown. The Raman spectrum of liquid chlorine trifluoride and the infra-red absorption spectrum of the gas have been studied by E. A. Jones T. F. Parkinson and R. B. Murray,117 using fluorothene (polychlorotrifluoro- ethylene) apparatus. Their results suggested either an unsymmetrical structure for chlorine trifluoride or association in the liquid.From con- sideration of the orbitals likely to be involved in bonding anything but a planar structure for the single CIF molecule seems unlikely ; and in view of Schmitz and Schumacher’s investigation on the gasYQQ a high degree of association in the liquid phase appears probable. The infra-red spectrum of chlorine monofluoride has also been examined ; the location of the centre of the fundamental band at 772 cm.-l is in agreement with the calculated position from the ultra-violet absorption spectrum of the molecule.l18 The electrical conductivity of a number of the interhalogen compounds raises the problem of their structures in the solid state. Absence of a decomposition potential is usually a characteristic of electronic conduction ; this however normally differs little in magnitude in the solid and the liquid state and must be ruled out for iodine monochloride and bromine tri- fluoride,68 since for these compounds the conductivity of the solid is very much less than that of the liquid.The discovery of the ionic structures of solid phosphorus pentachloride 119 and phosphorus pentabromide 12* suggests that some of the interhalogen compounds may also be found to possess ionic lattices. A wide range of molecules and ions is now available for use in the dis- cussion of the relation between electronic configuration and sfereochemistry. These include IF ; (IF& ; IF5 and BrF ; (El4)- and (BrF4)- ; ICI, BrF, and CIF ; (ICl2)- etc. ; ClF IC1 etc. ; (BrFz)+ and perhaps (ICl2)+.The observed configurations of the (IC14)- and (ICl2)- ions are usually described 49 lZ1 122 by saying that the valency shells of the iodine atoms in these ions contain twelve and ten electrons respectively and that the structures result from the occupation by unshared electron pairs of two trans-positions in an octahedron and of three equatorial positions in a tri- gonal bipyramid respectively. It has been suggested in a similar way that the structure of iodine pentafluoride may be derived from an octahedron by postulating occupation of one position by an unshared electron 122 It may be pointed out however that from G. Kimball’s systematic calcu- lations of bond type and electronic configuration,l23 the observed shapes of (IC14)- (IClz)- and IF may be derived by allotting to the iodine atoms valency configurations of 5p45d4 5p25d2 and 5p65d4 since p2d2 pd and 11’ J .Chem. Physics 1949 17 501. 11* T. F. Parkinson E. A. Jones and A. H. Nielsen PhysicaZ Rev. 1949 76 199. ll*@ N. N. Greenwood unpublished. llS D. Clark H. M. Powell and A. F. Wells J. 1942 642. 120 Clark and Powell Nature 1940 145 971. 121 L. Pauling “ The Nature of the Chemical Bond ” Cornell 1940. 12* A. F. Wells “ Structural Inorganic Chemistry ” Oxford 1945. 12* J. Chem. Phymb 1940 8 194. The I-F distance in IF is 1.75 A. 130 QUARTERLY REVIEWS p3d2 bonds are calculated to give planar linear and tetragonal pyramidal configurations respectively. On the basis of either method of approach (BrF2)+ unlike (1CI2)- is expected to be angular. What influence then has an " inert pair " of electrons on the stereochemistry of an ion or mole- cule ? From the discussion of this question by N.V. Sidgwick and H. M. Powell 124 it is evident that for most molecules it matters little if we assume that '' the mean positions of the electron pairs are the same whether they are shared or not " though the observed bond angle in hydrogen sulphide (92") does appear to require pure p-bonding in this compound. With valency groups of twelve or less electrons symmetrical configurations easily reconciled with one and often with more than one reasonable com- bination of orbitals are available ; and they frequently bear simple relation- ships to other symmetrical structures e.g. the octahedron to the square. For atoms with valency shells containing more than twelve electrons the position seems to be different.In the ions (SeCl,)- (SeBr,)- and (TeCl,)- which have a central atom with a valency shell of fourteen elec- trons but only six-fold co-ordination the configuration is octahedral but the observed bond lengths 125 1z6 are unexpectedly high by about 0.25 A. J. Y. Beach 127 has made the suggestion that such an increase in bond length is to be correlated with the existence of the 49 orbital (in the case of selenium) as a separate energy level the bonding orbitals in (SeCI6)- being 4p54d259 orbitals. The (IF6)- ion would be effectively isoelectronic with (TeCI,)- and it has therefore been suggested 64 that this ion (unless it differs from all other known finite complex six-co-ordination ions by not being octahedral) should have an interhalogen distance appreciably greater than the sum of the covalent radii.The structure of IF then has to be considered separately; in this compound the iodine atom would possess no inert electron pair and there is no reason to predict a long bond here. The observed symmetrical con- figuration of iodine heptafluoride lies beyond the scope of Kimball's treat- ment but it is not to be expected that the structures of IF and (IF6)- will be simply related. It now appears that the (ZrF7)-- and (NbF,)- ions,f28* 129 and the IF molecule which might have been expected to be similar all have different structures. The technical difficulties in the way of further structural investigations on the substances under discussion are considerable chemical reactivity and differences in the scattering powers of the halogens being especially important.If however these difficulties can be overcome the determina- tion of the structures of the series of interhalogen ions and molecules seems likely to provide results of considerable theoretical significance. The author thanks Dr. W. G. Palrnor and Dr. P. Gray for reading through this Review and making a number of helpful suggestions. 1 4 4 Proc. Roy. SOC. 1940 A 1'96 153. 1~ J. L. Hoard and B. N. Diekinson &id. 1933 84 436. 12' Quoted in ref. (121) p. 184. 188 G. C. Hampson and L. Pauling J . Amer. Chem. Soc. 1938 60 2702. lido J. L. Hoard ibid. 1939 61 1252. 125 G. Engel 2. Krist. 1935 90 341.

ISSN:0009-2681    

DOI:10.1039/QR9500400115

出版商:RSC    

年代:1950

数据来源: RSC

摘要:

ROTATION SPECTRA By D. H. WHIFFEN M.A. D.PHIL. (ROYAL COMXISSION FOR THE EXHIBITION OF 1851 SENIOR STUDENT THE PHYSICAL CHEMISTRY LABORATORY OXFORD UNIVERSITY) * WORK on the fine structure in ultra-violet electronic spectra and infra-red vibration spectra has in the past led to a considerable knowledge of rota- tional energy levels but only recently has it become convenient to make direct spectroscopic measurements a t the longer wave-lengths which correspond to simple changes in the rotational energy alone. This present Review covers this field of pure rotation spectra both for gases and for condensed systems and also the related subject of inversion spectra. For gases the basic principle namely that the energy absorbed from the electromagnetic radiation field is used to excite individual molecules to higher energy states is the same as that for other forms of absorption spectroscopy.The individual quanta of energy are much smaller and can be measured with very great accuracy so that fine details of molecular behaviour are exposed which are normally lost by comparison with the larger total energy changes involved at shorter wave-lengths. Most of the work on gases has been done in “ the microwave region,” that is from 5 to 0.5 cm. wave-length. Information is obtained about the accurate moments of inertia bond distances isotopic masses dipole moments nuclear spins and quadrupole moments internal electric fields in molecules and collision frequencies. For condensed phases the quantised nature of the rotational energy levels disappears as a result of the strong intermolecular interactions.The knowledge given by the absorption spectrum does not then concern indi- vidual molecular properties apart from dipole moments but rather inter- molecular forces ; potential barriers opposing rotation in crystals the structure of polymeric materials and the pseudo-crystallinity of liquids thus become matters for study. Experimental Methods The recent advances have followed from the general technical develop- ment consequent on the use of microwave frequencies in military applica- tions both for radar detecting systems and for communication. Variable- frequency monochromatic Klystron oscillators are the usual source of the radiation and no prism or grating is needed. The use of wave-guide for containing both the radiation and the material under investigation is common while resonant cavities and other circuit components have their special applications.At longer wave-lengths coaxial cables and Lecher * Present address The Chemistry Department The University Birmingham. 131 132 QUARTERLY REVIEWS wires are used besides resonant circuits with filled condensers; there is also a " thermometer " method in which the energy loss is measured by the resulting rise in temperature of the sample. Those interested in further &tails must refer elsewhere.1-4 Spectral positions are described in various units and it may be noted that 1 cm. is 10,000 p. and also lo8 A. and corresponds in frequency to 1 cm.-l (wave number) or 3 x 1O1O cycles/second that is 30,000 Mc/s. Frequency Mc/e. . whichis . . . . Gases Linear Molecules.-For a linear molecule with a moment of inertia I the allowed quantised energy states for rotation are given by where J is the rotational quantum number.Each energy state is (2J + 1)- fold degenerate in the absence of external fields and the terms with higher powers of J(J + 1) which correct for centrifugal stretching are negligible a t low J values. There is a selection rule which requires that interaction with electromagnetic radiation can only lead to a change in the quantum number J by one unit and so the energy involved in a spectral transition will be the difference in energy between two neighbouring levels namely h2(J + 1)/(4n21) where J is now the quantum number in the lower state. Since the energy is related to the absorption frequency Y by AE = hv the spectrum of a linear molecule will consist of a set of evenly spaced lines with frequencies in the ratios 1 2 3 4 .. . corresponding to the J values 0 1 2 3 . . Such a series of lines with J values from 4 to 10 was first found by M. Czerny5 for hydrogen chloride between 40 and 100 p. and subse- quently in the other hydrogen halides. Special importance was attached to the initial experiment since the old quantum theory had predicted lines a t frequencies given by h(2J + 1)/(8n21) which would be in the ratios 1 3 5 7 . . . ; this prediction could not be reconciled with the experiment but rather the new quantum theory was confirmed. Recent work has provided more accurate confirmation of the integral ratios between absorption lines. For example three transi- grot. = hV(J + 1)/(8n21) 24,325-92 48,651.7 60,814.1 2 x 12,162.96 4 x 12,162.9 6 x 12,1624 TABLE I F.H. Miiller Ergebn. Exakt. Naturwiss. 1938 17 164. W. Gordy Rev. Mod. Physics 1948 20 668. 2B. Bleaney Rep. Prog. Physics 1946-47 11 178. 4 M. Freymann It. Freymann and J. Le Bot J . Phys. Radium 1948 9 ID. 6E. Physik 1926 84 227. IW. 1927 44 235. WHIFFEN ROTATION SPECTRA 133 tions in carbon oxysUlphide7,* are given in Table I. The very small discrepancies are probably real and are due to centrifugal stretching of the molecule whose moment of inertia consequently increases with J . Moments of Inertia and Bond Lengths.-The fact that microwave fre- quencies can be measured to an absolute accuracy of 0.02 Mc/s. in 25,000 Mc/s. means that the moment of inertia could be determined to the same degree of accuracy 1 part in 106 were Planck's constant known to this degree of accuracy.The deduced moment will be the mean moment in the ground vibrational state and a correction must be applied to con. vert this figure into the equilibrium moment that is the moment of inertia pertaining to a structure in which the nuclei are clamped in their equi- librium positions. The correction arises partly because although the mean internuclear distances are unaltered by the zero-point motion of a simple harmonic vibration the mean of the squares of the distances from the centre of gravity is altered and it is on this latter quantity that the moment of inertia depends. And secondly there is a correction of the same order of magnitude for anharmonicity in the vibrations which alters the mean internuclear distances. The total correction which may be positive or negative and as much as l% can be made from measurements of the moments of inertia of the higher vibrational states and extrapolation to zero vibrational energy ; this frequently involves the detection of very weak absorption lines and is not generally attempted so that only ground state moments of inertia are available.For diatomic molecules the internuclear distance can be directly deter- mined from the moment of inertia and the atomic masses. For linear triatomic molecules there are two bond lengths to be found and so one moment of inertia is insufficient and recourse must be had to isotopic molecules. If then it is assumed that the interatomic distances are the same in the isotopic molecules the two different moments of inertia lead to two equations in two unknowns the bond lengths and these equations are soluble.The equilibrium moments of inertia should be used for the equilibrium configuration depends on the force field which is justifiably assumed to be unchanged by isotopic substitution. The ground state moments which are in fact commonly used are less suitable for the mean distances will change with zero-point amplitude and consequently with isotopic substitution. The correction may be important as the computed bond lengths are very sensitive to any change of bond length on isotopic substitution; for instance a real change of mean bond length of 0.001 A. on substitution of mass 33 for mass 32 may lead to an error of 0.03 A. in the value of the computed length. Consequently unless due allowance is made for the zero-point effects bond lengths derived from isotopic moments of inertia should not be trusted to better than 0.01 A.Apart from this difficulty errors of 0.005 A. may be introduced by uncertainty of the exact isotopic masses and of 0.001 A. by experimental errors. Carbon oxyselenide ' T. W. Dakin W. E. Good and D. K. Coles PhysicaZ Rev. 1947 71 640. a R. E. Hillger M. W. P. Strandberg T. Wentink jun. and R. L. Kyhl ibid. 1947 C. H. Tomes A. N. Holden a4d F. R. Merritt ibid. 1948 74 1113. 12 157. 134 QUBRTERLY REVIEWS to quote but one of several examples,l* has been investigated 11 in this way and found to have C=O 1.159 A. and C=Se 1.709 A. Isotopic Mrrsses.-If the moment of inertia of a third isotopic molecule is available a knowledge of the two interatomic distances will enable the unknown isotopic mass of the third isotope to be found.Even though ground-state moments of inertia are used the detailed nature of the calculation is such that the third mass is obtainable with considerable accuracy.12 For the rare 36S isotope W. Low and C. H. Townes l3 find 35.97834 & 0*0004 mp. using carbon oxysulphide with only the natural sulphur a bundances . Non-linear Molecules.-Spherically symmetrical molecules such as carbon tetrachloride have no permanent dipole moment and cannot have a pure rotation absorption spectrum. For a symmetrical top molecule such as methyl chloride in which two of the moments of inertia are equal I* = I +Ic and the dipole moment lies along the symmetry or C axis the rotational energy levels are given by where K is a second quantum number.For the rotation spectrum the selection rules are AJ = & 1 AK = 0 and the absorption frequencies are given by h(J + l)/(4n21A) which is the same expression as for linear molecules with I replaced by I,. Only this moment of inertia and not Ic can be found and a number of isotopic molecules are required to solve completely for the bond distances and angles. Amongst others,1° the J1-+2 transition of chlorosilane has been observed 1 4 with each of the isotopes s5C1 and 37Cl and an Si-H distance of 1.456 A. being assumed the Six1 distance is 2-035 A. and the HSiH angle 103" 57'. For each value of J values of K from - J to + J are allowed but the simple expression for the absorption frequencies shows these not to depend on H. This degeneracy is disturbed by the fact that IA depends slightly on K as the result of centrifugal stretching.For molecules without any symmetry in the moments of inertia asym- metric tops the rotational energy levels form a complicated set. Con- sideration of the selection rules and the various permitted transitions leads to an even more complex set of absorption lines and analysis of the spec- trum is dificult. In favourable cases the three moments of inertia can be determined and the structural parameters evaluated with the aid of isotopic substitution. After the introduction of deuterium into ethylene oxide G. L. Cunningham A. W. Boyd and W. D. Gwinn l5 have solved for the distances and find C-0 1.436 A. C-c 1.47 A. C-H 1.08 A. LHCH 116" 50' and that the C-C bond makes an angle 158" 5' with the CH2 plane. Water Erot. = h'J(J + 1)/(8n21~) + ( l / I c - ~ / I A ) ~ ' .K ~ / ( ~ Z ' ) 10 For tables of bond distances and other properties actually determined by micro- 11 M. W. P. Strandberg T. Wentink jun. and A. G. Hill Physical Rev. 1949 12 C. H. Townes A. N. Holden and F. R. Merritt ibid. 1947 72 513. 13 Ibid. 1949 75 629. 15J. Chem. Physics 1949 17 211. wave measurements see ref. (3) which is complete to July 1948. 75 827. laA. H. Sharbaugh jun. ibid. 1948 74 1870. WEEIFFEN ROTA" SPEUTRA 135 is an important asymmetric top molecule and lines have been identified l7 to 18 p. while there is one weak line in the microwave region due to the 5-1 -+ 6- transition and this has been found both a t low pressures l8 and in the atmosphere.19 isoThiocyanic acid as made from potassium thio- cyanate by liberation with phosphoric acid is interesting for lines which correspond to JI+2 in a symmetric-top approximation have been observed 2o for H14NW32S H14Nf3C32S D14N12C32S and DI4N13C32S and the four moments of inertia could not be made to give reasonable values of the bond distances on the assumption that the material was NCSH whereas the values found for HNCS are H-N 1.2 A.N-C 1.21 A. C-S 1.57 A. and LHNC 112". Absorption lines have also been found in methyl alcohol,21-23 but complete interpretation is uncertain ; it is an interesting molecule since internal rotation about the C-0 bond is possible and some at least of the lines will involve this degree of freedom. Excited Vibrational States.-As already stated there is a change of mean moment of inertia with vibrational quantum number owing to a change of vibrational amplitude.If then one of the vibration frequencies is SUE- ciently low t o be excited in an appreciable number of molecules a t room temperature absorption due to the vibrationally excited molecules should be detectable near the ground-state absorption lines. The lines of excited molecules can be distinguished by their smaller intensities and by the fact that they are even weaker relative to the ground-state lines at low tem- peratures where the population of the excited states is less. A quantita- tive investigation of the change of intensity with temperature 11 gave the vibrational frequency of the bending mode of carbon oxyselenide as 474 cm.-l whereas direct measurement subsequently gave 464 cm.-l. For most linear triatomic molecules the lowest vibration frequency is that of this degenerate bending mode; when this is excited each of the rota- tional energy levels is split into two as also is each rotational absorption line.24 This phenomenon of I-type doubling has been obser~ed,~ and in carbon oxysulphide the lines are 25 Mc/s.apart and in agreement with the correct theory.25 Stark Effect.-The (2J + 1)-fold degeneracy of the rotational levels is partly destroyed by an external electric field in the presence of which there are (J + 1) levels for each value of J in linear molecules. These are char- acterised by a quantum number M which runs from 0 to J while the levels M + 0 are still doubly degenerate and correspond to levels + M and - X . The transitions which are permitted in the absorption spectrum depend on 16 H. M. Randall D. M.Dennison N. Ginsburg and L. R. Weber Physical Rev. 18 C. H. Tomes and F. R. Merritt Physical Rev. 1946 70 658. 19 It. H. Dicke R. Beringer R. L. Kyhl and A. B. Vane ibid. p. 340. 20 C. I. Beard and B. P. Dailey J . Chem. Physics 1947 15 762. 21 €3. P. Dailey Physical Rev. 1947 72 84. 22W. D. Hershberger and J. Turkevich ibid. 1947 71 554. 23 D. K. Coles ibid. 1948 74 1194. 24 H. H. Nielsen and W. H. Shaffer J . Chem. Physics 1943 11 140. 26 H. H. Nielsen Physical Rev. 1949 75 1961. 1937 52 160. l7 H. Hopf 2. PhysiE 1940 116 310. 136 QUARTERLY REVIEWS the relative orientations of the static electric field and the high-frequency measuring field of the radiation. When these electric fields axe parallel the selection rule is AM = 0 and when they are perpendicular it is Carbon oxysulphide is one case where the two transitions J1-+2 M,,+, and Ml+l have been observed.26 In this case the frequency separation is given theoretically by in which p is the dipole moment and E is the static field strength; E appears only in a squared term and the Stark effect is of the second order.The frequency separation which is of the order 6 Mc/s. for a field of 1000 volts/cm. can be measured with considerable accuracy. The greater experimental difficulty is to obtain known uniform static fields inside a wave-guide and the ultimate accuracy of dipole-moment measurements from the Stark effect separation has not been reached. Even so an absolute accuracy of 0.01 D. is obtained and accuracy in relative moments where the same apparatus is used is considerably higher as the same field strength can be used in the two cases.The advantages of the method are large for no density determinations and no corrections for atomic polarisations are involved and low pressures lows mm. may be used. Further the measurement is made on one vibrational state and the change of dipole moment with vibrational energy can be measured ; thus values of 0.754 D. for the ground-state dipole moment and 0.728 D. for a molecule singly excited in the lowest stretching mode have been reported for carbon oxy- selenide.ll Similarly different isotopic species may be studied even in mixtures and the change of dipole moment consequent on the change of zero-point amplitude observed ; for 16012C32S and 16013C32S the dipole moments are 0.732 and 0.722 D. re~pectively.~' For asymmetric molecules the Stark effect may be of either the first or the second order and is of great assistance in identifying the lines of a complex spectrum; the dipole moment can still be evaluated.J. K. Bragg and A. H. Sharbaugh 28 have found a value of 2.17 D. for formalde- hyde whereas dielectric-constant measurements are severely handicapped by the tendency of formaldehyde to polymerise. The effect of a high-frequency Stark field has been studied by C. H. Tomes and 3'. R. Merritt 29 up to 1.2 Mc/s. For low frequencies the absorption corresponds to the instantaneous value of the field but as the frequency is increased to the order of magnitude of the collision frequency a complicated pattern is obtained while a t the highest frequency the Stark field reverses many times during the absorption process and the pattern corresponds to the average of the square of the field strength that is to a static field of the strength of the root mean square of the high-frequency field amplitude.A Zeeman effect that is a shift of absorption frequency with a static 26 T. W. D a b W. E. Good and D. K. Cobs Physical Rev. 1946 70 560. a7 M. W. P. Strandberg T. Wentink jun. and R. L. Kyhl ibid. 1949 75 270. A M = & l . (3/20 - 1/84)8n*I,~~8'/h~ Ibid. p. 1774. 89 Ibid. 1947 72 1266. WEIFPEN ROTATfON SPECTRA 137 magnetic field has been found in methyl chloride 3O but the interaction depends in a complex way on the nuclear spins. Intensity and Line Shape.-The intensity and shape of an absorption line are best considered together although they are governed essentially by the dipole moment of the molecule and a collision frequency respec- tively.J. H. Van Vleck and V. F. Weisskopf 31 have shown that the Lorentz formula satisfactory for visible spectroscopy requires modification when the actual time of duration of a molecular collision is short compared with the reciprocal of the measuring radiation frequency. Their formula for the power attenuation coefficient may be written where v is the measuring frequency vij the resonance frequency for the ij transition ni the number of molecules per C.C. in the lower state I pij I is the matrix element of the dipole moment corresponding to the transition and Fij is a structure factor and it has been assumed that hvij < kT which is justified in the microwave region. For rotation bands I pij I is the pro- duct of a numerical factor depending on J and K and the dipole moment of the molecule and it can be seen that the intensity depends on the square of this moment.In particular if the dipole moment is zero as for homo- polar diatomic molecules and for spherical-top molecules then the intensity is nil and there is no absorption spectrum. The structure factor is given by Av 83 23 n:&vij - v ) 1 + (Av)' + (vij + v>s + (Av> where Av is the collision frequency which is also given by 1/(2nz) where z is the time between collisions which are effective in terminating the absorption process. At low pressures where Av <vij only the first term is important and this is only large when v * vij which implies that the absorption consists in a sharp line. Neglecting the second term and the effect of overlapping lines due to other transitions the shape of any line is given by as is shown in Fig.1. The curve has a maximum proportional to l/Av a t v = vij and has half its maximum height a t v = vij & Av so that the width a t half height is 2Av. Thus if collisions are infrequent the line is narrow and conversely. An increase of collision frequency may be brought about by an increase of gas pressure and on a simple collision theory there is a direct proportionality. The absorption per unit path length is also proportional to R~ which too is directly proportional to the pressure if there is no foreign gas present. Consequently the maximum absorption will be quite independent of pressure since it depends on the ratio nj/Av. This is confirmed by experiment though there is an upper limit of pressure about 50 mm.of mercury where this independence no longer holds since triple collisions overlapping of lines and the second term in the structure factor become important. Also there must be a lower limit 8n3v vij I pij I a nj B'ij/(3klT~) Av F.. = - Av/[(vij - Y)' + (Av)'] ao C . K. Jen Physical Rev. 1948,74,1396. a1 Rev. Mod. Physics 1945,17 227. 138 QUARTERLY REVIEWS to this independence since there can be no absorption when no gas is present the limit of the pressure-independent absorption is about mm. when processes other than gas col- 11 lisions-contribute materially to the line breadth. Especially important are the Doppler effect and collisions with the walls of the containing vessel. In practice this lower limit can be reached and is required for the resolution of close lines ; in one case two lines of cyanogen chloride only 0.14 Mc/s.or 5 x 10-6 cm.-I apart have been resolved.32 r - - Fra. 1 Line breadths have been studied 33-36 mostly in the ammonia four times that for b.) inversion spectrum (see below) where the same formula for the intensity applies. Por one particular line the 3,3 line the collision frequency Av is given by 28P Mc/s. where P is the pressure in mm. of Hg. This corresponds to an effective collision diameter of 14 A. and similar values are found for other lines in the spectrum. This is large compared with the usual model with van der Waals radii but even a t 14 A. the dipole-dipole interaction energy between the two molecules is of the same order as the energy involved in the transition and it is hardly surprising that interaction of such strength should be able t o terminate the absorption process.The addition of polar foreign gases is found to cause more broadening of an absorption line than an equiv- alent pressure of a non-polar gas in which case only the weaker dipole- polarisability interactions are important. Further if the appropriate value of the transition moment I pij I is inserted in the absorption expression the maximum absorption is known in terms of the collision frequency. Comparison with experiment 33 gave good agreement for ammonia and in other cases it may be possible to reverse the procedure and find dipole moments from the intensity of the absorption lines. It has been found experimentally 37-39 that lines decrease in peak intensity and thereby increase in width if a large amount of measuring power is used at low gas pressures.Under such conditions so much power is absorbed by the gas that the molecules which have been excited by the radiation cannot return to the initial state by means of collisions suffi- ciently rapidly to maintain the population of this state a t its thermal equilibrium value. There is then a maximum capacity of the gas for 32 C. H. Tomes A. N. Holden J. Bardeen and F. R. Merritt Physical Rev. 1947 71 644. 34 Idem Proc. Physical Soc. 1947 59 418. 36 C. H. Townes Physical Rev. 1946 70 665. 87 B. Bleaney and R. P. Penrose Proc. Physical SOC. 1948 60 83. 38 T A. Pond and W. F. Cannon Physical Rev. 1947 72 1121. 39 R. L. Carter and W. V. Smith ibid. 1948 73 1053. ~ f a s absorption lines. (Pressure for a 33 B. Bleaney and R.P. Penrose Proc. Roy. SOC. 1947 A 189 358. 36 Idem ibid. 1948 60 540. WEIIFFEN ROTATION SPECITM 139 absorbing power at the given frequency and this maximum is controlled by the rate of attainment of thermal equilibrium; the absorption coeffi- cient or proportion of the power absorbed will therefore tend to zero &s the power of the measuring radiation is increased. From detailed con- siderations 37 40 the frequency of collisions which are effective in main- taining thermal equilibrium can be obtained from the results and when the saturation is apportioned 41 properly over the individual unresolved Zeeman components of the line this frequency is found to be exactly the same as that of collisions which interrupt the absorption process as measured from the half width a t low powers.Temperature is one of the quantities involved in the intensity formula both explicitly and through n and Av. The exact effect of a change of temperature will depend on the values of the relevant quantities but a reduction of temperature normally enhances the strength of a line in the microwave region and a fourfold increase for a reduction to - 70” from room temperature is general. Also for any one system the lines at higher frequencies will be stronger by virtue of the explicit vvi3 factors in the intensity expression and of an increase of nj the population factor which reaches a maximum at intermediate J values. Nuclear Quadrupole Coupling.-The considerations of the absorption frequencies already outlined fail to account for all the observed lines in most cases. Thus in the region from 23,860 to 23,910 Mc/s.which covers the J1-t2 transition of 35ClCN there me five lines 32 none of which can be attributed to 37C1CN or to vibrationally excited states the lines due to which are known outside this range. Under higher resolution still two of the lines in the region 23,883-23,887 Mc/s. are resolved into eight clear lines some of which have further shoulders. The origin of this hyperfine structure lies in the coupling of the nuclear spins to the molecular rota- tion through their electrical quadrupole moments in the above example the major splitting is due to the C1 nucleus and the finest structure to the lines is due to the weaker N interaction. To make the nature of the interaction plainer it must be realised that an atomic nucleus or more precisely its distribution of positive charge is not necessarily spherical in shape 42 except when the nuclear spin is zero or one-half.The spin Spin axis i, + + a. Posifive Spin axis. - + + _. 6. Negative. FIG. 2 Nuclear quadrupole moments. axis must be a completely symmetrical rotational axis but it remains pos- 4o R. Karplus and J. Schwinger Physical Rev. 1948 73 1020. 41 R. Karplus ibid. p. 1120. 48 8. Fluegge “ Nuclear Physics Tables,” by J. Mattauch Interscience p. 38. 140 QUARTBIRLY REVWWS sible for the ellipsoid of positive charge representing the nucleus to be either prolate as in Fig. 2a or oblate as in Fig. 2b. Either distribution can be regarded as the sum of a spherical distribution and a quadrupole distribu- tion which is indicated in terms of + and - signs in the lower part of Pig.2. The size of the quadrupole may be numerically the same in the two cases which may be distinguished by the addition of a sign by con- vention the prolate type of nucleus (Fig. 2a) is taken to have a positive quadrupole moment. If an electrical quadrupole is placed in an inhomogeneous electric field it has a potential energy which varies with its orientation in a manner similar to that in which a dipole has a potential energy varying with its orientation in a uniform electric field. For a nucleus in a molecule the possible directions of its spin relative to the axis of rotation of the whole molecule are limited by quantum conditions. In each of the allowed orientations the nucleus may have a different potential energy owing to interaction of the quadrupole moment with the electric field of the mole- cule and there will be a series of energy states.The pattern of theqe states depends on the values of the nuclear spin and of the molecular rotational quantum numbers while the energy scale factor is determined by the nuclear quadrupole moment and the degree of inhomogeneity of the field. For linear molecules the relative energies depend on Qa2V/&2 where Q is the quadrupole moment and V is the electric potential at the position of the nucleus so that a2V/az2 is the divergence of the field along the molecu- lar axis z. The product &PV/az2 may be either positive or negative and is effectively independent of the rotational quantum number. Provided the theoretical values of the energy levels are known it remains only to investigate the selection rules which are likewise governed by quantum conditions in order to predict the absorption spectrum.The mathematical details 4% 44 of the theory are somewhat lengthy but good agreement between theory and experiment has been obtained even in cases where more than one nucleus has a quadrupole moment 32 45 and in asymmetric molecules.*6 The parameter Q a 2 B / a x 2 must be chosen in sign and magnitude to fit the observed results but the number of lines to be expected their relative frequency separations and their relative intensities are determined by the nature of the change of rotational quan- tum number and the values of the relevant nuclear spins. In some cases if the spins are not known they may be found from the absorption spec- trum; for instance the spin of 1OB has been shown to be 3 and of llB to be 3/2 from the microwave spectrum of borine ~arbonyl.~' The energy parameter Qa2V/az2 which is found from the experiments does not lead to the quadrupole moment i$self unless the field divergence is known although for isotopic molecules the field is unchanged and the ratio of the two quadrupole moments can be obtained.The field diver- 48 J. Bardeen and C. H. Townes Physical Rev. 1948,73,97. 44 Idem i b a . p. 627. 46 A. G. Smith H. Ring W. V. Smith and W. Gordy ibid. p. 633. 48 J. H. Goldstein and J. K. Bragg ibid. 1949 75 1453. 47 W. Uordy H. Ring and A. B. Burg ibid. 1948 74 1191. WHZFFEN ROTAmON SPEUTRA 141 gence can be calculated from a satisfactory molecular wave function but the hydrogen molecule 48 is one of the few cases in which even moderate accuracy is possible.C. H. Townes 49 has suggested a rough approxima- tion in cases where the atom in question is bonded to the rest of the molecule by a pure p-type orbital and in these cases it is thought to be accurate to 20%. Alternatively if the quadrupole moment can be accurately deter- mined in other ways such as by atomic-beam techniques an accurate value of the field divergence can be obtained from the spectrum which may lead to interesting conclusions about the electronic structure of the molecule. Inversion Spectra For any non-planar molecule there are always two possible configura- tions which correspond to the dextro- and lawo-rotatory forms when these exist and in other cases to the corresponding configurations which would differ in a similar way if identical nuclei could be distinguished.Together they may be considered either as separate molecular species each with their own identical set of energy levels or as one species in which there is an extra double degeneracy of each energy level. Quantum theory prefers the latter view with a duplicity of levels whenever the two equivalent structures can be converted into each other by crossing a finite potential- energy barrier. The two energy levels of a pair have slightly different energies if the potential barrier is finite and it is transition from one of these levels to the other which is the transition involved in the inversion spectrum. It is not possible to correlate the energy levels with the dextro- and lavo-forms or their equivalents. Heisenberg's uncertainty principle shows that the two questions " In which energy state is the molecule 1 " and " Is it in the dextro- or the lzevo-form ? " cannot receive simultaneous answers for the energy state can only be found if the molecule is observed for a time longer than the reciprocal of the inversion frequency which frequency corresponds to the energy difference whereas the two structures are only defined over shorter times since they transform themselves into each other with this frequency by means of the tunnel effect.If the asymmetry resides in a four-covalent atom as in tartaric acid interconversion is only possible by the rupture of a chemical bond and the potential barrier opposing racemisation is very high and the inversion frequency correspondingly low probably a matter of reciprocal centuries. But when no bond rupture is required the barrier may be lower as in pyramidal molecules such as NH, where interconversion is achieved when the nitrogen atom passes through the plane of the hydrogen atoms and for NH the inversion frequency lies at 23,000 Mc/s.or 0.8 cm.-l in the ground state. The NH3 inversion fundamental is the only one already observed all others being predicted a t lower frequencies. It was fist found by C. E. Cleeton and N. H. Williams 5* in 1934 and has recently been reinvestigated by many workers 33 36 61 who use the high resolving power available with A. Nordsiek PhgsicaJ Rev. 1940 58 310. Ibid. 1934 45 234. '9 Ibid. 1947 71 909. 61 W. E. (hod ibid. 1946 70 213. L 142 QUABTERLY REVIEWS modern microwave techniques. It is now found that the band can be resolved into a large series of fine structure lines which correspond to the various rotational energy levels of the molecule except those with K = 0 which are inactive.The dependence of the exact inversion frequency on the rotational level arises from the distortion of the molecule by centri- fugal forces and the consequent alteration of the height of the potential energy barrier to which the inversion frequency is very sensitive. The detailed pattern agrees very well with the theoretical predictions of D. M. Dennison and G. E Uhlenbeck 52 and other^.^^-^^ The lines have been very accurately measured 56-58 and are useful as secondary frequency standards. The frequency of the 3,3 line has been used 59 to control the rate of an electric clock to an accuracy of 1 in 107. Further 14N has a spin of 1 and a quadrupole moment and there ia a consequent hyperfine structure 51 6 o y 61 to many of the fine structure lines while the absence of this fine structure with 15NH3 confirms that the spin of l5N is 1/2.A perturbation to the hyperfine structure arises 62 from the energy of the nuclear magnetic moment in the magnetic field due to the rotation of the whole molecule; the corrections under this head are only of the order of 30 Kc/s. or crn.-l which indicates the fineness of detail obtainable. Stark and Zeeman effects have also been For other molecules in general the inversion frequency may be taken as zero and reference to the intensity formula above shows that as vij -+ 0 64 v Av vijpij + :(F+iiv) Half the molecules are in the lower state and I pij 12 = p2/3 and a factor c/v is introduced to give the attenuation per wave-length which is then where n is the number of molecules/c.c.The last parentheses contain the frequency-dependent factor which is a very broad peak with a maximum when Y = Av that is when the measuring frequency equals the collision frequency. This will be a t about 1 em.-' for most gases at atmospheric pressure. Experimental verification of this formula is diflicult since there Physical Rev. 1932 41 313. H. Sheng E. F. Barker and D. M. Dennison ibid. 1941 60 786. 64 L. N. Hadley and D. M. Dennison ibid. 1946 70 780. 6 6 H. H. Nielsen and D. M. Dennison ibid. 1947 72 1101. 66 W. E. Good and D. K. Coles ibid. 1947 71 383. M. W. P. Strandberg R. L. Kyhl T. Wentink jun. and R. E. Hillger ibid. Idem ibid. p. 639. p. 326. bB Chem. Eng. News 1949 27 162.6o B. P. Dailey R. L. Kyhl M. W. P. Strandberg J. H. Van Vleck and E. B. 61 J. W. Simmons and W. Gordy ibid. 1948 73 713. 62 R. 8. Henderson ibid. 1948 '74 107. O3 D. K. Coles and W. E. Good ibid. 1946 70 979. 64 C. K. Jen ibid. 1948 74 1396. Wilson Physical Rev. 1946 70 984. WEW'FEN ROTATION SPECTRA 143 is an implied assumption that the rotational transitions do not contribute to the absorption at these frequencies and pressures which is not justified except for the lightest gases. However much of the microwave absorption of polar gases 65 a t atmospheric pressure must be formally attributed to the inversion spectrum while for ammonia under several atmospheres pressure 66 the inversion frequency falls effectively to zero but yet the inversion spec- trum is still distinct from the rotation spectrum.Oxygen.-Since the oxygen molecule has no electric dipole moment the simple rotational absorption is absent. However the ground state is a triplet state with two unpaired electrons the three levels corresponding to the different orientations of the parallel electron spins and consequently of the magnetic dipole moment with respect to the axis of the molecule. Transitions between the three levels can be induced by a high-frequency magnetic field and so there is a series of absorption lines 67 which lies near 2 crn.-l i.e. 5 mm. The different lines are due to different rotational states but only partial resolution 68-70 has so far been possible. Condensed Systems Debye Fomula.-For some purposes liquids may be considered as the limit of gases a t high pressures and it is interesting to see how the absorp- tion of liquids can be obtained as a limiting case from the formula of Van Vleck and Weisskopf quoted above.The detailed quantum levels of rota- tional energy are lost in the liquid state and so all vij -+ 0. This condition is similar to that postulated for the inversion spectrum of heavy molecules except that the rotational levels are also involved and in all cases I ,uij I = p2 md the rotation and inversion spectra are no longer to be differentiated. The corresponding attenuation per wave-length is 8 n W ( V A V .> 3kT v2 + (Av) which is just three times that due to the inversion spectra alone. These are not the most common symbols for describing the absorption of liquids and the frequency may be replaced by the angular frequency co which is 3zv the collision frequency by the time of relaxation z which is l/(2n Av) and the attenuation coeEcient per wave-length by the imaginary part of the dielectric constant E" which introduces a factor 1/(27z) ; with thesc substitutions the formula is This differs slightly from that first obtained by P.Debye'l? 72 for ths dielectric loss of liquids since it has been tacitly assumed that the real 65 66 67 68 J. E. Walter and W. D. Hershbsrger J . Appl. Physics 1946 17 814. B. Bleansy and J. H. N. Loubser Nature 1948 161 522. J. H. Van Vleck Physical Rev. 1947 71 413. R. Beringer ibid. 1946 70 53. 99 H. R. L. Lamont ibid. 1948 74 353. 70 M. W. P. Strandberg C. Y. Meng and J. G. Ingersoll ibid. 1949 75 1524. 71 " Polar Molecules " 1929 Chemical Catalogue Co. 72 P. Debye and H.Sack Marx's " Handbuch der Radiologie " 1934 VI Part 2,69. 144 QUARTERLY REVIEWS part of the dielectric constant is unity an assumption entirely justified for gases at low pressures but not for condensed systems. The equivalent Debye formula which takes this into account is 73 (e0 + 2)(~ + 2)4nnpa or 27kT (1 + ,.,a> &“ = where E~ is the dielectric constant a t zero frequency and E that at a fre- quency high compared to the absorption range the previous case is the special case when q,-hecO = 1 and the formula relates to the Clausius- Mosotti form of the internal field. By virtue of the relationship between the dipole moment and the dielectric constants an alternative form is E” = (Eo - E U J ) U ) t / ( l + o 2 t 2 ) Further the absorption is necessarily related to the dispersion that is the change of dielectric constant with frequency and the formula which corresponds is &‘ = E + ( E o - &UJ)/(l + c o 2 z 2 ) and these two formula may be considered as the constituent parts of the complex equation & = &’ - i&f/ - 8 + (eo - &,)/(1 + iot) This formula which applies to all condensed systems can be obtained in a variety of ways,71$ 74-76 each.of which throws into relief a different aspect of the parameter z. Although effects due to the molecular moment of inertia are usually negligible the more exact formula is T7# T8 where zf = Il(2kTz). In the above forms of the loss equations it has been assumed that only one relaxation mechanism and so one relaxation time is operative; for simple systems this is often true but if it is not the several contributing terms must be summed so that & = Em + c Kp/(l + iwtp) where K? describes the amount of the dispersion associated with the relaxation time z,,.If the dielectric constant is of this form there is an identity 79 E = &UJ + (Eo - E U J ) / ( l + icuz)(l + iot’) r 1; (e”/w)do = n(eO - eUJ)/2 which relates the area under an 8”-lnw curve to the total dispersion (e0 - e),. This last equation may be a useful check on the accuracy of any loss measurements and it can also be made to show if a sufficient 73 Compare the introduction of a similar factor in ultra-violet absorption inten- sities N Q. Chako J. Chem. Physics 1934 2 644. 74 W. A. Yager Physh? 1936 7 434. 75 W. Kauzmann Rev. Mod. Physics 1942 14 12. 76 R. E. Powell and H. Eyring “ Advances in Colloid Science ” 1942 Interscience 77 V.A. Dmitriev and S. B. Gurevich J. Exp. Theor. Physics U.S.S.R. 1946 16 7 9 G . M. L. Sommerman J . FrankZin Inst. 1935 219 433. Vol. I 213. 937. 78 J. G. Powles Trans. Farmlay SOC. 1948 44 802 844. WHIFFES ROTAmON SPEOTRA 145 part of the frequency range has been covered to obtain all the rotational absorption region. In some cases this area may be more accurately known than the total dispersion and may be used to calculate this quantity and therefrom dipole moments. This method for obtaining moments has the advantage that no determinations of the electronic and atomic polarisation are required. The value 0.19 D. for the dipole moment of p-cymene 80 is considered as good81 as the static dielectric constant value. Determination of ‘Relaxation Times.-In principle if there is only one relaxation time it can be obtained from the measurement of E’ or E” at a single frequency as i t is the only unknown in the equations for these quantities if c0 and E are known when using E” there may be ambiguity arising from the two solutions of the quadratic equation for z .but this may be resolved if the sign of de”/dcu is known.At low frequencies which were all that were available to many early investigators E’ is not very sensitive to the relaxation time but cf‘ or the loss is directly proportional to it. There is no way of deciding from low-frequency measurements if one relaxation time is sufficient to dercribe the loss or whether a distri- bution is required in which case the mean time 11 Krzr/lc K, is obtained. When the whole dispersion region can be covered i t is possible to com- pare the experimental curve with the theoretical and determine the distri- bution of relaxation times.Fig. 3 shows the variation of E‘ and E” with r r frequency for material with a single relaxation time ; the curves are drawn to a logarithmic frequency scale which makes the loss curve sym- metrical about the frequency at which wz = 1. The same curves are ob- tained if el and E” are plotted against In z for a fixed measuring frequency ; this is effectively obtained by measur- ing 8’ and e” as a function of tem- perature since l n z is often propor- tional to the temperature although some distortion is introduced from the variation of the dispersion (cO - E,) with temperature. Providing these curves are sharp that is t varies rapidly with T it is fair to assume that the maximum of the curve is given by COT = 1 and many relaxa- tion times have been obtained in this log cot FIG.3 (d - em) and E“ plotted against log 07. W&Y since it is easier experimentally to change the temperature than to cover a wide frequency range. From the height of such curves some idea of the distribution of relaxation times may be obtained since if there is but one relaxation time the maximum of E” should be (eo - ~ ) / 2 . D. H. Whiffen and H. W. Thompson Trans. Farday Soc. 1946 42 A 122. R. J. W. Le F6vre ibid. p. 162. 146 QUARTERLY REVIEWS Several types of distribution for z when there is more than one relaxa- tion time have been proposed including a Gaussian but in his discussion Kauzmann 75 has shown that they all predict very similar absorption curves and no one form is to be preferred on experimental or theoretical grounds.In some systems two or more dispersion regions are observed.82s 83 Molecular Rela'xation Times.-The dielectric relaxation times as intro- duced above describe the decay of induced electric polarisation and must be related to molecular behaviour if knowledge of intermolecular forces is to be obtained. An axis fixed in a body such as a rigid molecule subject to frictional forces and to the fluctuations of Brownian motion has a relaxa- tion time which describes the average rate of randomisation of the axis after thbremoval of an aligning mechanism. Quantitatively the amount of alignment or randomisation is measured by the length of the projection of the axis in question on the external direction.In particular if the internal axis coincides with a molecular dipole moment and if the projec- tion of the dipole moment is a measure of the electric polarisation then the relaxation time of the axis which may be called the molecular relaxation time and the relaxation time of the dielectric polarisation are the same. This is not true with the Clausius-Mosotti assumption for the internal field according to which the polarisation is not so simply related to the alignment of dipoles and in this case the molecular relaxation time is smaller by a factor ( E + 2)/(c0 + 2). For nitrobenzene this factor is 1/8 and for dilute solutions when E ~ ~ E i t is 1 and so the two times are the same. L. Onsager's dielectric theory,8* in which the Clausius-Mosotti internal field is abandoned requires the molecular and dielectric relaxation times to be the same even for strongly polar liquids.Other modifica- tions 8 5 p 86 have been proposed and it is uncertain 87 which theory is to be preferred although experimentally the Onsager theory accounts closely for the values of the dielectric constants of polar liquids.88 Consequently it seems better to accept the experimental values of the dielectric relaxation times and discuss them as if they were molecular relaxation times in all cases. For dilute solutions there is no error and in polar liquids the difference can scarcely be greater than a factor of ten. Even the above discussion is somewhat simplified since a molecule with a most general frictional ellipsoid may have three relaxation times 89 cor- responding to relaxation of the three principal frictional axes and unless the dipole moment lies along one of them the three times are involved in the dielectric loss.In many cases especially for symmetrical rigid mole- cules the dipole moment coincides with one axis and so only one relaxation rate is active even when this is not strictly true i t is nearly so or else 8 2 A. Schallamach Trans. Faraday SOC. 1946 42 A 180. 83 P. Girard and P. Abadie Compt. Tend. 1943 216 44. 84 J . Arner. Chem. SOC. 1936 58 1486. Debye and W. R a m Ann. Physik 1937 28 28. 86 R. H. Cole J . Ohm. Physics 1938 6 385. $7 See discussion in Trans. Farday SOC. 1946 42 A 3-40. 88 C. J. F. Bottcher Physica 1939 6 59 F. Perrin J . Phys. Radium 1934 5 497. WHXFFEN ROTATION SPECTRA 147 the three times are almost equal and no rigid molecule has yet been shown to have two or three distinguishable relaxation times on this account.A large range of relaxation times are possible according to the nature of the material and the temperature and Table I1 gives a representative se1ection.lo7 The columns show respectively the material the temperature TABLE I1 Material. Polyvinyl acetate . . Polymethyl acrylate . Ice . . . * . . Pentachlorotoluene Cetyl palmitate in para- Glycerol . . . . . , (10,000 atm.) . Octyl alcohol . . . Ethyl alcohol . . . Acetic acid in dioxan . Nitrobenzene . . . Nitrobenzene in benzene (crystal) flfin wax Water . . . . . 5.0~4-NaCl solution . . Heavy water (DaO) . . Toluene . . . . . Chloroform . . . . Chloroform in n-heptane Temp. 60' 25 - 45 - 20 - 5 - 42 10 - 40 20 0 0 20 20 20 20 20 75 20 21 20 - 80 20 25 25 ty sees.3 x 10-3 3 x 10-3 2 x 10-3 1.9 x 10-4 4 x 10-6 1.6 x 1 0 - 4 1.6 x 5 x 10-8 8 x 6 x lo-' 1-4 x 3 x 10-1' 5 x 10-11 1-2 x 10-11 3.2 x 9-6 x 7.4 x 10-12 1.2 x 10-11 4 x 10-11 7 x 10-12 7 x 10-12 3 x 10-12 1.6 x 10-4 1.6 x 10-9 vmar.. 60 c/s. 60 c/s. 80 c/s. 1 Kc/s. 40 Kc/s. 100 Kc/s. 3 Mc/s. 20 Mc/s. 250 Kc/s. 100 Mc/s. 1000 Mc/s. 5500 Mc/s. 3000 Mc/s. 14,000 Mc/s. 49,000 Mc/s. 17,000 Mc/s. 21,000 Mc/s. 13,000 Mc/s. 4000 Mc/s. 23,000 Mc/s. 22,000 Mc/s. 50,000 Mc/s. 1 Kc/s. 1 Kc/s. &.. 1 Ref. 5000 km. 5000 km. 4000 km. 300 km. 7 km. 300 km. 3 km. 300 km. 100 m. 15 m. 1200 m. 3 m. 31 crn. 6 cm. 9 em. 2.2 cm. 6.1 mm. 1.8 cm. 1.4 cm. 2-3 cm. 7 cm. 1.3 em. 1-3 cm. 6 mm. 90 91 92 92 93 94 94 110 95 96 97 98 99 100 83 101 102 103 104 104 105 104 80 80 106 80 D.J. Mead and R. M. FUOSS J . Amer. Chem. SOC. 1941 63 2832. 91 Idem ibid. 1942 64 2389. O 2 E. J Murphy Trans. Electrochem. SOC. 1934 65 133. 93 J. Lamb Trans. Paraday SOC. 1946 42 A 238. 9 4 A. H. White B. S. Biggs and S. 0. Morgan J . Amer. Chem. SOC. 1940 62 16. 96 W. Jackson Proc. Roy. SOC. 1935 A 150 197. O 6 S. Mizushima Physikal. Z. 1927 28 418. 97 W. E. Danforth jun. Physical Rev. 1931 38 1224. 98 P. Girard and P. Abadie Compt. rend. 1942 215 84. 99 H. C. Bolton Proc. Physical Soc. 1948 61 294. loo G. Potapenko and D. Wheeler jun. Rev. Mod. Physics 1948 20 143. lol W. Jackson and J. G. Powles Trans. Faraday SOC. 1946 42 A 101. lo2D. H. Whiffen and H. W. Thompson ibid. p. 114. lo3 F. J. Cripwell and G. B. B. M. Sutherland ibid. p. 149. lo4 C. H.Collie J. B. Hasted and D. M. Ritson Proc. Physical SOC. 1948 lo6 J. B. Hasted D. M. Ritson and C. H. Collie J. Chem. Physics 1948 16 1. lo* W. P. Connor and C P. Smyth J . Amer. Chem. SOC. 1943 65 382. lo' For fuller bibliographies see refs. 1 75 76 108 109. lo8 W. Ziegler Physikal. Z. 1934 35 476. lo* S . Glasstone K. J. Laidler and H. Eyrhg " Theory of Rate Processes," 1941 60 145. McGraw-Hill 110 R. W. Sillass Proc. Roy. Soc. 1938 A 169 66. 148 QUAR!FERLY REVIEWS the dielectric relaxation time the mean value being taken if there is a distribution the frequency of the maimum in E" the corresponding wave-length and the reference. Amorphous Solids.-Some of the longest relaxation times are found with various amorphous solids and constructional materials such as wood paper ebonite etc.Very often there is a series of relaxation times and the inter- pretation is complicated by the existence of ionic conductivity and Maxwell- Wagner dispersion as well as dispersion due to rotating dipoles. More definite information is to be obtained from measurements on a series of esters such as cetyl palmitate dissolved in paraffin wax.95' 110 Such solutions show sharp maxima in the absorption if the loss is plotted against temperature at a fixed frequency indicating a rapid change of relaxation time with temperature. The solvent wax is non-polar and so it can have no loss at any frequency. The heights of the maxima show that the spread of relaxation times is small and that as regards the molecular picture it is possible to ignore such a narrow distribution and discuss the relaxation time as if all the solute molecules behaved identically.The characteristics of this system namely a narrow distribution of relaxation times and a large temperature coefficient are also shown by solid chlorinated diphenyls 111' 112 and supercooled isobutyl bromide in a glassy state.113 The inverse OF the relaxation time may be considered as the average rate of rotation of the dipoles and the rate of rotation in its turn may be treated as a unimolecular reaction rate and reaction-rate equations applied.114 Consider a chain molecule such as cetyl palmitate with a polar group for which the resolved part of the dipole moment perpendicular to the chain length is large. The solvent wax has a quasi-crystalline packing of chains and it must be supposed that the long cetyl palmitate molecule can replace one of the wax chains to give effectively a mixed crystal.The dipole moment may point in one of two equally good directions controlled by the packing of the hydrocarbon chains but intermediate positions are unfavourable since in them the quasi-crystalline structure is disturbed If an electric field is applied one of the two positions of the dipole moment will be favoured by the field and molecules from the less favoured position will tend to rotate into the other position so that the induced electric polarisation is established. To reach the new position the molecule rnvsi pass through the intermediate most unfavourable state of higher energy. To surmount this potential energy barrier of height E the molecule will require this amount of thermal energy; energy derived from the electric field does not enter this part of the discussion since the measuring field is always weak.The chance of any molecule having this amount of thermal energy is exp. (- E/RT) So the rate of rotation which is proportional to the probability of a molecule reaching the top of the barrier will have an exponential dependency on the height of the barrier and the absolute 111 W. Jackson Proc. Roy. SOC. 1935 A 153 158. 112 A. H. White and S . 0. Morgan J . Franklin Imt. 1933 216 635. 113 W. 0. Baker and C. P. Smyth J . Ohem. Physics 1939 7 574 114 F. C. Frank TTWM. Pamday Soc. 1936 32 1634. WHIFFEN ROTA!PION SPEUTRA 149 temperature. Indeed a linear relation between Inz and 1/T is found and the barrier height is of the order of 15-30 kcals./mole for the esters dissolved in paraffin wax.The neighbouring molecules to that concerned in rotation d probably be displaced in the activated state and some of the thermal energy will be distributed among these neighbours and what is effectively a local melt- ing occurs. An increased reaction rate results for the probability of the energy E being distributed in several degrees of freedom is greater than its probability of appearing in one. The approach of the statistical theory of reaction rates 7 % log is slightly different and this requires where AF* is the increase of free energy of activation and AH* is the increase of heat content of activation corresponding essentially to E while the fact that several molecules must be disordered in the activated state requires there to be a large increase of entropy on activation and thus a large positive value of AS*.Both aspects of the theory require a rate faster than the height of the barrier would indicate and indeed the figures require76 about half a dozen molecules to be involved in the rotation process. -stals.-For most crystals the energy barrier preventing molecular rotation is very high no rotation occurs and the static and high-frequency dielectric constants are the same showing that no dispersion or absorption exists. In a few crystals rotation of polar molecules is possible and in these cases high static dielectric constants are found which have the same values as for the molten liquid apart from differences arising from the change of density. Studies of the change of dielectric constant with tem- perature a t low frequencies often show that the rotation remains facile until some low-temperature transition point below which rotation is impos- sible as is indicated by the low dielectric constant.Examples of this behaviour are hydrogen sulphide and the hydrogen halides.l16 It is neces- sary to use high frequencies to find dielectric loss and values of the relaxation time above the transition temperature but this has been done with camphor and some of its derivatives 116 117 and with a series of methylchlorobenz- ene~.~* Interpretation as in the last section applies and the energy barriers are about 10 kcals./mole and the entropies of activation are small indicating that the neighbours of the molecule which rotates are scarcely disturbed. Ice shows only a small change of static dielectric constant on formation from water a t O" but at high frequencies and lower temperatures absorp- tion and dispersion are observed.g2* e3 118 It is not known if the align- ment of dipoles by an electric field is brought about in ice by molecular rotation or by proton jumping that is by movement of a proton to a second potential minimum in the hydrogen bond.C. P. Smyth and C. 8. Hitchcock J . Amer. Chem. Soc. 1933 55 1830. 116 W. A. Yager and S 0. Morgan ibid. 1935 67 2071. 117 A. H. White and W. S. Bishop ibid. 1940 62 8. 11* C. P. Smyth and C. S. Hitchcock W. 1932 54 4631. 150 QUARTERLY REVIEWS Polymers.-Since various polymers are useful dielectric materials for condensers cables and other electrical equipment a large number have been examined among which polythene and polyfiuoroethylene stand out as having specially low loss at all frequencies.In general there is a spread of relaxation times arising from the inhomogeneity of the material on a microscopic scale while the area under the loss curve gives the total amount of polar substance present. W. G. Oakes and R. B. Richards l 1 9 have used this area to distinguish between one -CC12- group and two -CHCl- groups in the chain of chlorinated polythene and have shown how the relative amounts of these two types depend on the conditions of chlorina- tion. R. M. Fuoss and others in a series of papers 120 on the electrical properties of solids have measured a number of polymers. Among their conclusions are the facts that plasticisers reduce the relaxation times con- siderab1y,l2l that the mean relaxation time for polyvinyl chloride 122 and polyvinyl acetate is proportional in each case to the degree of poly- merisation and that the loss peaks of polyvinyl acetate are sharp which indicates a narrow distribution of relaxation times.Liquids.-There is no very sharp division between liquids and some amorphous solids and there is a wide range of relaxation times down to 3 x 10-l2 sec. as indicated in Table 11. However there is difficulty in choosing a satisfactory molecular model as a basis for the interpretation of the results. The application of a rate equation such as any of those used for solids to mobile liquids is danger- ous for in the derivation of such equations is the assumption75 that the exponential term is large and that the energy barrier is what chiefly controls the rate. Further the Eyring expression lo9 (kT/h)e-Ap*R/T is the rate of crossing the energy barrier while l/z to which it is equated is the rate of randomisation ; these two are only equal if the angular posi- tion of a molecule after passing the barrier is independent of its initial position before activation.The assumption that this is so is more easily justified for high barriers than for small ones. In contrast to the assumptions which presume an appreciable energy barrier to exist experimental results leave doubt as to whether there is any potential barrier to be crossed for the most mobile liquids. The relaxation times for solutions of polar molecules in n-heptane 80 all show an activation energy E defined by z = A exp. (E/RT) of 1.8 kcals./mole which is obtained from the slope of a lnr-l/T graph.Arising out of the factor kT/h which contributes to the slope E must be reduced by an amount RT to give AH* of the Eyring expression which means AH* = 1-2 kcals./mole for the hep- tane solutions. It is this quantity which should indicate the height of the potential barrier opposing rotation and it is indeed small. But further it ll@ Trans. Faraday Soo. 1946 42 A 197. 190 For Part S V see D. J. Mead and R. M FUOSS J . Arner. Chern. Soc. 1945 67 l*lD. J. Mead R. L. Tichenor and R. M. Fuoss ibid. 1942 64 283. 122 R. M. Fuoss ibicE. 1941 65 2401. 1566 and thence the earlier papers. WBIFE’EN ROTATION SPECTRA 151 is assumed that the barrier height is invariant with temperature and this is a condition which is more likely to be fulfilled at constant volume which implies constant intermolecular distance than at constant pressure which was the condition for the experiments.On any mechanism relaxation times will almost certainly increase with pressure as is observed for glycer01,~7 and so it must be expected that at constant volume the temperature coefficient of the relaxation time will be even smaller and in fact correspond to prac- tically zero as the value of the height of the potential barrier. Further it is diEcult to reconcile such small temperature coefficients with the existence of patches of crystalline order in the liquid state unless these patches are destroyed and re-formed in a time which is short compared to the relaxa- tion time of a molecule in the crystalline lattice; under these conditions the molecules would be able to rotate when the crystallite is destroyed and need never cross the potential barrier.The same small value about 1-2 kcals./mole or less of any presumed energy barrier is found for the pure liquids toluene and o-xylene,*O and in the latter case persists unchanged right down to the freezing point. The opposite extreme from considering relaxation as a discrete jump or series of jumps is the consideration of a molecule as a sphere rotating in a viscous medium. Debye’s 7 1 derivation of a relaxation time requires it to be given by z = c/(2kT) where c is a frictional constant. For macro- scopic spheres of radius a in a liquid with a viscosity coefficient q c = 83tqas whence z = 4np3/(kT) and Debye suggested that the introduction of some reasonable molecular dimension for the radius of the molecule and the macroscopic coefficient of viscosity should give the right order of magni- tude for the time of relaxation.As an order of magnitude calculation this is often satisfactory but attempts to force the expression into an exact calculation have led either to the introduction of a microscopic viscosity coefficient defined so that the formula shall be exact or else to the con- clusion that the molecular radius differs widely with solvent and differs somewhat with temperature. An example is nitrobenzene 1 whose apparent radius varies from 0.31 A. in Shell oil to 1.88 A. in hexane as calculated from the macroscopic viscosity coefficients. But despite this impossibility of correlating the macroscopic viscosity and the relaxation time directly the two processes must depend on similar intermolecular shear forces and some parallelism is to be expected.A gross increase of viscosity is accom- panied by an increase of the relaxation time which is relatively small. E’or instance benzophenone,l01 which has a relaxation time of 1.6 x 10-11 sec. in benzene has a mean time of 30 x sec. in a paraffin whose viscosity i8 300 times greater. The parallel nature may often be found in the tem- perature coefficients for the two processes as in toluene *O where the activa- tion energies are 1.9 and 2.0 kcals./mole for relaxation and viscous flow respectively. Alcohols too have the same activation energy about 6 kcals*/mole for the two processes and the relation between relaxation time and viscosity seems to be closer with alcohols 12% than with other liquids both the relaxation time and the viscosity are abnormally great lZs R.Goldammer PhysQaE. Z. 1932 33 361. 152 QUAR!l'ERLY REVIEWS since the intermolecular forces are high as the result of hydrogen-bond formation. This same close correlation appears to hold for water and Collie Ritson and Hasted104 have found that the relaxation time and the ratio of the viscosity to the absolute temperature are closely proportional both for water and for heavy water ; the derived radius with the Debye expression is 1-38 A. Extension of the measurements to salt solutions 105 shows that the effect of dissolved ions is to reduce the dielectric constant as a result of the removal of the water molecules in the solvation shell of the ions from the orientation process and also to reduce the relaxation time of the remainder of the water.The latter effect may be due to the increased ease of destruction of small crystalline patches as a consequence of the disturb- ance due to the ion impurity; if then relaxation occurs only when the molecule is outside the crystallite it is facilitated. Schallamach 82 has observed the loss in some mixtures of polar compounds and finds that even though the relaxation times of the two components are very different yet the mixture usually shows but one region of absorption lying in frequency between the absorption of the two components separately. If one of the components is associated and the other non-associated there is a greater tendency for two separate regions to be observed as exempli- fied by n-propyl alcohol and isoamyl bromide in contrast to the one region as shown by n-propyl alcohol and glycerol mixtures.From this review of the knowledge of rotation spectra it can be seen that the work with gases rests on a firm quantum-theoretical basis and that it remains to measure the properties of further simple molecules with as many isotopic species and as great an accuracy as possible The results of special interest to chemists will be accurate bond distances accurate dipole moments and possibly electric field gradients a t nuclear positions. The interpretation of the spectra of asymmetric top molecules is often possible and further work should lead to conclusions concerning molecules in which some internal rotation is present. For condensed systems the relaxation process of molecules in the solid state can be described in terms of energy barriers but for liquids an adequate molecular picture seems to be lacking.

ISSN:0009-2681    

DOI:10.1039/QR9500400131

出版商:RSC    

年代:1950

数据来源: RSC

摘要:

LIMITING DENSITIES By R. WHYTLAW-GRAY F.R.S. (EMERITUS PROFESSOR LEEDS UNIVERSITY) THE limiting density principle due to Lord Rayleigh and to D. Berthelot,2 postulates that a t sufficiently low pressure all gases and vapours obey the gas laws. This ideal state in which PV = RT exactly lies far below atmospheric pressure. If instead of working with gases a t pressures not far removed from atmospheric we could investigate the change in PV over a range say from 0.25 to 1 mm. pressure it would for the majority of gases be exceedingly difficult to detect any deviations even supposing that measurements could be made with the same percentage accuracy as a t higher pressures. In this state all gases would obey Boyle's law all would have the same coefficient of expansion and since Avogadro's law would be strictly true a comparison of their densities would give an exact measure of their relative molecular weights.Although this ideal state is in a pressure region in which exact measure- ments of density or compressibility are not practicable precise data are available for a large number of gases between the pressure limits of 800 and 200 mm. of mercury. Very few have been obtained below 100 mm. Now for all the permanent gases i.e. those gases which a t normal temperatures are well above their critical temperatures and for other gases as well it has been found that over this pressure range PV is a strictly linear function of P so that if we suppose this linearity to persist downwards we can find the value of PV for highly rarefied gases right down to the limit when P approaches zero.That is to say that although we cannot make direct measurements on rarefied gases with high percentage accuracy we can measure the deviation a t higher pressures and from this calculate the behaviour a t very low pressures. Its normal density is 1.42896 g. per litre at 0" and 1 atmosphere a t sea level and latitude 45". Hence 32 g. occupy 32/1.42896 = 22.3939 litres under these conditions. That is PV for this gas at 1 atmosphere is 22.3939 1. atm. mole-I; on expan- sion PV increases linearly and on extrapolation to the limit becomes 22.4145 1. atm. mole-I. Since the gas is now in the ideal state it is clear that 1 g.-mol. of a perfect gas (molecular weight = 32) would a t N.T.P. occupy 22.4145 litres instead of 22.3939 litres actually occupied by 1 g.-mol. of oxygen. One g.-mol.of any other gas would in the limit when P = 0 have the same PV product but of course each would have its own charac- teristic value when P = 1 atmosphere. This is shown graphically in the figure. Now for oxygen the ratio of PV at 1 atmosphere to PV at zero pressure at 0" i.e. (PV),/(PV),=22~3939/22.4145=0~99908=(1-0~OOO92)=(1 - A ) 163 Consider for example the gas oxygen. Proc. Roy. SOC. 1892 SO 448. Cmpt. rend. 1898 126 954. 154 QUARTERLY REVIEWS which is the deviation from Boyle’s law when oxygen of unit PV a t zero pressure is compressed to 1 atmosphere; A is called the compressibility coefficient and from the above relationship is clearly [ ( P V ) - ( P V)&’(P V), and since PV varies linearly with P the PV a t any intermediate pressure can be calculated. Thus we get the general expression pv = (ppv),(l - A p ) where pv and (pv) are the values for any one mass of gas a t any pressure in the vicinity of 1 atmosphere and at zero pressure p being expressed in atmospheres.The density of a perfect gas of molecular weight 32 is then the density of oxygen x 22.3939/22-4145 = 1*42896(1 - A ) = 1.42765 which is the limiting density of oxygen or the density oxygen gas would have if it behaved as a perfect gas. Gram- molecular volume. - U.005 22.428 Hydrogen. 22.394 Oxygen. 0000 22.4 145 Perfect gas. 0.005 22- 263 CaPbQf? dioxide. 22.256 Nitrous oxide. 0010 22.194 Silkon tetraf/uor/’dee. i 22477 Ammonia. 0.015 0020 2 1.891 Sulphur dioxide. 0,025 21.856 Dimethyl oxide. (methyl efher} 0 0.2 0.4 0.6 0-8 1.0 Atmospheres. Hence if LA L and Lz are the normal densities of three gases of molecular weights M, M, and M, and their respective compressibility coefficients are A, A, and A, then their limiting densities Lim.Lima and GL. are Li(1 - A,)? Li(1 - A 2 ) and Lz(l - A3) so that Limiting density Normal density = l - A - - ._____._____ - ___ Suppose L M, and A refer to oxygen then the molecular weights of the other two are given by and M = 32Li(1 - A3)/Ll(l - A ) or since 32/Li(1 - A ) = 22.4145 1. atm. M = 32Li(1 - A2)/L1(1 - A,) M = 22.4145L;(l - A,) and M = 22.4145Li(1 - As) WHYTLAW-GRAY LIMM"G DENSITIES 155 Frequently the compressibilities are expressed in another notation introduced by Ph. A. Guye,3 in which PoVo/PIVl = 1 + A instead of PIV1/PoV = 1 - A for the pressure range of 1 to 0 atmosphere. Hence if the 0" isothermals are linear 1 - A = 1/(1 + A) and A = A f ( 1 + A).For gases which exhibit small deviations from ideality A is numerically nearly equal to A but for others the difference is significant. As W. Cawood and H. S. Patterson 4 have pointed out 1 + A refers only to the ratio of PV at 1 atm. to PV at infinite dilution and is inapplicable for intermediate pressures for which Berthelot's expression is more accurate. Guye's notation being used the molecular weight M of a gas is given by and M = 22*4145LB/(l + A ) Normal density 1 1 + A --______ -___ - Limiting density - 1 - A - The question might be asked whether the correction factor 1 + A or 1 - A could not be calculated with sufficient approximation from an equation of state such as that of van der Waals ; for it can be shown that 1 + A = (1 + a ) ( l - b) where a and b are the well-known '' constants " hence M = 22.4145 x L,/(1 + a)(l - b).Unfortunately a and b are not constants and vary with temperature and pressure and when calculated from the critical constants and inserted in this expression give erroneous values for M . Guye in 1905 proposed a method of correction which he named " Reduction des Elements Critique " which gave for the permanent gases a satisfactory approximation but it has not proved of general applic- ability. More accurate values of 1 + A or 1 - A can be obtained from the Beattie-Bridgeman equation as J. B. M. Coppock has shown but the various constants in this equation have to be found for each gas from high-pressure measurements. It is clear then that with the help of the limiting density principle we can determine (a) the relative molecular weights of gases directly in terms of the standard oxygen ; (b) the volume of a gram-mole of a perfect gas at the ice point at a pressure of 1 atm.i.e. the value of PV per mole in the equation (c) if the exact value for To the temperature of melting ice is known The temperature of melting ice on the absolute or thermodynamic scale is defined in terms of the coefficient of expansion of a perfect gas which can be deduced from the coefficients of expansion of real gases. Because of the deviations from the ideal the coefficients of expansion at constant pressure (K) and constant volume (/I) differ somewhat when for any one gas they are measured at the same initial pressure and over the same temperature range. When o( and 18 are determined at lower and lower PV = RT,; the fundamental constant R can be found.J . Chim. physique 1908 6 769. J . Physical Chem. 1933 37 995. 'J. 1933 619. 156 QUARTERLY REVIEWS initial pressures they approach each other and on extrapolation to the limit become identical giving the one value y the expansion coefficient of a perfect gas. The exact evaluation of y is of primary importance in thermometry. Starting with the work of Chappuis much careful research of high accuracy has been carried out to determine this important coeffi- cient in order to find the ice point on the absolute scale and to correct temperatures measured by gas thermometers. D. Berthelot in his classical memoir " Sur les Thermometres 21 Gaz " defined the general relationships between the compressibility coefficients of real gases and oc and /? but accurate thermometry is a field in itself and cannot be discussed here.It must suffice to state that the most recent work according to R. T. Birge,e fixes the ice point on the absolute scale at 273.16" & 0.01" K. In what follows the application of the limiting density principle to the determination of molecular and atomic weights will be considered. The range of the method is however restricted for although upwards of 100 stable gases are known these contain only 22 elements. If it could be extended to include vapours it would be possible to add another 10 or more elements to the total. Its great advantage over gravimetric analysis is that it provides a direct link with oxygen and thus avoids cumulative errors. Of course its use as a check on chemically determined atomic weights depends upon the accuracy with which densities and compressibilities can be measured.Up to the present in spite of much laborious research data for only a limited number of gases are known with sufficient precision to give molecular weights accurate to 1 part in 10,000 or more. This is due in part to errors inherent in gasometric operations and also to the difficulty of preparing many gases in a high state of purity. Then again until recently the form of the PV-P isothermals for the easily condensable gases was doubtful which made exact extrapolation to zero pressure uncertain. The development however of more exact methods of measurement combined with improved ways of low-temperature fractionation and puri- fication now make it possible to obtain more data of high precision.Applications.-Gases the densities and compressibilities of which have up to the present been determined with accuracy are (1) The inactive gases with the exception of krypton and radon. (2) H2 N2 0 2 c o NO N20 CsH4 C2H2 C3Hf3 C3H6 (CH3)20 cHacl CH3F CF, SiF, HCl HBr NH, PH3 SO, H,S and SiH,. Thus apart from (I) for which no analytical data are possible this list covers only the elements 0 (standard) H N C S Si F C1 Br and P. It could be extended to B As Sb Ge Se Te and I by utilising the hydrides or fluorides which with the exception of H,Se7 have not been investigated with the object of finding molecular weights. Notwithstand- ing this limitation physicochemical methods have played an important part in fixing accurate values for many atomic weight ratios.Rep. Progr. Physics 1941 8 110. 7 P. Bruylants and A. Bytebier Bull. SOC. chim. Belge 1912 866. WHYTLAW-GRAY LIMM!lXGl DENSITIES 157 When after the discovery of argon Rayleigh and afterwards Leduc found that the density of pure nitrogen indicated an atomic weight for this element very close to 14 an error in Stas's value 14.044 was suspected. Limiting-density determinations of nitric and nitrous oxides were made shortly afterwards by Ph. A. Guye and by R. VV. Gray independently which together with the analysis of the same oxides definitely proved that this atomic weight could not exceed 14-01. It was then pointed out by Guye lo that this error in Stas's nitrogen indicated errors in many of the fundamental atomic weights including that of the sub-standard silver.These were only corrected after many years of careful research begun by Richards and his school a t Harvard and completed by G. P. Baxter 0. Honigschmid and others. Later the limiting-density method applied to carbon monoxide 11 shortly after the discovery of I3C indicated a proportion of about 1% of this isotope and an atomic weight of 12.01 for the mixed element signifi- cantly higher than 12.00 the then-accepted value. Mass spectroscopy as well as gravimetric analysis has since confirmed the higher value. Again accurate measurements carried out on highly purified specimens of the inactive gases have with the exception of krypton given atomic weights in close concordance with mass spectroscope values. Then there is the case of phosphorus the atomic weight of which when first computed from Aston's earlier estimate in 1927 was found to be 30.978 a value so much lower than 31.02 well established by chemical methods that the existence of higher isotopes was suggested.A careful determina- tion of the limiting density of very pure phosphine by M. Ritchie l2 in 1930 led however to 30.977 which was confirmed in 1937 by 0. Honigschmid and W. Menn 13 by the analysis of POCl which gave 30.978 and in 1940 by Honigschmid and 3'. Hir~chbold-Wittner,~4 who analysed POBr and found 30.974. Subsequent and more accurate mass-spectrum determina- tions proved the element to be a simple one and gave the value 30.975. The accepted International value is now 30.98. Usually limiting-density methods give results in close agreement with mass-spectroscope measurements ; another instance is fluorine for which the limiting densities of CF and CH,F indicate a value slightly below 19 in agreement with Aston's latest measurements.There is one striking instance however in which the data obtained by the mass spectrograph are in disaccord both with the results of chemical analysis and with limiting-density determinations. It is for the common element silicon for which the value 28.063 has been accepted for many years and which is based on the analysis of SiCl and SiBr (by G. P. Baxter P. F. Weatherill and E. 'W. Scripture) 15 by the standard method of silver titration. The limiting densities of SiH and SiF give respectively 28.112 * Compt. rend. 1905 141 826 ; Mern. SOC. Physique GenAvve 1908 35 615. lo J .Chim. physique 1906 4 181. l l M . Woodhead and R.Whytlaw-Gray J. 1933 846. le Proc. Roy. Soc. 1930 A 128 551. l3 2. a w g . Chem. 1937 235 129. l4 Ibid. 1940 243 355. l6 Proc. Nat. A&. Sci. 1923 68 245. J. 1905 87 1601 ; Thesis Bonn,.1907. M 158 QUARTERLY REVIEWS and 28.105 whilst the most recent and reliable mass-spectrograph values are 289087 l* and 28.086.l' The latter are in close agreement with X-ray and density measurements made recently by T. Batuecas using crystalline quartz who finds Si = 28.081. It might be argued that both limiting- density values are effected by errors which make them too large were it not that Honigschmid and M. Steinheil l8 in 1924 using the same method as Baxter and his colleagues obtained 28.105 and Weatherill and D. K. Brundage l9 in 1932 from the ratio SiC1,/Si02 deduced Si = 28.103.This interesting case clearly requires further investigation. The Measurement of Densities and Compressibilities with High Accuracy Densities.-More attention has been paid to the determination of the density of oxygen than to that of any other of the permanent gases. The relationship between its normal and limiting densities provides data from which the gas constant is calculated as well as the gram-molecular volume of a perfect gas of molecular weight 32. The weight of a normal litre of this gas is hence an important standard. Apart from the early work of Regnault in 1848 up to 1921 ten different workers in various countries have made 170 determinations of the weight of a normal litre. E. Moles 20 has recalculated all these data and reduced them to a common standard and from all he obtains a mean value of 1-42892 & 0.0003 for the weight of a litre of gas at N.T.P.at sea level and latitude 45" (g = 980.616). The mean values of different workers exhibit an extreme variation of 22 x g. but the variations among the single results in any one series are much greater than this. Rayleigh's 21 investigation of the densities of the principal gases which appeared in 1893 set a standard of accurate measurement not always reached by subsequent workers. If we consider his 16 weighings of oxygen prepared by three different methods (chlorate permanganate and electrolysis) we find an extreme variation of 25 x g. and the mean value for a normal litre was 1.42905 but Moles on recalculation reduces this to 1.42894 rfi 0.00003 by making small corrections to the gravity factor the vacuum contraction and the coeEcient used for the density of water for cal- culating the bulb volume.Work carried out since 1921 has furnished another 113 determinations of this important constant from two different laboratories. Thus from Harvard G. P. Baxter and H. W. Starkweather 22 find La = 1.42898 & 0.00003 (mean of 65 determinations) Ln = 1.42897 & 0.00003 (mean of 6 determinations) whilst E. Moles Z3 and his collaborators in Madrid in an extended series of researches from 1934 to 1937 find L = 1.42894 & 0.00001 (mean of 42). lag. anorg. Chem. 1924 141 101. l6 D. Williams and P. Yuster Physical Rev. 1946 69 556. 1' M. G. Inghram ibid. 1946 70 653. J . Amer. Chem. SOC. 1932 54 3932. 2o J . Chim. physique 1921 19 100 ; 2. anorg. Ghem. 1927 167 40. 21 Proc. Roy.SOC. 1893 53 144. 22 Proc. Nat. Acad. Sci. 1924 10 479 ; 1926 12 699. 23 " Poids Molec. et Atomique des Gaz " Collection Scientifique Paris 1938 ; Instit. Internat. Co-oper. Intellect. pp. 30-35. WHYTLAW-GRAY LIMITING' DENSITIES 159 This last value has been corrected for adsorption and Moles contends that a similar correction should be applied to the Harvard results bringing their final figure down to 1.42895 and he claims that the normal density of oxygen is now established with an accuracy of 1 part in lo5. This certainly represents the extreme limit of accuracy of which the method is capable when carried out with every modern refinement. However much one may be disposed to question this claim it must be admitted that the individual values in his various series show a degree of concordance greater than any that has hitherto been obtained.Further the admirable experi- mental methods he has developed and perfected over a number of years support his contention that systematic errors have been eliminated. Compressibilities and Limiting Densities.-Two general methods are in use for determining the deviation from Boyle's law a t constant temperature below one atmosphere. In (a) the isotherm or volumetric method an unknown mass of gas is confined in a volumeter over mercury and con- nected with the lower chamber of a manometer carrying a point and forming the dead space. The volume is varied by withdrawing mercury from the volumeter and the corresponding pressure read after defining the volume of gas in the dead space by setting the point to the lower meniscus.The volumeter may be in the form of three or more bulbs connected by capillary tubing bearing marks which define predetermined volumes or it may be a single bulb and the volume increment is measured by weighing the mercury. Provided the temperature of the volumeter and the manometer is kept constant the main error is in the pressure readings for it is very difficult even with a good cathetometer and scale to reach an accuracy greater than & 0.01 mm. Hence PF determinations are not usually attempted much below 200 mm. In (b) the densities are determined in the usual way at a series of pres- sures say at 1 0.75 0.5 0.33 atm. and a set of values obtained of W/PV where ?V is the mms of gas filling a density bulb of volume F at pressure P. In this way a series of values of density per unit pressure is obtained from which on extrapolation W/PV for P = 0 which is the limiting density can be obtained directly.This method due originally to Guye has been largely used by recent workers. Thus Baxter and Starkweather have employed it with success for neon argon oxygen and nitrogen and Moles for a whole series of gases. Its accuracy is limited not only by the pre- cision reached in reading pressures but also by the difficulty of weighing gases a t low pressures especially those of low molecular weights such as methane and ammonia. The various values for oxygen obtained by the two methods by different workers are shown in table on next page. There is close agreement in the mean values although method (a) shows a divergence of 1.2 parts in 10,000 between the extremes.In addition to the two foregoing methods the 1 + il values can be deduced by extrapolation from measurements made at high pressures between for example 20 and 100 atmospheres. At first sight this proce- dure would appear of doubtful validity on account of the large extrapolation 160 QUARTERLY REVIEWS Compressibility of oxygen at 0" at pressures below 1 atm. Observer. 24GrayandBurt . . . . 25 Jaquerod and Scheuer . . 2* Guye and Batuecas . . . 27 Batuecas Maverick and Schlatter . . . . asHeuse and Otto. . . . 29Casado . . . . . . Mean Isotherm method (a). 1.00097 1.00097 1*00085 1-00087 1.00097 1-00089 1.00092 Observer. 29aMoles and Toral . . . 29a Moles and Roquero. . . 22 Baxter and Starkweather . 29Casado . . . . . . Mean Density method (b). 1.000917 1.000933 1.000934 1.00091 1 1.00093 necessary but it must be remembered that the actual deviations from the ideal state are so muchzlarger at high pressures that their magnitude can be determined with a higher percentage accuracy.Actually as W. Wild 30 showed in 1931 the compressibility coefiicients of the permanent gases from high-pressure data agree very closely with those measured directly a t low pressure and recently C. S. Cragoe 31 in a critical analysis has recsl- culated the 1 + A values from high-pressure data for He A N, 02 and H and assessed their probable errors in comparison with the data of Baxter and Starkweather obtained directly by the density method. His results are striking and have a lower probable error than the low-pressure measure- ments. Thus for nitrogen from the data of A.Michels H. Wouters and J. de Boer,32 the values of il x 105 for nitrogen by different mathematical treatments are 45.42 45-25 45.32 & 0.31 ; 45-32 & 0.35. For oxygen he finds 1 + il = 1.000953. One important point which is brought out by tJhese high-pressure extrapolations is the very close approximation to linea- rity of the PV isotherms at low pressures. Thus if the high-pressure data are expressed by an expansion formula of the type PV = A + Bp + cp2 . . . the cp2 term for the permanent gases is so small that its effect is negligible. For the more compressible gases high-pressure data of suffi- cient accuracy have not been available with the exception of those for CO and C,H, for which we have the measurements of Michels and his colleagues. An extendea investigation of compressibiljties over medium pressures say 1-5 atmospheres which with one exception,a has not been attempted since the time of Regnault would be very valuable and for easily liquefiable gases would furnish data on the form of those isotherms a t the lower pressures.Taking then the mean values for oxygen we have normal density 24 J. 1909 1635. 26 J. Chim. physique 1923 20 308. 28 Ann. Physik 1929 Band 2 VIII Heft 1012 ; 1930 Band 4 VI Heft 778. 28 Thesis University of Santiago. 29a Ref. 23 p. 47. 31 J . Res. Nat. Bur. Stand. 1941 26 495. 25 Mern. SOC. Physique Qedve 1908 35 659. 27 Ibid. 1925 22 131. 30 Phil. Mag. 1931 12 42. 32 Physica 1934 1 587. WHYTLAW-URAY LIMITINQ DENSITIES 161 L; = 1.42896 ; compressibility per atmosphere at 0" 1 + R = 1.00092 or A = ~ / ( i + a) = 0.00092 Limiting density Lli,.=I 1.42896 (1 - 0*00092) = 1.42765 Molecular volume at 0" of perfect gas under normal conditions 32 (g = 980.616 lat. 45O) V, - - - = 22.4145 1. atm. mole-f = RT It may be noted that the values for V46 obtained by different workers show small variations ; thus Moles gives 22.4147 -& 0.0001 Batuecas 22.4150 & 0.0007 and Casado 2 2 4 5 & 0.002 and the value considered most probable by Birge 22.4151 & 0.0006 is based on 1 + R for oxygen = 1.000953 from Cragoe's calculation from high-pressure data. Further if the Coefficient of expansion of a perfect gas is 1/273.16 where 273.16" is the ice point on the absolute scale R = 22.4145/273-16 = 0.082056 1. atm. mole-l Having fixed then the limiting density of oxygen within narrow limits the molecular weights of other gases can be determined from the formula M = 22*4145Ln(1 - A) where L represents the weight of a normal litre of any gas M its molecular weight and A its compressibility coefficient between 0 and 1 atmosphere or 1-42765 where &&.and L; represent the limiting and normal densities of gas X AX its compressibility coefficient and A:a and and AOa ths corre- sponding values for oxygen. Turning from oxygen to other gases numerous determinations of densities and compressibilities have been made from about 1900 onwards. Not all of these can rank as high-precision data or can compare in accuracy with Rayleigh's densities of N, CO CO, and N,O which even to-day are very close to modern values. D. Berthelot in his papers on limiting densities used Leduc's determina- tions which agreed closely with Rayleigh's but both these workers deter- mined compressibilities at room temperatures and correction to 0" involves uncertainties (see Patterson and Cawood Zoc.cit.). Guye in Geneva from 1903 onwards made a great contribution to accurate gas work and with his collaborators Jaquerod Scheuer Baume Moles Batuecas and others he studied H, 0, N, CO, N20 NO NH, HC1 HBr SO, (CH,),O and CH,Cl and the work of this centre certainly laid the foundation on which much recent research has been built. Elsewhere notably in Ramsay's laboratory new methods of gas mani- pulation purification and measurement were developed and special studies were made of the physical constants of gases by Travers Gray Burt and others (H, O, NO HC1 He). In more recent times the very accurate measurements carried out a t the Reichsanstalt on the low-pressure compressibilities of H, N, 0, He Ne and A by Heuse and Otto call for special mention as do also the 162 QUARTERLY REVIEWS Densities and cmpressibilities of gases at 0" Observer as Moles and Clavera 1927 .. . . 84 Moles and Salazar . Moles and Garrido . 28 Heuse and Otto . a 2 Michels VVouters and de Boer oalc. by Cragoe from high press. . . 95Baxter and Stark- weather . Baxter and St'ark weather 1925-28 Baxter and Stark- weather . . . Baxter and Stark- weather . . . 28He~se and Otto 1930 . a4 Moles and 'Saiazai 1934 . . . . 95 Batuecas Maverick and Sohlatter 1929 . . . 87 Schlatter 1930 . a 8 M ~ l o ~ and Tori 1936 . . . Guye and Batuecas 40Moles Toral and Escribano . I Batuecas 1935 . . 41 Batuecas 1934 .. 42Casado 1943 . . 88M~les and Toral 1936 . . . 42Casado 1943 . . 8 Guye 1905. . . Gray 1906 . . . dB Jaquerod and Scheuer . . . 44Batuecas . . . Moles Sancho and Roquero . . . Batuecas and Moles 1930 . . . . 40M01es Toral and Escribano . . Moles Toral and Escribano . . Moles and Toral . 45 Whytlaw-Gray and Burt 1909 . . 47 Scheuer. . . . 48 RecaIc. by Moles . 4% Batuecas and Moles 1926 . 6 0 Guye Mo1es,61 Riemans2 . . 12Ritchie 1930 . . Gas. (4 1 + 1. 1.00046 1.00048 1.000453 D-99948 1*00058 1.00048 1*0070( 9) 1 1*0078(0) 1.02574 1.0071 1 1.00117 1*00111 1.00748 1.00 7 37 1-00932 1.009 1 (2) ~ - - 1.00043 1.00041 1~00090 0.99941 1.00040 1-00694 1.00730 1 -0203( 0) 1.02520 1 *OO 7 3 7 1.0070 1.01521 1*0155( 7) 1-01186 1.0240 1*01004 1.0242 Normal Density. 1.25049 1.25036 1.78364 0*89990 0.17846 1*25001 1.97693(8) 1.97686 1.26035(8) 1.26041 1.9148(5) 2.1079 1*97821(5) 1.9775 1.3402 1.34027 0.77 140( 2) 0.77170 1*53842( 6) 2*92654(7) 4.69049 1.63915 1.63909 1.63911 2.3070 3-64421 1.53072 Mol.wt. 28.0164 28.0 149 28.0152 39.944 20.183 4.0002 28.0065 44,007 43*999( 5) 28.046 28.032 42.067 46.068 46-087 44*016(7) 44-0 18 30.006 17.0328 17.0318 34.079 64.062 .04*084 36.469 36.465 50-491 6 80.962 34.000 At. wt. 14.0082 14.0075 14.00 76 C = 12.0065 C = 12,007 , = 12.000 , = 12.007 , = 12.000 , = 12.006 , = 12.010 , = 12.019 N = 14.0083 , = 14.009 , = 14.006 , = 14.009 , = 14.0078 S = 32.063 , = 32.062 Si = 28.104 c1 3 35.457 , = 35.458 Br = 79.918 P = 30-977 33 2. anorg. Chem. 1927 167 49. 35 Proc. Nut. Akad. Sci. 1926 12 20 703; 1928 14 50 57; 1925 11 231.95 J . Chim. phy&que 1929 28 648. 38 Sitzungsber. Akad. Wks. Wien 1936 145 948. 34 Anal. SOC. Xsp. Pis. Quim. 1934 32 954. 37 Ibid. 1930 27 44. W3IYTI;BW-GRAY LlMITING DENSITIES 163 density and limiting-density data on 0, Nz He Ne and A furnished by the work of Baxter and Starkweather at Harvard. From 1930 onwards Moles has been the most active worker in this field and a constant advocate of gasometric as opposed to gravimetric methods for the determination of atomic weights (see reports on atomic weights of the International Committee). In conjunction with Batuecas and others he has revised much of the earlier work. For a number of years in Madrid he devoted all the resources of his laboratory to the development of a refined and improved technique and he claims to have obtained a greater precision than any that has been reached hitherto.Quite recently impor- tant work of the same type has come from the laboratory of Batuecas a t the University of Santiago. All this laborious and careful work developing slowly over many years has been carried out by methods outlined in discussing oxygen. Alterna- tive methods depending on the buoyancy principle have also been used which offer many advantages. In the table on preceding page are collected well-established data obtained by the orthodox methods. Although the results obtained by different workers do not show as close an agreement as might be expected the mean values of the atomic weights vix. C = 12.007 N = 14.008 S = 32.063 Cl = 35.457 Br = 79.918 P = 30.977 agree closely with International values.In the case of both C1 and Br this close agreement may be fortuitous for the experimentally determined isothermals of HC1 HBr and CH3C1 exhibit a slight curvature and in consequence extrapolation to zero pressure is to some extent uncertain. 53 Small discrepancies of the order of a few parts in 104 are apparent especially in the 1 + A values obtained by method (a). Many of these were determined by following the Geneva technique which is open to criticism because in each experiment three points only were obtained in the PV-P graph and reliance was placed on a large number of experiments to give an accurate final mean though individual experiments exhibited wide deviations. The exact determination of pressure too is a matter of extreme delicacy and difficulty when the highest possible accuracy is required; for apart from errors inherent in scale and cathetometer capillarity has to be taken into account unless manometer tubes of wide bore are employed as in Baxter and Starkweather's work in which the diameter was 3 - 4 cm.If this is done dead-space errors in method These will be described later. 39 J . Chim. physique 1923 20 308. 4 1 Bol. Uniu. de Santiago 1935 0ct.-Dec. ; J . Chim. physique 1934 31 165. d3 University of Santiago Thesis 1943. J . Chim. physique 1925 22 109. 40 J. 1909,95 1633. 48 2. physikal. Chem. 1925 115 Heft 1/2 61. 49 Arch. Soc. Phys. Hist. Nut. Genkve 1925 42 120. 6o J . Chim. physique 1919 17 141. 68 J. Chim. physique 1917 15 293. s8 See Batuecas loc. cit. ; Baxter J . Arner. Chem. SOC. 1922 44 595. 40 Trans. Faraday Soc. 1939 35 1439. 43 Compt.rend. 1905 140 1384. 45 2. anorg. Chem. 1938 236 225. *? Sitzungsber. Akad. Wiss. Wien 1911,123 IIA 169,52. 61 2. physikal. Chem. 1925 115 61. 164 QUARTICRI;Y REVIEWS (a) due to meniscus volume fluctuations may be considerable. There is little doubt that in the past insufficient attention has been given either to the height or to the shape of the mercury meniscus in accurate mano- metry. This has recently been emphasised by Whytlaw-Gray and Teich.54 It is surprising too that little use has been made of Rayleigh’s point manometer in which a steel measuring rod carrying points a t predetermined distances apart is applied directly to the two mercury surfaces. As Rayleigh showed the accuracy with which a point can be set to a mercury surface is many times greater than the accuracy of measurement by the orthodox method of scale and cathetometer.Moreover refraction errors are eliminated. in their work on acetylene used a multiple-point manometer and also kept the manometer volumeter and connecting tubes strictly at the same temperature by immersion in melting ice and obtained a very accurate series of PV values strictly linear with respect to P. This admirable technique was used later by the Reich- sanstalt workers Reuse and Otto. In the work of Moles and his associates as well as in that of Baxter and Starkweather the density method was used for the determination of both the normal and the limiting densities. For many interesting details of refined experimentation the original papers must be consulted. In addition to the orthodox method of weighing gases in bulbs of different volumes Moles used the volumeter method in which the purified gases were confined and measured in large vessels or volumeters surrounded in melting ice and weighed in much smaller vessels after adsorption on char- coal cooled in liquid air a procedure employed previously by many workers but in his case used successfully as a proof of the avoidance of systematic errors.In one respect Moles’s results 56 differ from those of the majority of previous experimenters in that he in conjunction with Crespi has measured the adsorption of his gases on glass surfaces and applied a correction to his final values. For liquefiable gases this correction is of significance. Thus for CO and SO in a 500-C.C. soda-glass bulb the correction a t 1 atm. to the normal litre weights amount to 1 and 4.5 parts in lo* respectively.Although the agreement in the densities determined in bulbs and volumeters of various capacities after correction is very satisfactory it may be ques- tioned whether the data obtained in another apparatus though made from the same glassware are directly applicable to his various bulbs and volu- meters. There is also an error due to the solution of gas in the stopcock grease and its subsequent evolution in a vacuum which nobody up to the present has investigated systematically and which in some cases may be as great as or greater than the adsorption error. One very interesting feature of Moles’s work is the closely linear rela- tionship found between the ‘‘ litre weights ” Le. the densities determined at a series of pressures from 1 to 0.25 atm.and reduced to unit pressure and pressure for the gases SO, SiFF, H,S NH,-strong evidence that a linear extrapolation is valid for the liquefiable gases as it is for the J. T. Howarth and F. P. 54 Trans. Farraday Xoc. 1948 44 770. 56 Bull. Xoc. chim. Belg. 1938 4’9 405. 55 Ibid. 1925 20 544. WHYTLAW-BRAY LIlWTfNGI DENSITIES 165 permanent gases. In the past much discussion has ranged around this question of extrapolation and of the validity of different modes of mathe- matical treatment of experimental data.67 Earlier results obtained with methyl ether and methyl chloride and also with hydrogen chloride by the isotherm method (a) give PV-P curves concave towards the pressure axis ; method ( b ) apparently usually gives straight lines though the experimental evidence is not sufficiently accurate to exclude a slight curvature.Theoreti- cally a curvature in the direction observed (i.e. concave towards the pressure axis) is to be expected. W. H. K e e s ~ m ~ ~ on the basis of van der Waals's equation has calculated the curvature for the gases SiF4 NH, and CO, for which Moles finds ib linear relation and contends that it is great enough to raise by significant amounts the values of the molecular weights. The evidence for the linearity of the low-pressure isotherms of liquefiable gases was discussed by Moles 59 in 1938 and further work since then has gone far to confirm his conclusions. Working with a differential compressibility apparatus similar to that described by C. G. Addingley and Whytlaw-Gray 60 but designed to eliminate capillarity and meniscus volume errors G.A. Bottomley D. Massie and R. Whytlaw-G;ray6l have obtained data for O, CO N,O CO, C,H, and C,H over the pressure range of 800-100 mm. and in every case the deviation from a straight line was too small to be detected although the error of measurement of individual points did not exceed a few parts in lo5. Just recently SO has been studied from atmospheric pressure down to 15 mm. and no defhite evidence of curvature obtained. It seems clear then that even with such gases as sulphur dioxide and propane the critical temperatures of which are respectively 157" and 96" the use of an equation of the second degree (such as pu = A + Bp + Cp2) is unnecessary and that a significant departure from the linear does not occur a t ordinary temperatures. One comment may be made in considering the data in the table and that is to draw attention to the fact that they are expressed in absolute measure and hence depend on an exact knowledge of the gravity factor in the laboratory where the measurements were made the true height of the manometric column the density of mercury the density of water etc.Possibly the slight differences obtained by various workers for the normal density of the same gas may in part be due to the use of factors which are not identical. Since molecular weights are purely relative it is cer- tainly better to compare gases with oxygen in the same apparatus and under precisely the same conditions as has been done in the latest work in this field. Up to this point no mention has been made of hydrogen though before 57 R. T. Birge and F.A. Jenkins J . Ghem. Physics 1934,2 179 ; see also Batuecas (ref. 41). 68 " Determination Physico-chimique des Poids Molec. e t Stomique des Gaz " Collection Scientifiqus Institut Internat. Co-operation Intellectuelle Paris 1938 p. 165. 6s Ibid. p. 187. 61 Proc. Roy. SOC. 1949 in the press. Trans. Paraday SOC. 1928 24 378. 166 QUARTERLY REVIEWS the discovery of deuterium many determinations were made of its density (Morley Rayleigh Leduc) and compressibility (Chappuis Jaquerod Guye and others). It is a melancholy fact that all the determinations of the O/H atomic weight ratio made by either gravimetric analysis or physico- chemical methods are in error because the hydrogen used was deficient to a greater or lesser extent in the heavier isotope. No modern worker has had the courage to attempt to measure the density of chemical hydrogen or to redetermine its atomic weight.The present highly accurate value rests on mass-spectrograph measurements. The combining volumes of oxygen and hydrogen however are probably unaffected by a variation in deuterium content and the very accurate and beautiful work of P. P. Burt and E. C . Edgar,62 published in 1916 is one of the very few investiga- tions which can be used to check the limiting-density principle. They found the ratio of the volumes containing equal numbers of molecules to be H/O = 1.00144/1 whilst if we calculate it from (1 + A)os 1.00092 we get 1*00150/1 and a direct differential method gave 1*00148/1 differ- ences sufficiently small to be attributed to experimental error. Other instances in which a direct measurement of combining volumes has been made with any claims to accuracy are N2/N,0,63 3H/N,64 and H,/HCl = 1*00790,48 whilst ~ - _ _ _ _ (1 f A)H - 0.99942 but on account of various experimental difficulties none of these can be regarded as free from errors.Limiting Pressures.-If purely comparative measurements are to be made with high accuracy there is much to be said for the use of a silica- fibre buoyancy microbalance which consists of a small balance beam of fused silica swinging about a horizontally placed fibre at right angles to the beam and from one end of which a sealed buoyancy bulb is suspended. The microbalance suitably mounted in a glass case is observed with a low-power microscope and its zero position defined in relation to a fixed pointer. The balance is in connection with a sensitive manometer and in comparing gases the pressures a t which the balance floats exactly about its zero point are determined.Thus the pressures at which two or more gases have the same density are compared instead of densities at the same pressure as in the older method. These microbalances can be made capable of detecting a change of density corresponding with a change of 0.001 mm. of air pressure or less so that the error in the final result is determined mainly by the error in the pressure readings. Thus the many sources of error inherent in the weighing of gases in glass bulbs can be eliminated to a large extent. Further the comparison of densities becomes a much quicker and simpler operation and requires relatively small amounts of 62 Phil. Trans.1916 A 218 413. 64 Mern. SOC. Phys. Hist. Nat. GenBve 1908 35 594. 63 J . Chim. physique 1905 3 562. WHYTLAW-GRAY LIMITMCI DENSITIES 167 gas. To get the ratio of the limiting densities it is necessary to measure pressure ratios a t two or more densities which is done by altering the weight on the beam by means of small silica riders suspended above the buoyancy bulb. The theory of the baIance used in this way follows from Berthelot’s definition which in molecular and with balancing but since balancing of- compressibility coe6cient A i.e. PV = (PV),(l - AP) gram-mols. = nRT(1 - AP). If w g. is the weight of gas of weight M displaced by the bulb then since n = w/M, W PIVl = L ( 1 - AIPl)RT . MI another gas of molecular weight M, compressibility pressure P, W P,V2 = $-(1 - A2P2)RT .2 the volume of the bulb is fixed V = Y, and since at pressures both gases have the same density w1 = wz. * (1) A, and * (2) the two Hence dividing (1) by (2) we have At other densities we get other values of r r, r, r3 etc. which when graphed against P give the limiting value of r for P = 0 which as can be seen from the equation is the inverse ratio of the molecular weights. If 8s Berthelot assumed and subsequent experience has confirmed PV is a strictly linear function of P for the permanent gases then it is true that the ratios also vary linearly with PI (in practice usually oxygen). Another advantage of the method is that it allows gases to be compared at relatively low pressures where an error in the compressibilities is not so important but conversely the calculation of compressibilities from ratios obtained at low pressures would not be so reliable as those deduced over a wider pressure range.The main disadvantage of the method is that adsorption errors might be more serious for the buoyancy bulbs used are necessarily small their volume varying from 2 to 10 c.c. and the ratio of surface to volume correspondingly large. To compensate for adsorption an open bulb or other suitable surface is used as a counterpoise at the end of the beam remote from the buoyancy bulb in such a position that their surface moments are as nearly equal as possible. T. S. Taylor e5 was the first to construct and use a fibre suspension quartz microbalance in 1917 with which he compared the densities of H, He and 0,; later it was developed and modified in the Leeds Chemistry Department and has been used for accurate atomic weight determinations extending over a number of years.In order to attain greater accuracy in temperature control and also because the compressibilities decrease as temperatures rise the workers in Leeds use thermostated water-baths for microbalance and manometer. Hence their results for relative densities at 1 atm. pressure are not directly 6 b Physical Rev. 1917 10 653. 168 QUARTERLY REVIEWS comparable with measurements made a t 0" 0.-although of course the limiting-pressure ratios should be identical with limiting-density ratios for in the limit all gases have the same expansion coefficient. The values obtained by this method for the molecular weights of Xe CO CO, C,H, C,H, CH,F CF, H,S and SiH are shown in the following table.I Limiting-pressure ratios Observer. 66 Whytlaw-Gray Patterson andCawood . l1 Woodhead and Wiytiak Gray 1933 . . . . . 67 Patterson and Cawood 1936 68 Cast20 Massie and Whytlaw -Gray} 67 Patterson and Cawood 1936 9 9 7 9 9 Y t 9 9 ? $ 9 7 9 9 9 ? 68 Robks 70 Cawood and Whytlaw-Gray} Gas com- pared with 0 Limiting ratio p p . 4.1020 0.87535 1*3753(6) 0-87673( 5) 1.37807 2.74966 1 *06349 1.37542 1.06504 1.00451 Mol. wt. 131.26 28-01 1 44.01 0 1 28.0556 44.098 87.989 34.03 18 44.0135 34-081 32-144 Atomic weight. 131.26 c = 12.011 , = 12.010 , = 12.012 , = 12.011 F = 18.995 , = 18.997 N = 14.007 S = 32.065 Si = 28,112 Temp. 18" 19.8 21 21 20-76 21 21 21 21 21 It will be seen that the atomic weights determined by this method differ by small amounts from the corresponding figures in the table on p.162. The mean value for carbon is here 12.011 as against 12.007 a difference of 3-3 in lo4. Moles attributes this to an adsorption error in the microbalance but this criticism is certainly not valid for the CO/O ratio. An examination of the work of Cawood and Patterson does show that adsorption may lead to erroneous results especially if the surfaces of the balance are not entirely free from silica bloom. Since the publica- tion of their paper careful measurements of the adsorption on clean freshly blown surfaces of vitreous silica have been made by T. H. Henry G. A. R. Hartley and R. Whytlaw-Gray71 by a mercury-displacement method which shows that with these surfaces for the gases examined adsorption is remarkably small about 20 times less than on soda glass and provided the surfaces of the bulb and its counterpoise on the micro- balance are similar to the surfaces of the larger bulbs on which adsorption measurements were made no significant error is likely.In the case of sulphur dioxide adsorption was measured by direct weighing on small bulbs similar to those used on the buoyancy balance and again the correction was found to be beyond the limits of experimental error. With some gases however silica behaves in an abnormal way. Thus hydrogen diffuses into J. 1949 1746. 68 A. L. Roberts Thesis Lee& 1930. 66 Proc. Roy. Soc. 1931 A 134 7. 70 cc Poids Molsc. et Atomique des Gaz " Paris 1938 see ref. (58). 71 Trans. Paraday Soc. 1939 35 1452. 67 Phil. Trans. 1936 A 236 77. WHY!l!LAW-QRAY LIlMTING DENSITIES 169 vitreous silica and it has not been found possible to get accurate comparisons between this gas and oxygen with the microbalance.With helium the effect is even greater. Ammonia was found in recent unpublished measure- ments to be nearly as strongly adsorbed on silica as on glass arid silicon tetrafluoride has an adverse effect on silica surfaces even after prolonged desiccation. With many gases however vitreous silica is undoubtedly preferable to glass and as the appended table shows it could be used with advantage for high precision determinations of densities compressibilities and coefficients of expansion of many gases. Gas. SO, direct weighing small bulbs 72 . . SO, large bulbs mercury displacement . N,O displacement . . . . . . . co, . . . . . . C2H49 N2 co . . . . . . . A . . . . . . . 0 2 SiF, direct weighing72 .. . . . r n 3 7 2 ?¶ . . . . . 3 9 7 7 ? ¶ 9 . . . . . . . . . . . . . . . . . . . . . 9 ) Silica 1 Soda glass 1 Jena glass C.C. ( x adsorbed at 1 atm. press. per om.*. T = 21" C. ::: } 1.61 1.44 1.51 0.74 0.44 0.42 0.20 45 310 T = 15" C. 290 63 89 63 23 41 96 490 - 130 20 150 (The above figures for glass are taken from a paper by E. Moles Bull. Xoc. chim. Those for silica are from Trans. Paraday Soc. 1939 35 1457.) Belg. 1938 47 423. Returning to the atomic weight of carbon it is remarkable that the Spanish workers have obtained results by the orthodox methods so con- sistently lower than those found by the microbalance. In view of the known variation in isotopic composition in carbon from different sources it might be supposed that the difference could be attributed to this cause.The maximum variation however found by Nier in carbon from a wide variety of sources was from 82.7 to 92.5 for the ratio of 12C to 13C corre- sponding to C = 12.0120 and C = 12.0105 respectively on the chemical scale a variation only about half as great as the difference in question. It is difficult to believe that the gases used by Moles and his collaborators could-have had such an abnormally high proportion of 12C. In view of the known variation in isotopic composition of other elements such as sulphur it would seem desirable to use only material of definite isotopic composition for atomic-weight determinations. The question might well be asked whether limiting-density measurements could be made with sufficient precision to detect definitely isotopic changes. It is certainly difficult when using the orthodox methods to be sure of small differences of the order of a few parts in lo5.Measurements with the 7 Driver Thesis Leeda 1946. 170 QUARTERLY REVIEWS microbalance would appear to offer a greater chance of success provided of course that a sufficiently high degree of chemical purity in the gases examined could be ensured for by this method errors inherent in weighing gases in bulbs are to a large extent eliminated. Recently a very accurate investigation on the limiting densities of very pure nitrogen and methane has been made by B. Lambert and his co-worker~,~~ in the course of which they have studied in great detail the errors in the silica-fibre microbalance and also in the exact measure- ment of pressure. Using a very beautiful form of microbalance they have been able to reach a remarkably high degree of precision.Although the object of the work was rather to demonstrate the degree of accuracy possible with the technique developed yet the concordance of the measurements was such that isotopic variations could well have been detected. They find for the atomic weights of carbon and nitrogen 12.0112 and 14.0078 respectively. A method of directly measuring small differences in gaseous densities was worked out by E. R. Roberts H. J. Emelhus and H. V. A. Briscoe 74 in 1939 in their work on ethyl- and dimethyl-deuteramines. In this the two gases to be compared were contained in two vessels and adjusted to exactly the same density by means of a buoyancy microbalance. By an ingenious arrangement the pressure difference between the two was measured at constant volume on an oil gauge a pair of constant-volume indicators or capastats being used for this purpose.By this device the pressure difference between the two gases at precisely the same density could be measured accurately without contact with oil. A modification of this method has been used lately by B. Leadbeater and Whytlaw-Gray75 to compare the densities of nitrogen and carbon monoxide. Instead of the capastats the microbalance itself was used to measure the small difference in pressure which for this pair of gases at a filling pressure of about 350 mm. was of the order of 0.025 mm. After the densities of the nitrogen and carbon monoxide in the two vessels had been adjusted to equality the connecting stopcock was opened and a small amount of carbon monoxide flowed through and mixed with the nitrogen until pressure equality was established.The pressure difference was measured by the deflection of the balance which had been calibrated previously in terms of pressure. This method of working which is only applicable to gases of very nearly the same molecular weight proved capable of measuring density differences up to 2 parts in 106. Since air oxygen is heavier than water oxygen by nearly 7 parts in 106 a direct measurement of this small difference is within the limits of experiment and there is little doubt that the method could be improved to detect still smaller changes. In the case of carbon monoxide and nitrogen measurements were made at two different densities so that the limiting ratio could be calculated.This was found to be press. CO/press. N = 1-0001360 so that if we assume 7 3 Private communication. 7 5 Leadbeater Thesis Leeds 1946 ; Proc. Int. Cong. Pure and Appl. Chem. 1947. 74 J. 1939 41. WHYTLAW-GRAY LIMITING DENSITIES 171 for carbon the value 12*0112 that of nitrogen is 14-0075. Birge calculates the most probable mass-spectrograph values for these two elements to be C = 12.0114 and N = 14-0074 so that in this case at least the limitjng- density measurements have reached an accuracy as great as that of stoicheio- metric determinations. To summarise the general conclusions of this Review it is evident that the principle of limiting densities when applied to the determination of molecular atomic weights is capable of giving results of the highest accuracy especially when applied to the permanent gases.For the liquefiable gases it would seem that modern work has resolved the difficulties with which the measurement of these much larger deviations from the ideal state is fraught and that uncertainties in the extrapolation to zero pressure have been much reduced so that now even if more precise data should bring to light a slight curvature in the isothermals it will be possible to fix the limiting values within narrow limits. Very few attempts have been made to apply the principle to vapours. W. Ramsay and B. D. Steele76 in 1903 using ii modified and improved apparatus of the Hofmann type determined the densities and the PV-T variations down to pressures of about 40 mm. for a number of organic vapours but the limiting values differed considerably and to varying extents from the expected values.Batuecas 77 has recalculated all the results obtained by Ramsay and Steele and concluded that the accuracy of the experimental results was not as great as was supposed and that the contention that the limiting-density method was inapplicable to vapours was by no means proved. Since 1903 very few vapour densities have been measured with the necessary accuracy to decide this question ; the data only of A. Magnus and E. Schmid 78 need be considered. These workers obtained with chloroform and benzene values for the compressibilities and densities which as Batuecas 79 has shown give molecular weights in close agreement with standard numbers. It seems likely then that when the considerable experimental difficulties are overcome the principle can be extended to a wider range of elements. 8. physikal. Chem. 1903 44 348 ; Phil. Mag. 1903 6 492. 77 8. physikal. Chem. 1939 A 183 438. 78 8. anorg. Chem. 1922 120 232. 79 Ibid. 1941 246 158.

ISSN:0009-2681    

DOI:10.1039/QR9500400153

出版商:RSC    

年代:1950

数据来源: RSC

摘要:

ISOTOPIC TRACER TECHNIQUE By H. R. V. ARNSTEIN PH.D. D.I.C. and R. BENTLEY PH.D. D.I.C. (NATIONAL INSTITUTE FOR MEDICAL RESEARCH LONDON N.W.7) ISOTOPES were first used as tracers in 1913 when G. Hevesy and F. A. Paneth l applied the inseparability of the isotopes lead and radium-D to determine the solubility of “ insoluble ” lead salts. Later Hevesy and L. Zechmeister demonstrated the lack of exchange between the lead atoms of tetraphenyl-lead and lead ions. The first biological problem studied with this tracer method was the uptake and release of lead (labelled with thorium-B) by the roots of bean plants.3 Following the discovery of deuterium and later of the heavy isotopes of nitrogen and carbon important advances were made in our understand- ing of metabolic processes. Deuterium was immediately applied to investi- gate the behaviour of water in biological systems.With the use of D,O the rapid exchange of the body water of a goldfish and its surroundings could be observed.* Water metabolism in humans was also studied and the “average life ” of water molecules determined as about 14 days.5 Synthesis of organic compounds containing stable C-D bonds enabled fat and fatty acid metabolism to be studied.6 These results taken with those7 from studies of the metabolism of amino-acids labelled with 15N provided the basis for the new concept of the “dynamic state of body constituents ”.8 The introduction of the carbon isotopes llC and 14C in 1939 and 1940 respectively made possible a new approach to problems of photosynthesis carbon dioxide fixation,lO and fatty-acid metabolism.11 Isotopes of other elements of biological significance (e.g.32P 1311) are now readily available and have found extensive and no less important applications. The field of tracer chemistry is now too broad to be considered in a single Review ; this Review will describe practical considerations in the use of stable isotopes the methods available for the synthesis of compounds containing stable isotopes (except that 1% will be included to give a com- lZ. anorg. Chem. 1913 82 322. Ber. 1920 53 410. G. Hevesy Biochem. J. 1923 17 439. G. Hevesy and E. Hofer 2. physiol. Chem. 1934 225 28. Idem Klin. Wochenschr. 1934 13 1524. R. Schoenheimer and D. Rittenberg Science 1935 82 156 ; J . Biol. Chern. It. Schoenheimer and S. Ratner ibid. 1939 127 301. R. Schoenheimer “ The Dynamic State of Body Constituents ” Harvard Univer- sity Press Cambridge Massachusetts 1946.S. Ruben W. Z. Hassid and M. D. Kamen J. Amer. Chern. Soc.,1939 61 661 ; S. Ruben and M. D. Kamen Proc. Nat. Acad. Xci. 1940 26 418 ; A. W. Frenkel Plant Physiol. 1941 16 654. lo H. G. Wood C. H. Werkman A. ICemingway and A. 0. Nier J. Biol. Chem. 1940 135 789. l1 H. A. Barker and M. D. Karnen Proc. Nat. Acad. Sci. 1945 31 219. 1935 111 163. 172 ARNSTEIN AND BENTLEY ISOTOPIU TRACER TECHNIQUE 173 plete account of syntheses with isotopic carbon) and a discussion of the limitation of the methods. Analysis of Stable Isotopes The detection and analysis of radioactive isotopes are already well described12 and only D 15N and 1 8 0 will be discussed here. In early work with deuterium analyses were based on the difference in refrac- tive index l3 of D20 and H20 or more conveniently the difference in density.14 Of the many densimetric methods available the most useful is probably the falling-drop method,l5 16 an accuracy of 1 p.p.m.being possible. In the case of organic compounds a representative sample of the hydrogen and deuterium must be obtained as water. The dried compound is burnt in hydrogen-free dry oxygen over copper oxide. The water is condensed and rigorously purified by repeated vacuum- distillation. l6 In the case of 15N and 13C mass-spectrometric analysis is necessary. Aston's mass spectrographs were hardly suited to the routine determina- tion of isotope content and it was preferable to use an electrical rather than a photographic method for the detection of ion beams.Such a spectrometer was first constructed by D. Rittenberg and his colleagues,17 being modelled after W. M. Bleakney's instrument .I8 Burther advances in mass spectrometer design particularly by A. 0. Nier 19 and R. L. Graham et aZ.,20 have led to commercially available instruments. In operation a gas sample a t a pressure of a few cm. of mercury is allowed to leak through a fine capillary into the spectrometer tube which is held a t a high vacuum (10-7 to lo-* mm. Hg). The methods used to prepare suitable gas samples will now be described briefly. Deuterium.-The mrtss-spectrometric analysis of deuterium has the special advantage that very much smaller samples may be analysed than by the other methods. As little as 5 mg. of an organic compound is burnt in a micro-combustion tube.The water is condensed in a cold trap and then distilled over hot zinc at 385". The hydrogen-deuterium mixture is la M. D. Kamen " Radioactive Tracers in Biology " Academic Press Inc. New York 1947 ; G. Hevesy " Radioactive Tracers Their Applications in Biochemistry Animal Physiology and Pctthology " Interscience Publishers Inc. New York 1948 ; M. Calvin C. Heidelberger J. C. Reid B. M. Tolbert and P. E. Yankwich " Isotopic Carbon" J. Wiley & Sona Inc. New York 1949. l3 R. H. Crist G. M. Murphy and H. C. Urey J. Amer. Chem. Soc. 1933 55 5060; J . Chem. Physics 1934 2 112. l4 D. Rittenberg and R. Schoenheimer J. Biol. Ohem. 1935 111 169. l6 A. S. Keston D. Rittenberg and R. Schoenheimer ibid. 1937 122 227. l8 M. Cohn " Preparation and Measurement of Isotopic Tracers " J. W. Edwards l7 D.Rittenberg A. S. Keston F. Rosebury and R. Schoenheimer J. Biol. Chem. l8 Physical Rev. 1929 34 157; 1932 40 496. le Rev. Sci. Imtr. 1940 11 212 ; 1947 18 398. Ann Arbor Michigan 1946 p. 51. 1939 127 291. R. L. Graham A. L. Hmkness and €I. G. Thode J. Sci. Imtr. 1947 24 119. N 174 QUARTERLY REVIEWS collected in a vacuum system and compressed into a sample bulb with the aid of a Toepler pump.21 Some special difficulties in the mass-spectrometric determination of deuterium have been discussed recently by H. W. Washburn.22 The analyses are made by comparison of the intensities of the ion beams of mass 3 (HD) and 2 (H,) and are accurate to 0.002 atom yo of deuterium. Nitrogen.-The ideal gas for 15N assay is nitrogen itself. intensity of ion beam mass 28 [14N14N] intensity of ion beam mass 29 - [laW5N] The ratio - R = By consideration of the equilibrium is measured.(for which K = 4 at room temperature) it follows that 14N14N + 15N15N + 214N16N atom yo 15N = 100/(2R + 1) Nitrogen of organic compounds is readily obtained as gaseous nitrogen after a Kjeldahl digestion 23 and subsequent oxidation of the ammonia with hypobromite. The oxidation is carried out in vacuo using a two- legged Y-tube fitted with a ground-glass cap. Nitrogen gas is thus released in the absence of diluting air. Air leakage giving rise to incorrect ratios is indicated by ion beams of mass 32 (oxygen) and 40 (argon). Providing the leakage is not more than about 3% i t is possible to correct for this dilution. Routine 15N analyses are accurate to &- 0.003 atom-%. Casbon.-Although several volatile carbon compounds are available in practice 13C is assayed as carbon dioxide.The ion beams used are those due to 12C160160 (44) and 13C160160 (45). The contribution of Wl6O17O to the 45 beam is usually ignored since it is the same in both normal carbon dioxide and the sample. If intensity of ion beam mass 44 - [f2C1601e0] R = _ intensity of ion beam mass 45 [13C1s0160] it can be shown that atom yo 13C = 100/(R + 1). It is more important to obtain a sample of carbon dioxide whose carbon atoms are truly representative of those of the parent compound than to obtain a quantitative conversion. Either combustion 24 or wet oxidation are suitable methods. By using D. D. Van Slyke and J. Folch’s mixture g5 the oxidation may be carried out in simple apparatus 26 or in a vacuum- tube.,‘ In either case the carbon dioxide is collected as barium carbonate 21D.Rittenberg Report of a Symposium on the Use of Isotopes in Biological Research American Cancer Society 1947 p. 27 ; D. B. Sprinson and D. Rittenberg U.S. Naval Medical BUZZ. Supplement March-April 1948 p. 89. 22Ibid. p. 75. 23 D. Rittenberg “ Preparation and Measurement of Isotopic Tracers ” J. W. Edwards Ann Arbor Michigan 1946 p. 31; D. B. Sprinson and D. Rittenberg U.X. Naval Medical Bull. Supplement March-April 1948 p. 82 ; J . Biol. Chem. 1949 180 707. 24 D. Rittenberg op. cit. p. 39. 26 J . Biol. Chem. 1940 136 509. 26 A. Lindenbaum J. Schubert and W. D. Armstrong Analyt. Chem. 1948 20 27 H. A. Barker quoted in “ Isotopic Carbon ” Calvin et al. 1949 p. 93. 1120. ARNSTEIN AND BENTLEY ISOTOPIC TRACER TECHNIQUE 175 and subsequently decomposed with mineral acid the two-legged tube being used.A method for purification of carbon dioxide without absorption in baryta has been described.28 Apart from these general methods special techniques are available for some compounds e.g. the or-carboxyl group of an amino-acid is oxidised to carbon dioxide by the ninhydrin reagent and fatty acids may be decarboxylated to carbon dioxide by several methods. In routine work analyses accurate to rfrr 0.01 atom % 13C are obtained. Carbon dioxide is pumped away relatively slowly from the spectrometer. Oxygen.-Samples of oxygen gas may be directly analysed by com- parison of the intensities of the ion beams of mass number 32 (lsOl6O) and 34 (lOlaO). When large numbers of oxygen analyses are performed the life of the spectrometer filament is considerably reduced.Carbon8 dioxide may also be used in the analysis of 180 the ion beams of mass 44 (12C1601eO) and 46 (12C160180) being used. As mentioned earlier carbon dioxide may be obtained from many decarboxylations and analysed directly. It is also employed for the analysis of H,l*O samples by using the exchange reaction z9 C1602 (g.) + H2180 (1.) The equilibration is conveniently catalysed by the addition of a little car- bonic anhydrase and by using a tube of small volume as little as 5 mg. of Hz180 may be successfully analysed. If €2 is the abundance ratio (mass 44 mass 46) of the carbon dioxide after equilibration where K is the equilibrium constant of the exchange reaction (= 2*076).30 + CfsOl6O (g.) + H2160 (1.) Atom% lSO in water = lOO/[#X(ZE - 1) + 11 !l!he Synthesis of Labelled Compounds The incorporation of an isotope into a compound depends largely on the nature and availability of the isotope.The isotopes of carbon and nitrogen are still rare and expensive and the desired high conversion of isotope can be achieved by appropriate modifications of the usual reaction conditions such as the use of excess of the non-isotopic reactant intro- duction of the isotopic material a t the latest possible stage of a synthesis and recovery of any unused isotopic material. Isotopic carbon is available as NaTN containing up to 60 atom % excess of 13C and Ba14C0, usually of 1 mc./mM. specific activity. The maximum permissible dilution of 13C is 6000 as 0.01 atom yo excess is still detectable with a mass spectrometer.31 14C (half-life about 5000 years) 32 emits soft @-rays of approx.0.15 MeV. maximum energy. Bal4CO3 of a 28 S. Weinhouse ‘‘ Preparation and Measurement of Isotopic Tracers ” 3. w . Edwards Ann Arbor Michigan 1946 p. 43. M. Cohn and H. C. Urey J. Arner. Chem. Soc. 1938 60 679. 30 R. Bentley Nucleonics 1948 2 (2) 18. 32 L. D. Norris and M. G. Inghram Physical Rev. 1946 70 772 ; 1948 73 350 ; 31 Idem Research 1949 2 378. A. F. Reid J. R. Dunning S. Weinhouse and A. V. Grosse ibid. 1946 70 431. 176 QUARTERLY REVIEWS specific activity of 1 mc./mM. giving a counting rate of about 8 x 106 counts per minute on the usual end-window Geiger-Muller counter can still be accurately counted after 400,000-fold dilution. 15N available as l5NH,NO3 (containing up to 62 atom yo excess in the ammonium radical) K15N0, and potassium phthalimide can be measured with an accuracy of 0-004 atom yo and its maximum detectable dilution 81 is therefore 15,500.Deuterium is produced as heavy water of almost lOOyo purity and can be diluted to a greater extent than the other stable isotopes. The above figures are the maximum dilutions which the isotope can be permitted to undergo during synthesis and subsequent isolation in a chemical or biological experiment. Dilutions in animal experiments are usually considerable ( lo3 and greater) and radioactive carbon compounds for such work should have a specific activity greater than lO-Zpc./mg. but for chemical investigations much lower activities (10-3 to 10-4 pc./mg,) will suffice. Since a hundred-fold dilution of 14C is usually permissible during an organic synthesis from starting material of high specific activity the carrier technique can be used advantageously.With this technique intermediates often need not be isolated and manipulations are then reduced to a mini- mum. Finally a large excess of non-isotopic end product is added to facilitate the isolation of the labelled compound. Even optical resolu- tions can be avoided by addition of the appropriate non-isotopic optically active isomer to the 14C-~~-mixture and crystallisation of the labelled stereoisomer as shown by the recent preparation of 14C-~-cystine and 1 4 C - ~ -me thionine .s3 Purity of Isotopic Compounds.-The criteria of purity used in organic chemistry are often sufficient for tracer work but the result may be in error even when all the conventional criteria of purity are applied.How- ever additional techniques are available for checking the purity of labelled substances. The molecule may be degraded chemically and the isotope content of the degradation products compared with that of the original substance. A radioactive impurity may be detected by a combination of paper chromatography and radio-a~tography.3~ If the counter-current distribution method is used a number of partition coefficients can be determined by measuring the isotope content and by a chemical method. If the values obtained by the two methods are in agreement the com- pound can be considered pure. Counter-current extraction has been used to separate 14C-hippuric acid from 14C-cc-acetamido-y-phenylbutyric acid of ten times the specific Recently the difference in vapour pressure of solutions containing dif- ferent amounts of solid has been applied to the determination of the purity 3 3 J.L. Wood and H. R. Gutman J . Biol. Chem. 1949 179 535. 84 R. M. Fink C. E. Dent and K. Fink Nature 1947 160 801 ; R. M. Fink and K. Fink Science 1948 107 253; R. M. Tomarelli and K. Florey ibid. p. 630; W. Stepka A. A. Benson and M. Calvin ibid. 1948 108 304. 85 H. R. V. Arnstein and A. Neuberger Bwchsm. J. 1949 45 iii. ARNSTEIN AND BENTLEY ISOTOPIC TRACER TECHNIQUE 177 of amino-acids. By means of a simple differential tensimeter it was possible to detect 0.2% of D-glutamic acid in L-glutamic acid.36 Location of Labelled Atom.-Tracer work often necessitates the loca- tion of isotopic atoms in a molecule. It is then necessary to degrade the compound by an unequivocal method and to determine the isotope content of the degradation products.Sometimes the isotope content of some atoms can only be determined by difference but analysis of all the fragments provides a useful check of the purity of the compound and the validity of the degradation procedure. * For example decarboxylation of W-methyl- labelled barium acetate 2 1 2 1 b25O 1 2 1 2 2 (CH,*CO,),Ba ___+ BaCO + CH,*CO*CH + CHI + CH,*C02H unexpectedly gave barium carbonate containing approximately 1*5-2% of the original isotope but the iodoform had the same specific activity as the starting Biological as well it5 chemical methods have been used to degrade glucose obtained from rat-liver glycogen 38 L. msei fermentation 1 2 3 4 5 6 CH( OH)*CH( OH)*CH( OH)*CH( OH)*CH*CH,*OH I 0 I 1,6 2,5 394 1,6 2,5 3,4 NaOI 1,6 2,5 ZCH,*CH(OH)*CO,H -+ CH,.CHO + CO -+ CHI + H-CO,H alternative 1,6 2,5 3,4 H.G. Wood et also degraded glucose by purely chemical reactions MeOH 1 2 3 4 5 6 HI0 Glucose ,-F MeO*CH*CH(OH)*CH(OH)*CH(OH)*CH*CHz*OH 2 I 0 I 1 2 4 5 6 3 HIO MeO*CH*CHO CHO*CH*CH,*OH + H*CO,H -+ I 1- 0- 1 2 4 5 1,2,4,5 6 __+. 4H*COsH + CHZO + MeOH + MeOH The two procedures permit calculation of the isotopic concentration in positions 1 6 3 and 4 separately as well as the sum of positions 2 and 5. Specific and Non-specific Labelling.-In most cases the isotope is incor- porated in a definite position of the molecule but when the synthesis involves a symmetrical intermediate the final product will be labelled in s6 I. W. Hughes and G. T. Young Nature 1949 164 603.17 B. A. Fries and M. Calvin J . Amer. Chem. Soc. 1948 70 2235. as13[. G. 'Wood N. Lifson and V. Lorber J. Biol. Chern. 1945 169 475. ** S. Aronoff H. A. Barker and M. Calvin {bid. 1947 169 459. 178 QUARTERLY REVIEWS more than one position ; e.g. synthesis of 15N-isoguanine 40 NH2 CNBr + l5NH -+ CN*16NH + 16NH,Br NC.CH,.CN 16NJj! Cl.CH,*CO,H CN*16NH2 + H2S -+ 15NH2*CS*15NH2 -+ ._____3 HS''l6N NH2 NH2 NH2 1Biosyntheses.-A number of compounds can be prepared conveniently by biosynthesis but this procedure usually gives a non-specifically labelled compound. 14C-Glucose and 14C-fructose have been prepared from W O by photo- synthesis using Turkish tobacco leaves,41 and 14C-labelled sugars have also been obtained from barley seedlings and I4CO2. In this case the highest activity was found in the 3 and 4 positions the lowest in the 1 and 6 positions of the hexose.39 14C-Labelled sucrose has been obtained by photosynthesis using the leaves of Canna indim 41 and by an ingenious enzymic synthesis which gives sucrose labelled independently in the glucose or fructose moiety 42 enzyme from 14C-fructose + sucrose ______j.glucose l-14C-fructoside P . saccharophzla (sucrose) enzyme 14C-glucose 1 -phosphate + fructose ___ 14C-glucose 1 -fructoside (sucrose) lYB has been used for the biosynthesis of starch by bean leaves,43 and Geiling et aE.44 have prepared labelled nicotine and digitoxin by growing plants in 14C02. 1%-Radioactive silk has also been made biosynthetically,45 and 14C- bufagin has been obtained 46 from toads which were fed with alga previously exposed to 14C02.A. Bendich J. F. Tinker and G. B. Brown J . Arner. Chem. SOC. 1948 70 3109. E. W. Putnam W. Z. Hassid G. Krotkov and H. A. Barker J . Biol. Chem. 42 H. Wolochow E. W. Putnam M. Doudoroff W. Z. Hsssid and H. A. Barker 48 L. G. Livingstone and G. Medes J. Qen. Physiol. 1947 31 75. 4p E. M. K. Geiling F. E. Kelsey B. J. McIntosh and A. Ganz Science 1948 45 P. C. Zamecnik R. B. Loftfield M. L. Stephenson and C. M. Williams Science 4~ J. Doull K. P. DuBois and E. M. K. Geiling Fed. Proc. 1949 8 286. 1948 173 785. ibid. 1949 180 1237. 108 658; T. E. Kimura and E. M. K. Geiling Fed. Proc. 1949 8 308. 1949 109 624. ARNSTEIN AND BENTLEY ISOTOPIC TRACER TECHNIQUE 179 Barker et al.47 showed that Cloatridium acidi-urici incorporated 14C02 into acetic acid during uric acid fermentation.About 50% of the original radioactivity was found in the acetic acid two-thirds being located in the methyl group. Another micro-organism Clostridium thermoaceticum utilised 80% of the 14C02 for acetic acid synthesis the activity being equally dis- tributed between the two carbon atoms.11 W02 is converted into acetic and butyric acid by Butyrobacterium rettgeri,48 and 14C-acetic acid is in- corporated into n;-butyric and n-hexoic acids when ethanol is fermented by Clostridium k l u y ~ e r i . ~ ~ Recently yeasts have been used for the biosynthesis of 15N-labelled nucleic acids,5* and s5S-radioactive penicillin has also been prepared by biosynthe~is.~~ Since the scope of the tracer technique is hindered to a large extent by the limited number of labelled organic compounds available the synthetic methods which have been developed will now be discussed.Synthetic 13C and 1*C-Cornpound~.-One-carbon compounds. Many one- carbon compounds (see Table I) have been prepared from 14C02 usually as intermediates for the synthesis of more complicated molecules though some compounds (e.g. urea guanidine formic acid) have also been used for biochemical studies. Since 13C is available as cyanide it is usually introduced directly into more complicated molecules. W02 has been generated from BaWO by treatment with an acid (usually sulphuric acid) and roasting of the salt a t high temperatures (l100°).52 By fusion of barium carbonate with lead chloride or lead chlorjde-silver chloride (1 1) at 400" a quantitative yield of 14C02 has been 47 H. A.Barker S. Ruben and J. V. Beck Proc. Nut. Acad. Sci. 1940 26 48 H. A. Barker M. D. Kamen and V. Haas ibid. 1945 31 355. 49 H. A. Barker M. D. Kamen and B. T. Bornstein ibid. p. 373. 6o F. J. DiCarlo A. J. Schdtz P. M. Roll and G. B. Brown J . Biol. Chem. 1949 180 329. 61 D. Rowley J. Miller S. Rowlands and E. Lester-Smith Nature 1948 161 1009; S. F. Howell J. D. Thayer and L. W. Labaw Science 1948 107 299. 62 M. G. Inghram U.S. Atomic Energy Commission MDDC 60 June 1946. s3N. Zwiebel J. Turkevich and W. W. Miller J . Amer. Chem. SOC. 1949 71 477. 376. 180 QUARTERLY REVIEWS TABLE I Compound. Na214C0 . . . . NaHf4C0 . . . . KHl4C0 . . . . 1 4 ~ 0 . . . . . . lQCOCl . . . . . K14CN . . . . . NaWN . . . . . Bd4CNa. . . . . NH2*14CS*NH2 . . . (NH2),14C:NH,HN0 . 14CCN.NH*14C:NH*NH2 .H*l%O,H . . . . H“%O,CH . . . 14CH,*OH . . . . (NH 2) ,14C:NHyHCl NH2*14C0,C,H a . 1 4 ~ ~ ~ 1 . . . . . “CH,*NO2 . . . . H*l4CH0 . . . . 14CH4 . . . . . 14cc1 . . . . . 64W. B. Leslie U.S. Starting material and method. BaWO ~a14c0 ~a14c0 14co a W O (exchange reaction) CaPCO (heated with Zn) 14C0 (heated with Zn) 14co WO (heated with NH3-K) 14C0 Ba14C0 (NaN reduction) Ba14C0 (NH a t 850’) ~ 8 1 4 ~ 0 1 4 ~ 0 WOCI BaWN BaWN BaWN Bal4CN2 14CN*NH*14C:NH*NH BaWN NH ,*14CO*NH KH14C0 (catalytic reduction) NaWN HWO2H WO (catalytic reduction) H*14C02CH (catalytic reduction) 14CH,*OH 14CH,*OH 14CH,*OH 14CH,I WH,*OH (via CH3*C0,14CH2CI) WH,*OH (catalytic oxidation) 14CH,*OH (catalytic oxidation) 14C02 (Ni-Tho catalyst 330’) 14CH4 H*14C0,H B&14C03 itomic Energy Commission MD Yield %.- 100 - 100 - 100 91 61 83 40 98-99 50-60 90 95 88 70 60 50-60 60-70 Yield baaed OD co, %. - 100 - 100 - 100 - 99 95 90-96 80 78 73.5 80 90 67 95 61 36 - 98 - 50 - 90 85-90 - 76 45-55 96 93 C 674 June 1< D. B. Melville J. R. Rachele anld E. B. Keller J. Biol. Chem. 1947 J. T. Kummer J . Amer. Chem. SOC. 1947 69 2239. 66 R. B. Bernstein and T. I. Taylor Science 1947 106 498. 68 S. Weinhouse ibid. 1948 70 442. 69 J. L. Huston and T. H. Norris ibid. p. 1968. Ref. 54 55 55 56 57 58 69 60 59 60 61 62 63 64 65 66 67 68 65 66 65 69 65 65 70 55 60 71 55 72 65 65 72 73 74 75 76 77 78 78 1947. 69 419. D. B. Melville J. G. Pierce and C. W. H. Partridge J. Biol. Chem. 1949 180 29% 61 “Isotopic Carbon” Calvin et al. 1949 pp. 160 161. 62 R. B. Loftfield Nucleonics 1947 1 (3) 54. u3 A. W. Adamson and W.K. Wilmarth J . Amer. Chem. SOC. 1947 69 2564. 6 4 ‘‘ Isotopic Carbon ” Calvin et al. p. 158. 66 S. H. Zbarsky and I. Fischer Canadian J. Res. 1949 27 81. 66 A. Murray and A. R. Ronzio J . Amer. Chem. SOC. 1949 71 2245 67 “Isotopic Carbon” Calvin et al. 1949 p. 167. G8Ibid, p. 158. Ibid. p. 159. aRNSTEM AND BENTLEY ISOTOPIC TRACER TECHNIQUE 181 Functionally labelled compozcnds reluted to carboxylic acids. Carboxylic acids (see Table 11) are usually prepared by carbonation of a Grignard reagent with isotopic carbon dioxide (side reactions being avoided by modi- fying the reaction conditions),79 or by reaction of a suitable halide with isotopic cyanide. Carbonation of organo-metallic compounds has also been used.80 W-Phenylacetic acid has been prepared by the Arndt-Eistert reaction 81 At3 C6H,*13COC1 f CH2N2 + C6H,*13CO*C~ + CsH,*CH,*13C0,H TABLE I1 Functionally labelled compounds Compound.jMf3thod.' I- CH,*13C0,H . . . . CH,*WO,H . CH,*13CO*CH,*13C0,H CH,*CO*CH,*13C0,H . CH,*CH,*l3C0,H. . . CH,*CO*13C0,H . . . C6H,J4C0,H. . . . p-MeO*C,H,*14C0,H. . i~o-C4H,*"CO,H . . . 3 4-(MeO),C,H,*'*CO,H p-NH,*CeH,*l'CO,H. . C,H,*CH,*14C0,H . . CBH,*CH,*14CO*NH,. . 14C- 1 -Naphthoic acid . a a a a b b d a a C Ref. 82 83 84 82 82 86 82 64 80 86 87,88 89 80 81 90 91 92 Compound. 14C -2 -Naphthoic acid n-C7Hl,*1SC0,H. . . n-C1oH,1J4CO,H . . n-C11H,,*"CO,H . . ~-C1,Hg1*"CO,H . . n-C1gH,,*WO,Me . . n-C,gHgg*"CO,H . n-CllH,,*'4CN . . . 14C-Tripalmitin . . . Fl~orene-9-~~C-carboxylic acid. . . . . . 14C-Nicotinic acid . . l4CO ,H*CHpJ4C0 ,H' 14C0,H*C:C*14C0,H ."CO g H W O aH . l4CO ,H*CH,*CHS."COaH Kethod.* a a a cb a a b b e b b Ref. 93 94 95 96 96 95 95 96 97 98 80 99 100 101 101 * a By carbonation of Grigna 1 reagent. b By carbonation of orgtmometallic compound. c By oxidation of ketone. d By Arndt-Eiatert reaction. e From nitrile. 7O H. E. Skipper C. E. Bryan and 0. S. Hutchinson U.S. Atomic Energy Com- mission Circular (3-8 September 1947. " Isotopic Carbon " Calvin et al. 1949 p. 165. la B. M. Tolbert J. Amer. Chem. SOC. 1947 69 1529. l 8 R. B. Regier and R. W. Blue J. Org. Chem. 1949 14 505. 76A. R. Jones and W. J. Skraba Science 1949 110 332. 77 H. R. V . Arnstein Nature 1949 164 361. 78 W. H. Beamer J. Amer. Chm. Soc. 1948 70 3900. 70 W. G. Dauben J. C. Reid and P. E. Yankwich Analyt. Chem. 1947 19 828. soA. Murray W. W. Foreman and W.Langham J. Amer. Chem. Soc. 1948 70 J. C. Sowden J. Biol. Chem. 1949 180 65. C. Heidelberger J . Biol. Chem. 1949 179 139. 1037. 81 C. Huggett R. T. Arnold and T. 1. Taylor ibid. 1942 64 3043. W. Sakami W. E. Evans and S. Gurin ibid. 1947 69 1110. 83 M. E. Swendseid R. H. Barnes A. Hemingway and A. 0. Nier J. Biol. Chem-. 84 L. B. Spector U.S. Atomic Energy Commission MDDC 532. H. G. Wood C. H. Werkman A. Hemingway A. 0. Nier and C. G. Stuckwisch 8' J. C. Reid Science 1947 105 208. 1942 142 47. J . Amer. Chem. SOC. 1941 63 2140. 86 0. K. Neville ibid. 1948 70 3499. J. C. Reid and H. B. Jones J. Biol. Chern. 1948 174 427. " Isotopic Carbon " Calvin et al. 1949 p. 183. Ibid. p. 180. Ibid. p. 181. 182 QUARTERLY REVIEWS Non-functionally labelled carboxylic acids. These compounds can often be prepared by carbonation of labelled Grignard reagents with non-isotopic carbon dioxide e.g.methyl-labelled 14C-sodium acetate from methyl iodide in 7 0 4 0 % yield (see Table 111). Methyl-labelled 13C-acetic acid has been made from cyanide by an ingenious synthesis 103 NaWN + Zn( WN) $-+ HO*C,H4*f3CH:NH,HCI + phenol Zn CrO HO*C,H4* 13CH3 --+ 13CH3*C0,H Doubly-labelled 14C-acetic acid is available by the following series of reactions 1°4 Mg-H H*O KOH on at 700-7 50" asbestos Ba14C03 + BalPC + 14CHi14CH TABLE I11 Non-functionally labelled compounds Compound. 14CH,*C02H . . . . 13CH,*C0,H . . . . l4CH,J4CO2H . . . CH,*13CH,*COzH . . . l3CH,*CR,*CO,H . CH,*CH,*13CH,*C02H' 13CH,*13CO-C02H . . . CH,*14CO*C0,H . . . CloH21*14CH,*[CH2] ,420 2H 14CHi14CH. . . . .14CH,:14CH . . . . CH,*CHWH,. . . . CH,*CH2J3CH . . . CH,JWH,*OH . . . WH,*CH,*OH . . . CH,*lWH,Br . . . . l*CH,*CH,Br . . . . Ref. 102 103 104 105 105 105 82 99 106 107 95 108 109 110 111 112 112 112 112 Compound. 14CH,J4CH,I . . . . . 14CH,*14CH2MgI. . . . . CH3*CH2*14CHz*OH. . . . C6H,*14CO*CH . . . . . C6H,*14CO*CH0. . . . . C6H,*14CH(OH)*C02H . . . '"-Biotin . . . . . 1 6-13C-Dehydroisoandrosterone acetate . . . . . . . 2 1 -14C-Progesterone . 1 7 - ( 14C-Methyl) - testosterone . 3-14C-Testosterone . . . . l-14C-~-Glucose . . . . . l-14C-~-Mannose. . . . . 1-l4C-Methadone. . . . . C6H,*14C0.CHBr . . . Ref. 113 113 110 86 86 86 86 60 114 115 116 116 117 74 74 112 92W. G. Dauben J. Org. Chem. 1948 13 313. 93 C. Heidelberger P. Brewer and W. G. Dauben J. Amer. Cltena. Soc. 1947 9 4 S.Weinhouse G. Medes and l!?. F. Floyd J . Biol. Chem. 1944 155 143. 95 W. G. Dauben J. Amer. Chem. SOC. 1948 70 1376. g6 H. J. Harwood and A. W. Ralston J . Org. Chem. 1947 12 740. D7 E. Hines and A. Gemant Science 1949 110 19. B8 C. J. Collins J . Amer. Chem. SOG. 1948 70 2418. g9 D. M. Hughes and J. C. Reid J . Org. Chem. 1949 14 516. 69 1389. loo " Isotopic Carbon" Calvin et al. 1949 p. 191. lol Y. J. Topper J . Biol. Chem. 1949 1'77 303. lo2 B. M. Tolbert ibid. 1948 173 205. lo3 H. S . Anker ibid. 1946 166 219. ARNSTEIN AND BENTLEY ISOTOPIC TRACER TECHNIQUE 183 Benzene derivatives labelled in the 1 3 5-14C-Toluene has been Ring-labelled aromatic compoundr. ring have been made by three procedures. synthesised from sodium pyruvate via uvitic acid 99 CH3 I CH3 C02H CH3 I NaOH = fi-gL-& CO2H * 4CH3*WO-C0,Na + H02C C0,H H0,C M.Fields et al. have labelled toluene in the 1-position and also prepared benzoic acid and benzene 118 80- 8 6 ___jl C02H 85 % cH6H [CH nls(MgBr)* Ba14C03 CHa*l*CO,Et - CH3 CH3 I (2J 90-95% ~ f!J 90-95%+ 0 ~ 6 1 3 5-14C-Mesitylene has been made from acetone 119 Most other aromatic compounds have been prepared by cyclisation of labelled carboxylic acids. Thus 9J4C-1 2-5 6-dibenzanthracene and 20-methyl-ll-14C-cholanthrene have been made Q2 93 from the 14C-carboxyl- 104 R. Abrams Experientia 1947 3 488. 105 " Isotopic Carbon " Calvin et al. pp. 192 193. lo6 H. S. Anker J . Biol. Chem. 1948 176 1333. 10' M. Calvin and R. Lemon J . Amer. Chem. Soa. 1947 69 1232. 108 W. J. Am01 and R. Glascock Nature 1947 159 810. loS Idem i b g .1948 161 932. 110 B. A. Fries and M. Calvin J . Amer. Chem. SOC. 1948 70 2235. 111 0. Beeck 5. W. Otvos D. P. Stevenson and C. D. Wagner J . Chem. Physics 1la B. M. Tolbert F. Christenson F. Nai-Hsuan Chang and P P. T. Sah J . Org. lla W. J. Am01 and R. Glascock Nature 1949 163 61. 11* E. B. Hershberg E. Schwenk and E. Stahl Arch. Biochem. 1948 19 300. ll6 B. Riegel and F. S. Prout J . Org. Chem. 1948 13 933. 116 H. B. MacPhillamy and C. R. Scholz J . Biol. Chem. 1949 178 37. 117 R. B. Turner Science 1947 106 248. ll*M. Fields M. A. Leaffer and J. Rohan Wd. 1949 109 35. llS A. V. Grosse and S. Weinhouse &id. 1946 104 402. 1948 16 255. Chem. 1949 14 525. 184 QUARTERLY REVIEWS labelled 1- and 2-naphthoic acids e.g. 14C02H __3 __+ ________j. CS,-AICI 80% 2-Acetamido-9- 14C-fluorene was made by a similar series of reactions from 2-iododiphenyl 120 Grignard HO,W 0 Reduction 9- 14C-Phenanthrene has been prepared by the Wagner rearrangement of 9- 14C-hydroxymethylfluorene 98 Arnino-acids (see Table IV).Carboxyl- or a-labelled glycine has been prepared by reaction of chloro- 121 or bromo-acetic acid lZ2 with ammonia. Other syntheses of carboxyl-labelled glycine have been carried out with 13C or 14C cyanide *% 82 2H-CHO + K1*CN --+ CH,:N*CH,*14CN + HzN*CH,*14CN 4 H,NCHa*14C02H x2O F. E. Ray and C. R. Geiser Science 1949 109 200. l a x R. Ostwald J . Biol. Ch.e.m. 1948 178 207. lZ4 N. Olsen A. Hemingway and A. 0. Nier ibi&. 1943 148 611. ARNSTEM AND BENTLEY ISOTOPIU TRACER TECHNIQUE 185 Carboxyl-labelled alanine has been made from isotopic cyanide 6% 123 CH,*CHO + H W N + NH -+ CH,*CH*l*CN -+ CH,*CH14C0,H I i NH NH* Cyanide has also been used in a novel synthesis of cc-labelled alanine c6H5*cHo + Cl*CH,*CO,Na + Na13CN -+ C,H5*CH:Cm1WN -+ I C0,H Li salt LiMe C,H5*CH:CH*13CN -+ C,H,*CH:CH*13C0,H -+ Ctl(MnO O i C,H5*CH:CH*13CODCH + oxime -+ C,H5*CH.CH-13CH*CH8 ____3.1 121 82 122 62 62 123 124 125 77 126 55 127 128 128 hH2 DL-TJTOS~~~ . DL-3 4-Dihy- droxyphenyl- alanine . . DL-Tryptophan DL-Lysine . . ~-Lysine . . DL-Leucine . Anthranilic acid I NH*COPh TABLE IV Carbon-labelled amino-acids Amino-acid. Glycine. . . DL-Alanine . . DL-Serine . . L-Serine. . . DL-Methionbe . L-Methionine . DL-Phenyl- alanine Position of label. 1 Ref. I I Amino-acid. I- /I 1-14c 2-14c 1 -13c 1 -14c 1 . 1 4 ~ 1-13c 2-13c 16N-1-13C 3-"C 34s-3 4-1q3 Methyl-"C cc -Carboxyl- l4C a-Carboxyl-14C 1 3 5-14c Position of label.p4c # P C p-14c Carboxyl-lsC 4 4 C €-'4C p c yJ4C 16N-fl.14c Carbox yl-lac! Ref. 87 88 129 130 76 131 132 133 134 135 135 136 laS S. Gurin and D. W. Wilson Fed. Proc. 1942 1 114. 1*4 J. Baddiley 0. Ehrensvgrd and H. Nilsson J. Biol. Chem. 1949 178 399. 126D. Shemin ibid. 1946 162 297. 126 G. W. Kilmer and V. du Vigneaud ibid. 1944 154 247. 12'S. Gurin and A. M. Delluva ibid. 1947 170 545. 128 B. Schepartz and S. Gurin ibid. 1949 180 663. lsS "Isotopic Carbon " Calvin et at. 1949 p. 225. lao C. Heidelberger J. BWZ. Ohem. 1948 175 471. lS1 H. W. Bond ibid. p. 531. ls3 P. Olynyk D. B. Camp A. M. GrifBth S. Woislowski and R. W. Helmkamp lS4 H. Borsook C. L. Dewy A. J. Haagen-Smit G. Keighley and P.H. Lowy ls6 M. J. Coon S. Gurin and D. W. Wilson _Fed. PTOC. 1949 8 192 ; M. J. Coon l** J. F. Nye H. K. Mitchell E. Leifer and W. H. Lmgham ibid. 1949 179 783. R. W. Schayer C. L. Foster and D. Shemin Fed. Proc. 1949 8 248. J. Org. Chem. 1948 18 465. J. BioE. Chena. 1948 176 1383. and S. Gurin J. Biol. Chem. 1949 180 1159. 186 QUARTERLY REVIEWS Syntheses with Heavy Nitrogen The introduction of 15N from available starting materials (see p. 176) involves the formation of a carbon-nitrogen linkage and the synthetic reactions used for this purpose are relatively few ( a ) Gabriel phthalimide synthesis e.g. preparation of 15N-glycine 137 HC1 f$('>-> Cl.CH,.CO ,E t Kco> l6N*CH2*CO2Et 4 co 0 15NH2*CH,*C0,H Two equivalents (three in the case of dicarboxylic acids) of 15NH are used and the isotope which does not react is re~0vered.l~' (c) Cyanamide syntheses.Labelled cyanamide has been prepared from ammonia and cyanogen bromide 138 Z15NH3 + CNBr -+ 15NH2*CN + 16NH,Br Condensation of 15N-cyanamide with 15N-sarcosine has been used to prepare creatine 138 15NH,*CN + CH,*l5NH*CH2*CO2H -+ 15NH:C(NH2)*15N(CH,)*CH,*C0,H Cyanamide has also been converted into isoguanine 40 and g~anidine.13~ ( d ) Syntheses with urea. from 15N-urea,140s 141 e.g. (b) Knoop-Oesterlin reduction of keto-acids. king-labelled uric acid has been synthesised C0,Et 7"\ M ~ O K 16NH CH*NHAc Hcl 15NH2*CO*16NH2 + CH*NHAc . + I I .) I I C0,Et 82--90% co co 90-95'$'0 \ / 6NH Po\ 15NH CH*NH I I co co 15NH \ / KCNO _____ /""\ 16NH CH*NH*CO*NH I I co co 15NH \ / HCI --+ 80 % (e) 15N-Sodium thiocyanate prepared in good yield from ammonia,142 137 R.Schoenheimer and S. Ratner J . BioZ. Chem. 1939 127 301. 138K Bloch R. Schoenheimer and D. Rittenberg ibid. 1941 138 155. 139 a. B. Brown P. M. Roll and A. A. Plentl Fed. Proc. 1949 6 517. 140K. Bloch and R. Schoenheimer J. BioZ. Chem. 1941 138 167. I4l L. F. Cavalieri V. E. Blair and G. B. Brown J . Am.er. Chem. Soc. 1948 70 143 C. Tesar and D. Rittenberg J . Biol. Chem. 1947 170 35. 1240. ARNSTEIN AND BENTLEY ISOTOPIC TRACER TECHNIQUE 187 has been used to label histidine by an interesting synthesis from natural optically active histidine 142 HC--C*CH,*CH*CO,Me C,H,.COCI HC=C*CH2*CH*CO2Me H,O I I - 1 I I ___ k NH NH Na&Os NHBq NHBz NHBz b H ’ NaSCI6N Fea(SOJs H,N*CO*CH,*CH*CO,H -% HC===AWH,*CH*CO,H ____+ I i I NH N 16NH NH LC/ I SH HC=C*CH,*CH*CO,H N 16NH NH I I I \CH/ (f) lW-Formamidine has been prepared 139 and converted into 15N- (9) Proline has been labelled by an interesting method involving sub- adenine.143 stitution of a heterocyclic oxygen with heavy ammonia 144 ,c1 H,O-HC1 ____+ f5N 15NH 62 5NH \ CO2H I5NH2 Other similar syntheses are tabulated in Table V page 188. Syntheses with Deuterium In recent years 13C and 14C have to some extent superseded deuterium in biological tracer work. However its low cost and ready incorporation into organic compounds partly outweigh its chief disadvantage vix. loss of label by exchange and its importance in the study of reaction mechanisms is undiminished. The hydrogen atoms of organic compounds can be broadly divided into three groups with respect to their stability to exchange under normal conditions (1) Stable e.g.hydrogen atoms of paraffin hydrocarbons ; 1 4 3 L. F. Cavalieri J. F. Tinker and G. B. Brown J . Awaer. Chem. XOC. 1949 14* M. R. Stetten and R. Schoenheimer J . Biol. Chem. 1944 153 113. 71 533. 188 QUARTERLY REVIEWS TABLE V ?"Labelled cmpounds Amino-acid. D L - A I ~ ~ ~ o . . . . L(+)-Alanine. . . . DL-Arghhe . . . . DL-Aspartic acid . . . DL-Glutamic acid. . . L( +)-Glutamic acid . . Glycine . . . . ~(+)-Histihine . . . DL-Hydroxyprohe . . leucine. cine. . . . leucine. cine. . . . D L - L Y S ~ ~ (a-l5N). . . L( +)-Lys+e (a-l5N) . . DL-Norleucine. . . . DL-Phenylalanine . . DL-Phenylaminobutyric acid . . . . . . L( + ) -Phenylaminobut yric acid . . . . . . D( - )-Phenylaminobutyric acid .. . . . . L(-)-Proline . . . . Sarcosine . . . . . DL-Serine . . . . . L(+)-Argi?ine . . . DL-LOUChe . . . . D(-)-LySlIlO (a-16N) . . D(-)-LyShe (a-"N) . . DL-PX'Olhe. . . . . L(-)-Serine . . . . D(+)-Serine . . . . DL-Tyrosine . . . . Ref. 137 125 125 140 140 137 137 125 125 145 142 146 137 147 147 148 149 149 149 150 137 137 151 151 151 144 144 125 152 153 154 154 154 137 Compound. l5NH,*CN . . . . . . . NaSClSN . . . . . 15NH,*CH16gH,HCi . . . . 16NH,*CO*16NH . . . . . f6NH:C(OMe)*16NH . . . . 16NH,*CS*16NH . NH,-C:(NH)*16NMe*CH,*C0,H . HN*C:(NH)*16NMe*CHa*C0 . . 1 6 ~ ~ ,a ( ~ ~ N H ~ N M ~ ~ C H ,GO ,H' I I 16NH,*C:(16NH)*1SNH . . . . NH,*C:(NH)*16NH*CHz-C0,H . . NH,*CO*16NMe*CR,~C0,H . . . Betaine . . . . . . . . Choline . . . . . . . . Ethanolamine . . . . . . Aniline .. . . . . . . 2-Phenylindole . . . . . . 1 . 3-l5N-Thymine . . . . . 1 3-16N-Uracil . . . . . . NH**CO*"NH*CH,*COgH * 1 3-16N-Uric acid . . . . . 9-lW-Uric acid . . . . . . 1 3-16N-Xanthine . . . . . 1 :2:3-lSN-Guanine . . . . 1 3-16N-isoGuanine . . . . Orotic acid . . . . . . . 1 3-1sN-2 6-Diaminopurine . . 1 3-l'N-Adenine . . . . . Ref. 138 140 142 139 143 140 141 140 156 138 152 152 156 140 140 140 157 145 140 158 140 159 159 156 156 141 141 143 156 156 53 160 40 V. du Vigneaud S. Simmonds J. P. Chandler and M. Cohn J. BioZ. Chem. R. Schoenheimer S. Ratner and D. Rittenberg J. BioZ. Chem. 1939 130 703. 1946 165 639. ld6 M. R. Stetten Fed. Proc. 1949 8 256. lP8 S. Ratner R. Schoenheimer and D. Rittenberg ibid. 1940 134 653. 14@ N. Weissman and R. Schoenheimer ibid. 1941 140 779.160 R. M. Fink T. Enns C. P. Kimball H. E. Silberstein W. F. Bale S. C. Maddens lS1V. du Vigneaud M. Cohn G. B. Brown 0. J. Irish R. Schoenheimer and 15 K. Bloch and R. Schoenheimer ibid. p. 111. 163 V. du Vigneaud S. Simmonds and M. Cohn ibid. 1946 166 47. 154 D. Stetten ibid. 1942 144 501. lS6 A. A. Plentl and R. Schoenheimer ibid. 1944 153 203. lS7 D. Stetten ibid. 1941 140 143. 15@ C. F. H. Mlan and C. V. Wilson J. Arner. Chem. Soc. 1943 65 611. l 6 0 S. Bergstrom H. Arvidson E. Hammersten N. A. Eliasson P. Reichard and and G. H. Whipple J . Exp. Ned. 1944 80 456. D. Rittenberg J. BioZ. Chem. 1939 131 273. G. E. Boxer and D. Stetten ibid. 1944 153 617. H. v. Ubisch J . BWZ. Chem. 1949 177 495. ARNSTEM AND BENTLEY ISOTOPIC TRACER TECHNIQUE 3.89 (2) semi-labile e.g.the methylene hydrogen atoms of glycine,161 and the enolisable hydrogen atoms of a ketone whose exchange with yater is catalysed by alkali ;I62 (3) labile e.g. the carboxyl hydrogen of a carboxylic acid. The lability of a hydrogen atom is increased by an " activating '' factor (such as unsaturation or a high concentration of positive charges) on the adjacent carbon atom and it must be ascertained in each case. Since exchange may occur during the isolation procedure as well as in a biochemical reaction the effect of the isolation technique should be checked independently. Exchange Reactions.-Under sufficiently vigorous conditions ordinarily stable hydrogen atoms become labile or semi-labile. Aromatic compounds exchange ring hydrogen atoms when heated with DC1 and aluminium chloride l 6 3 or with concentrated D,S0,.164 When heated with concen- trated D,SO, several amino-acids 1 6 1 and fatty acids 165 exchange the a-hydrogen atoms but under different conditions (D,O and platinum a t 130") fatty acids take up more deuterium which is evenly distributed throughout the ~ h a i n .1 ~ ~ Cholesterol has been labelled by heating it with heavy water acetic acid and platinum ; 166 more deuterium was introduced into the side chain than the rings. The chief disadvantages of this method are the unavoidable dilution of the heavy isotope which can be minimised by using a large excess of D,O and the difficulty of labelling specific positions in some cases e.g. cholesterol. Addition Rea&ons.-Deuterium gas can be added to CGC C=C C H C=N or C=O bonds but only the carbon-bound atoms are resistant to subsequent exchange.Reductions are carried out catalytically with D, or chemically with D,O or reagents such as lithium aluminium deuteride. Amino-acids have been labelled by reductive amination 137 167 and by reduction of the oximes and phenylhydrazones of a-keto-acids.la8 Hydrogenation reactions give compounds labelled in a t least two posi- tions ; if one of the positions is " activated " the label can be " washed out" by exchange and a specifically labelled compound is obtained. Another method of labelling only one position is by addition of DBr to a double bond. Dry DBr is made by heating deuterium with bromine 169 or reaction of D,O with thionyl bromide.17O D. Rittenberg A. S. Keston R. Schoenheimer and G. L. Foster J . Biol. Chem. 1938 125 1. 162 M.Anchel and R. Schoenheimer ibi&. p. 23. 16* C. K. Ingold C. C. Raisin and C. L. Wilson J. 1936 915. 165 W. E. van Heyningen and D. Rittenberg J. Biol. Chem. 1938 125 495. lB6 K. Bloch and D. Rittenberg ibid. 1943 149 505. 16' D. Rittenberg S. Ratner and H. D. Hoberman J. A w r . Chem. SOC. 1940 A. Kilt and A. Langseth 2. physikal. Chem. 1936 176 A 65. 62 2249. D. Shemin and R. M. Herbst ibid. 1938 60 1951. C. L. Wilson and A. W. Wylies J. 1941 596. 170 R. C. Elderfield W. J. Gensler F. Brody J. D. Head S. C. Dickerman L. Wiederholt C. B. Kremer H. A. Hageman F. J. Kreysa J. M. Griffing S. M. Kupchan B Newman and J. T. Maynard J . Amer. Chem. SOC. 1946 68 1679. 0 190 QUARTERLY REVIEWS Sometimes water can be added to an unsaturated system e.g. prepara- tion of acetaldehyde from acetylene and D20.1711 172 Replacement Reactions.-Decarboxylation of an acid labelled with deuterium in the carboxyl group results in the transfer of the label to the a-carbon atom.For example acetaldehyde labelled with deuterium in the aldehyde group can be prepared by decarboxylation of CH,*CO*C02D and deuterobenzene has been made by decarboxylating calcium mellitate in the presence of Ca(OD),.173 Halogen atoms are easily replaced with deuterium by decomposing the Grignard reagent with heavy water by reduction with deuterium and palladium-black or by reaction with Raney nickel containing deuterium. Desulphurisation with Raney nickel containing deuterium 174 results in the replacement of a thiol or thio-ether group with deuterium. This method has recently been used to prepare 7-deuterocholesterol from 7- bromo- cholesterol and 7 7-dideuterocholesterol from 7-ketocholesterol 175 (after conversion into the mercaptol by reaction with ethanedithiol) as well as deuterode thiopenicillin.174 Limitations of the Tracer Technique Interpretation of Tracer Experhents.-The outlines of the tracer tech- nique may be summarised as follows A compound X containing an excess above normal of one or more isotopic atoms is converted into Y. If after purification Y also contains an excess of the isotope one or more atoms of Y must have been derived from X. It would seem therefore easy to decide whether X is a direct precursor of another substance Y simply by a determination of the isotopic content of Y. There are however several factors which may affect more or less seriously the interpretation of the results especially of experiments in wiwo.For example the labelled sub- stance though a normal intermediate in an animal may be unstable under the conditions of administration or may be destroyed before absorption or may not be transferred across cell membranes. If the labelled com- pound is readily oxidised in wivo the labelled atom may also be incor- porated into compound Y by an alternative pathway often involving fixa- tion of carbon dioxide. I n witro experiments may be used advantageously in such cases as one of the pathways may be selectively blocked by suit- able enzyme inhibitors. It is often difficult to make valid quantitative or even qualitative interpretations of in vivo experiments and some effects peculiar to isotopic compounds and their reactions are discussed below.The isotope eflect. Since isotopes differ in mass and zero-point energy physicochemical Merences may be expected to become marked in a con- tinuous process where repeated fractionation can occur. In 1939 A. 0. Nier and E. A. Gulbranson176 found that the 12C-isotope was enriched in 171 K. BIoch and D. Rittenberg J. BioE. Chem. 1944 155 243. 172 J. E. Zanetti and D. V. Sickman J . Amer. Chem. Soc. 1936 58 2034. 173 H. Erlenmeyer and H. Lobeck Helv. Chim. Acta 1935 18 1464. 174 " The Chemistry of Penicillin " Princeton University Press 1949 p. 267. 176 D. K. Fukushima and S Lieberman Ped. Proc. 1949 8 200. 17* J . Amer. Chern. Soc. 1939 61 697. ARNSTEIN AND BENTLEY ISOTOPIC TRACER TECHNIQUE 191 plants ; the same effect has recently been explained by the faster utilisa- tion of 12COg compared with W02 for photosynthesis by barley seed- lings.177 Since the enhanced 1 8 0 content of the atmosphere could not be due to photosynthetic reactions 178 a possible explanation was fractionation by bacteria.Fractionation by soil bacteria was observed but it was in- sufficient to explain the enrichment quantitatively. 179 Recently it was found that in the hydrolysis of 14C-labelled urea with the enzyme urease the carbon dioxide produced early in the reaction had a higher specific activity than the later fra~tions.17~~ With carbon the 12C-12C bond seems to be less stable than either the 12C-13C or the l2C-l4C bond. When l-13C-propane was pyrolysed it was found that the 12G12C bonds were ruptured 8% more often than the 12C-f3C bonds. l80 During the decarboxylation of malonic and bromomalonic acid the 12C-12C bond was ruptured 1.12 and 1.4 times respectively as fre- quently as the 12C-W bond.181 A similar effect has now been observed in the alkaline hydrolysis of 1%-labelled ethyl benzoate the rate of hydro- lysis of the 14C-labelled compound was slower than that of the normal ester.l81O Isotopic asymmetry. Attempts to prepare compounds possessing optical activity as a result of isotopic asymmetry have usually been unsuccessful. In 1936 G. It. Clem0 and A. McQuillen 182 claimed to have synthesised and resolved a-pentadeuterophenylbenzylamine An isotope effect has also been observed in chemical reactions. C6H C$&*CO*C6D6 “6H6>:N*H + 6)CH*NH* C6D5 C,D but the observed rotation was very small. Attempts by H.Erlenmeyer and H. Schenkel lS3 to resolve a similar compound CIID,*CHPh*C02H were unsuccessful. Recent experiments on the preparation of an optically active acid R*CHDC02H (R = benzyl ethyl or n-butyl) by the Marckwald asymmetric synthesis also failed in spite of elaborate precautions to pre- vent racemisation during the reaction.184 Nevertheless it is believed that failure to obtain an isotopically asymmetric compound is due to practical rather than theoretical reasons especially as L. E. Young and C. W. Porter 185 have shown that replacement of hydrogen by deuterium in an optically active compound changes the specific rotation. * 177 J. W. Weigl and M. Calvin J. Chem. Physics 1949 17 210. 178 S. Ruben M. Randall M. D. Kamen and J. L. Hyde J . Amer. Chem. SOC. M. Dble R. C. Hawkings and H.A. Barker ibid. 1947 69 226. 179a F. Daniels and A. L. Meyerson Science 1948 108 676. l** D. P. Stevenson C. D. Wagner 0. Beeck and J. W. Otvos J . Chem. Physics lS1 P. E. Yankwich and M. Calvin ibid. 1949 17 109. l8la W. H. Stevens and R. W. Attree Canadian J . Res. 1949 27 B 807. lea J. 1936 808. Helv. Chim. Ada 1936 19 1169. 18* D. J. G. Ives and M. R. Nettleton J. 1948 1085. IEti J . Amer. Chem. Soc. 1937 59 328 437. * [Added in proof.] E. L. Eliel ( J . Amer. Chem. Soc. 1949,71 3970) has recently prepared optically active Ph*CHD*CH by reduction of optically active Ph*CHCl*CH with lithium duminium deuteride. 1941 61 877. 1948 16 993. 192 QUARTERLY REVIEWS In the interpretation of biochemical tracer experiments it has been assumed that if the administration of a compound (X) labelled in a specific position gave rise to a compound (Y) in which the labelled atoms are equally distributed in two or more positions the conversion of X into Y must have taken place through a symmetrical intermediate (2) * * * * * R,*CH,*CH,R -+ R,*CH,*CH,R -+ R,*CH,*CH,R (XJ (2.1 (Y4 conversely if the distribution of the labelled atoms in Y was unequal it was concluded that a symmetrical compound (Z) could not have been an intermediate.This argument has been applied to the biochemical conver- sion of serine into glycine 186 where the glycine isolated after administration of serine labelled with l5N and 13C (in the carboxyl group) had the same 15N 13C ratio ; it was concluded therefore that aminomalonic acid could The tricarboxylic acid cycle (after H. G. Woodl87) a a a a b a a CH,*CO,H succinste oxidation inhibited by malonate b a a1 b HO,C*CH,*C( OH) *CO,H Citrate 11 a CH,*CO,H b a a l b HO,C*CH==C C0,H cis- Aconitate 11.a CH,*CO,H b a a/ b HO,C*CHOH*CE*CO,H isocitrate 11 b 6H3*80*X + CO 2 b a ‘ b H02C*CH2*60*C0,H Oxaloacetate a - a b $ CH,*COCO,H + Pyruvate b HO 26*dH,*5H (OH )*CO,H Malato 11 la8 D. Shemin J . Biol. Chem. 1946 162 297. Is7 Physiol. Rev. 1946 26 198. ARNSTEIN AND BENTLEY ISOTOPIC TRACER TECHNIQUE 193 not have been an intermediate for in this case the expected 15N lSC ratio would be twice that of the starting material. A more important example of this type is the relationship of citric acid and the tricarboxylic acid cycle. The a-ketoglutaric acid isolated during studies on the fixation in vitro of WO by pigeon liver contained llC almost exclusively in the a-carboxyl group.188 In identical experiments with W02 the addition of non-isotopic citric acid caused no decrease in the isotope content of the a-ketoglutaric acid.lS9 On the basis of the dis- tribution of isotope in a-ketoglutaric acid citric acid (a symmetrical com- pound) was excluded as a direct participant of the tricarboxylic acid cycle as shown on previous page. However it has been pointed out by A. G. Ogston lQO that some sym- metrical molecules might behave like an asymmetric compound if three-point attachment to an asymmetrical enzyme were a pre-requisite for reaction. The main condition for the recognition of such a complex by the tracer tech- nique is the possibility of '' isotopic asymmetry " in the <' symmetrical '' com- %\c/C02H R /aO,H pound; thus fulfils this requirement whereas C R2/ \CO,H R/ \CO,H does not.Both aminomalonic acid (labelled in one carboxyl group) and citric acid (labelled in one of the terminal carboxyl groups) could exist in two " isotopic modifications " which can in theory at least be derived from D- and L-forms e.g. * / E. A. Evans and L. Slotin J . Biot. Chem. 1941 141 439. Is9 H. G. Wood C. H. Werkman A. Hemingway and A. 0. Nier ibid. 1941 188 483; 1942 142 31. l90 Nature 1948 162 963. 194 QUARTERLY REVlEWS These ‘‘ isotopic modifications ” have not yet been resolved chemically ; however in a biological system containing a presumably asymmetrical enzyme such stereochemical differences have been observed as for example in the following experiment.lsl Citric acid biosynthesised in vitro from oxaloacetate pyruvate and l4COa was found to be isotopically asym- metrical Le.only one of the terminal carboxyl groups contained 14C (as expected from the tricarboxylic acid cycle). After isolation and purifi- cation by the carrier technique the labelled citric acid was converted enzymically into or-ketoglutarate. The or-ketoglutaric acid was found to contain 14C almost exclusively in the a-carboxyl carbon atom thus pro- viding experimental proof for Ogston’s views. Ogston’s original concept concerning biosynthesis through an enzyme substrate complex has recently been extended 192 to include any partial asymmetrical synthesis involving the reaction of a symmetrical compound labelled with an isotope in one position with an asymmetric reagent. lgl V. R.Potter and C. Hsidelberger Nature 1949 164 180. IQ2 P. E. Wilcox ibid. p. 757.

ISSN:0009-2681    

DOI:10.1039/QR9500400172

出版商:RSC    

年代:1950

数据来源: RSC

摘要:

SO= ASPECTS OF FmtAN AND PYRAN CHEMISTRY By D. G. JONES PH.D. D.I.C. A.R.I.C. and A. W. C. TAYLOR B.Sc. PH.D. A.R.I.C. (RESEARCH DEPARTMENT IMPERIAL CHEMICAL INDUSTRIES LIMITED BILLINGHAM Co. DURHAM) THE chemistry of furan and ppan compounds has been of academic interest for over a hundred years but during the last thirty years has become of increasing commercial importance mainly owing to the establishment of suitable manufacturing processes for furfuraldehyde from pentosan-contain- ing agricultural materials.1 It is not intended to deal here with all the many aspects of furan and pyran chemistry as comprehensive reviews have recently appeareda2 This Review deals mainly with those aspects in which there have been noteworthy advances in the last five years namely the hydrogenation and oxidation of furan compounds and the chemistry of the tetrahydrofurans and di- and tetra-hydropyrans which has in the main followed from the ready accessibility of 2 3-dihydro-4-pyran (11) by catalytic dehydration of tetrahydrofurfuryl alcohol (I) (14 (11.) A striking advance has also been made in the conversion of furfur- aldehyde into the parent compound furan (111) which until recently could be made only in small quantities by the decarboxylation of furoic acid thus (111.) In the vapour phase at 200-280° in the presence of catalysts especially nickel furfuraldehyde decomposes to furan and the yield is improved to 65% by presence of a limited quantity of hydr~gen.~ R.Paul obtained furan among the products of decomposition of furfuryl alcohol over Raney nickel at 150". The vapour-phase decomposition of furfuraldehyde to furan over lime at 350450" was the first major advance towards a satis- factory preparation of furan being a distinct improvement on the passage (a) F.N. Peters ibid. 1936 28 755 ; 1939 81 178 ; 1948 40 200 ; A. Wacek Angew. Chem. 1941 54 453; L. N. Owen Ann. Reports 1945 42 157; (b) 0. W. Cass Id. Eng. Chem. 1948 40 216 ; (c) G. F. Wright and H. Gilman $bid. p. 1817. R. Paul Bull. Soc. chim. 1933 [iv] 53 1489 ; B.P. 647,334 ; C. H. Kline and J. Turkevich J. Amer. Chem. Soo. 1945 67 498. Org. Synth. Coll. Vol. I 2nd Edn. 1941 p. 274. C. L. Wilson J. 1945 61 ; B.P. 553,175. R. Paul Bull. SOC. chim. 1938 5 1592 ; 1941 8 607. 195 1 H. J. Brownlee and C. S. Miner Id. Eng. Chem. 1948 40 201. ' U.S.P. 2,337,027. 196 QUARTERLY REVIEWS of furfuraldehyde through fused caustic alkali or over soda-lime.* More recently details have been published of the catalytic vapour-phase decom- position of furfuraldehyde in the presence of steam and a dehydrogenation catalyst such as zinc manganese chromite at a temperature of 350450" with a steam-furfuraldehyde molar ratio of between 2 1 and 6 1.The overall reaction in which 85-90% yields of furan are obtained is This reaction has now been applied commercially as a continuous process.2b Hydrogenation Reactions of Furm Compounds By reason of their structure furan and compounds containing the furan ring can be hydrogenated under various conditions to give a wide variety of products. Only the more important types and technically interesting examples are dealt with here. Information on these subjects has been summarised on several occasions,1° but such important advances have been made in the last 3-5 years as to make it desirable to review the present position.The reactions are conveniently classified into those involving 1. Hydrogenation of a side chain only ; 2. Hydrogenation of the nucleus and a side chain; 3. Hydrogenolysis of the ring. 1. Side-Chain Hydrogenation.-Hydrogenation of a side chain attached to a furan nucleus without saturation of the nucleus itself has proved possible. Catalysts containing Group VIII metals can be employed under liquid- phase conditions if care is taken to use moderate temperatures otherwise ring saturation will also occur. For instance furfuraldehyde may be con- verted into furfuryl alcohol in 90% yield by use of a cobalt-on-kieselguhr catalyst at 50 atm.hydrogen pressure and 80-100" l1 or by use of a for- aminate cobalt catalyst * l2 at 10-30" ; at temperatures above 120" there is a tendency for tetrahydrofurfuryl alcohol to be formed. Similarly furfurylideneacetaldehyde (3-furylacraldehy de) (IV) in the presence of 8 C. D. Hurd A. R. Goldsby and E. N. Osborne J . Amer. Chem. Xoc. 1932 53 2532. B.P. 575 362. 10 (a) H. Adkins " The Reactions of Hydrogen with Organic Compounds over Copper Chromium Oxide and Nickel Catalysts " Univ. of Wisconsin Press Madison 1937; (a) L. N. Owen Ann Reports 1945 42 166; (c) B. H. Wojcik Ind. Eng. Chem. 1948 40 210. l1 B.P 605,922. * Foraminate catalysts are prepared by extraction of a non-catalytically active metal from the surface layers of granules of alloys of that metal with other cata- lytically active metals.Examples of such alloys are nickel-aluminium and -mag- nesium cobalt-aluminium and copper-aluminium -silicon and -zinc ; those most commonly used contain aluminium. The extraction is done by either acid or alkaline solutions as required. The foraminate catalysts are different from the Raney catalysts which are in powder form and in which the extraction of the non-catalytically active metal is virtually complete. For fwrther information see B.P. 611,987 621,749 623,595 624,035 and 628,405. 1 B.P. 627,293. JONES AND TAYLOR FURAN AND PYRAN CHEMISTRY 197 Raney nickel in ethanol a t 23" gives 48% of 3-(2-furyl)propaldehyde (V) and 30% of 2-(3-hydroxypropyl)furan (VI) (W.) (V. 1 (VI.) whereas at 80" the main product is (VI) and at higher temperatures nuclear hydrogenation begins.l3 The extent of hydrogenation depends to some extent on the nature of the substituents with aldehyde and amino-groups decreasing the extent of reaction. Thus furan is more readily saturated than is furfuraldehyde or furfurylamine. A similar suppressing effect is achieved by the addition of ammonia or an amine; thus using a nickel catalyst at 50" and at 50 atm. pressure and in presence of aqueous ammonia a 95% yield of 2-ethylfuran (VII) is obtained from 2-vinylfuran (VIII) which without the ammonia under otherwise similar conditions gives an 85y0 yield of 2-ethyltetrahydrofuran (IX) l4 (VII.) (VIII. ) (IXJ Similarly the complete hydrogenation of 2-(2-cyanovinyl)furan is suppressed and with Raney nickel at 50" mixtures of 3-(2-furyl)propylamine and di-3-( 2-fury1)propylamine are obtained.Also 2-fury1 cyanide in the presence of ammonia gives furfurylamine over Raney nickel at room temperat~re.1~ An alternative method of hydrogenating the side chain and not the ring is to employ copper-containing catalysts either under high pressure in the liquid phase or at somewhat higher temperatures a t atmospheric pressure in the vapour phase. However here again care is necessary as too drastic conditions lead to hydrogenolysis of the furan ring as described below. The use of copper chromite catalysts 1 0 ~ for the conversion of furfuralde- hyde into furfuryl alcohol at 75-250 atm. and at 150-175" is well known and these conditions have been used in a commercial batch process.lOO Recent improvements in catalysts have made possible continuous processes which with a foraminate copper catalyst under 250 atm.at 80" give sub- stantially complete conversion of furfuraldehyde into furfuryl alcohol. l2 A high yield of furfuryl alcohol is also obtainable when working in the vapour phase below 150" given careful choice of the form of the copper catalyst ; l6 free alkali reduces the activity of the catalyst and an 85% yield of furfuryl alcohol 1 7 is obtained up to 265" whereas otherwise 2-methyl- furan would be the major product. Further work on the conversion of furfuraldehyde into furfuryl alcohol and 2-methylfuran with particular copper catalysts and of furfuryl alcohol into 2-methylfuran has recently H. E. Burdick and H. Adkins J . Amer. Chem. Soc. 1934 56 438. l4 B.P. 596,880. l6 W. Huber J. Amer. Chem.Xoc. 1944 66 876. l'H. D. Brown and R. M. Hixon Id. Eng. Chem. 1949 41 1382. l7 B.P. 621,743. 198 QUARTERLY REVIEWS been published.18 This control of the direction of hydrogenation of furfuraldehyde to either furfuryl alcohol or 2-methylfuran depending on the presence or absence of free alkali has been discussed by J. G. M. Bremner and R. K. F. Keeys.l9 Another recent example of side-chain hydrogenation using copper catalysts in either the liquid or the vapour phase is the conversion of 2-vinylfuran in high yield into 2-ethylfuran.20 Two " chemical " methods of effecting side-chain reductions may be mentioned treatment of furfuraldehyde with formaldehyde and alkali gives a 90% yield of furfuryl alcohol,21 and reduction of furfuryl alcohol by sodium and ethanol in liquid ammonia gives a 20% yield of 2-methylfuran with 38% of the unchanged alcoho1.22 2.Nuclear and Side-Chain Hydrogenation.-Saturation of the nucleus in furan compounds proceeds very readily under moderate pressure in presence of platinum nickel or similar catalysts and in general no difficulty is experienced in obtaining the tetrahydro-derivative. Examples are too numerous for specific mention. The effect of substituents such as 4 H 0 or -NH2 or of the presence of amines has been mentioned earlier but satis- factory hydrogenation can be obtained by modifying the catalyst.1Oa 14 l5 In particular direct hydrogenation of furfuraldehyde to tetrahydrofurfuryl alcohol in high yield has been difficult to achieve in a reasonable time with a nickel catalyst under pressure in the liquid phase.One method which avoids the difficulty is two-stage hydrogenation over a copper chromite catalyst for the formation of furfuryl alcohol which is then converted with ease over a nickel catalyst into the tetrahydro-derivative. A mixture of the two catalysts is now claimed loC to be more economical operating at 170-180"/75-100 atm. An earlier method avoided a two-stage process by using a nickel chromite catalyst under pressure in the presence of a very large excess of hydrogen-sufficient to maintain vapour-phase conditions. However the difficulty of direct conversion of furfuraldehyde into tetra- hydrofurfuryl alcohol has now 23 been overcome by hydrogenating in the liquid phase in the presence of a foraminate nickel catalyst a t a pressure greater than 50 atm. between 130" and 200". Similar results are obtained with cobalt catalysts.12 The direct hydrogenation of furfuraldehyde to tetrahydrofurfuryl alcohol has recently been described 24 a ruthenium catalyst being used but it is unlikely that this will be preferred to direct continuous hydrogenation over nickel owing to the cost and limited avail- ability of the catalyst.This section of furan chemistry cannot be left without remarking that 80 far no satisfactory conditions have been established either for the partial hydrogenation of the nucleus to give a dihydrofuran or for the complete hydrogenation of the nucleus without hydrogenation of unsaturated 18 (a) U.S.P. 2,445,714; L. W. Burnette I. B. Johns R. F. Holdren and R. M. Hixon I n d . Eng. Chem. 1948 40 502; U.S.P. 2,456,187; (b) C. L. Wilson J. 1945 61.J. G. M. Brernner and R. K. F. Keeys J. 1947 1068. 20 B.P. 621,744 ; 627,492. 21 A. M. Berkenheim and T. F. Darkova J. Gen. Chem. Russia 1939 9 924. 22 A. J. Birch J. 1945 809. 23 B.P. 608,540. 24 U.S.P. 2,487,054. JONES AND TAYLOR FURAN AND PYRAN CHEMISTRY 199 substituents. For instance it has not been possible directly to convert fur- furaldehyde into tetrahydrofurfuraldehyde. Poor yields have been obtained by the saturation of either furfuraldehyde diethyl acetal or diacetate fol- lowed by the removal of the protecting 18b Recently the oxida- tion of tetrahydrofurfuryl alcohol with air over a silver gauze has proved to be a convenient method of producing this hitherto inaccessible material in 60% yield.26 On storage tetrahydrofurfuraldehyde forms a dimer the structure of which has not yet been fully el~cidated.~' 3.Hydrogenolysis of the Furan Ring.-The opening of the furan ring by hydrogen can be effected in both the liquid and the vapour phase by employing either copper- nickel- or platinum-containing catalysts under rather more strenuous conditions than those previously mentioned for side- chain hydrogenation. The chief difficulty is to avoid the alternative reaction giving the tetrahydrofuran compound which takes no further part in the hydrogenolysis 28 and in general this difficulty is greatest when nickel or platinum catalysts are used. Thus L. W. Burnette 29 found that in the vapour-phase hydrogenation of 2-methylfuran over a partly activated Raney nickel catalyst 2-methyltetrahydrofuran was formed in !%yo yield at ZOO" together with appreciable quantities of pentan-%one and pentan- 2-01.Later work 28 with a more active catalyst showed that at 100" 2-methyl-tetrahydrofuran was the main product but that above that tem- perature pentan-%one was formed in increasing yield the maximum being 75% at 185" ; small quantities of pentan-2-01 were also formed. The earlier work of H. Adkins loo indicated that hydrogenolysis of 2-methylfuran was the preferred reaction when a copper chromite catalyst was used in the liquid phase at 200 atm. and 250" ; a 30% yield of pentan- 1-01 a 33% yield of pentan-2-01 and only a 15% yield of Z-methyltetra- hydrofuran were obtained. The reaction has been extended to furan which gave n-butanol in up to 70% yield,30 and t o 2 5-dimethylfuran which gave hexan-2-01. This work showed that the presence of acids e.g.acetic wid in the reaction medium was desirable as alkaline conditions led to a major amount of the saturated ring compound at the expense of the product formed by hydrogenolysis. Similarly hydrogenolysis occurs with substituted furan compounds in which the substituents contain oxygen e.g. furfuraldehyde furfuryl alcohol etc. Hydrogenation of furfuraldehyde or furfuryl alcohol to mixtures of Z-methylfuran pentan-1-01 and the pentane-1 2- and -1 5-diols with copper chromite catalyst at 250° i.e. under more severe conditions than necessary for the formation of furfuryl alcohol from furfuraldehyde is well known.loa A full description of this method of preparation of pentane- 1 5-diol has been p~blished.~l High yields of these diols are also claimed 32 for a process using a nickel catalyst at 40 atm.and 200-220". 26A. Hinz G. Mayer and G. Schucking Ber. 1943 76 676. 26 J. G. M. Bremner R. R. Coats A. Robertson and (Miss) M. L. Allan J. 1949 28 C. L. Wilson J . Amer. Chern. SOC. 1948 70 1313. 29 L. W. Burnette Iowa State Coll. J . Sci. 1944 19 9. 525. 27 J. G. M. Bremner and A. Robertson ibid. p. S27. ** B.P. 586,222. 81 Org. Synth. 26 83. D.R.-P. 555,405. 200 QUARTERLY REVIEWS The effect of the presence of water in the reacting system is marked ; from alkylfixans are obtained diols and keto-alcohols and from furfur- aldehyde or furfuryl alcohol mixtures of diols and triols. Thus K. S. T~pchiev,~~ using a palladium catalyst hydrogenated 2-methylfuran (X) at room temperature in the presence of water and a limited amount of an acid such as hydrochloric acid to give 4-ketopentan-1-01 (XI) ; at higher temperatures pentane-1 4-diol (XII) is formed.The mechanism proposed is as follows 11 OH-CH %*CH 2*CH ,*CWMe*OH 4 5 . HO *CH ,*CH **CH ,*COMe (XII. ) (XI. 1 A further study of this reaction 34 has shown that with a nickel catalyst and dioxan as solvent (X) and water in the presence of small amounts of acetic acid are converted in 61% yield into (XII) in 8 hours at 150°/80 atm. In the absence of dioxan the yield of (XII) is lower and more Z-methyl- tetrahydrofuran is formed. If however thereaction is stopped after 14 hours 30% of (XI) is found together with similar amounts of unchanged (X) and 2-methyltetrahydrofuran and only about 4% of (XII). The hydrogenolysis of furfuraldehyde or furfuryl alcohol in the presence of water to give mixtures of diols and triols has also been described.35 Furfuryl alcohol water and glacial acetic acid with a Raney nickel cata- lyst at 160°/80 atm.gave a 40% yield of pentane-1 2- and -1 5-diols and a 45% yield of pentane-1 2 5-triol. Liquid-phase Oxidation of Furans 1. Autoxidation.-G. 0. Schenck was the first to describe systematic work on the course of this reaction and on the products obtained.36 Complete experimental details have yet to be published but he found that peroxides were formed from air and furan compounds and he distinguished between two types of reaction. One called the “ unsensitised ” reaction was best effected by irradiating the compound in presence of oxygen and an inorganic salt such as calcium chloride ferrous nickel or manganese sulphate ; and the other the “sensitised” reaction was carried out by irradiating a solution of eosin in the furan.The products formed are shown in the annexed Scheme. 33 Russian Patent 48,104 (Chem. Abs. 1937 31 8549) ; K. S. Topchiev Compt. 34 L. E. Schniepp H. H. Geller and R. W. Van Korff J . Amer. Chern. Soc. 1947 a5 U.S.P. 2,097,493. 36 (a) F.I.A.T. Review of German Science (1939-1946) Preparative Organic Chemistry Pt. 11 1948 p. 188 ; (b) G. 0. Schenck Ber. 1944 77 662 ; (c) idem ibid. 1947 80 289; Natwrwiss. 1943 31 387. rend. Acad. Sci. U.R.S.S. 1938 19 497 69 672. JONES AND TAYLOR FURAN AND PYRAN CHEMISTRY 201 Un.sensitised oxidation Sensitised oxidation (1) -c-c- -c=c- (4) -c-c- -c=c- /I /I -+ I I /I /I -+ I I -c c- -GO oc- -c c- -c c- lPy)/l ‘O/ ‘0’ 0-0 (or a poly- meric form) (2) 2 moles of furan + 0 + a compound (2 moles of furan + 0,) (3) -4-c- -c-c- It /I -+ It I>O \O/ ‘O/ -c c- -c c- These peroxides are of course reactive and can be hydrogenated to The reactions of 2 5-dimethylfuran and its oxidation products are 3% give saturated dicarbonyl compounds.C E C H Similarly of course 2-methylfuran gives 4-ketopent-2-en-1-a1 and thence lzvulaldehyde. Another important reaction of the peroxide formed in the sensitised reaction is the rearrangement in alcoholic solution giving a +-ester (XIII) of an unsaturated keto-acid ; the peroxide from %methylfuran reacts in this way (XIII. ) 0-0 ‘ JOH- JOH- HO ,C*CH:CH*COMe HO IIC*[CH2]2*COMe The autoxidation of furfuraldehyde in an alcohol containing eosin is particularly interesting.36u The reaction products are formic acid (formed in up to 80% yield) and the $-ester of /?-formylacrylie acid (XIV) of which the yield depends on the alcohol used.Schenck considers the reaction to take place by an oxidation followed by rearrangement of the aldehyde 202 QUARTERLY REVIEWS group to a formate group which is then decomposed by the alcohol present. The overall reaction is (XIV.) The yield of ester is about 75% when butanol or ethanol is used. When furfuraldehyde is shaken with oxygen at room temperature 8-formyl- acrylic acid can be detect.ed in the product ; its polymerisation is thought to lead to a mixture of resinous acids which are responsible for the dark colour developed in furfuraldehyde on storage in air and light .37 The formation of @-formylacrylic acid from furfuraldehyde had already been observed when sodium chlorate was used as the oxidising agent in presence of vanadium pentoxide ; the yield was 55% and formic acid was also formed.3s If the oxidation was done in presence of osmium tetroxide tartaric acid was obtained in 48% yield.Some furan compounds which give reactions similar to (1) and (4) of the Scheme on p. 201 are given in the Table.36Q The reactions of most of the compounds were not examined in detail but the Table may be a useful guide to the sort of products that might be expected. A utoxidation of furan compounds -- -=__. -co co- -+ -h,i- 0-0 2-Methylfuran 2 5-Dimethylfuran 2 -Methyl - 5 -e thylfuran 2 5-Diphenylfuran 2 5-Dibromofuran Methyl furoate Furfuryl alcohol Furylpropylcarbinol Furan 2-Methylfuran 2 5-Dimethylfuran Furoic acid Furfuryl alcohol Furfuryl acetate 3 4-Dichlorofuran Furylacrylic acid 2.Miscellaneous Oxidations.-Oxidation of furans in the liquid phase gives useful results with furfuraldehyde as in the experiments described above or with particular furans which may react favourably under these conditions. Thus 2 5-diphenylfuran gives an 80% yield of 1 2-dibenzoyl- ethylene on oxidation with concentrated nitric and 3 4-dichlorofuran behaves similarly giving dichloromaleic acid.40 There have been several studies of the action of hydrogen peroxide or peracids on furans perhaps the earliest being that of C. F. Cross E. J. Bevan and T. Heiberg.41 Later workers 42 found that although addition 87 A. P. Dunlop P. R. Stout and S. Swadesh Id. Eng. Chern. 1946 38 705; A. P. Dunlop ibid. 1948 40 204.N. A. Milas J . Amer. Chern. SOC. 1927 49 2005. 39 R. E. Lutz and F. N. Wilder ibid. 1934 56 978. *O A. F. Shepard N. R. Winslow and J. R. Johnson &id. 1930 52 2083. *l C. F. Cross E. J. Bevan and T. Heiberg J. 1899 75 747. 42 J. Boeseken C. 0. Vermij H. Bunge and C. Van Meeuwen Rec. Trav. chim. 1931 50 1023. JONES AND TAYLOR FURAN AND PYRAN CHEMISTRY 203 did occur with furan 2-methylfuran and furfuryl alcohol by far the greater part of the product was resinified in the acid conditions used ; furfuraldehyde gave an estimated 40% yield of #?-formylacrylic acid. Recent work has now shown that fmans can be oxidised to 2 5-dialkoxy- or 2 5-diacyloxy-2 5-dihydrofurans in good yields. The products are in fact acetals or ketals of unsaturated 1 4-dicarbonyl compounds and can in turn be used for further syntheses.The general method is to oxidise the compound with bromine or chlorine in a hydroxylic solvent such as an aliphatic alcohol or acid ; better results may sometimes be obtained by adding a base to the reactants so that there is no chance of resinification by acid formed in the reaction. The general reaction can be represented + 2HCl In this way furan 2-methylfuran and furfuryl acetate have given the corresponding dimethoxydihydrofurans in 50-60% yields ; 43 with bromine in acetic acid furan gives a 70% yield of 2 5-diacetoxy-2 5-dihydrofuran.** Furfurylidene diacetate gives the same type of compound (XV) 43 (XV.) The dimethoxydihydro-compounds can be hydrogenated to give the acetals of the corresponding saturated 1 4-diketones 45-a very convenient route because of the ready availability of many furan derivatives.A remarkable reaction leading to similar products has been described by L. Vargha J. Ramonczai and P. Bite.46 They found that on shaking the oxime O-toluene-p-sulphonate of 2-acetylfuran in alcohols the furan ring opened giving an unsaturated diketo-acetal (XVI) in 80% yield. This -+ Me*CO*CO*CH:CH*CH(OR) (XVI.) I J c . M e ‘0 1 N*O *SO ,*C BH 4Me Me*CO*CO*CH,*CH,*CH(OR)~ ’ (XVII.) product is what would have been expected from the oxidation of 2-acetyl- furan by chlorine in alcohol. On treatment with dilute acid even in the 43 N. Clauson-Kaas Kgl. Dawke Vicknskab. SehJcctb Mat.;fys. Medd. 1947 24 18 (Chem. Abs. 1948 42 1930) ; N. Clauson-Kaas and F. Limborg Acta Chim. Scand. 1947 1 619 (Chem. Abs. 1948 42 5902) ; N.Clauson-Kam and J. Fakstorp ibid. p. 415 (Chem. Abs. 1948 42 5901); B.P. 595,041. 44N. Clauson-Kaas Acta Chim. Scand. 1947 1 379 (Chem. Abs. 1948 42 5447). 46 B.P. 610,876. 46 L. Vargha J. Ramonczai and P. Bite J . Amer. Chem. SOC. 1948 70 371. 204 QUARTERLY REVIEWS cold (XVI) loses acetic acid but it can be hydrogenated to the dihydro- compound (XVII) which in hot dilute sulphuric acid undergoes intra- molecular dehydration and cyclisation to give catechol. Di- and Tetra-hydro-furans and -pyrans Dihydro-furans and -pyrans 1. Dihydrofurans Synthesis.-@) 2 3-Dihydrofurans. 2 3-Dihydro- furans like the corresponding dihydropyrans are anhydrides of keto- alcohols from which they are formed on heating; T. R. Marshall and W. H. Perkin first identified 5-methyl-2 3-dihydrofuran which they synthe- sised by heating 4-ketopentan- 1-01 (prepared from ethyl acetoacetate and ethylene br~mide).~' Unsubstituted 2 3-dihydrofuran was first made in 24% yield by passing tetrahydrofurfuryl alcohol over a copper-nickel alloy ; 48 a better method is to treat 3-chloro-2-alkoxytetrahydrofurans (made from tetrahydrofuran) with sodium the yields being 50-?5y0 depending on the alkoxy-gro~p.~~ Early preparations of 2 5-dihydrofuran were by heating erythritol with formic and removing the elements of hydrogen bromide from 3-bromotetrahydrof~ran.5~ The latter method has been used to make 2 2-dialkyldihydrof~rans,~~ but according to a later reference 53 mixtures of 2 5- and 2 3-dihydrofurans are formed by this reaction (b) 2 5-Dihydrojurans.A more recent method links the preparation of such compounds with modern developments in acetylene chemistry ; but-2-ene-1 4-diol or methyl- substituted butenediols made by hydrogenation of the product of reaction of acetylene with formaldehyde or a ketone,54 give 2 5-dihydrofurans when passed in the vapour phase over solid dehydrating catalysts55 or when heated with sulphuric or phosphoric acid.56 Only the cis-form of but-2-ene- 1 4-dio1 (which is the isomer predominantly formed on hydrogenating but-2-yne-1 4-diol) gives 2 5-dihydrofuran while the trans-diol gives only crotonaldehyde.57 2. Dihydropyrans Synthesis.-(a) 2 3-Dihydro-4-pyrans (1 5-Epoxy- pent-l-enes). The preparation of dihydropyran by the catalytic dehydration of tetrahydrofurfuryl alcohol has already been mentioned. Tetrahydro- furylmethylcarbinol reacts similarly in the vapour phase giving 2-methyl- 5 6-dihydro-4-pyran (XVIII) but diphenyltetrahydrofurylcarbinol is 47 T.R. Marshall and W. H. Perkin J. 1891 59 880. 48 C. L. Wilson J. 1945 62 60 Henninger Ann. Chim. 1886 [vi] 7 216 217. 61 E. D. Amstutz J. Org. Chem. 1944 9 310. 63 J. Colonge and P. Gamier Bull. SOC. chim. 1948 432. lisH. Normant Compt. rend. 1948 227 283. 154 B.I.O.S. Final Report No. 367. 66 B.P. 510,949. 67 A. Valette Compt. rend. 1946 223 907 ; cf. J. R. Johnson and 0. H. Johnson 48 H. Normant Compt. rend. 1949 228 102. 68 B.P. 510,615. J . Arner. Chem. SOC. 1940 62 2615. JONES AND TAYLOR BURAN AND PYRAN (IHEMISTRY 205 dehydrated (although in the liquid phase) without ring enlargement ,58 to give 2-benzhydrylidenetetrahydrofuran (XIX) (XVIII.) (XIX.) This ring enlargement during dehydrafion is similar to that which occurs when cyclobutylcarbinol is dehydrated to give cycbpentene.59 The method would not however be of much use for the preparation of 2 S-dihydro- 4-pyrans with substituents in other than the 6-position since the required tetrahydrofurfuryl alcohols substituted in the nucleus are difficult to prepare. Substituted dihydropyrans can also be obtained by dehydration of the appropriate keto-alcohol ; 5-ketohexan-1-01 gives 2-methyl-5 6-dihydro- 4-pyran when heated go and 6-keto-2-methylheptan-2-01 gives 2 2 6-tri- methyl-:! 3-dihydro-4-pyran (XX) M e 2 0 M e -!- H20 OH-CMe ,*CH,*CH ,*CH ,*COMe -+ (XX.1 Trimethylene dibromide and its substitution products react with ethyl acetoacetate to give esters of 5 6-dihydro-4-pyran-3-carboxylic acids and on hydrolysis the parent acids are obtained.The reactions using 1 3-dibromobutane Me*CO*CH ,*CO ,Et + and ethyl acetoacetate are 62 Br*CHMe*CH,*CH,Br + The f e e acid is easily decarboxylated giving the alkyl-substituted dihydro- Pyran. The dimerisation of unsaturated carbonyl compounds t o give dihydro- pyrans has been extensively studied by XI. Alder et aLg3 The compound is heated in presence of about 1% of quinol and the reaction which takes place usually in good yield is v C- - jc C- % > + I - + -co -CO’ a. R. Paul BUG. SOC. chirn. 1938 5 919 ; A. L. Dounce R. H. Wardlow and Connor J . Arner. Chern. SOC. 1935 57 2556. m M. Dojarenko Ber. 1927 60 1536. 6o W. H. Perkin {bid. 1886 19 2557 ; A. Lipp Annaien 1896 289 187.61 Verley Chern. Weekbl. 1897 [iii] 17 185. saR. G. Fargher and W. H. Perkin J. 1914 105 1353. ssK. Alder H. Oppermans and E. Ruder Ber. 1941 74 23 906 920 926. P 206 QUARTERLY REVIEWS Thus acraldehyde gives (XXI) ; but-l-en-3-one gives (XXII) ; 2-methyl- but - 1 -en-3-one gives (XXIII) ; and crotonaldehyde gives (XXIV). This dimerisation of crotonaldehyde is interesting as an isomer (XXVI) is formed if the polymerisation is acid-catalysed. (XXI.) (XXII.) (XXIII. ) (XXIV. ) An example of another type of condensation is the preparation of methyl 6 6-dimethyl-5 6-dihydro-4-pyrone-2-carboxylate (XXV) by condensing methyl oxalate with mesityl oxide in presence of sodium and then cyclising the product by sulphuric acid 64 0 0 0 II II II + I --+ II 11 / H O /c\ * Me f i C 0 2 M e HC CH /c\ HC CH C02Me Me 2C C0,Me Me& C*CO,Me 2'0 (XXV.) (b) 2 3-Dihydro-6-pyrans (1 5-Epoxypelzt-Z-elzea).The unsubstituted dihydropyran has been made recently by R. Paul et aZ.65 in 75% yield by dehydrobromination of 4-bromotetrahydropyran with potassium hydroxide in ethylene glycol. 4-Bromotetrahydropyran was presumably made from 4-hydroxytetrahydropyran but the method used is not given. Another method for preparing 2-alkyl-substituted compounds described in greater detail later is the removal of the elements of hydrogen chloride from the trans-form of 3-chloro-2-alkyltetrahydropyrans.66 Apart from these there is only one general method described in the literature namely condensation of a carbonyl compound with an unsaturated alcohol in the presence of toluene-psulphonic acid an entraining agent such as benzene being used to remove the water formed 67 Me R R ' o M e 4- H20 RR'CO + CH,:CMe*CH2*CHMe*OH -+ When aldehydes are used the corresponding 4-hydroxytetrahydropyrans are also formed ; from acraldehyde or chloroacetaldehyde \CHO 2 3-dihydro-6-pyrans substituted by vinyl or chloromethyl M e O M e respectively can be made.The dimerisation of crotonaldehyde when heated to give (XXIV) was mentioned earlier ; if acid is used as the (xxVI.) catalyst (XXVI) is formed in about 25% yield.68 04 L. Claisen Annalen 1896 291 132. 6 6 R. Paul and S. Tchelitcheff Cmpt. rend. 1947 224 1722. 67 U.S.P. 2,422,648 ; 2,452,977. 0. Riob6 Ann. Chim. 1949 [xii] 4 593. M. De%pine Compt. rend. 1910 150 394. JONES AND TAYLOR FURAN AND PYRAN CHEMISTRY 207 3.Dihydro-fur4Lns and -pyrans Reac!tions.-The dihydro-furans and pyrans fall into two groups so far as their chemical reactions are con- cerned. 2 3-Dihydrofurans and 2 3-dihydro-4-pyrans which have the double bond in the a-position to the oxygen atom react very similarly to vinyl ethers while 2 5-dihydrofurans and 2 3-dihydro-6-pyrans behave as olefins in which the activity of the double bond is not affected by the oxygen atom. (a) 2 3-Dihydrofurun and 2 3-Dihydro-4-pyran (11). Mild hydrogena- tion in non-acidic solvents gives the corresponding tetrahydro-compound in excellent yield ; thus 2 3-dihydrofuran gives tetrahydrofuran,@ 2 3-di- hydro-4-pyran (11) gives tetrahydropyran,6Q and the substituted compounds react similarly. Under more vigorous conditions at 200" with cobalt or nickel catalysts cyclopentanone [formed by rearrangement of (11)] and gaseous products are formed as well as tetrahydropyran.70 If (11) is hydrogenated in aqueous emulsion over a copper catalyst at about 200° pentane-1 5-diol is obtained in nearly theoretical yield ; 7 1 2 3-dihydrofuran would prob- ably react in the same way giving butane-1 4-dio1 but has not been tried.The dihydro-compounds react vigorously with water and other hydroxylic compounds in presence of a trace of acid giving mainly 2-hydroxy- or 2-alkoxy-tetrahydro-compounds or esters. The reaction has been particu- larly well studied with (11). In dilute hydrochloric acid hydration is fairly rapid at low temperature and the products are 2-hydroxytetrahydropyran (XXVII) and di( tetrahydro-2-pyranyl) ether (XXVIII).72 The hydroxy- compound is the semi-acetal of 5-hydroxypentan-l-al and the equilibrium mixture consists mostly of the cyclic form.2 3-Dihydrofuran reacts (XXVII.) (XXVIII.) similarly to give 2-hydroxytetrahydrofuran the semi-acetal of 4-hydroxy- b~tan-l-al.*~ 49 When alcohols phenols or acids are the corre- sponding ethers or esters are formed; these are unstable in presence of aqueous acids giving the alcohol and 2-hydroxytetrahydropyran. These reactions are common to all 2 3-dihydro-4-pyrans and a more complex example where the final product is a diketone is the reaction of (XXIII) with acidified ethanol. The alcohol adds to the double bond in the usual way but the ring opens giving as a 6nal product 3-ethoxy-3 &dimethyl- 6g B.P. 565,175 ; Org. Synth. 23 90. 70 C.L. Wilson J. Amer. Chem. Soo. 1948 70 1311. 71 B.P. 621,735. 72 L. E. Schniepp and H. H. Geller J. Amer. Chem. SOC. 1946 68 1646 ; R. Paul 7s G. F. Woo& and D. N. Kramer J. Amer. Chem. SOC. 1947 69 2246. But%. Soo. chim. 1934 [v] 1 971. P" QUARTERLY REVIEWS 208 octane-2 7-dione (XXIII. ) (XXIX) 74 MeCH CH I Me EtOH Y'&)Me OEt (XXIX.) Because of its ease of reaction with hydroxyl groups (11) has been used for their protection in reactions in alkaline media.75 The bis-tetrahydro-2- pyranyl ether of catechol after reaction with butyl-lithium followed by carbonation and hydrolysis gave a 48% yield of 2 3-dihydroxybenzoic acid OH OP OH O O H + 2() + O O P hydrol. (XXX.) Hydrogenation of 2-hydroxytetrahydropyran over a nickel catalyst gives pentane-1 5-diol; if the hydrogenation is effected in presence of ammonia or an amine aminopentanols are formed in good yield 76 + NHzR + Ha + R*NH*[CHJ,*OH + H,O 00.2 3-Dihydroxy-tetrahydropyran (XXX) and -tetrahydrofuran can be made from the corresponding dihydropyran or dihydrofuran by reaction with osmic acid arid hydrogen peroxide in tert.-butanol77 or with lead tetra-acetate 49 respectively. Both diols behave as a-fiydroxy-aldehydes and give 2 4-dinitrophenyIosazones. 2 3-Dihydrofuran decomposes when heated above 375" to give mainly formylcycbpropane (XXXI) and crot~naldehyde.~~ Although formylcycb- propane was not isolated as such the yield a t 460" was estimated to be about 40% calculated on the dihydrofuran consumed but at higher tem- peratures increasing proportions of propylene and carbon monoxide were formed.Formylcyclopropane gives it small amount of dihydrofuran on pyrolysis at 500" so that the decomposition is reversible. (11) is decom- posed smoothly at about 500" giving very good yields of acraldehyde 7 4 J. Colonge and J. Dreux Compt. rend. 1949 228 582. 76 W. E. Parham and E. L. Anderson J. Amer. Chem. SOC. 1948 70 4187. 76 G. F. Woods and H. Sanders ibid. 1946 68 2111 ; I. Scriabine Bull. SOC. chim. 77 C. D. Hurd and C. D. Kdso J. Amer. Chem. SOC. 1948 70 1484. 78 C. L. Wilson ibid. 1947 69 3002. 1947 14 454; B.P. 676,087. JONES AND TAYLOR FURAN AND PYRAN CHEMISTRY 209 (XXXII) and ethylene ; 79 acraldehyde can also be produced from tetra- hydrofurfuryl alcohol in one stage by use of an aluminium silicate catalyst at 450°.80 Other work has shown that with this and similar catalysts much polymer may be formed at lower temperatures from tetrahydro- furfuryl alcohol.3-Chloro-5 6-dihydro-Q-pyran behaves similarly to (I) -+ CH,*CH:CH*CHO + CH,*C%CH2 + CO (XXXI. ) CH* \ CH I CH / (XXXII. ) CH2 0 * AH2 4- 0 and gives 2-chloroa~raldehyde~~ although the yield is less than that of acraldehyde from (I). These pyrolyses of dihydropyrans are similar to the decomposition of Diels-Alder adducts which (e.g. the decomposition of cyctohexene to buta-1 3-diene and eth~lene)~3 regenerate the original reactants when heated ; however no synthesis of 2 3-dihydro-4-pyrans by reaction between an ethylene and a substituted acraldehyde has been reported although cyclohexene has been made by a similar reaction from ethylene and buta-1 3-diene.84 The reactions of 2 3-dihydro-4-pyran (11) with halogens and halogen compounds have been well studied and the products have proved useful intermediates for further syntheses.Similar reactions of 2 3-dihydrofuran have been little investigated because it is not so available but it would probably react in a very similar way to (11). Dihydro-4-pyran readily adds chlorine bromine hydrogen chloride or hydrogen bromide,s5* st? giving 2 3-dihalogeno- or 2-halogeno-tetrahydro- pyrans. The halogen atom in the 2-position is removed as hydrogen halide on distillation of the product at atmospheric pressure giving S-chloro- or 3-bromo-5 6-dihydro-4-pyran (XXXIII). Addition of chlorine to these unsaturated compounds gives 2 3 3-trihalogenotetrahydro- pyrans.87 The halogen atom in the 3-position in all these compounds is relatively inert but that in position 2 resembles that in a-chloro-ethers and reacts readily with Grignard compounds giving 2-alkylfetrahydro- 79 J.G. M. Bremner D. G. Jones and 8. Beaumont J. 1946 1018. C. L. Wilson J . Amer. Chem. SOC. 1947 69 3004. B.P. 608,538. 82 B.P. 578,071. ssF. 0. Rice and M. T. Murphy J . Amer. Ohm. Soc. 1944 66 765. 84 L. M. Joshel and L. W. Butz ibid. 1941 63 3350. 86 B.P. 571,265. 86 R. Paul BuEI. Soc. chim. 1934 [v] 1 1403. U.S.A. Office of the Publ. Board Dept. of Commerce Report No. P.B.-803. 210 QUARTERLY REVIEWS pyrans 86 or 3-chloro- or 3-bromo-2-alkyltetrahydropyrans ; 88 with cuprous cyanide the corresponding nitriles are obtained.89 Q- Q"' OH Q"'(Br) (XXXrV.) (XXXIII.) The alkyl-chloro-compounds have ,been studied by 0.Riobh G6 who found that they could be easily separated into cis- and trans-forms by distillation. The configurations were assigned on consideration of the physical properties of the geometrical isomers and of the nature of the products formed on loss of the elements of hydrogen chloride. With alcohols or sodium salts of aliphatic acids the dihalogeno-compounds give 2-alkoxy- or 2-acyloxy-compounds and with water give substituted bis- tetrahydropyranyl ethers.88 Reaction of 2 3-dihydro-4-pyran or its deriva- tives with halogens in a hydroxylic solvent also gives the corresponding halogenated 2-hydroxy- or 2-alkoxy-tetrahydropyran in good yield ; thus 3-chloro-2-hydroxytetrahydropyran (XXXIV) can be made by chlorinating an emulsion of (11) in water,9o preferably under alkaline conditions so as to avoid formation of (XXVII) .Addition of hydrogen chloride to dihydrofuran gives 2-chlorotetrahydro- furan.49 2 3-Dichlorotetrahydrofuran which is formed by chlorinating 2 3-dihydrofuranY is most conveniently made by chlorinating tetrahydro- furan (see below) ; the chlorine atom in the 2-position is as reactive as that in 2-chlorotetrahydropyan and a corresponding series of hydroxy- alkoxy- and alkyl-chlorotetrahydrofurans can be made.49 Like the analo- gous t e trah y drop yrans the 3 - c hloro -2 - alkyl t e trah ydrofurans also exist in cis- and trans-forms which can be separated by di~tillation.~1 The halogen atoms remaining in these 2-alkyl- and 2-alkoxy-tetrahydro- compounds can be removed by various treatments giving interesting pro- ducts. Treatment of 3-chloro-2-hydroxytetrahydropyran with hydroxyl- amine gives tetrahydrofurfuraldoxime,88 formed by the following reactions Q C H O + OXirnO o:H + HO*[CH8'J8*CHC1*CH0 + If 3-chloro-2-alkyltetrahydropyrans are distilled from a solution of potassium hydroxide in diethylene glycol a mixture of two isomeric SsR.Paul Compt. rend. 1944 218 122; B.P. 606,107. 8* F.D. 3781/45 p. 000731. Obtainable from Techn. Information Document *lL. Crombie and S. H. Harper J. 1960 1714. Unit Board of Trade. @O B.P. 570,160 ; 698,080. JONES AND TAYLOR FURAN AND PYRAN CHEMISTRY 21 1 alkyldihydropyrans is obtained 66 (XXXV. ) The cis-isomers react very easily and give 80-90% yields of a mixture of the isomeric alkyl-dihydro-2- and -4-pyrans ; the proportions vary with the reaction conditions and the alkyl group.The trams-isomers react much more slowly and the products obtained in 90% yield consist almost entirely of the 2-alkyl-5 6-dihydro-2-pyrans (XXXV). The results of these experi- ments on the assumption that trans-elimination of the elements of hydrogen chloride is more rapid than cis- provided the chemical evidence for con- figuration mentioned earlier. The 3-chloro-2-alkyltetrahydrofurans react similarly. The ring fission of tetrahydrofurfuryl halides with sodium typical of the behaviour of #hhloro-ethers has been extended to the 3-chloro-2-alkyl- tetrahydro-pyrans and -furans to give @- or y-ethylenic alcohols respectively gym + 2Na -+ NaO*CH,*CH,*CK:CHPP + NaCl (XXXVI.) + 2Na + ~aO*CHa*CH,*CH,*CHCHPr~ + NaCI (XXXVII. ) in very good yields. Thus 3-chloro-2-propyltetrahydrofuran gives hept- 3-en-1-01 (cf.XXXVI),92-94 and 3-chloro-2-propyltetrahydropyran gives oct-4-en-1-01 (XXXVII).66 Some recent work has elucidated the stereo- chemical relationships in these reacti0ns.9~ The higher-boiling (&-)isomer of 3-chloro-2-methyltetrahydropyran reacted the more vigorously with sodium but both isomers gave the same alcohol shown by infra-red analysis to be trans-n-hex-4-en-1-01 this being confirmed by synthesis of the two stereoisomers by unambiguous routes. This ring fission leading to a tram- alk-4-en-1-01 is thought to be general for 3-chloro-2-alkyltetrahydropyrans. The trans-form of 3-chloro-2-methyltetrahydrofuran gave solely trans-pent- 3-en-1-01 but the cis-isomer gave a mixture of cis- and trans-alcohols detected by comparison of the infra-red absorption spectra and by chemical evidence.Similarly the two 3-chloro-2-ethyltetrahydrofurans gave the tram-form and the cis-tram-mixture of n-hex-3-en-1-01 These unsaturated alcohols can be converted into dienes by pyrolysis of their acetates at 550O.95 3-Halogenotetrahydrofurans react very slowly with magnesium giving compounds which re-arrange to give unsaturated primary alcohols but they react readily with lithium alkyls to give about 20% yields of the @a H. Normant Cmpt. rend. 1948 226 733. Yu. K. Yur’ev M. G. Voronkov I. P. Cragerov and C. Ya. Kondrat’eva J. Cen. O h . Russia 1948 18 1804. e4 S. H. Harper and I;. Crombie N&m 1949 164 1053. 96 0. Riob6 Compt. r e d . 1948 226 1626. 212 QUARTERLY REVIEWS corresponding 3-alkyltetrahydrof~rans,~~ 93 *6 this being probably the most convenient method of making them.The reaction of 3-chloro-2-alkoxytetrahydrofurans with sodium has already been mentioned as a method of preparation of 2 3-dihydrofuran ; when these compounds are heated with potassium hydroxide in glycerol both the chlorine atom and the ethoxy-group are lost and furan is formed in 80% yield.49 When heated with sodamide the chloroethoxytetrahydro- furan gives a 5070 yield of a 2-ethoxy-2 5-dihydrofuran. This is quite stable except in presence of acid where it yields furan and ethanol quan- titatively ; if 2 4-dinitrophenylhydrazine is present the derivative of 4-hydroxybut-2-en-1-a1 (XXXVIII) is formed.97 KOH h acid + derivative of - t==*) (XL. 1 3-Bromo-2-ethoxytetrahydropyran reacts similarly with sodamide 49 or with sodium in to give an ethoxydihydropyran probably (XXXIX).Acid hydrolysis of this compound in presence of 2 4-dinitro- phenylhydrazine gives a derivative of 5-hydroxypent-2-en- 1 -a1 (XL) but the parent compound could not be isolated. If (XXXIX) is steam-distilled from phosphoric acid a 55% yield of penta-2 4-diene-1-a1 is formed Q O E t -+ Q O E + CH,:CH*CH:CEI*CHO The pentadienal is also formed by treating 2 3-dichlorotetrahydropyran with potassium hydroxide. 2 3-Dihydro-4-pyran (11) reacts with carbonyl chloride ; 99 addition takes place at room temperature in the absence of added catalyst giving 5 6-dihydro-4-pyran-3-carboxyl chloride (XLI) in about 50% yield 0 __+ COCL @Cl -+ @OCl (XLI.) 0 0 c1 (11.) The acid was identified by the formation of much formic acid but no oxalic acid on ozonolysis and by its decarboxylation when heated to-give (11) and carbon dioxide ; it can easily be hydrogenated to the tetrahydro-compound.lo* 98Yu. K. Yur'ev and I. P. Gragerov J . Gen. Chem. Rzwlsia 1948 18 1811. 97F. Quennehen and H. Normant Compt. red. 1949 228 1301. 98 G. F. Woods and H. Sanders J . Amer. Chem. Xoc. 1946 68 2483. 99 B.P. 670,974. lo0 B.P. 612,314. JONES AND TAYLOR FmRAN AND PYRAN CHEMISTRY 213 The chemistry of this compound is at present undistinguished and is almost that of the unsaturated ethylenic bond. Oxidation with air in the vapour phase over a molybdenum-vanadium- titanium oxide catalyst gives maleic anhydride.lol The preparation of 3-chloro- and 3-hydroxy-tetrahydrofuran has been described lo2 and a mix- ture of 3 4-dichloro- 3-chloro-4-acetoxy- and 3-chloro-4-hydroxy-tetra- hydrofuran103 is obtained on reaction with chlorine in acetic acid.(c) 2 3-Dihy&ro-6-pyrans The properties of the two isomeric dihydro- pyrans have been compared by R. 2 3-Dihydro-6-pyran is hydro- genated easily to tetrahydropyran ; it reacts with 0.5~-sulphuric acid at 150" giving 3-hydroxytetrahydropyran and with perbenzoic acid to form the epoxide which gives 3 4-dihydroxytetrahydropyran on hydrolysis ; with chlorine 3 4-dichlorotetrahydropyran is formed. These products are stable and have the normal properties expected of say the corresponding cyclohexane derivatives ; the activity of substituents in the 2-position which is characteristic of derivatives of the dihydro-4-pyransY is of course absent. The behaviour of the isomeric dihydropyrans on pyrolysis shows another difference.2 3-Dihydro-4-pyran gives acraldehyde and ethylene but 2 3-dihydro-6-pyran a t 600" gives butadiene and formaldehyde.65 This reaction seems to be general for 2 3-dihydro-6-pyransY as the dimer (XXVI) obtained from crotonaldehyde by acid behaves in a similar way at 400--600".1°4 The 2-vinylcrotonaldehyde (XLII) first formed is unstable and forms penta-1 3-diene by loss of carbon monoxide ( b ) 2 5-Dihydrofuran. 27 CH CCHO + Me*CHO + li -j. CH ,:CH*CH:CHMe CHMe + co M o Q M e (XXVI. ) (XLII.) Oxidation of (XXVI) gives the acid (XLIII).105 When this acid is heated with Raney nickel and hydrogen in alkaline solution the double bond moves to give the isomeric acid (XLIV),106 together with some of the acid formed by hydrogenation.The re-arranged product can also be obtained in 60% yield by heating the sodium salt of the acid with Raney nickel on a water-bath ; the acid is quite stable if heated with alkali in absence of nickel. The product obtained on decarboxylation was identical with one synthesised earlier by Perkin by a different route. (XXVI.) (XLIII. ) (XLIV.) lol U.S.P. 2,215,095. lo2 U.S.A. Office of the Publ. Board Dept. of Commerce Report No. P.B.-42,455. loS B.P. 616,762. lo* U.S.P. 2,387,366. lo6 U.S.P. 2,378,996. L v b ^ ~ Dei'ephe &nci A. fi%reau c%mpt. rend lY& W %t; 214 QUARTERLY REVIEWS Tetrahydro-furans and -pyrans 1. Synthesis.-(a) Tetrahydrofurans. The most convenient synthesis of tetrahydrofuran is by hydrogenation of furan which is now commercially available ; similarly 2-methylfuran which is made by the hydrogenation of furfuraldehyde (p.197) gives 2-methyltetrahydrofuran. 2-Alkyltetra- hydrofurans can be made from the corresponding 2-alkylfurans them- selves made by condensing furfuryl bromide with the appropriate Grignard reagent.10' Another route is by dehydration of furylalkylcarbinols,1°8 followed by hydrogenation lo9 Tetrahydrofuran was made in Germany during the war by a method similar to that already described for 2 5-dihydrofuran. But-2-yne-1 4- diol was hydrogenated to butane-1 4-dio1 and when this was heated with 0.3% phosphoric acid a t 280"/100 atm. tetrahydrofuran was formed in 93-94% yield ; 54 tetrahydrofuran is also formed under milder conditions when butane-1 4-diol is heated with phosphoric acid 110 at about 165" or by dehydration in the vapour phase over Morden bentonite.111 The intra- dehydration of 1 4-diols is a general method of making tetrahydrofurans ; other examples are the dehydration of (XLV) 112 by hot potassium hydrogen sulphate and of (XLVI) 113 by hot sulphuric acid to the corresponding t e trah ydrofuran derivatives.JJ CH ,:CH*CiC*CMe( OH)*CH ,*CH,*CH z*OH -+ CH ,:CH*CszC Me 0 (XLV.) OH*CH,*CHMe*CH,*CH,*OH -+ Me- (XLVI.) Tetrahydrofurans with alkyl substituents in the ring have been made by cyclising unsaturated alcohols. The simplest example is that of pent- 4-en-1-01 which gave a 90% yield of 2-methyltetrahydrofuran when it was warmed with concentrated sulphuric acid. 114 The necessary y-unsaturated alcohols were made by reduction of the ketones formed in the reaction between ethyl acetoacetate and the appropriate chlorides and by this method several alkylated tetrahydrofurans have been made.1l5# 116 Addi- tion of chlorine or bromine to the unsaturated alcohol and then cyclisation in presence of quinoline gives alkylated 3-bromotetrahydrofurans 1 1 5 9 117 from which as mentioned earlier 3-alkyltetrahydrofurans can be made.(b) Tetrahydropyrans. Tetrahydropyran itself is easily made in good 10' R. Paul Bull. Soe. chirn. 1935 [v] 2 2227. lo9 Idem ibid. 1938 5 1053. 111A. N. Bourns and R. V. V. Nichols Canadian J . Res. 1948 26 B 81. l121. N. Nazarov and I. V. Torgov J. @en. Chem. Russia 1948 18 1480. 11* Yu. I(. Yur'ev and I. P. Gragerov ibid. p. 1811. 114 R. Paul and H. Normant Compt. rend. 1943 216 689. 116 J. Colonge and A. Lazier Bull. Soe. chim.1949 15 17. 108 Idem ibid. p. 2220. l10 U.S.P. 2,251,292. 116 H. Normant Compt. red. 1948 226 1734. 117 D.R.-P. 696,726. JONES AND TAYLOR FURAN AND PYRAN CHEMISTRY 2 15 yield by hydrogenating the available 2 3-dihydro-4-pyran in either the liquid or the vapour phase,69 and 2-alkyltetrahydropyrans are made by reaction of 2-chlorotetrahydropyan with Grignard reagents ; 66 others are available by hydrogenation of dihydropyrans described earlier. Dichlorodiethyl ether is a convenient starting material for preparing 4-substituted tetrahydropyrans by reaction with such compounds as ethyl malonate.118 However the interest in these compounds has been more in the introduction of the tetrahydropyran ring into say drugs than in the chemistry of the compounds prepared. 2. Reactions.-Ring fission of tetrahydro-pyrans and -furans with various reagents has led to numerous useful aliphatic compounds difficult to obtain by other routes.Since the last Annual Report,2 tetrahydropyran has been shown to give an 85% yield of 5-chloropentan-1-01 acetate when warmed with acetyl chloride in the presence of zinc chloride as catalyst the benzoate being similarly obtained by use of benzoyl ch10ride.l~~ The fission of tetrahydrofuran with thionyl chloride under the influence of numerous catalysts has been investigated. The use of zinc chloride favours the formation of 1 4-dichlorobutane whereas sulphuric acid leads mainly to 4 4'-dichlorodibutyl ether.12* The superiority of sulphuric acid as a catalyst in the fission of tetrahydrofuran with phosphorus oxychloride (to give a 70% yield of 4 4'-dichlorodibutyl ether) has been estab1ished,l2l but the method gives poor yields when applied to tetrahydropyran and 2 -met hyltetrahydrofuran .Tetrahydrofuran when caused to react with oxoniurn salt-forming sub- stances (R+X-) gives compounds of the type 122 These with water alcohols or aqueous acids give where Y = OH OMe OAc or Cl. Many combinations of compounds giving suitable catalytic cations are quoted. The products formed by inter-polymerising tetrahydrofuran with pro- pylene or ethylene oxide in presence of a mixture of zinc chloride and thionyl chloride and replacing the chlorine in the polymer by methoxyl have found some application as synthetic lubricants. Tetrahydro-furan and -pyran are readily attacked by oxidising agents. Tetrahydrofuran hydroperoxide (XLVII) has been isolated from tetrahydro- furan which has been exposed to air ; 123 it gives y-butyrolactone as a major product on decomposition together with 2-hydroxytetrahydrofuran.The oxidation of tetrahydrofuran with air is best done under pressure using a cobalt catalyst the main product being y-butyrolactone ; 12* tetrahydro- 118 U.S.P. 2,242,575 ; G. H. Harnest and A. Burger J . Amer. Chem. SOC. 1943 65 370. 119 M. E. Synerholm ibid. 1947 69 2581. 120 Ref. 102 Report No. P.B.-631. 121 K. Alexander and L. E. Schniepp J . Amer. c7hern. Soc. 1948 70 1839. la2F.D. 3781/45 p. 001047; F.I.A.T. Final Report No. 293. 128 A. Robertson Nature 1948 162 153. RO*[CH2],*(O~[CH2]4)n*O*[CH2}r~Y 124 B.P. 608,539. 216 QUARTERLY REVIEWS pyran treated similarly gives 8-valerolactone.y-Butyrolactone is also produced when tetrahydrofuran or simple 2-alkyltetrahydrofurans are (XLVII.) treated with nitrogen tetroxide and oxygen ; 125 as the conditions become more severe increasing amounts of succinic acid are formed.126 Very good yields of succinic acid can be obtained from tetrahydrofuran by oxidation with nitric acid,12' and tetrahydropyran gives glutaric acidjlZ8 but there is no mention of alkyl-succinic or -glutaric acids being made from appropriate alkyl tetrahydro-furans or -pyrans by this method. Tetrahydrofuran has also been oxidised electrolytically to succinic acid.129 Tetrahydrofuran is substituted by chlorine even at O" and the pro- ducts consist mostly of 2 3-dichlorotetrahydrofuran (XLVIII) with some 3- chloro-2- (4- chlorobutoxy ) tetr ah ydrofuran (XLIX) which is formed by fission of the tetrahydrofuran ring to chlorobutanol and reaction of this with the active 2-chlorine substituent in dichlorotetrahydrofuran.130 This (XLVIII.) (XLIX.) is certainly the most convenient method of making dichlorotetrahydrofuran. There is no published work on the chlorination of tetrahydropyran but of course 2 3-dichlorotetrahydropyran is readily available from the chlorina- tion of dihydropyraa. The inter-dehydration of tetrahydrofurans in the vapour phase with amines hydrogen sulphide or hydrogen selenide to give pyrrolidines thio- phans and selenophans has been recently reviewed by L. N. Owen; tetrahydropyran reacts in a very similar way but the yields are not usually as good as when tetrahydrofurans are used. Tetrahydrofuran reacts with nickel carbonyl under pressure in presence of nickel chloride or iodide giving carboxylic acids.A small continuous unit was operated in Germany during the war years in which adipic acid was made by the reaction of tetrahydrofuran nickel carbonyl and carbon monoxide in presence of nickel iodide at 270"/200 atm. Under certain conditions an 80% yield of adipic acid was claimed ; some 6-valerolactone and valerio acid were also formed.131 Under more drastic conditions at 300"/800 atm. over nickel chloride as catalyst a mixture of acids is formed among them being butyric and butane-2-carboxylic acid.132 lZE H. Schmid and A. Maschka Molzatsh. 1949 80 235. 128 D. G. Jones unpublished. 129 Canadian P. 450,353. Ref. 120 Report 131 B.I.O.S. Final Report No. 351. 132 U.S.P. 2,432,474. lZ6 B.P. 610,166. le7 Belg. P. 444,240. lSo H. Normant Compt. rend. 1948 226 185 ; D.R.-P. 703,956. NO. P.B.-675.

ISSN:0009-2681    

DOI:10.1039/QR9500400195

出版商:RSC    

年代:1950

数据来源: RSC