摘要:

QUARTERLY REVIEWS GEOMETRICAL ISOMERISM ABOUT CARBON-CARBON DOUBLE BONDS By L. CROMBIE PILD. A.R.I.C. (LECTURER IMPERIAL COLLEGE OF SCIENCE AND TECHNOLOGY S. KENSINGTON LONDON S.W.7) THIS review of geometrical isomerism which is written primarily from the point of view of an organic chemist deals only with the >C=C( system. It is divided into three sections these being examinations of (i) some general properties of the )C=C( system (ii) methods for the determination of geometrical configuration and (iii) methods for the preparation of geo- metrical isomers. Attention is drawn to the existing reviews ; lV3 those of more specialised aspects will be mentioned later. Some General Properties of )C=C( Systems As a corollary to the theory that the valencies of carbon are disposed tetrahedrally van't Hoff suggested a model (I) for eth~lene.~ The imaginary tetrahedra encompassing the directed valencies of the two carbon atoms are placed edge to edge and inspection shows that the six atoms involved are then coplanar that LHCH is 109" 28' that LCCH is 125" 16' and that the )C=C( bond distance is 0.577 x the C-C distance i.e.0.89 A. H\c .a*-.... C/H H/ 'v \H (1.1 Physical measurement has since shown that the )C=C( distance in ethylene and substituted ethylenes is actually 1-33 (a little longer when in conjugation). But as the van't Hoff model indicates ethylenes are planar except when bulky substituents interfere and cannot be accommo- dated without some twisting of the double bond. Modern theory shows that the two members of the double bond are not equal as in the original model.One is cylindrically symmetrical about the bond direction (a CT bond) the other consists of two inseparable streamers of electron density above A. Werner " Lehrbuch der Stereochemie " G. Fischer Jena 1904 ; M. Ramart Lucas and J. Hoch Ann. Chim. 1930 [XI 13 385 ; K. Freudenberg " Stereochsmie " F. Deuticke Leipzig 1933 Vol. I1 ; H. Gilman " Organic Chemistry " Vol. I J. Wiley New York 1943 (Contributor C. S. Marvel) ; G. W. Wheland " Advanced Organic Chemistry " J. Wiley New York 1949. * A . Michael J. Amer. Chem. SOC. 1918 40 704 1674. V. Grignard " Trait6 de Chimie Organique " Masson et Cie. Paris 1935 Vol. I 4 J. H. van't Hoff BUZZ. SOC. chim. 1875 [ii] 23 295 ; " The Arrangement of Atoms (Contributor C. Dufraisss). in Space" Longmans Green London 2nd Edn. 1898. 101 102 QUARTERLY REVIEWS and below the plane of the ethylene the latter (a n bond) is therefore not cylindrically symmetrical.Because of this lack of symmetry it is only when the molecule is planar that there is maximum overlap of the two atomic p orbitals which together constitute the n bond. The hybridisation of the carbon atoms involved is trigonal and the angle between the lobes of maximum electron density 120". Measurement shows that the bond angles in ethylenes in fact vary a little from this (LHCH in ethylene 110" -J= 5" LHCCl in dichloroethylene 116" -& 2"). As a consequence of the geometry of the system certain substituted ethylenes must exist in two isomeric but non-enantiomorphic formsY4 e.g. types Aa (cis) and Ab (trans). The substituents R and R need not be digerent.Types B and C are often encountered but D (e.g. fulvenes) and E (e.g. indigo) are less common. The criterion for the existence of geometrical isomers is that a rotation of one carbon atom through 180" gives a new planar arrangement of the substituents. It may be noted that certain trans-isomers i.e. Ab(R = R,) C(R = R, R = R,) or E(R = R,) have centres of symmetry. Type Aa This has useful consequences (see below). Type Ab R4 'c-c /R3 R,/ \R u u I 3 3 2 Type E * * The ring size is immaterial but the system must be unsymmetrical about the C =C axis. The double bond in an approximately planar ring system must be con- strained in the cis-configuration (e.q. cyclohexene) but when the ring is multiplanar cis- and trans-isomers exist (e.g. civetone '). Geometrical isomers about a single bond forming part of a ring system (e.g.cis- and trans-cyczohexane-1 2-diol) are very common and theoretically cis- and trans-isomers about ordinary single bonds (e.g. 1 2-dichloroethylene or buta-1 3-diene) can exist though because of the low energy barrier separ- ating the two forms they have not been isolated. Steric hindrance may raise the energy barrier and the two forms of (11) have been isolated and C. A. Coulson Quart. Reiews 1947 1 144. 6 Landolt-Barnstein " Atom und Molekularphysik " 2te Teil Molekulen I Springer-Verlag Berlin 1951. M. Stoll J. Hulstkamp and A. RouvB Helv. Chim. Acta 1948 31 543. CROMBIE GEOMETRICAL ISOMERISM 103 Configurations about single bonds will not be discussed are quite stable.* further except when they are directly relevant to the >C-C( system.Me Me When two double bonds are present as in CR,H.-::CH*CH==CHR, four isomers exist if R + R (truns-trans cis-trans truns-cis and cis-cis) and three if R = R,. The number of isomers possible increases rapidly as the number of double bonds increases. For a symmetrical conjugated com- pound with n double bonds the number of isomers possible ( N ) can be obtained from the expression N = 2n-1 + 2p-1 (when n is even n = 2p when n is odd n = 2p - 1). A long-chain truns-compound (111) is essentially linear and simulates the corresponding saturated one (IV) whereas the corresponding cis-isomer has a bent structure (V).* This bending is most marked when the double P. R. Shildneck and R. Adains J . Anzer. Chenz. Soc. 1931 53 2203. R. Kuhn and A. Winterstein Helv. Chim. Acta 1928 11 87.lo W M. Smith and K. C. Eberly 117th Meeting Amer. Chem. Soc. Detroit April 1050; H. Mark Research 1951 4 167. * The bonding caused by n cis-double bond is illustrated by t,he two polymers formed by esterifying fumaric acid with cis- and tmrrs-butenediol.lO. One polymer (the cis) is riibbory and of low softening point but the other (trans) is a high-melting crystailine solid. c o c 0 c c trarw-trccns,t fits easily into t,ho crystal lattico in long chains ; crystalline. co c 0 c c \J \o/ ‘0 C O C 0 c \o/ \(/ \cg \c/ Bc I C I trans-cis,t alignment of chains diflicult ; rubbery. t Angles are somewhat distorted by limitations of printers’ type. 104 QUARTERLY REVIEWS bond is a t the centre of the chain. In a polyene chain where there are many possible permutations and combinations of cis- and trans-arrange- ments about the double bonds a great diversity of molecular shape is possible.-AN\/ /vvc'3 \ Saturated (IV.) cis- t (V.) AN\/- trans- (111.) 5 Some Factors affecting the Stability of Geometrical Isomers.-The repul- sion of like dipoles oriented cis to each other (e.g. the two C=O groups in cis-dibenzoylethylene) contributes to the instability of this form whereas interaction between unlike dipoles may be expected to increase stability. In specific cases internal hydrogen bonding can be envisaged as a stabilising factor as in the cis-enol (VI).ll An extreme case of cis-stabilisation is pseudo-acid formation (VII) .12 Two bulky cis-substituents preventing planarity in one form increase its energy. Moreover when the double bond is in conjugation the departure from planarity prevents maximum delocalisation of the 7~ orbitals and the resonance energy is decreased.\I 0 (VII.) Scale drawings of planar projections of cis- and trans-dibenzoylethylene would illustrate the operation of steric hindrance in the cis-form [cf. (VIII)]. [The following values are used for the bond lengths C-C 1.54 ; C=C IIB. F. Eistert F. Weygand and E. csendes Chem. Ber. 1951 84 745. l2 R. E. Lutz and M. Couper J . Org. Chem. 1941 6 77. Cf. however H. Henecka Chem. Ber. 1948 81 189. CROMBIE GEOMETRICAL ISOMERISM 105 1.33 ; C-H 1.09 ; C=O 1.20 ; C-C (benzene) 1.40 A. The bond angles C-C-R and 0-C-C are taken as 123" and the atoms and groups are drawn by using van der Waals radii for oxygen 1.40 hydrogen 0-75 and methyl 2-00 A. Van der Waals radii are too large a measure of steric hindrance and covalent radii too small here a compromise is effected by the use of a low value for the van der Waals radius of hydrogen.] A rather more complicated case is that of the cis- and trms-methyl- dibenzoylethylenes l3 (see VIII and IX).It is known experimentally that the cis- is more stable than the trans-isomer. Prom an electrostatic point of view the cis-isomer would be expected to be the less stable. But there is steric hindrance in both forms. In the trans-isomer this can only be relieved by rotating both phenyl nuclei about the Ph-CO-C bonds. In the cis-isomer rotation of one Ph*CO-C system is enough to relieve the hindrance and this leaves the Ph*CO*CH=C(CH3)- system planar and in conjugation. The resonance energy contribution from this more than offsets any energy increase due to electrostatic repulsions.If the sterically hindered double bond lies between two long con- jugated systems a considerable loss of resonance energy may be caused by the fact that it breaks or decreases conjugation between them. L. Pauling has discussed the carotenoids.l4 In this case the system -C(CH,)CH=CH-CH= (X) is highly hindered in the cis- but not in the trans-form. It is clear from other examples that although it may result in an isomer being unstable hindxance-even considerable hindrance-does not in itself preclude the existence of a cis-isomer." However in the carotenoids and vitamin A the loss of resonance energy is large and investi- gators in the field have felt justified in discounting the existence of cis- isomers of system (X)-the so-called " sterically ineffective double bond ".l 3 L. P. Kuhn R. E. Lutz and C. R. Bauer J . Arner. Chem. Xoc. 1950 72 5058 ; R. E. Lutz and C. R. Bauer ibid. 1951 73 3457. l4 L. Pauling Portschr. Chern. Org. Naturstoffe 1939 3 203. * Recent careful work 16 has not however enabled the isolation of ciS-cis-j?- methylmuconic acid to be achieved all attempts giving a trans-cb-isomer. 106 QUARTEIltLY REVIEWS Caution is desirable in eliminating isomers in this way. The length of the resonating system and the position of the double bond are determining \ C / 0 C=C EJ \ C / / \ C factors as the loss of resonance energy is much less when the double bond is situated a t the end of the chain than when it is near the centre. The cal- culations of 1,. Pauling illustrate this.15 Stereomutation of Geometrical Isomers.-A pure geometrical isomer may be converted into an equilibrium mixture with its stereoisomer by thermal isomerisation.The proportions of each in this mixture depend on the relative thermodynamic stabilities under the conditions employed the more stable predominating. Strong illumination markedly affects the composition of the equilibrium mixture and the thermodynamically less stable isomer is often favoured. This is because the photochemically excited molecules spend the greater part of their time in phases of high potential and when such a molecule goes back into its normal state that isomer predominates whose configuration corresponds to these phases of higher potentia1.l The ease of stercomutation is governed by the activation energy E for the particular mechanism (Fig.1). This must be influenced by the electron density a t the double bond and consistently with this conjugated stereo- mutate more easily than do unconjugatecl compounds. The energy difference ( E l ) between cis- and trans-isomers about a double and a single bond may not be very different but in the former case the activation energy (E,) is much higher and the two forms exist separately. G. B. Kistiakowsky has discerned experimentally a t least two mech- anisms for thermal stereomutations.l8 cf. l9 Both are kinetically of the first l5 L. Pauling Helv. Ghem. Acta 1949 32 2241. l6 J. A. Elvidge R. P. Linstead and P. Sims J. 1951 3398. l7 A. R. Olsen and F. L. Hudson J . Amer. Chem. SOC. 1933 55 1410 ; A. R. Olsen and W. Maroney ibicl. 1934 56 1320 ; G.N. Lewis T. T. Magel and D. Lipkin 1940 62 2973. l8 G. B. Kistiakowsky and W. R. Smith ibid. 1935 57 269. 1@ J. L. Magee W. Shand and H. Eyring ibid. 1941 63 677. CROMBIE GEOMETRICAL ISOMERISM 107 order and their rate constants are ca. loll exp. - 40,000/kT and m. lo4 exp. - 25,00O/kT respectively. In order to effect geometrical interconversion the double bond must be ( ( dissociated " in some way so that rotation can occur and one of the carbon atoms be twisted through 180" relatively to the other. When the carbon atom has been twisted through go" rotation may proceed onwards to give the stereoisomer or may become reversed to yield the original isomer again. This " perpendicular ethylene " is the transition state in simple cases. By linear combination of the two atomic p orbitals of the carbon atoms besides the low energy n orbital in planar ethylene a second high-energy anti-bonding orbital (n*) can be formed.In " per- pendicular ethylene " the order of stabilities is reversed. Throughout the actual twisting operation the electrons occupy t'he lowest energy orbital formed by hybridisation of the two.22 The electrons in the ( ( perpendicular 9" A q / e o f rotation Fm. 1" * If steric hindraiico or other cnrises inhibit rotation in one direction ono of the energy Steric hindranco may also causo a displacement of the barriers will be higher than the other. energy minima from the 0" and tho 180" position. ethylene " formed by the continuous adiabatic rotation are anti-parallel (i.e. transition is from the singlet ground state to an upper singlet H.Eyring associates this with stereomutations having a fre- quency factor of loll and an energy of activation of ca. 40 kcals. Since in excited ethylene the two electrons occupy separate orbitals their spins can also be parallel (triplet state). But the probability of this reversal of spin occurring is low although the low activation energy favours this the second isomerisation mechanism (frequency factor lo4 energy of activation 25 kcals.). The triplet mechanism is of nearly the same height in the various molecules considered but the singlet level differs greatly from one molecule to another depending on the ability of the two electrons from the double bond t o enter into some other kind of binding. When a phenyl zoR. A. Harmann and H. Eyring J. Chem. Phys. 1942 10 557.21 H. M. Hulbert R. H. Harman A. V. Tobolsky and H. Eyring Ann. N.Y. Acad. Sci. 1943 44 371. 2 2 M. J. S. Dewar " Electronic Theory of Organic Chemistry " Oxford Univ. Press 1949. 108 QUARTERLY REVIEWS group is available there is resonance and stilbenes isomerise by the adiabatic singlet mechanism. Resonance stabilisation is smaller in the case of but- 2-ene or maleic ester and these use the triplet mechanism. Besides the purely thermal and the purely photochemical cis-trans- equilibrations there are many examples which are catalysed. The r6le of catalysts is still not clear and a number of different mechanisms are probably involved. The suggestions in the literature are largely concerned with schemes which allow the catalyst to add in some way-thus reducing the n electron density and permitting rotatiori-and then to disengage.Evidence indicates that the halogens cause stereomutation by it free- radical mechanism. 23 Thus the bromine -catalysed conversion of cis- stilberie into an essentially trans-product occurs only in the presence of light and is sensitive to the effects of peroxides and antioxidants. The following scheme involving bromine atoms 24 has been suggested cis-PhCH=CHPh + Br' + PhCHBr-bHPh cis-PhCH=CHPh tmns-PhCH--C'HPh -1 PhGHBr*6HPh etc. The PhCH in the intermediate is considered to be planar and the PhCHBr- pyramidal free rotation being possible. Both the cis- and the trans- isomer can pass through the intermediate and equilibrium is attained. It is likely that stereomutations catalysed by other atoms free radicals and paramagnetic substances proceed by similar mechanisms.The thermal and photochemical iodine-catalysed stereomutation of di-iodoethylene is viewed 25 as passing through the intermediate IH&-CHI rather than proceeding by a type of mechanism which A. R. Olson proposed (see later).26 A number of acid-catalysed stereomutations are known and an inter- mediate of the type RO2C*CH,*CH*CO2R has been invoked in the case of the maleic-fumaric system there is evidence against this. 27 An alterna- tive mechanism has been suggested by K. Nozaki and R. Ogg.28 HX is * on subse- considered to give an intermediate RO,C*CHX-CH=C quent elimination of HX the isomerised or the original compound may be formed. + /OH \OR A similar view (Wheland l) postulates a resonating system + + RO ,C*CH=CH-C*OH RO ,C*CH*CH=C-OH I 0-CH I O'CH 23 M.S. Kharasch J. V. Mansfield and F. R. Mayo J . Amer. Chem. SOC. 1937 69 1165 ; Y . Urushibara and 0. Shimamura Bull. Chem. SOC. Japan 1937 la 507 (Chem. Abs. 1938 33 1682). 24 A. Berthoud and C. Urech J . Chim. phys. 1930 27 291. 25 R. M. Noyes R. G. Dickinson and V. Schomaker J . Amer. Chem. Xoc. 1945 26 A R. Olsen J . Chem. Phys. 1933 1 418. 28 K. Nozaki and R. Ogg J . Amer. Chem. SOC. 1941 63 2583. 67 1319. C. Horrex Trans. Farachy soc. 1937 33 571. CROMBIE GEOMETRICAL ISOMERISM 109 Aluminium trichloride and boron trifluoride 29 acids in the wider sense also catalyse stereo-equilibrations but their action is not general. Catalysis by paramagnetic substances has been discussed by Eyring.20y 21 The numerous catalysts available for the process are listed on p.139. Nomenclature.-Compounds of type A present no difficulty in nomen- clature but for types B and C the groups to which the prefix cis or trans applies must be specified in some way. Rules 30 for the systematic naming of long-chain olefinic geometrical isomers were adopted by I.U.P.A.C. Ccmmission of Nomenclature of Organic Chemistry in New York Sept. 1951. The compounds are regarded as derivatives of the longest chain which con- tains the maximum number of double bonds the cis- or trans-prefix des- cribing the juxtaposition of the carbons of the main chain being placed immediately before the numbering of the double bond e.g. 3-tert.-butyl- cis-2 cis-4-hexadiene (XI). H H Me c-c Br I \ / Me \ / \ / \ C=C CMe H \ / c-c H / \ c1 (XI.) (XII.) Some of the better known pairs of isomers have trivial names e.g.angelic and tiglic acids citraconic and mesaconic acids and compounds may be named as derivatives of these. Sometimes the less stable isomer has been designated by the addition of the prefix is0 to the name of the stable com- pound (isocrotonic acid isostilbene) ; allo may also be used (allocinnamic acid). In order to name certain simple stereoisomers such as those of l-bromo- 1-chloro-2-iodoethylene it is suggested that the groups to which the prefix applies be italicised in cursive text e.g. (XII) is cis-l-bromo-l-chloro-2-iodo- ethylene or trans-l-bromo-l-chZoro-2-iodoethylene but there are manifest difficulties in indexing on this convention and it has no value in speech. The use of is0 is particularly undesirable. Methods €or the Determination of Geometrical Configuration As a result of the difference in shape and symmetry between corre- sponding cis and trans-isomers there are differences in a large number of physical and chemical properties.* The process of determining configura- 2g C.C. Price and M. Meister J . Amer. Chem. SOC. 1939 61 1595. 3* M. B. Epstein and F. D. Rossini Chem. Eng. News 1948 26 2959 ; cf. ibid. 1949 31 L. Crombie and S. H. Harper J. 1950 873. 32 W. Sanderman Seifensieder Ztg. 1941 68 41. * There are also marked differences in physiological properties. 27 1303. Thus cis-hex-3- en-1-01 has the odour of fresh grass but the trans-isomer smells of ~hrysanthemums.~~ Taste,32 vitamin A activity,33 serum reactions,34 enzyme reactions,35 reproduction of algae 36 and many more such properties are markedly dependent on geometrical configuration.Normally cis- or trans-isomers are synthesised stereospecifically in Nature cis-type-A forms are very common. I10 QUARTERLY REVIEWS tion is therefore one of detecting the differences and relating them to the configurations of the isomers. It is fhe latter step which presents the difficulty as the relation between molecular shape and symmetry aid physical or chemical properties may be indirect slid the results require very critical evaluation. Iiidiscrimiiiate use of rules or regularities without reference to their basis has sometimes led to erroneous conclusions. The solution of a more difficult configurational problem may come from the convergence of a number of lines of evidence the single lines each carry- ing in themselves insufficient weight.Some of the methods described below have been tested only on limited ranges of stereoisomers and these readily available rather than representative. Furthermore the possibility of stereomutation under experimental conditions must always be kept in mind. Many of the methods have limited scope those which give promise of determination of configuration when only one isomer is available are very valuable. I n all cases however it is desirable to examine both isomers if possible before making an assignment. Melting and Boiling Points and Related Phenomena.-It has been found empirically that usually (but not always) cis-compounds of type ,4 have lower melting points than their trans-isomers and this is generally related to the higher symmetry of the trans-type which results in better pack- ing and greater forces in the crystal lattice.Melting-point regularities within a series of the more complex types ( B and C) may also give indications of configuration. Since the' formation of an anisotropic liquid phase (" liquid crystal " or smectic state) demands among other factors a linear molecule in this type of case its formation has been cited to support trans- configurations 37 the compound melts sharply to a turbid liquid which melts equally sharply to a clear liquid e.g. trans- but not cis-p-methoxy- cinnamic acid. 38 It has been cited in support of the trans-configurations of one 1 4-di-p-inethoxyphenylbutadiene 39a and an anil of trans-p- aminostilbene. 37 Because of the similarity in shape trans-isomers frequently form solid solutions (" mixed crystals ") with the corresponding saturated compounds whereas the cis-isomers do not e.g.trans-crotonic acid (but not the cis- acid) with butyric acid trans-cinnamic acid (but not the cis-acid) with phenylpropionic a~id.39~ Substituted trans-stilbenes form mixed crystals York 1949. 33 L. Zechmeister in " Vitamins a i d Hormones " Vol. VII Acad. Press Inc. New 3 4 K. Landsteiner and J. van der Schier J . Exp. iWed. 1934 59 751. 35 A. Jung and H. Muller Helv. Chim. Actu 1922 5 239 ; W. Fabisch Biochent. Z . 1931 234 84; G. B. Crippa and S. Maffei Gazzetta 1940 70 212; D. Pressmann J. H. Bryden and L. Pauling J . Amer. Chem. SOC. 1948 70 1352. 36 I. M. Heilbron J . 1942 79. 37 C. Weygand and R. Gabler Ber. 1938 71 2474. 38 R. Stoermer Ber. 1911 44 637. 39a Y . Hirshberg E. Bergmann and F.Bergmann J . Amer. Clhem. SOC. 1950 72 5120. 39b G. Bruni and F. Gorni Atti R. Accad. Lincei 1S99 8 I 454 ; G. Bruni ibid. 1904 12 I 626; C. Dufraisse Ann. Chim. 1922 [ix] 17 133; G. B. Semeria. Atti Accad. Sci. Torino 1924 59 700 ; Gazzetta 1925 55 79 ; J. Timmermans Bull. SOC. CROMBIE GEOMETRICAL ISOMERISM 111 with the meso-forms of the corresponding saturated compounds but eutectics with the racemic. 39c Boiling-point and solubility regularities between the cis- and the truns- series have been claimed (i.e. that the cis-isomer has the higher b.p. and greater solubility). A number of factors is involved in the manifestation of these and similar properties and there are numerous exceptions. have compared the physical properties of a series of geometrical isomers with those of a series of comparably substituted aromatic compounds and on the assumption that similar alterations in the positioning of groupings cause similar changes in physical properties have assigned configurations by conzparison.A favour- able example 41 illustrates the principle A. Langseth 40 and K. von Auwers and L. Harres CH / CO,Et CH,-C-CO,Et CH /' CO,Et CH,*C*CO,Et II CH3 \ CH *OH (XV.) (XVI.) 55*5"/1 I H-C-CH I/ 0 0.11. (XIV.) (XIII. ) B.p./mm.. 118.5"/12 49"/11 134'/15 d$O . . 1.017 0.917 1,025 0.924 14308 1.5149 1.4353 20 T L I ~ ~ . . 1.5086 Compounds (XIII) and (XIV) have lower boiling points densities and refractive indices than their isomers (XV) and (XVI) and if the structures of the aromatic compounds are accepted as known then (XIV) and (XVI) are given the configurations shown.Unfortunately there are numerous exceptions. 41 Surface-film Properties.-Because of their shape molecules of long-chain trans-acids and -alcohols can form close-packed surface films this is not possible with the bent cis-forms. Also whereas the truns-forms can pack closely with the corresponding saturated ones the cis-compounds cause expansion when introduced into a film of saturated acid or alcohol. On collapse the cis-films usually give oil lenses but the truns-films tend to give solids or smectic liquids by chain adlineati0n.4~ Energy Contents.-By measurement of heats of combustion or hydro- genation an estimate of the difference between the energies of two geometrical isomers may be made. If this energy difference can be related to the juxtaposition of groupings then configurations may be assigned usually in type-A isomers the cis is the less stable but in other cases decision u prior; on relative stabilities may not be possible.Theoretically the measurements of heat of combustion should be made on t,he vapour because of lattice effects. However the cis- normally packs less well than the trans-isomer and for this reason the solid cis-compound chim. B e l g 1927 36 179 ; H. Keller R. Pctsternak and H. von Halban Helv. Chim. Acta 1946 29 512. 39c F. von Wessely and H. Welleba Ber. 1941 74 785. 40 A. Langseth 2. physikal. G'hem. 1925 118 49. 4l K. von Auwers and L. Harres ibid. 1929 143 1. 4 2 J. Marsden and E. K. Rideal J . 1938 1163; A. E. Alexander J . 1939 777. 112 QUARTERLY REVIEWS has an even higher heat of combustion than expected.Thus the heat of isomerisation (- AH)-the difference between the two heats of combustion- of maleic to fumaric acid is 4.4 kcal. in the gas phase but 6.9 in the crystalline state. A few examples are collected in Table 1 the different heats of TABLE 1 Dimethyl maleate solid (- 19") 4 3 . . - . . , fumarate , (- 19") 4 3 . . . . . cis-Cinnamic acid solid (m.p. 42") 4 4 . . . . 7 9 Y 9 9 9 , (m.p. 58") 4 4 . . . . 7 7 9 Y 9 , (m.p. 68") 4 4 . . . . trans- , 9 , (m.p. 133") 45 . . . . cis-4-Nitrostilbene solid (m.p. 61") 4F. . . . . trans- , , (m.p. 157.5") 4 6 . . . Heat of cornbus tion keal./mole 667.2 664.1 1043.8 1045.2 1047-6 1040- 9 1723.2 1716.2 Heat of isoinerisation - A H ) kcal./niole 3.1 2.9 4.3 6.7 7.0 TABLE 2 Heat of ' Heat of 1 hydrogenation isomerisation kcal./mole .~ kcal./mole trans- But - 2 -one (gas) 4 7 . . . . . . Methyl cis-cinnamate (11s.) . . . . . . , trans- , (liq.) 48 . . . . . . Diethyl maleate (ljq.) 48 . . . . . . Diethyl fumarate (119.) 48 . . . . . . cis- 9 9 9 7 (gas) . . . . . I 27.62 28.57 28.19 24-18 33.52 29-30 0.95 4.01 4.22 combustion for the three polymorphic cis-cinnamic acids are noteworthy. An inherent difficulty in the method is that the energy difference is a small quantity derived by difference of two large ones and is therefore subject to considerable error. Prom this point of view measurement of heat of hydrogenation gives a more reliable energy difference. Measurements are available for cis- and trans-butenes in the gas phase 47 and certain stereoisomers in the liquid state 48 (Table 2). It must be noted that the two isomers must give the same compound on hydrogenation-not an erythro- and threo-pair as these have different energies themselves.X-Ray Crystallographic Measurements.-A complete X-ray structural analysis can give an unequivocal decision on the cis- or trans-configuration 4 3 A. Wassermann 2. physikal. Chem. 1930 146 418. 4 4 F. Eisenlohr and A. Metzner ibid. 1937 178 339. P. Landrieu F. Baylcoq and J. R. Johnson Bull. SOC. chim. 1929 [iv] 45 44. 46 C. M. Anderson L. G. Cole and E. C. Gilbert J . Amer. Chem. Soc. 1950 72 1263. 47 G. B. Kistiakowsky J. R. Ruhoff H. A. Smith and W. E. Vaughn ibid. 1935 48 R. B. Williams ibid. 1942 64 1395. 57 876. CROMBIE GEOMETRICAL ISOMERISM 113 of a compound. Thus stilbene 49 (m.p. 124") and the side-chain double bond (C22:23) of calciferol 50 are truns and sorbic acid 51 (m.p.134.5") has the truns-trans-structure. But the less arduous earlier stages of X-ray crystallographic analysis may be used to determine molecular symmetry which in suitable cases enables an allocation of configuration to be made. Thus one of the isomers of diethylstilbcestrol dipropionate (XVII) has a E~.CO.O<%-CH~-CH~ CH,*CH,-C 3 0 . C O . E t (XVII. ) centre of symmetry and must therefore be The centre of symmetry in dimethyl trans-trans-muconate and dimethyl trans-fumarate shows that the accepted configurations are correct. 53 In the case of a$?-diethylidenedibenzyl (3 4-diphenylhexa-2 4-diene) (m.p. 101") the molecule has again been shown to have a centre of symmetry buttwo structures (XVIII) and (XIX) are possible and a decision them must be made on other evidence; (XVIII) is favoured.54 between (XVIII.) (XIX.) Amongst long-chain aliphatic compounds the structures of the synthetic polyenes prepared by R.Kuhn's procedure have been shown to be all- trans (as are many natural carotenoids). The distinctive X-ray diffraction " side spacings " of glycerides of trans-acids resemble those of the corres- ponding saturated acids but those of the cis-forms are different-another consequence of the bent ~hain.~5 For the field of high polymers mention must be made that a cis-structure for the isoprene units of natural rubber and a trans-structure for those of gutta percha which has totally different properties have been confirmed by X-ray measurements this has also revealed other interesting details. 56 The structures of cis- and trans-but-2-enes 57 and dichloroethylenes 58 have been determined in the gas phase by electron-diffraction measurement.Ultra-violet and Visible-light Absorption.-It has long been realised that 49 J. M. Robertson and I. Woodward Proc. Roy. Soc. 1937 A 162 568. 50 D. Crowfoot and J. P. Dunitz Nature 1948 162 609. 51 K. Lonsdale J. 81. Robertson and I. Woodward Proc. Roy. Xoc. 1941 -4,178,43. 5 2 C. H. Carlisle and D. Crowfoot J. 1941 6. 53 I. E. Knaggs and K. Lonsdale J. 1942 417. 5 4 G. A. Jeffrey H. P. Koch and S. C. Nyburg J. 1948 1118. 5 5 M. G. R. Carter and T. Malkin J. 1947 554. 56 C. W. Bunn Proc. Roy. Soc. 1942 A 180 40. 57 L. 0. Brockway and P. C. Cross J . Amer. C??zem. Soc. 1936 58 2407. 58 L. 0. Brockway J. Y. Beach and L. Pauling ibid. 1935 57 2693. 114 QUARTERLY REVIEWS differences exist between the ultra-violet light absorptions of cis- and trans- An empirical approach compares the absorption of a cyclic model having the chromophoric system in question held rigidly in one con- figuration with that of the unknown isomer.Thus the absorption spectrum of trans-stilbene (XX) resembles that of 2-phenylindene (XXI) whilst that of cis-stilbene (XXII) differs markedly. 6o However less empirical methods are now available. 59 (XXI.) (XXII.) The light absorption of polyenes has been extensively studied.61 If in a planar " all-trans "-carotenoid one of the double bonds is inverted to give a planar non-sterically-hindered cis-isomer three effects are observed. The long-wave-length K band is shifted to a slightly shorter wave-length (by ca.5 mp) its extinction coefficient is lowered and a new band called a " cis- peak" appears at considerably shorter wave-length. The K band arises (on classical views) from the absorption of light by a system of electrons oscillating from one end of the conjugated chain to the other (transition from the ground state to the first excited state). The intensity of the band a factor dependent on the dipole moment is proportional to the square of the distance between the ends of the conjugated system.62 As the truns- isomer is the most extended form a fall in intensity is therefore to be expected on conversion into cis-forms. The magnitude of the change is naturally dependent on the position of the cis-double bond. The alteration of the K band to shorter wave-lengths by the truns-cis-change is ascribed to an alteration in the energy difference between the ground and the first excited state.A '' cis-peak " arises from a transition from the ground to the second excited state this involves the oscillation of the electrons from the ends of the conjugated system towards the middle and from the middle towards the ends. In the case of an all-trans-polyene (which possesses a centre of symmetry) this oscillation causes no dipole change and no band appears in the cis-peak region. But cis-isomers have a dipole moment for this transition perpendicular t o a line drawn between the ends of the conjugated system hence cis-peaks arise. When the cis-double bond is situated a t the centre of the polyene chain the dipole change and therefore the cis-peak reaches 59 V. Henri and J.Errera Compt. rend. 1925 180 2049 ; 1925,181 548 ; J. Errera J. Phys. Radium 1926 [vi] 7 215. 60 C. Wiegand and E. Merkel Med u. Chem. Abhandl. med. chem. Fortschungstiitten I.G. Farbenind. 1936 3 320 (Chem. Abs. 1937 31 5797) ; U. V. Solmssen J. AmeT. Chem. Soc. 1943 65 2370. 6 1 L. Zechmeister Chem. Reviews 1944 34 267 ; Ann. N . Y . Acad. Sci. 1947 49 220 and references cited therein. 6 2 L. Zechmeister A. L. Le Rosen W. A. Schroeder A. Polghr and L. Pauling J. Amer. Chem. Soc. 1943 65 1940. CROMBIE GEOMETRICAL ISOMERISM 115 maximum intensity.61 62 The difference between the light absorption of a trans- and a cis-carotenoid (not homogeneous) is shown in Fig 2. L. Zech- meister has made subtle use of the three effects mentioned in assigning probable stereochemical configurations to carotenoids.Similar considerations apply to the light absorptions of shorter con- jugated systems of double bonds. In certain cases (e.g. dienes or @-un- saturated acids) the length of the chromophore is identical in both the cis- and the trans-form but even here progress may be made as another factor-partial or complete steric inhibition of resonance-may supervene. Often the spectral differences arise from the superimposition of effects due to both chromophore length and resonance inhibition. If in one form of the geomet- rical isomers the cis-arrangement of bulky substituents prevents the achievement of coplanarity and thus full conjugation of the chromophore system a marked fall in extinction coefficient (com- pared with that expected for an unhindered case) occurs.In more extreme cases the wave- length is shifted to shorter regions as well. E. A. Braude 63 has discussed the effects as follows when the steric inhibition of planarity is small the electronic transitions may be restricted to vibrational states in which as a result of the various stretching and bending vibrations of a poly- atomic molecule a sufficient de- 75 I- 300 400 500 Wave - /ength (mp) FIG. 2 A All-trans-y-curotem in hexane. B Mixture of stereoisomers of y-carotene formed by iodine- and light-catalysed stereomutation. gree of coplanarity is attained. In such a case although the intensity falls the wave-length is not appreciably shifted. When the degree of steric hindrance is large though not large enough to prevent resonance altogether the energy level of the excited state (which is more infiuenced by ionic resonance contributions than is the ground state) will be raised relative to the latter.The effect of this is to displace the K band to a shorter wave- length in addition the fall in intensity is pronounced. In both cases bands characteristic of the partial chromophores may also obtrude. Some typical examples of the light absorption of cis- and trans-isomers are given in Table 3 (cf. also refs. 64 65) and these can be interpreted on the 63 E. A. Braude E. R. H. Jones H. P. Koch R. W. Richardson F. Sondheimer and J. B. Toogood J . 1949 1890. e 4 H. P. Koch Chem. and Ind. 1942 273 ; A. G. Caldwell and E. R. H. Jones. J. 1946 599 ; A. W. Nineham and R. A. Raphael J. 1949 118 ; M. E. Herr and F. W Heyl J. Amr. Chern. SOC. 1950 72 1753; R. Kuhn and M.Hoffer Ber. 1932 65 651. 6 5 B. Arends Ber. 1931 64 1936. I 116 QUARTERLY REVIEU7S lines indicated above. In type-A compounds steric hindrance is possible only in the cis-forms and as the latter from chromophore-length considera- tions have lower extinction coefficients and sometimes main absorption at slightly shorter wave-lengths than the trans-isomers the two effects reinforce each other. Other types must be considered individually in conjunction with models classes may be envisaged in which steric hindrance and chromophore-length effects work in opposition. The use of vacuum ultra-violet spectra in stereochemical diagnosis has been mentioned by certain authors but is as yet of little importance.66 TABLE 3 CHPhzCHPh 6 5 67 . . . CPhMe:CHMe* 68 . . . CHPh:CH*CHMe.OH 69 .Ph*CO*CH:CH.COPh l3 . PhC0CH:CHPh l3 . . CHPh:CH.CH:CH,’o . . CHPh:CH*CO,H71. . . cis ’max. 278 235 240 253 { 2; 268 261 ‘mar. 9 350 8 200 12 000 17 800 14 000 8 900 14 300 10 500 trans %I,t%X. 294 243 {E 266 (225 1298 280 272 ‘ma*. 23 400 12 100 19 500 21 400 12 200 23 700 23 000 19 500 *cis and trans refer t o P h and Me (not t o t h e two Me). The light absorption within pairs of geometrical isomers in the solid state differs as might be expected. Thus trans-dibenzoylethylenes are yellow but the cis-forms are colourless.3~72 Infra-red and Raman Spectra.-In favourable examples infra-red and Raman spectroscopy can distinguish between cis- and trans-isomers because of symmetry differences. trans-Dichloroethylene has a centre of symmetry (CZh) whereas the cis-compound has not (C,,).Because of this the rule of “ mutual exclusion ” holds for the former but not for the latter.73 This means that for trans-dichloroethylene all frequencies present in the infra-red spectrum are absent in the Raman spectrum and vice versa As a corollary the C=C stretching frequency present in the infra-red spectrum of the cis- is absent in that of the trans-form. These features are illustrated in the following extract from the spectra of the two dichloroethylenes 73 6 6 E. P. Carr and H. Stuckeln J . Amer. Chem. SOC. 1937 59 2138; I. I. Rusoff J. R. Platt H. B. Klevens and G. 0. Burr ibid. 1945 67 673. 67 A. Smakula and A. Wassermann 2. physikal. Chem. 1931 A 155 353. 68 D. J. Cramm J . Amer. Chem. SOC. 1949 71 3883. E. A. Braude and J. A. Coles J. 1951 2078 2085. 70 0. Grummitt and F.J. Christoph J . Amer. Chem. SOC. 1951 73 3479. 71 E. Havinga and R. J. F. Nivard Rec. Trav. chim. 1948 67 846. 7 2 J. B. Conant and R. E. Lutz J . Amer. Chem. SOC. 1923 45 1303. 7 3 G. Herzberg “ Infra-Red and Raman Spectra of Polyatomic Molecules ” D. Van Nostrand Inc. New York 1945 pp. 330 256 and refs. cited therein. CROMBIE GEOMETRICAL ISOMERISM cis Infra-red Raman (cm.-l) (cm.-l) 1179 1303 1591* 1587* 1689 3086 3077 3160 1200 3089 trans Infra-red (cm.-l) Raman (cm.-') 1270 1576* 1626 1692 307 1 117 3142 * Denotes the C=C stretching frequency. The spectra of cis- and trans-butene show similar features (here the C=C stretching is a t 1663 cm.-l in the infra-red).74 In the case of cis- and trans-1 2-dichloroprop-l-ene an argument on these lines has been employed to support assignment of structure.75 The C-H of dichloroethylene is here replaced by C-CH which has a similar electric moment.The C=C frequency does not disappear in the infra-red spectrum of the trans- but it is much weaker than in that of the cis-form. There are also more lines in the cis-spectrum. The method is limited though the extent of the limitations has not been fully explored. Support for the assignment of configuration to cis- and trans-selachyl alcohols has been claimed because of the intensity differ- ences of the C=C stretching frequency (1652 cm.-l) in the infra-red.76 M. Bourguel B. Grhdy and L. P i a ~ x 7 ~ have established empirically that the strong frequency associated with the C=C stretching mode in the Raman spectrum is lower in the cis- (by ca. 15 cm.-l) than in the trans-form e .g cis-hept-2-ene 1658 cm.-l trans- 1674 cm.-l. Though both frequencies vary with substitution the difference is upheld over a considerable variety of c0mpounds.7~ Much use has been made of the effect for assignment of structure and quantitative estimation. 79 As the frequency difference is small examination of both members of the cis-trans-pair is needed for decisions to be made with confidence. The shift has also been noted in the infra-red.aO The Raman spectra of cis- and trans-isomers are normally rather similar though the cis-forms are said to contain more lines at shorter wave-lengths.al Displacement of a strong frequency from ca. 1300 em.-' for the 7 4 H. Gershinowitz and E. B. Wilson J . Chem. Phys. 1938 6 247. 7 5 H. J. Bernstein and J. Powling J .Amer. Chem. SOC. 1951 73 1845. 7 6 E. Baer H. 0. L. Fischer and L. J. Rubin J . Biol. Chem. 1947 170 337; cf. J. W. McCutcheon M. F. Crawford and H. L. Welsh Oil and Soap 1941 18 9. 7 7 M. Bourguel B. Grhdy and L. Piaux Compt. rend. 1932 195 129. 78 B. Grhdy Bull. SOC. chim. 1935 [v] 2 1029 ; 1936 [v] 3 1093 1101 ; 1937 [v] 4 415; Compt. rend. 1936 202 322. 79 B. Grhdy and L. Piaux ibid. 1934 198 1235 ; G. Goethals Bull. SOC. chirn. Belg. 1937 46 409. P. Couvreuier and A. Bruylants ibid. 1950 59 436 ; L. Crombie J . 1952 in the press. A . Dadieu A. Pongratz and K. W. F. Kohlrausch Sitzungsber. Akad. Wise. Wien Math. Naturw. Klasse Abt. IIa 1931 140 353 (Chem. Abs. 1932 26 1615). 118 QUARTERLY REVIEWS trans- to m. 1270 cm.-l for the cis-form has been noted on a number of occasions.78~ 82 83 A medium-strong band a t ca.970 cm.-l (10.35 p) occurs in the infra-red spectra of trans-compounds of type R*CH,*CH=-CH*CH,R' but is weak or absent in the cis-forms. It is ascribed to the out-of-plane deformation of the two hydrogen atoms attached to the truns-double bond.84 85 The correlation has been proved valid for hydrocarbon^,^^ acids,s6 87 esters,86 87 alcohols,31 88 halides,s9 ketones,g0 and steroids 91 and is proving of great value in both diagnostic and quantitative 87 Only very small samples are needed. An increase in the intensity of infra-red absorption at 675-715 cm.-l (14-15 p) has been recorded on passing from the trans- to the cis-isomer. Although stereoisomers with greater substitution a t the double bond can be distinguished by their infra-red spectra correlations with their stereochemistry have not been Dipole Moments.-Interpretation is simplest when the ethylenic com- pound possesses one polar substituent on each ethylenic carbon atom and when the substituents have their axes of symmetry in the direction of bonding e.g.di~hloroethylene.~~ I n such cases the results are decisive for in the cis-compounds the two C-C1 moments add vectorially whereas in the trans-compounds they cancel out p = 0.00 D ,u = 1-74 D Similarly one q9-dichlorostilbene (m.p. 144") has zero moment whilst the isomer (m.p. 60') has a moment of 2.69 D and is therefore cis.94 When the two polar substituents are not the same the moments of the two stereoisomers must be calculated by vector addition so that comparison with the experimental values may be made.Complexities however arise. 8 2 L. Ruzicka H. Schinz and B. P. Susz Helv. Chim. Acta 1944 27 1561. 8 3 L. Crombie and S. H. Harper unpublished. 8 4 R. S . Rasmussen R. R. Brattam and P. S. Zucco J. Chem. Phys. 1947 15 135. 8 5 N. Sheppard and G. B. B. M. Sutherland Proc. Roy. Soc. 1949 A 196 195. 86 P. C. Rao and B. F. Daubert J . Amer. Chem. SOC. 1948 70 1102. 0. D. Shreve M. R. Heather H. B. Knight and D. Swern 116th Meeting Amer. Chem SOC. Atlantic City N.J. 1949 ; Analyt. Chem. 1950 22 1261. Crombie and S. H. Harper J. 1950 1707 1714. L. F. Hatch and 8. S. Nesbitt J . Amer. Chem. Soc. 1950 72 727. Stedman and D. Thompson J. 1951 2445. R. N. Jones J . Arner. Chem. Soc. 1950 72 5322. Sutherland J. 1950 915. 90 L. Crombie and S. H. Harper J. 1952 869; L. Crombie S. H. Harper R.O1 J. H. Turnbull D. H. Whiffen and W. Wilson Chem. and Ind. 1950 626 ; D. Barnard L. Bateman A. J. Harding H. P. Koch N. Sheppard and G. B. B. M. O 8 Q. P. Mikhailov and D. V. Tischenko J . @en. Chem. U.S.S.R. 1939 9 782. s4E. Bergmann J. 1936 402. CROMBIE GEOMETRICAL ISOMERISM 119 Thus trans-dichloropropylene has a higher moment (0.84 D) than would be expected and this is attributed to contributions by resonance structures of the type (XXIII).95 trans-l-Chloropropylene actually has a higher moment (1.97 D) than the cis- isomer (1.718 D) and this also has been explained by recourse to r e s ~ n a n c e . ~ ~ metry free rotation complicates the system. In cal- ( ~ ~ 1 1 1 . ) culations of the dipole moment two extreme values must be obtained corresponding to the limits of rotation the experimental value lying somewhere between them.A recent case is cis- and trans-2-chloro- vinyldichloroarsine 97 H& \ -/ c-c / \ When the substituents no longer have axial sym- c1 H cis trans 2.14 limits of pmlc* { 3-20} rotation 2.03 Found 2 . 6 1 ~ 2.21 In the trans-compound free rotation does not affect the calculated value as the As-C and the C-C1 bond are parallel. E. Bergmann has employed the method for kssigning geometrical struc- tures to compounds such as the ctp-dibromocinnamic acids the methyl B-bromocinnamates 4-bromo- and 3 4-dibromo-1 4-diphenylbut-3-en-2- one etc.94 98 99 Sometimes assignment has been made on only one isomer e.g. the foregoing 3 4-dibromo-compound (p = 3-17 D). The argument is that the dipole moment is close enough to that of benzylideneacetophenone (p = 2.93 D) to justify the conclusion that the two bromine atoms cancel each other.E. Bergmann also found that cis-p-chlorodiphenylvinyl bromide and certain other compounds have moments about double those of the trans-is~mers.~~~ loo He suggests that the latter are molecular com- pounds of cis- and trans-forms in 1 1 proportions though a different opinion has recently been expressed.lo1 Among long-chain compounds oleyl and elaidyl alcohols have been examined but show little difference (1.72 and 1.70 D respectively).lo2 The dipole moments of more than twenty pairs of geometrical isomers have now been measured.lo3 In some cases results give clear-cut decisions but in many others the complexities cause the results to be equivocal. 95 H. A. Smith and W. H. King J .Amer. Chem. SOC. 1948 70 3528. 9.5 N. B. Hanny and C. P. Smyth ibid., 1946 68 1005. 97 C. A. McDowell H. G. Emblem and E. A. Moelwyn-Hughes J. 1948 1206. 98 E. Bergmann and A. Weizmann Chem. Reviews 1941 29 529; E. Bergmann 99 E. D. Bergmann " Isomerism and Isomerisation " Interscience Publ. New 100 E. Bergmann L. Engel and H. Meyer Ber. 1932 65 446. 101 D. Y. Curtin and E. E. Harris J . Amer. Chem. Sac. 1951 73 2716 4519. 102 A. I. Wildschut Physica 1932 12 194. 103 A number of values are collected in " Dipole Moments " by L. G. Wesson J. 1935 987. York 1948. [The Technology Press (M.I.T.) Mass. 19481. 120 QUARTERLY REVIEWS Kinetic Methods.-(i) C. Paal lo* found that cis-ethylenes of type A are hydrogenated faster than the corresponding transforms when a supported palladium catalyst is used and this " rule " has sometimes been used in - cis --- trans (Pairs o f carboxylic acids) I .. I I Amt. of catalyst _it FIG. 3 deciding geometrical structures ; lo5 recently Z. Csuros has demon- strated its restricted validity.lo6 The rate of hydrogenation of both cis- and truns-isomers in the pres- ence of supported palladium and platinum catalysts does not vary in a linear manner with the amount of catalyst. This is shown for palladium in Fig. 3. Only when (under the experimental conditions) the amount of catalyst is in the region A does Paal's rule hold. Z. Csiiros suggests that the rule should be modified to the form in the presence of palladium cata- lysts the maximum rate of hydro- genation is always attained by the trans-compound when smaller concentrations of catalyst are used.For platinum catalysts this statement is true for cis- rather than for trans- isomers. (ii) Differences in the speeds of- esterification ascribed to steric effects were used to support the formulations of tiglic (XXIV) and angelic (XXV) acids.lO7 Tiglic acid with no methyl cis to the carboxyl group is esterified the faster. Me-C-H H-C-Me !I Me-C-CO,H (XXIV. ) II Me-C-CO ZH (XXV.) (iii) Dehydrohalogenation occurs more rapidly when the hydrogen and the halogen a.tom are juxtaposed trans about the double bond than when they are cis.lo8 Thus chlorofumaric acid trans-2-bromobut-2-ene truns-3- lo4 C. Paal and H. Schiedewitz Ber. 1927 60 1221 ; 1930 63 766 ; C. Paal H. Schiedewitz and K. Reuscher Ber. 1931 64 1521 ; M. S. Platonov J . Russ. Phys. Chem. Soc. 1929 61 1055 (Chem.Abs. 1933 27 539); M. S. Platonov Yu. A. Borgman and G. Ya. Salaman ibid. 1930 62 1975 (Chem. Abs. 1932 26 16). lo5 I. E. Muskat and B. Knapp Ber. 1931 64 779; A. K. Plisov and V. P. Golendeev Rep. U.S.S.R. Fat and Margarine Inst. 1935 No. 2 12 (Chem. Abs. 1936 30 4465) ; contrast however C. Weygand A. Werner and W. Lanzendorf J . pr. Chem. 1938 151 231. lo6 Z. Csiiros K. Zech and I. GBczy Hung. Chim. Acta 1941 1 1 (Chem. Abs. 1947 41 109); Z. Csuros Muegyetemi K6zleme'nyek 1947 110 (Chem. Abs. 1948 42 3726) ; idem Research 1951 4 52. lo' J. J. Sudborough and M. J. P. Davies J. 1909 95 975. lo8 S. J. Cristol N. L. Hause and J. S. Meek J . Amer. Chem.Soc. 1951 73 674; S. J. Miller and R. M. Noyes ibid. 1952 74 629. CROMBIE GEOMETRICAL ISOMERISM 121 chloroprop - 2 -en - 1 - 01 and cis- dichloroethylene are deh ydrohalogenat ed faster than their stereoisomers 2$ log ; Ill - b repid c slow I / \ B B X B / \ X An example in which the differences are small and complicated by inter- conversion during reaction has been noted.1l0 In other cases the expected acetylenic compound can only be obtained from the form allowing trans- eliminations and different products are obtained by treatment of its stereoisomer with alkali (cf.p. 13O).l1l (iv) Attempts have been made to decide the configuration of allylic halides by studying the reactivity of the halogen atorn.ll2 113 L. F. Hatch et ul. examined the rates of hydrolysis of cis- and trans-1 3-dichloropropene and the rates of the replacement reaction with potassium iodide and con- cluded that the low-boiling isomer is the trans-form.l12 Their interpreta- tions are in error as later work has shown that the latter is cis.114 115 Differences in the reaction rates of cis- and trans-1 3-dichlorobut-2-ene with OEt- and OH- are too small to be of value.lf5 (v) G.F. Wright 116 has found that cis-ethylenes are mercurated by alcoholic mercuric acetate faster than are the trans-isomers the reaction may be used diagnostically or for quantitative e~timati0n.l~' If the rate of addition is slow it may be catalysed by nitric acid or boron trifluoride. The excess of mercuric acetate is determined by thiosulphate or dithizone titration. J. Chatt 11* has suggested a mechanism for the stereospecific truns-addition of the reagent H R rH R - / \ H R' (cf. bromination) = ' I C-HgX / \ H R' N H R \ / R.0-C I + H+ 61-HgX / \ H R log A.Michael J. pr. Ghem. 1895 [ii] 52 289 ; J. Wislicenus and P. Schmidt Annulen 1900 313 210; G. Chavanne Compt. rend. 1912 154 776; L. F. Hatch and A. C. Moore J . Amer. Chem. SOC. 1944 66 285 ; R. E. Lute D. F. Hinkley and 122 QUARTERLY REVIEWS (vi) Another method of diagnosis and estimation consists of bromination of the ethylene addition of potassium iodide and titration with thiosul- phate.1199 120 121 Iodine is liberated faster from the erythro-dibromide derived from the trans-ethylene than from the threo-dibromide derived from the cis-ethylene R-CHBr-CHR'Br + 31- -+ RCH-CH-R' + I,- + 2Br- Resolvability into Enantiomorph.-In some cases because of restricted rotation caused by steric hindrance one of the geometrical isomers is resolv- able and the other not.Thus the compound assigned structure (XXVI) should be dissymmetric as coplanarity of the ring and carboxyl is prevented it is in fact resolvable and the structure is thus supported.122 Cl Me 1 CO,H Me (XXVI.) OMe U " E C 0 H Me I Me Me I Br (XXVII. ) Compound (XXVII) is so formulated because it is not resolvable models show that its geometrical isomer like (XXVI) would be dissymetric. A negative result from a resolution is always rather uncertain however and the enantiomorphs might be very easily racemised. Intramolecular Cyclisation.-This method-one of the earliest to be used-applies in its simplest form to pairs of geometrical isomers only one of which can be cyclised under mild conditions which do not cause stereo- mutation. Thus only one (and hence the cis-) o-aminostilbene (XXVIII) undergoes the Pschorr reaction to yield phenanthrene (XXIX).123 12* Similar reactions which have been applied to decide configuration are lactone R.H. Jordan J . Amer. Chem. Xoc. 1951,73,4647 ; L. J. Andrews and R. E. Kepner ibid. 1947 69 2230. 110 L. F. Hatch and P. S. Hudson ibid. 1950 72 2505. 111M. Fitzgibbon J . 1938 1218. 112 L. F. Hatch L. B. Gordon and J. J. RUSS J . Amar. Chern. Xoc. 1948 70 1093. 113 L. F. Hatch and G. B. Roberts ibid. 1946 68 1196. 114 L. F. Hatch and R. H. Perry ibid. 1949 71 3262. 115 L. J. Andrews and R. E. Kepner ibid. 1948 '70 3456 ; 1947 69 2230. 116 G. F. Wright ibid. 1935 57 1993. 117 W. H. Brown and G. F. Wright ibid. 1940 62 1991 ; M. H. Thomas and F. E. W. Wetmore ibid. 1941 63 136; T. Connor and G. F. Wright ibid.1946 68 256. 118 J. Chatt Chem. Reviews 1951 48 7. 119 R. T. Dillon W. G. Young and H. J. Lucas J . Amer. Chem. SOC. 1930 52 1953 ; W. G. Young D. Pressmann and C. D. Coryell ibid. 1939 61 1640. 120 C. E. Wilson and H. J. Lucas ibid. 1936 58 2396. 121 W. G. Young S. J. Cristol and T. Skei $bid. 1943 65 2099. 122R. Adams and C. W. Theobald ibid. p. 2383. 123 R. Stoermer Annalen 1915 409 13. 124 J. M. Gulland and C. J. Vernon J . 1928 1478; P. Ruggli et aE. Helv. Chim. Acta 1936 19 1288; 1937 20 37; 1941 24 173. CROMBIE GEOMETRICAL ISOMERISM 123 formation from suitable cis-hydro~y-acids,~~~ formation of anhydrides from cis-dibasic furans from cis-dibenzoylethylenes,126 dihydrofurans from cis-diols,127 128 carbostyrils from cis-aminocinnamic acids 123 indenones from cis-cinnamic a ~ i d s 1 ~ ~ and coumarins from coumarinic acids.If the sequence cis -+ cyclic -+ cis-compound can be carried out under mild conditions the case is further strengthened. H H (XXVIII. ) (XXIX.) Sometimes cyclisation occurs very readily as for carbostyril coumarin and some cases of lactol formation so it may not even be possible to isolate the cis-acid. As rather drastic methods are needed on other occasions stereomutation may complicate the issue though by adjustment of con- ditions a decisive result can often be obtained For instance A. Valette finds that whereas one butenediol (trans) gives a mixture of 2 Fi-dihydro- furan and crotonaldehyde the other (cis) gives only the dihydrofuran.128 The cyclisation of both cis- and trans-compounds (XXX) as shown gives the same mixture of products but other cyclisation conditions can be found which differentiate between them.130 131 95% EtOH clc- CCZ I1 + Ph*CO*CHCl*CHCl*COPh CPh Ph.CO*CH=CCl*COPh I1 HC1 PhC (XXX.) (XXXII.) \O/ (XXXI. ) The method of intermolecular cyclisation may also be applied to stereoisomeric pairs which are cyclised under the same conditions but give different products. C. F. Koelsch 132 assigned the structures o f some sub- stituted triphenylacrylic acids by indenone formation one geometrical isomer (XXXIII) gave (XXXIV) and the other gave (XXXV) (structures demonstrated by synthesis). II II I I I1 (XXXIII.) (XXXIV. ) (XXXV.) Ph-C-CO2H Ph-C-CO OC-C-Ph (R = C1 or OMe) lZ5 L. J. Haynes and E. R. H. Jones J. 1946 954 ; J. English and J. D. Gregory lZ6 R. E. Lutz W. G. Revely and V. R.Mattox ibid. 1941 63 3171 ; R. E. Lutz 12' R. Lespieau Compt. rend. 1907 144 144; J. Salkind Ber. 1923 56 187. 128 A. Valette Ann. Chim. 1948 [xi;] 3 644. 128 R. Stoermer and E. Laage Ber. 1917 50 981. 130 R. E. Lutz and F. N. Wilder J. Amer. Chem. SOC. 1934 56 1193. 131 R. E. Lutz A. H. Stuart F. N. Wilder and W. C. Connor ibid. 1937 59 2314. 138 C. F. Koelsch ibid. 1932 54 2487. J . Amer. Chem. SOC. 1947 69 2123. and C. E. McGinn ibid. 1942 64 2583. 124 QUARTERLY REVIEWS Another interesting example is the cyclisation of the substituted iso- itaconic (XXXVI) and itaconic (XXXVII) acids.133 Further examples are known.134 (XXXVI.) (XXXVII.) Stereospecific Additions.-Certain reagents add stereospecifically in a ‘‘ trans-” fashion and produce from a cis-ethylene a threo- and from a trans- trans-Addition to a trans-double bond.Y (2) - erythro cis-Addition to a trans-double bond. €3 A2 (xxxrx) V ( 2 ) - threo * The diagrams do not necessarily imply synchronous addition although this will explain Bromine addition is considered to proceed through an intermediate ,,Br+L or Br+ the process being completed by attack by Br- from the other retentions of configuration. RCH-AHR RCH=I=CHR side. 133 W. S. Johnson and A. Goldman J . Amer. Chem. SOC. 1944 66 1030 ; 1945 134 H. Stobbe Ber. 1904 37 1619 ; M. Ramart-Lucas and J. Hoch Ann. Chirn. 67 430. 1930 [XI 13 385. CROMBIE GEOMETRICAL ISOMERISM 125 ethylene an erythro-configuration. Conversely other reagents add " cis " and produce from a cis-ethylene an erythro- and from a trans-ethylene a threo-form. To illustrate this the process is shown diagramatically in two cases for a symmetrical addendum.If A = A then the (-j-)-erythro- becomes a meso-compound. When the addendum is unsymmetrical two (*)-pairs can if there is no orienting influence be produced in each case. If stereospecific addition occurs in one sense only an optically active product can result e.g. Y -+ (XXXVIII) or Z + (XXXIX). Conditions giving rise to this may be envisaged. Thus if the geometrical isomer must be adsorbed with a t least three of its atoms uniquely attached to certain corresponding atoms of a surface before reaction can occur attack may be directed in one sense only. The appropriate oriented adsorption and cis- attack is not possible in the sense required to produce the diastereoisomer A2 a2 As a minimum requirement adsorption on one centre other than the two reaction centres making three in a.11 is sufficient.If an olefinic compound can be converted by known stereospecific reactions into a saturated compound of known configuration then its own stereochemistry can be deduced. This is the basis of the method on p. 121. Reagents 1359 136 which on present evidence add largely or entirely trans are the halogens,* hypohalous acids performic acid the iodine-silver ben- zoate complex hydrogen peroxide in the presence of pervanadic or pertung- stic acid and 2 4-dinitrobenzenesulphenyl chloride. Alkaline perman- ganate and osmium tetroxide-hydrogen peroxide cause cis-additions. The finding that pent-2-en-4-yn-1-01 gives (-j)-erythropent-4-yne-l 2 3-trio1 (configuration demonstrated by degradation to known reference substance) with performic acid shows that the double bond is trans.135 The addition of a dienophile to a diene in the Diels-Alder reaction is stereospecific cis-addition being the rule.Consequently the diene can be converted into a cyclic compound the configuration of which may be 135 R. A. Raphael J. 1949 5.44. 136 D. Swern J . Amer. Chem. SOC. 1948 '70 1235; M. Mugdan and D. P. Young 13' R. E. Buckles W. E. Steinmetz and N. G. Wheeler J . Amer. Chem. SOC. 1950 * It has been shown that bromine will convert a (f)-dibromide into a meso- Further there are certain other irregularities in bromine J. 1949 2988. 72 2496. compound in certain cases.137 addition although explanations have been given. 126 QUARTERLY REVIEWS known or easily determined. towards geometrical isomers provides evidence of their configuration.as follows Moreover the reactivity of a dienophile Butediene reacts with cis- and trans-isomers (e.g. the cinnamic acids 138) cis cis trans trans Similarly cyclopentadiene gives stereoisonieric adducts with cis- and trans- dibenzoylethylene~,~~~ and a series of trans-acids when allowed to react with anthracene give products stereochemically different from those derived from the cis-anhydrides.140 H. L. Holmes et aZ.141 support a truns-configuration for the yellow form of B-benzoylacrylic acid because it gives a trans- adduct (XL) with 2 3-dimethylbutadiene. trans-Dienes of t,he type R*CH=CH*CH=CH react readily with dieneophiles but cis-forms react much more sluggishly (often with resinification) or not at all. This is exemplified by the reactivity of trans-piperylene 142 and trans-1 -cyanobuta-1 3-diene 143 towards maleic anhydride or acrylic acid under conditions such that the cis-isomers are un- affected.Separations of cis- and trans-isomers may thus be effected. The differences are attributed to hindrance in the cis-form to the approach of the dienophile to the ends of the conjugated system. The failure of natural pyrethrolone to give adducts with dieno- philes supports the trans-structure indicated by other evidence .144 The stereochemistry of the addition of maleic anhydride to cis- and trans- piperylenes is shown be10w.l~~ 6 t- COPh (xL; A = co,‘EI) Me /-\-Me 1-1 ‘0’ oc co Rapid + ‘O/ I I I=====[ oc co ‘0’ ‘0’ oc co trans cis In the series CHR=CH*CH=CHR’ similar differences in reactivity axe trans-trans-1 4-Diphenylbutadiene reacts with maleic anhydride to shown.138 K. Alder H. Vagt and W. Vogt AnmLm 1949 565 135. 139 R. Adams and M. H. Gold J . Amer. Chem SOC. 1940 62 56. 140 W. E. Bachman and L. B. Scott ibid. 1948 70 1458. 141 H. L. Holmes R. M. Husband C. C. Lee and P. Kawulka ibid. p. 141. 142 R. F. Robey C. E. Morrell and H. K. Wiese ibid. 1941 63 627 ; R. F. Robey 143 H. R. Snyder J. M. Stewart and R. L. Meyers ibid. 1949 71 1055 ; H. R. 144 L. Crombie S. H. Harper and D. Thompson J. 1951 2906. 145 D. Craig J . Amer. Chem. SOC. 1950 72 1678. Science 1942 96 47 ; D. Craig J. Amer. Chem. SOC. 1943 65 1006. Snyder and G. I Poos ibid. 1950 72 4104. CROMBIE GEOMETRICAL ISOMERISM 127 yield (on hydrolysis) (XLI). The trans-cis- (XLII) and the cis-cis-isomer (XLIII) do not react under conditions which avoid inversion to the trans- trans-compound.146 P h - n - P h oc co I=I ' O / + Ph I I Ph C0,H C0,H (XLI.) p) Ph Ph (XLII.) C L - P h Ph (XLIII. ) As expected trans-trans-l-methyl-4-phenylbutadiene and 1 4-dimethyl- butadiene are more reactive than their cis-trans-isomers though adducts can be obtained from both 146 147 t C Ph C0,H C0,H Ph.CH=CH*CH=CHMe + 1- I -1-1" I 3 2 Me Conversion into Compounds of Known Configuration.-If suitable geo- metrical isomers of well-established configuration are available for reference this method is highly attractive though it depends on use of reactions which do not cause stereomutation. Samples of crotyl alcohol,sg psnta-2 4- dien-1-01,~~~ and hex-3-en-1-01 31 have been shown to be cis. trans and trans respectively because they can be derived from the appropriate known acids by reduction with lithium aluminium hydride.Similarly the lower- boiling isomer of 1 3-dichloroprop-2-ene is cis as it gives the known cis-l-chloro- prop-l-ene on reduction with this reagent.l15 The configuration of the ketone Pr*CH=CH*COMe as trans was demon- strated by deriving it from trans-hex-2-enoic acid via the acid chloride and methylzinc iodide.148 A well-known example is the conversion of one form of 3 3 3-trichlorocrotonic acid on the one hand into (trans)-fumaric acid and on the other into one of the crotonic acids which is therefore trans.149 There are other examples scattered in the literature though the method has not been much used. Certain reactions described in the next section are known to give almost exclusively cis- or trans-isomers and the configuration then needs only confirmation.Methods €or the Preparation of Geometrical Isomers The required stereoisomer may be synthesised by a stereospecific process or separated from the mixture of cis- and trans-isomers made either by a non-stereospecific synthesis or by stereomutation of the other isomer. In 1 4 6 K. A. Alder and M. Schumachsr Annalen 1951 571 87 122. 147 K. Alder and W. Vogt ibid. p. 137. 148 E. N. Eccott and R. P. Linstsad J. 1930 905. 1 4 9 K. von Auwers and H. Wissebach Bey. 1923 56 715. 128 QUARTERLY REVIEWS the following section some generalisations have been attempted in certain cases these must be accepted with reserve because of the fragmentary nature of the evidence. Results obtained must be evaluated with the following points in mind (i) that stereomutation may have occurred under the conditions of reaction or isolation (ii) that one of the stereoisomers may have been eliminated during the purification or not specifically looked for (iii) that one of the isomers may have reacted further e.g.a cis-unsaturated hydroxy-acid may have lactonised and (iv) that the reaction conditions may be critical and the observations not general. Only detailed and extensive investiga- tions can settle these points. Addition of Reagents to Acetylenes.-The hydrogenation of acetylenes with a palladium catalyst gives good yields of olefins if continued until one mol. of gas is absorbed.l5O9 151 Sometimes hydrogenation stops spontan- eously here if the catalyst is weak but usually the ethylene absorbs hydrogen as fast as does the acetylene.Ordinary platinum catalysts are said to give mixtures containing acetylenic ethylenic and saturated compounds though there have been favourable reports of their spe~ificity.~~ C. Paal 152 and other early investigators found that the ethylenes pro- duced were essentially cis though some succeeding workers reported stereo- isomeric mixtures. M. Bourguel attributed some of the confusion to the instability of the cis-products and impurities in the protecting agents for the ~ata1ysts.l~~ To combat this he introduced a colloidal palladium-starch catalyst though nowadays palladium on such supports as calcium strontium or barium carbonate barium sulphate charcoal or polyvinyl alcohol is used. All these give essentially cis-ethylenes as do Raney n i ~ k e l l ~ * - l ~ ~ iron powder and platinum poisoned with carbon monoxide.156 Careful examination by physical methods (Raman 157 and infra-red freezing-point methods 59) has shown that in many hydrogenations carried out in the usual way the cis-isomer is contaminated with some of the truns- Generally this is not serious in preparative work except when high purity is essential.There are indications that temperature 160 and speed of hydrogenation 161 may influence the configuration of the product. It has been suggested that catalytic hydrogenations involve simultaneous 150 K. N. Campbell and B. K. Campbell Chem. Reviews 1942 31 77. 151 A. W. Johnson " Acetylenic Compounds " Vols. I & 11 Arnold London 1946 152 C. Pad Ber. 1909 42 3930. 153 M. Bourguel Bull.SOC. chim. 1929 [iv] 45 1067. 154 G. Dupont ibid. 1936 [v] 3 1030. 155K. N. Campbell and L. T. Eby J . Amer. Chem. SOC. 1941 63 216 2683. 156 K. Ahmad F. H. Bumpus and F. M. Strong ibid. 1948 70 3391. 15' G. Smets Acad. roy. Belg. Classe Sci. 1947 21 3. 158 F. Sondheimer J. 1950 877. 159 A. L. Henne and K. W. Greenlee J. Amer. Chem. Soc. 1943 65 2020. 159aD. E. Ames and R. E. Bowman J. 1952 677. 160 S. Takei Ber. 1940 73 950. 161 E. Ott and F. Schiirmann Ber. 1928 61 2119. p. 1950. CROMBIE GEOMETRICAL ISOMERISM 129 &-addition of hydrogen from the catalyst to the acetylene.la2 la3 On such a theory no trans-isomer is to be expected and its occurrence is ascribed to secondary isomerisation. During the catalytic hydrogenation a concomitant rupture of the C-H bond of the olcfin occurs and the radical subsequently reunites with another hydrogen atom formed by dissociation of a hydrogen molecule to give either the original olefin or its stereoisomer la2 H H H H X The preferential reduction of the acetylene is ascribed to its preferential adsorption on the catalyst to the exclusion of the ethylene.This method is the most valuable general one available for the synthesis of cis-type-A ethylenes of many kinds. In contrast with catalytic hydrogenation " chemical " methods often yield mainly trans-i~omers.1~0 151 A liquid-ammonia solution of sodium reduces acetylenic hydrocarbons and alcohols (and sometimes the sodium salt of an acetylenic acid la4) to stereochemically pure trans-eth~1enes.l~~ A mechanism has been suggested.165 Zinc and acetic acid have been used to reduce acetylenic acids to trans-ethylenic acids 166 though oleic acid (cis) from stearolic acid has been claimed to result from its use in the presence of titanous ch10ride.l~' Zinc and ammonia react with phenylpropiolic acid to yield trans-cinnamic acid laS but when ammonium chloride is present some cis-isomer is obtained (10%).E. Ott et aZ.la9 have studied this reaction using various metals in the presence of ammonia and ammonium chloride. Manganese gives mainly cis-cinnamic acid (14% trans). Metals more noble than zinc used as couples yield predominantly trans (Zn/Ag 24% cis Zn/Cu 37% cis). Metals more electropositive than zinc e.g. magnesium barium or sodium give mixtures of saturated and unchanged materials. Hexa-amminocalcium Ca(NH,), 170 and lithium aluminium hydride 171 reduce disubstituted acetylenes to trans-olefins.Electrolytic reduction of acetylenes a t spongy nickel cathodes gives cis-ethylenes in good yields but lead lead amalgam and platinum cathodes are not effecti~e.l~2~ l'j24. Farkas and L. Farkas Trans. Paraday Xoc. 1937 33 837. 163 R. P. Linstead W. E. Doering S. B. Davies P. Levine and R. R. Whetstone l'j4D. R. Howton and R. H. Davis J. Org. Chem. 1951 16 1405. l'j5 K. W. Greenlee and W. C. Fernelius J . Amer. Chem. Xoc. 1942 64 2505. 166 L. Aronstein and A. F. Holleman Ber. 1889 22 11 81 ; A. Holt Ber. 1892 25 961 ; A. Gonzalez Anal. Pis. Quim. 1926 24 156 (Chem. A h . 1926 20 2310). 167 G. M. Robinson and R. Robinson J. 1925 127 175. 168 E. Fischer Annalen 1912 394 361. 169 E. Ott V. Barth and 0. Glemser Ber. 1934 67 1669.170 K. N. Campbell and J. P. Mc.Dermott J. Amer. Chem. Xoc. 1945 67 282. 171 V. M. Mitchovitch and M. L. Mihailovic Compt. rend. 1950 251 1238. 1720K. N. Campboll and E. E. Young J. Amer. Chem. SOC. 1943 65 965. J . Amer. Chem. SOC. 1942 64 1985. 1 30 QUARTERLY REVIEWS Early investigators considered that halogens or halogen acids added cis to a triple bond but later it was suggested that truns-addition was normaL2 Considerations of space prevent adequate discussion but it seems that the experimental facts do not warrant a generalised statement and a thorough investigation is needed. Often mixtures of stereoisomers are obtained but sometimes only one isomer has been isolated. It is unfortunate that many of the examples chosen give rather easily stereomutable products. An excellent summary of the information available on the addition of halogens and halogen acids to acetylenic acids and alcohols is a~ai1able.l~~ Addition of hydrogen bromide or hydrogen iodide to long-chain acetylenic acids followed by reductive elimination of the halogen with zinc and acid or sodium and alcohols is said to yield trans-isomers.This indicates that the halogen acid has added trans.l7Zb Elimination Reactions.-The elimination of bromine from a diastereo- isomeric dibromide with metallic zinc or iodide ions proceeds by trans- stereospecific reaction. An E2-type elimination mechanism has been suggested with the transition state as shown in (XLIV).173 The literature contains a number of other elimina- tion reactions which yield geometrical isomers stereospecifically when one diastereoisomer is used.Tiglic acid (XLV) can be converted into its pure stereo- isomer angelic acid (XLVI) by (trans-)addition of bromine (XLIV-) elimination of hydrogen bromide with methanolic potassium hydroxide and removal of the vinyl halogen with sodium amalgam.174 If trans-elimination and replacement a t the trigonal carbon atom without inversion as seems reasonable are allowed then the reaction is explained thus cyr >TTc< Br t I- (XLV) (XLVI) Sodium in liquid ammonia may also be used to replace such vinyl halogen at0ms.l7~ The a- and /?-stilbene dichlorides give cis- and trans-monochlorostilbene respectively on treatment with alcoholic potassium hydr~xide,~'~ and meso- 172b F. Rrafft and R. Seldes Ber. 1900 33 3571 ; A. Arnaud and S. Posternack Comnpt. rend. 1910 150 1130; 1916 162 944; C.Collaud Helv. Chirn. Acta 1943 26 1064. 173 S. Winstein D. Pressman and W. G. Young J. Amer. Chem. Xoc. 1939 61 1645. 17* H. P. Kaufmann and K. Kuchler Ber. 1937 70 915 ; R. E. Buckles and G. V. Mack J. Org. Chem. 1950 15 680. 1 7 5 M. C. Hoff K. W. Greenlee and C. E. Boord J. Amer. Chem. SOC. 1951 73 3329. 176 T. W. J. Taylor and A. R. Murray J. 1938 2078. CROMBIE GEOMETRICAL ISOMERISM 131 cc cc' 4 4'-tetrachlorostilbene with the same reagent gives only one of the trichlorostilbenes (probably cis) .17 The elimination of the elements of hydrogen bromide and carbon dioxide from ccp-dibromo-acids occurs stereospecifically and the evidence supports trans-elimination. An illuminating series is trans- m.p. 256" ~ (),,=CHBr -NaOAc + m.p. 203-204" alc.NaOH+ ,((E;CH -+ m.p.30° cis very rapid cis- m.p. 137" -+ m.p. 154-155" -+ m.p. 86" trans ____+ alc.NaOH L ) C H = C H B r veryslow ' ~ 0 2 ~ Similarly addition of hydrogen iodide to trans-a-ethylcrotonic acid yields a-ethyl-p-iodobutyric acid which on treatment with sodium carbonate gives the trans-isomer as expected 179 180 2cx M&J..co2H. $.*.. C02H Na*coJEt 1.e. Me - I I I Me H I i Me H trans trans The cis-acid gives the cis-hydrocarbon. Another case is the bromination of trans-crotonic acid followed by treatment with sodium carbonate which yields the expected cis-propenyl bromide. 6 9 9 181a In certain cases different dehydrohalogenating agents give ethylenes of different stereochemistry from one diastereoisomer.181b This may be due to stereomutation or change of reaction mechanism.The stereochemistry of the Boord olefin synthesis is not certain as there seems to be no information on pure diastereoisomers. As would be expected there are reports of the isolation of cis-trans-mixtures of olefins,lsO 182 e.g. Me,CH*CH( 0Et)CHMeBr __+ Me,CHCH=CHMe Zn threo + erythro 177 E. E. Fleck J . Org. Chem. 1947 12 708. 178W. Davies B. M. Holmes and J. F. Kefford J. 1939 357. 179 H. J. Lucas and A. N. Prater J . Arner. Chem. SOC. 1937 59 1682. 180 M. L. Sherrill and E. S . Matlock ibid. p. 2135. 1*1a J. Wislicenus and H. Langbein Annalen 1888,248 318 ; G. Chavanne Cornpt. 181* L. N. Owen and M. U. S. Sultanbawa J . 1949 3105. 182 C. G. Schmidt and C. E. Boord J . Arner. Chem. SOC. 1932 54 751 ; C. R. Noller cis f trans _ _ _ _ ~ _ ~ _ _ _ _ _ _ _ - ~ ~ - . _ _ _ rend.1914 158 1698. and R. A. Bannerdot ibid. 1934 56 1563. K 132 QUARTERLY REVIEWS The thermal elimination involved in the Tschugaev reaction is considered Thus the erythro-racemate yields largely cis- and to be a cis-elimination. the threo-racemate largely truns-oleh. 68 SMe SMe When elimination occurs at a pair of adjacent carbon atoms which are not involved in diastereoisonier formation the product if capable of geo- metrical isomerism should be a mixture of cis- and trans-isomers. Thus when tetrabromoethane is kreated with zinc in ethanol a mixture of cis- and trans-dibromoethylene is obtained. 183 Br /I 7 H Br I Br Br 1 Br Br ". 1 "H Br"" ~ '.*Br 7 Br Br ! Br \.v \:/ \\/ J H J H 4 I$,*= ~ '*br H 4 (XLVII.) (XLVIII.) (XLIX.) Three conformations are possible by rotating the rear (dotted bonds) carbon atom.trans-Dehalogenation being assumed two conformations (XLVII) and (XLVIII) will give the cis- and (XLIX) the trans-form though in the latter case two eliminations are possible. A well-known case of this type of elimination is the preparation of cis-crotonic acid in which a mixture of the two stereoisomeric chloro-esters is obtained and separated MeCO-CH,-CO,Et + Me*CC1,.CH,.CO2Et -+ Nrt- Me-CCl=CHCO ,E t + MeCH-CHCO tH Hg Cis 8( + 4 the ,!?-chloro-cis-crotonic ester is formed in only poor yield.8gs In some of the eliminations of this type conformational specificity has been inferred. Thus only trans-CMe,*CH=CHCl is obtained when 1 2- dichloro-3 3-dimethylbutane is treated with alkali. 184 Also alkaline 183 A. Pongrata Ber. 1936 69 1267. 184 G.G. Eke N. C. Cooke and F. C. Whitmore J . Amer. Chem. SOC. 1950 72 1511. CROMBIE GEOMETRICAL ISOMERISM 133 dehydrochlorination of Me*C'Cl,*CH,Cl will give trans-MeCCl-CHCl 185 whereas when Me*CCl,*CHCl*CO,H is treated with a base cis-Me*CCl=CHCl is obtained.ls6 The evidence here is unsatisfactory but if such cases can be substantiated then the indication is that certain conformations are more favourable to elimination reactions than others the effect should be reflected in the relative proportions of cis- and trans-isomers in other reactions. With a wide variety of reagents cis-trans-mixtures have been isolated when hydroxy-compounds are dehydrated ; lSo ls8 lB9 occasionally only a truns-isomer has been isolated 7O9 lgO but this may be due to stereo- mutation or to experimental difficulties.Many of the hydroxy-compounds have been derived from the Reformatski reaction or the reaction of aldehydes or ketones with Grignard reagents. They are of two types racemic non- diastereoisomers and mixtures of racemic diastereoisomers. The isolation of mixtures of geometrical isomers from both classes is to be expected even if the dehydration is stereospecific. The latter point does not seem to have been tested-pure diastereoisomers would be needed for this in any case dehydration a t high temperature over a contact catalyst would be expected to proceed by a different mechanism from the lower-temperature chemical elimination. It is clear from the literature that mixtures of stereoisomers are frequently obtained and that the type of dehydration catalyst and the conditions influence the relative proportions of each.lsg lY1 This may be due to stereomutation or elimination from different preferred orientations in different cases.(bg + R*CH=CH*[CH,],.OH A H (L.) The ring scission of 2-alkyl-3-chlorotetrahydropyrans (L) may be classed The cyclic compound exists in cis- and 185 E. Huntress and F. Sanchez-Nieva ibid. 1948 70 2813 ; cf. 1950 72 3459; 186 W. Szenic and R. Taggesell Ber. 1895 28 2665. 18' K. von Auwers Annalen 1923 432 46. 18* R. Stoermer F. Grimm and E. Laage Ber. 1917 50 959; W. Chalmers Trans. Roy.Soc. Canada 1928,22,111,69 ; G. A. R. Kon R. P. Linstead and J. M. Wright J . 1934 599 ; H. Burton and C . W. Shoppee J . 1935 1156 ; H. van Risseghem Bull. Xoc. chim. Belg. 1938 47 4 7 ; H. Okazaki J . Chem. Soc. Japan 1942 63 368; R.C. Fuson and P. L. Southwick J . Amer. Chem. SOC. 1944 66 679 ; M. Rubin A. Kozlowski and M. R. Salmon ibid. 1945 67 192 ; D. T. Mowry and A. G. ROSSOW ibid. 1945 67 926; R. Neher and K. Miescher Helv. Chim. Acta 1946 29 449 ; S. Miron and G. H. Richter J . Amer. Chem. SOC. 1949 71 453 ; R. E. Buckles and G. V. Mock J . Org. Chem. 1950 15 680. here as an elimination reaction. 1951 73 1843. 18D S. H. Harper and J. F. Oughton Chem. and Ind. 1950 575. 190L. F. Fieser and M. Fieser Experientiu 1948 4 2 8 5 ; R. E. Lutz and R. S . Murphey J . Amer. Chem. Soc. 1949 71 479. lol R. Kuhn and M. Hoffer Ber. 1932 65 651 ; C. F. F. Bergmann M. Weizmann E. Dimant J. Patai and J. Szmuskowicz J . Amer. Chem. SOC. 1948 70 1612; C. Cauquil M. H. Barrera and G. Tufpin Compt. rend. 1950 231 779. 134 QUARTERLY REVIEWS trans-forms and on treatment with sodium each gives the same trans- ethylenic alcohol.But in the case of 2-alkyl-3-chlorotetrahydrofurans (alkyl = Me and Et) the trans-isomer gives a largely trans-alk-3-en-1-01 and the cis-isomer a mixture of cis- and truns-alcohols (R = Et ; 55% cis). A mechanism for this has been suggested.88 A method has been developed for the synthesis of geometrical isomers as follows R*CO,Me + R’*CO,Me __+ R*COCHR’*OH + R‘*COCHR*OH Na Pondorff + or Raney Ni +Ha HO*CHR*CHR‘*OH CHRBrCHR’Br -+ CHR==CHR’ HBr Zn HZSOA HOAC erythro - threo . cis on0 inversion t rans-elimn. +. erythro ---+ trans sepd. This acyloin method has been applied to the synthesis of long-chain acids of the type R*[CH2],*C€I=CH*[CH2]n*C02H (where R = alkyl or [CH,I,*CO,H) in both the cis- and the truns-forms the erythro- and the threo-glycols being separated by cry~tallisation.~~~ One step involves a single inversion this can also be made use of in passing from a cis- t o a truns-olefin as follows 120~193 Me H \ / epoxylation Ac c=c Me*CH-CH*Me - H,O+ I I HBr H / \ Me O H O H Me Me \ / Br I Zn Me*CH-CH*Me ---+ c=c H / \ I (trans-elimn.) Br H * One inversion.It has been mentioned above that zinc debromination gives essentially trans-elimination. It has been pointed out however that if a pure dibromide is treated with zinc and the resulting hydrocarbon is then brominated a proportion (3-20%) of the unexpected diastereoisomer is formed.121 lg4 The amount increases with chain length. There is some evidence that the partial inversion occurs at the zinc debromination stage but it is probably a secondary effect due to the zinc halide formed.121 Careful scrutiny of the products of the acyloin method by physical examination is therefore desirable since limitations reminiscent of the route concerning semi- hydrogenation of acetylenes may be involved.1g2 L. Ruzicka P1. A. Plattner and W. Widmer Helv. Chim. Acta 1942 25 604 1086 ; P. Baudart Bull. SOC. chim. 1946,13 87 ; D. E. Ames and R. E. Bowman J. 1951 1079 1087; B. W. Boughton D. E. Ames and R. E. Bowman J. 1952 671. lg3 W. G. Young Z. Jasaitas and L. Levanas J . Amer. Chem. SOC. 1937 59 403; S. Winstein and H. J. Lucas ibid. 1939 61 1576 1581. lg4 H. van Risseghem Bull. SOC. chim. Belg. 1938 47 194. CROMBIE GEOMETRICAL ISOMERISM 135 Condensation Reactions.-The orp-unsaturated acid isolated in the Doebner reaction is generally considered to be trans.195 The same assign- ment has been made for the &acid derived from the similar triethanolamine- catalysed condensation.31 However the oily and neutral materials are normally discarded in these reactions and the presence of cis-acid or a product of subsequent reaction is possible. It has been found that depen- dent on conditions 6-17 yo of hexenolactone-which might have been derived from a cis-precursor-can be isolated from preparations of sorbic acid by the Doebner method.lg6 The condensation of benzaldehyde and ketones is recorded as yielding a trans-a#?-unsaturated ketone.197 E. N. Eccott and R. P. Linstead 148 have described an unusual case. When n-butaldehyde and acetone are condensed without check to the exothermic reaction a cis-ketone PPCH=CH*COMe is obtained.But if the reaction is controlled a hydroxy- acid is isolated which on dehydration with various reagents gives the trans- isomer. The two ketones are resistant to configurational change but on Ponndorf-Meerwein reduction yield the same a l ~ o h o l . ~ ~ ~ The Perkin reaction,124 lg97 200 as applied t o the synthesis of substituted stilbenes gives a product witch the aryl groups cis to each other e.g. o-N0,C,H4 Ph O-NO Z'CsH4 Ph CuCrO \ / / \ c___ c-c \ / / \ c=c H CO,H H H By use of trans-o-nitrocinnamaldehyde a trans-cis-diene has been ob- tained. 200* The configuration of the products from the Kuhn-Winterstein modification of the reaction is less certain though trans-ethylenes have been isolated. 200 Scission of Ring Systems containing Double Bonds.-Under suitable conditions ring systems containing olefinic linkages which must be con- strained in the cis-configuration in an approximately planar ring can be opened with retention of configuration.Maleic anhydride or a-methyl- aconitic anhydride may be hydrated to the cis-acids and unsaturated lg5 W. H. Lauer and W. J. Gender J . Amer. Chem. SOC. 1945 57 1171. lg6 R. Joly and G. Amiard Bull. SOC. chim. 1947 139. IB7 M. T. Bogert and D. Davidson J . Amer. Chem. SOC. 1932 32 334 ; but cf. B. Koechlin and T. Reichstein Helw. Chim. Actu 1944 27 549. ls8C. L. Arcus and J. Kenyon J. 1938 698. T. W. J. Taylor and C. E. J. Crawford J. 1934 1130 ; E. D. Amstutz and E. R. Spitzmiller J . Amer. Chem. SOC. 1943 65 367 ; E. F. M. Stephenson J.1949 655. G. B. Bachman and H. I. Hoaglin J . Org. Chem. 1943 8 300. * Recent work (S. Ishraelashvili Y. Gottlieb M. Imber and A. Habas J . Org. Chem. 1951 16 1519) indicates that contrary to expectation a number of dienes with unsubst>ituted aryl groups which have been thus prepared have entirely trans-trans configurations. A methoxynaphthylacraldehyde however gave a mixture of trans- t runs - and trans -cis -isomers. 136 QUARTERLY REVIEWS lactones can also yield the cis-acids in this way trans-cis-muconic half ester has been prepared. 201 NaOMe t C MeO,C*CH,*CH*CH=CH Me0 ,C.CH=CH*CH=CHCO,Me I I o--co Peracetic acid oxidation of an o-quinoneJ202 pheno1,203 204 or catechol 204 yields cis-cis-muconic acid p-Benzoquinone is oxidised by chloric acid to maleic acid 205 0 0 The ring scission of 2-ethoxy-5 6-&hydro-%pyran however seems to give trans-penta-2 4-dienal 144 (probably because of subsequent stereomutation by H+).Nitric acid oxidation of substituted furans has been used repeatedly to prepare cis-aroylethylenesY2O6 e.g. MeC-CBr Me Br II I/ PhC CPli 'O/ \ / / \ c=c Ph*CO' COPh Thiophen derivatives may also be oxidised to similar compounds. *07 Rearrangement Reactions.-The conversion of secondary allylic alcohols into bromides by phosphorus tribromide gives under the ordinary pre- parative conditions an equilibrium mixture containing mainly the primary bromide HO*CHRCH=CH -+ Br*CHR*CH=CH + CHR=CH,*CH,Rr 201 J. A. Elvidge R. P. Linstead P. Sims and B. A. Orkin J. 1950 2235. 202 J. Boeseken and G. Sloof Proc. Acad. Sci. Amsterdam 1929 32 1043.,03 J. Boeseken and R. Engelberts ibid. 1931 34 1292. 204 J. Boeseken ibid. 1932 35 750. 205 A. Kekul6 and 0. Strccker Annulen 1884 223 170. 206 R. E. Liitz and C. $1. McGwin J. Amer. Chem. SOC. 1942 64 2553. 207 A. Angeli and G. Ciamician Ber. 1891 24 74 1347. CROMBIE GEOMETRICAL ISOMERISM 137 the latter has the truns-configuration. 208 Acid-catalysed anionotropic rearrangement of cis-phenylpropenylcarbinol (like that of the trans-isomer) gives largely or exclusively trans-methylstyrylcarbinol. But cis-propenyl- vinylcarbinol unlike its trans-isomer rearranges with retention of con- figuration to cis- butadienylmethylcarbino1.69 An explanation involving a pseudo-cyclic intermediate (LI) annelised because of " n hydrogen bonding " has been suggested 69 cis The conditions and mechanisms by which a compound such as cis-crotyl chloride (which rearranges less readily than does the corresponding bromide) retains and loses configuration when it reacts raises interesting possibilities but the subject has not yet been investigated.The rearrangement of (LII) with sodium methoxide in ether yields a trans-by-ester with a smaller amount of ketone,209 but the similar rearrange- ment of (LIII) gives a mixture of cis- and trans-esters.210 NaOMe (LII.) trans MeCHBr-CMeBrCOMe Me*CH=CMeCH ,*CO ,Me C,H,*CMeBr*CO-CH,Br - C,H,CMe-CH*CO ,Me (LIII.) cis -t trans The carbinol (LIV) rearranges to yield a mixture of cis- and truns- isomers (LV). 212 p-Me*C,H,CPh( OH)*C+zCPh p-Me-C,H:,.CPh:CH*COPh (LIV.) (LV.) Replacement Reactions at a Trigonal Carbon Atom.-A. R. Olson 26 suggested that the replacement reaction e.g.between iodide ion and iodine bonded to a trigonal carbon atom proceeds with inversion at each replace- ment. This view does not accord with later evidence it ha,s been demon- strated that some one hundred exchanges occur for each inversion of con- 208M. Bouis Ann. Chim. 1928 [XI 9 403; s. Winstein and W. G. Young 209 R. B. Wagner ibid. 1949 71 3214. 210 R. I3. Wagner and J. A. Moore ibid. 1950 72 074. a l a M. Badosche Bull. SOC. chim. 1928 43 337. J . Amer. Ghem. SOC. 1936 58 104. 138 QUARTERLY REVIEWS figuration. 25 It is also considered that metalation and carboxylation or replacement of the metal by hydrogen does not involve inversion 69 101 Me Br Me Li c-c + c-c \ / L i \ / / \ / \ H H H H cis cis p -ClC ,H Br \ / Li then CO Me C0,H co 2 ‘\ / / \ - C-C H H cis c-c .‘c=c’ P h and CuCrO / \ Ph CiS Ph ’ . \ P h cis (LVI.) (LVII.) Carboxylation of the Li derivative of cis-(LVI) gives the expected acid with the carboxyl and the p-chlorophenacyl group in cis-relation decarboxyla- tion with copper chromite yields cis- (LVII) again indicating absence of inversion.lo1 A number of other examples of decarboxylation at a trigonal carbon atom are known to proceed with retention of configuration.124 176 l80 Reductive elimination of a vinyl halogen atom by sodium amalgam or sodium in liquid ammonia is also best interpreted in these term~.~74,~753 l 8 7 213 A. N. Nesmeyanov et aZ. record that reactions of the type (Cl*CH=CH),Sb + HgC1 + Cl*CH=CH*HgCI SnCl (Cl-CH-CH) ,SnCl (Cl.CH=CH) ,Hg proceed with retention of configuration a t all stages when either stereoisomer is used.214 The Meerwein 215 reaction yields in most recorded examples trans- stilbenes though this may be due to the fact that the starting materials are trans PhCH=CH*CO,H + Ph*N,Cl --+ Ph*CH=CHPh + CO + N + HCl It has been found that the two isomers of p-Br*C,H,*CPh=CH*CO,H react with diazotised aniline to give the same isomer.More information on the mechanism is needed. Miscellaneous Reactions .-The re act ion of unsymmetrical diary le t h ylene s with oxalyl chloride which is formally a t least a substitution reaction 213 A. Michael and 0. Schulthess J . pr. Chem. 1892 [ii] 46 ”236. 214 A. N. Nesmeyanov A. E. Borisov and A. N. Abramova Izvest. Akad. Nauk. S.S.S.R. Otdel. Khim. Nauk. 1947,647 (Chem. Abs. 1948 42 6316) ; A.N. Nesmeyanov and A. E. Borisov Doklady Akad. Nauk. S.S.S.R. 1948 60 67 (Chem. Abs. 1949 43 560) ; A. N. Nesmeyanov A. E. Borisov and A. N. Gus’kova Bull. Acad. Sci. U.R.S.S. Clmse Sci. chim. 1945 639 (Chem. Abs. 1946 40 4659). 215 F. Bergmann J. Weizman and D. Schapiro J . Org. Chern. 1944 9 408. CROMBIE GEOMETRICAL ISOMERISM 139 generally gives mixtures of cis- and trans-/l/l-diarylacrylic acids though certain examples have only yielded one isomer. 216 Ph C-CH + (COCl) + \ / p-Br*C,H Ph Ph C-CHCO ,H N a $0 \ C-CHCOCl --+ \ / p -Br*C,H A well-known preparative method for trans-diaroylethylenes is the Friedel-Crafts reaction between fumaroyl chloride and a hydrocarbon 217 a similar reaction between aromatic hydrocarbons and maleic anhydride yields trans-aroylacrylic acids.Catalytic hydrogenation of aromatic or/l-diketones over a platinum catalyst in methanol gives mainly a cis-stilbenediol but if the hydrogenation is continued the conditions cause stereomutation to the trans-diol.218 In another example if the hydrogenation is conducted in methanol the cis- form is obtained but if in light petroleum the product is the trans-i~omer.~~~ Stereomutation (cf. p. 106).-A wide variety of catalysts for thermal equilibration of geometrical isomers has been used and a selection is as follows nitrous acid (or Poutet's reagent mercury and nitric acid),220 sulphur,221 selenium black,222 platinum black alkali metals halogen sulphuric acid 224 phosphoric a ~ i d 2 ~ ~ red phosphorus and water 225 ammonia primary and secondary amines,226 and tetranitr~methane.~~~ Recent patents z2* recommend inter alia alkanethiols thiuram disulphides salts of dithio-acids thiazoles thiazolines and thioamides.For photochemical equilibrations halogens-usually iodine-are commonly used as catalysts. The choice of solvent influences the equilibrium in some cases. 216 F. Bergmann M. Weizmann E. Dimant J. Patai and J. Szmuskowicz J . Amer. Chem. SOC. 1948 '70 1612. 217 J. B. Conant and R. E. Lutz ibid. 1923 45 1303. 218 R. C. Fuson S. L. Scott E. C. Horning and C. H. McKeever ibid. 1940 62 219 R. C. Fuson C. H. McKeever and L. C. Behr ibid. 1941 63 2648. 220 H. N. Griffiths and T. I?. Hilditch J. 1932 2315 ; J. C. Smith J. 1939 974. 221 G. Rankoff Ber. 1931 64 619 ; E. Rosemann Chenz. Umschau Fette Ole Waschse Harz 1932 39 220 (Chem. Abs. 1933 27 702). 222 G.Rankov Annuaire univ. Sofca Facultk phys. math. 1941 38 Livre 2 133 (Chem. Abs. 1948 42 2453) ; J. P. Kass and G. 0. Burr J. Amer. Chem. SOC. 1939 61 1062 ; D. Swern E. F. Jordan and H. B. Knight ibid. 1946 68 1673. 223 F. H. Malpress Nature 1946 158 790 ; R. E. Lutz and R. H. Jordan J . Amer. Chem. SOC. 1950 '72 4090. 224 P. Ruggli and H. Zaeslin Helv. Chim. Acta 1935 18 853. 2 2 5 G. Rankov Ber. 1936 69 1231. 2 2 6 G. R. Clemo and S. B. Graham J. 1930 213. 227 H. P. Kaufmann Ber. 1942 75 1201. 228 U.S.P. 2,404,103/1946 ; 2,414,066/1947 ; 2,454,385/194 (Chem. Ah. 1946 209. 40 6500; 1947 41 2437; 1949 43 1798). 140 QUARTERLY REVIEWS The valke of the method turns on the availability of satisfactory pro- cedures for isolation. A solvent may be chosen from which one isomer con- tinually crystallises on irradiation thus effecting complete conversion.Sometimes a trace of catalyst is added to the distillation-vessel and the lower-boiling isomer removed continuously. by distillation. Fractional crystallisation and distillation have often been used and great success has attended chromatographic separations particularly in the polyene series where stereomutation is often the only way in which many geometrical isomers may be obtained. The Reviewer is indebted to Professor R. P. Linstead C.B.E. F.R.S. and Dr. E. A. Braude for criticism of the manuscript.

ISSN:0009-2681    

DOI:10.1039/QR9520600101

出版商:RSC    

年代:1952

数据来源: RSC

摘要:

THE POLYlVLERISATION OF ALDEHYDES By 3. C. BEVINGTON M.A. PH.D. (DEPARTMENT OF CHEMISTRY BIRMINGHAM UNIVERSITY) THIS Review is concerned primarily with the formation and properties of cyclic and linear polymers of aldehydes. Compounds of the aldol type might be considered in this connection since their molecular formuls can be written as (RCHO), and similarly the simple carbohydrates which have been prepared from formaldehyde and certain hydroxy-aldehydes might be treated as polymers of the aldehydes since they have the same empirical formuh ; the aldols and carbohydrates are not true addition polymers however and only substances having the repeating unit *CHR*O* are dealt with in this Review. The formation of these polyoxymethylene derivatives can be represented by overall equations which are formally similar to those for the polymerisation of olefinic monomers but the two types of reaction differ profoundly in many respects.A chapter was devoted to the polyoxymethylenes in Walker’s recent monograph on formaldehyde but information on polymers of other alde- hydes seems never to have been collected and correlated adequately. Poly- mers of formaldehyde were described by Staudinger 2 in one of the earliest systematic accounts of high polymers ; it appears that in 1932 .almost as much was known about the polymers of formaldehyde as about those of olefinic compounds but the position is very different now since knowledge of the latter type is in a much more advanced state. High polymers of aldehydes are of no importance as fibres plastics or rubbers because they depolymerise readily.Precise work on these polymers is difficult since in only a few cases has it been shown quite unambiguously what catalyst if any is responsible for initiating polymerisation. It has not been possible to establish a single set of conditions under which all aldehydes can be converted into high polymers and so a direct comparison of their reactivities cannot be made. Another difficulty in detailed studies of the polymerisations is associated with determinations of molecular weights ; in the case of formaldehyde all but the low polymers are insoluble and linear polymers of other aldehydes decompose a t appreciable rates a t room temperature. In some cases the formation of high polymers seems to occur by an ionic mechanism and this is an added difficulty in the way of a complete understanding of the reactions since even for olefinic monomers much less is known about the ionic type of polymerisation than about the radical type.Aldehydes show a great tendency to form cyclic polymers of low molecular weight. I n one respect the aldehyde polymers are simpler than their olefinic counterparts as they invariably consist of a head-to-tail arrangement of monomer units. This is associated with the fact that the carbonyl bond is 1 “ Formaldehyde ” Reinhold New York 1944 Chapter 7. a “ Die Hochmolekularen Organischen Verbindungen ” Springor Berlin 1932. 141 142 QUARTERLY REVIEWS strongly polar so that its carbon and oxygen atoms differ markedly in reactivity. Cyclic Polymers Trimem.-Many aldehydes form cyclic trimers of general formula (I). /CHR-0 0 ‘CHR \CHR-O/ (1) Among the trimers which have been described are those where R = H CH, CHCI, CH,*CHBr and CH,=[CH,I9.The trimer of formaldehyde is usually called trioxan and not trioxymethylene as the latter name has long been used for a material now known to be a linear polymer of fairly high molecular weight. The best known of the trimers is paraldehyde derived from acetaldehyde. The conversion of an aldehyde into its trimer is catalysed by traces of acids and in many cases it proceeds readily at low temperatures. Special conditions are necessary for the preparation of trioxan since formaldehyde forms linear polymers very easily but the preparations of other trimers present little difficulty. There have been few studies of the kinetics of this type of polymerisation but the formation of paraldehyde by using phosphoric acid as catalyst has been followed dilatometrically both for the pure aldehyde and for solutions in benzene.Chloral is polymerised by sulphuric acid to metachloral which is almost certainly a cyclic polymer and may be a trimer ; the reaction can also be promoted by small amounts of the oxidation pro- ducts of the aldehyde.* The trimers can be depolymerised by acids although they are stable to alkalis ; substances such as ferric chloride which are electron-acceptors and therefore acidic in the broad sense also catalyse the depolymerisation. Mixtures of trioxan and acids can be used as sources of monomeric formalde- hyde. The kinetics of the depolymerisations of trioxan and paraldehyde have been studied both in the gas phase and in solution ; both the thermal and the acid-catalysed depolymerisation of paraldehyde proceed at greater rates than those of trioxan under the same conditions.Mixed Trimers.-A number of mixed trimers of two aldehydes are known examples being (11) and (111) /CH(CH,)-O\ /CH( CHCl2)-0\ 0 CH-CCl 0 CH2 (11) (111) It appears that in these mixed trimers one of the aldehydes must be halo- genated. These polymers can be made from mixtures of the appropriate monomeric aldehydes with traces of hydrogen chloride as catalyst. Some \CH( CH,)-O/ \CH(CHCI a )-o/ a Hatcher and Brodio Cainadian J. Res. 1931 4 574. Moureu Dufraisse and Berchet Bull. SOC. chim. 1928 43 942 957. Bell and Burnett Tram. Paraday SOC. 1937 33 355 ; 1938 34 420 ; 1939 35 474. BEVINGTON THE POLYMERISATION O F ALDEHYDES 143 of the mixed polymers decompose slowly on storage and all are depoly- merised when warmed with acids ; alkalis have no effect.Mixed trimers of three aldehydes have not been prepared.6 Tetramers and Dimem-Cyclic tetramers of formaldehyde and acetalde- hyde are known the accepted structures being (IV) and (V). Tetra- oxymethylene is rather uncommon and the conditions necessary for its formation have not been closely defined. Metaldehyde is well known ; it is prepared from acetaldehyde containing traces of acid and the reaction can take place a t temperatures only just above the freezing point of the monomer vuix. - 123". Metaldehyde and paraldehyde are usually formed simultaneously but the yield of the former is increased as the temperature is reduced. During the formation of metaldehyde at low temperatures by use of alcoholic sulphuric acid as catalyst another polymer is also formed ; this has a molecular weight corresponding to the formula (CH,*CHO) and its boiling point is about 74" which is intermediate between those of the monomer and the trimer.This polymer decomposes readily to acetaldehyde and it is possible that it is the dimer (VI). There have been no reports of similar dimers of other aldehydes. /O\ 'O/ O-CH,-O-CH O-CHMe-0-CHMe M e O H HGMe I I I I CH,*OCH,*O CHMe-0 CHMe-0 (IV) (V) (VI) The Mechanism of the Formation of Cyclic Polymers.-The alcohol CH,Br*CHBr*OH loses hydrogen bromide on standing to give the substituted paraldehyde (VII). This reaction has been cited as evidence for the theory /CH( CH,Br)-0 \CH( CH,Br )-O/ 0 ' H.CH,Br ( V W that the formation of cyclic trimers of aldehydes normally involves the compound R-CHX-OH formed by the addition of a molecule of acid HX to the aldehyde.The polymerisation can then be represented by the set of equations RCH /x + R C H O \OH /X HOCHR /OH + R C H O -+ RCH \O R*CH /" \O--CHR \o--cHR/ /X HO*CHR\ \O --CHR/ /O RCH \O /O--CHR -+ R-CH \O-CHK/ Hibbert Gillespie and Montonna J . Amer. Chem. SOC. 1928 50 1950. Travers and Sollers Trans. Faraduy Soc. 1936 32 248. 144 QUARTERLY REVIEWS This scheme involves the addition either of the acid HX or of a hydroxy- compound across a carbonyl bond. It is known that such additions involve charged bodies as intermediates so the whole reaction effectively proceeds by an ionic mechanism and the occurrence of polymerisation a t low temperatures is understandable.It is probable that other cyclic polymers of aldehydes are formed by similar processes. The polymerisation of chloral in the presence of air is inhibited by small amounts of certain substances e.g. phenols and aromatic a m i n e ~ ~ which are effective in stopping free-radical reactions. This is not evidence for a radical type of polymerisation for this aldehyde because the inhibitors probably prevent the oxidation which furnishes the catalysts for the poly- merisation and t,hey have only an indirect effect upon the polymerisation. Depolymerisation of Cyclic Polymers.-The ability of acids to initiate depolymerisation of cyclic polymers of aldehydes suggests that these sub- stances can cause the rupture of carbon-oxygen bonds of the type present in these polymers. This reaction will be referred to again in the next section.High Polymers of Formaldehyde Polymerisation at Low Temperatures.-Liquid formaldehyde polymerises very readily even at temperatures only just above its freezing point (- 1 1 8 O ) and in the preparation of the monomer stringent precautions are necessary to remove traces of impurities which may initiate reaction. Even the purest samples of liquid formaldehyde slowly gelatinise at low temperatures ; the gels can be evaporated giving monomeric gas and only very small amounts of solid,s showing that they consist of small quantities of a very high polymer dissolved in the monomer. No effective stabiliser is known for the aldehyde although polymerisation may be retarded slightly by quin01.~ At low temperatures solutions of the monomer in non-polar solvents are more stable than the pure monomer but they still exhibit a strong tendency to polymerise.The effect of oxygen is not clear ; Spence and Wild 8 attributed the instability of the monomer in part to oxygen adsorbed on the walls of the vessel but Staudinger 2 stated that the addition of oxygen decreased the tendency of the liquid to poly- merise. Although effective inhibitors are unknown a number of substances act as powerful catalysts for the polymerisation of the liquid aldehyde. Traces of hydrogen chloride or boron trifluoride cause almost instantaneous polymerisation of the liquid aldehyde at about - 80° and small amounts of normal aliphatic amines cause very rapid polymerisation of solutions of formaldehyde in ether a t that temperature.lo Staudinger considered that the molecular weight of the polymer formed from the pure aldehyde at low temperatures exceeds 150,000 but confirma- tion is not possible since the polymer is insoluble.Polymers prepared by using traces of polar catalysts are evidently of much lower molecular weight since they do not possess the plastic properties exhibited by the other polymers a t about 160". * J. 1935 338. Spence J. 1933 1193. lo Walker J. Arner. Chem. SOC. 1933 55 2821. BEVINGTON THE POLYMERISATION OF ALDEHYDES 145 The mechanism of the polymerisation a t low temperatures is not certain but a number of points indicate that it is ionic ; these points are (i) the nature of the catalysts since substances such as boron trifluoride and hydrogen chloride are unlikely to act as sources of free radicals ; (ii) the extreme rapidity of the reactions a t low temperatures ; (iii) the polarity of the carbonyl bond.In the case of catalysis by boron trifluoride a polar complex is likely to be formed and growth of the polymer may be represented thus -RF3 +- 0.CH2+ + O:CH + -BF3 +- OCH2*OCH2+ Such a process would require little activation and so could occur readily a t low temperatures; i t involves the separation of opposite charges but as the dielectric constant of liquid formaldehyde is probably quite high this is not improbable. Polymerisation in Solution at Normal Temperatures.-Formaldehyde dissolves readily in water polymerisation occurring to give polyoxymethyl- ene glycols of general formula HO*[ CH,*O],*CH,*OH where n probably can have values up to 12 ; if n exceeds this value the polymer is insoluble in water.Cryoscopic measurements have shown that the degree of poly- merisation decreases as the concentration of the solution is decreased or as the temperature is raised and that the process is reversible. The velocities of the polymerisation and depolymerisation reactions have been measured by cryoscopic interferometric and dilatometric methods (the results have been summarised by Mark and Raff 11) and have been found to be com- paratively small. In aqueous solution formaldehyde is believed to form the hydrate CH,(OH) very readily ; evidence for the existence of this hydrate is that the corresponding hydrate of chloral is stable and that by using the isotope l80 it has been shown that oxygen is exchanged quite readily a t 20" between acetaldehyde and water,12 probably by means of the equilibrium CH,.CHO + H20 CH,*CH(OH) The growth of the formaldehyde polymer may result from a polycondensa- tion; the simplest type of reaction is represented as HO*CH,*OH + HO*CH,*OH + HO*CH,*O.CH,-OH + H20 and the general reaction as HO*[CH,*O]p*CH2*OH + HO*[CH,*O],CH,*OH -+ Both polymerisation and depolymerisation are catalysed by hydrogen ions and the former may occur by the following mechanism HO*[CH2*O],+,+1*CH,*OH + H20 H0.[CH,*O]p.CH2*OH + H30+ -+ HO.[CH,*O],.CH,+ + 2H,O HO*[CH,.O],CH,.OH + H,O -+ HOfCH2-O],*CH2-O- + H30+ HO*[CH 2.O]p*CH 2+ -t HO*[CH ,*O],*CH,*O- + HO~[CH2~O]p*CH2-0*CH2*[O~CH,]11 " High Polymeric Reactions " Interscience New York 1941 pp.374 et seq. l2 Herbert and Lauder Trans. Faraday SOC. 1938 34 432. 146 QUARTERLY REVIEWS The net reaction clearly can be represented by the general equation above.The mechanism of the depolymerisation is discussed in a later section. The common polymer paraformaldehyde consists of a mixture of poly- oxymethylene glycols the average degree of polymerisation being about 30. The terminal hydroxyl groups of the polyoxymethylene glycols can be replaced by other groups. Monomeric formaldehyde is freely soluble in the lower alcohols and polymers can be recovered from the solutions ; if an alcohol R-OH is used polymers of the series R*O*[CH,*O],*CH,*OH are formed. Polymers of this type can also be prepared by controlled treat- ment of the glycols with alcohols e.g. HO*[CH,*O],.CH,*OH + CH,.OH + H0.[CH2*O],*CH2*0*CH3 + H,O The heat of polymerisation of formaldehyde calculated from the heats of combustion of the monomer and its polynier is approximately equal to the heat of solution of the aldehyde in water and alcohols ; Walker lo related the heats of polymerisation and solution with saturation of the aldehyde molecule.Dimethyl ethers of general formula CH,*O*[CH,*O],*CH,*O*CH3 can be prepared by the action of methyl alcohol on the glycols in the presence of traces of sulphuric or hydrochloric acid. Treatment of the glycols with acetic anhydride yields diacetates of general formula Ac*O*[ CH,*O],*CH,*OAc. The diacetates and the dimethyl ethers are more stable than the glycols from which they are derived. Individual members of the diacetate series having values of n up to about 20 have been isolated and fractionation of the dimethyl ethers has also been achieved.The physical properties change progressively with the molecular weight. The molecular weights of fairly low polymers can be determined satisfactorily by the cryoscopic method and those of higher polymers by end-group assay ; in the latter method the polymer is hydrolysed and the liberated acetic acid or methyl alcohol is determined. Polyoxymethylene diethyl and dipropyl ethers have also been prepared. Poberisation in the Gas Phase.-If gaseous formaldehyde is in contact with a surface a t a temperature below about go" deposition of polymer occurs a t an appreciable rate. The reaction was studied by Spence and by Carruthers and Norrish,13 the latter workers making a detailed kinetic study. They used an apparatus the whole of which was heated to about 100" while mixtures of formaldehyde and other substances were made ; then part of the surface was cooled to about 20" and the ensuing polymerisa- tion was followed manometrically.It was found that formic and acetic acids are powerful catalysts and that the catalysed polymerisations have the characteristics of a branching-chain reaction. The rate of polymerisa- tion could be expressed by an equation of the form [aldehyde] [ cat a1 yst ] Ic'[aldehyde] - k"[catalyst] ____._- ~ Rate = ~ so that the rate increased markedly with rising catalyst concentration. A set of chemical equations consistent with the observed kinetics was con- structed. The reactive centres are assumed to be present on the cold surface l3 Trans. Faraday SOC. 1936 32 195. BEVTNGTON THE POLYMERISATION OF ALDEHYDES 147 and to grow by the successive addition of aldehyde molecules from the gas phase.The formation of primary centres is believed to occur thus when formic acid is present H \C*O*CH,*OH O Y H\C*OH + CH,O + O Y The growth of polymer is believed to involve the hydroxyl group according to the equation H*CO*O*CH,*OH + CH,O + H*CO.O*CH,*O-CH,*OH The kinetics indicate that deactivation of a centre occurs by reaction with an aldehyde molecule the suggested equation being H*CO*[OCH,],*OH + CH,O + HCO*[O*CH,],*CHO + H,O the polymer molecule losing its hydroxyl group and therefore its capacity for growth. According to the kinetics branching occurs by the reaction of a centre with a molecule of formic acid the process being represented thus H ‘CO --+ H\C*IO*CH,],*O.CH(OH) O Y H\C*[O*CH,I,*OH + O Y HO/ The two hydroxyl groups can engage in growth reactions so that both structural and kinetic branching develop.A similar set of equations can be written for catalysis by acetic acid. Hydrogen chloride stannic chloride and boron trifluoride are even more powerful than formic and acetic acids as catalysts for the gas-phase poly- merisation and they promote reactions which appear to be branching-chain processes.14 It is believed that the polymerisation catalysed by hydrogen chloride involves the formation of the compound Cl*CH,*OH and that growth and deactivation of centres occur in the same way as in the case of catalysis by formic acid. A branching reaction of the type envisaged for formic acid cannot apply for hydrogen chloride since that substance does not contain an unsaturated group and cannot be incorporated in a growing chain.There is good evidence that hydrogen chloride can cause the rupture of carbon-oxygen bonds in polyoxymethylene polymers and it is believed that such a reaction occurs during polymerisation. The catalyst is thought to split the polymer chain into fragments each of which possesses a hydroyxl group and is therefore able to act as a centre for polymerisation the suggested equation being ClfCH 2*O],.CH2*O*[CH2.0]m*CH2*OH + HC1 + Cl*[CH ,.O],.CH ,*OH + Cl.[CH ,.O],*CH ,*OH As a result of this reaction kinetic branching occurs but only straight- chain polymers are produced. The suggested equations indicate that in each polymer molecule there is one molecule of hydrogen chloride combined l4 Bevington and Norrish Proc. Roy. SOC. 1951 A 205 517.L 148 QUARTERLY REVIEWS and this relationship can be used to determine the average chain-length of the polymer. The nature of the catalysts indicates that the mechanism of the catalysed polymerisation of formaldehyde in the gas phase may be essentially ionic. The suggested scheme involves the addition of hydroxy-compounds across the polar carbonyl bond and such processes normally involve charged intermediates. The conflicting reports on the effect of oxygen on the polymerisation of the liquid monomer at low temperatures were mentioned on p. 1 4 4 ; in the gas phase it appears that oxygen exerts no specific effect on the reaction and behaves only as an inert diluent like nitrogen.14 This is evidence against a free-radical mechanism since polymerisations of that type are usually affected by oxygen.Depolymerisation of Formaldehyde Polymers.-The fact that the de - polymerisation of the cyclic polymer trioxan is catalysed by acids indicates that carbon-oxygen bonds of the type present in aldehyde polymers can be broken by acids and further that in the acid-catalysed depolymerisation of linear polymers of formaldehyde the bonds which are broken are not necessarily near the ends of the polymer chains. The depolymerisation of polyoxymethylene glycols and diacetates is catalysed both by acids and by alkalis but with the dimethyl ethers and trioxan only acids are effective. Staudinger suggested that there are two distinct mechanisms for depolymerisation of linear polymers. Under alkaline conditions a step-wise degradation is believed to occur units con- taining only one formaldehyde molecule being split off; it is thought that polymers with methoxy end-groups are stable under these conditions because these groups resist the initial attack.Acidic catalysts are thought to be able to cause any carbon-oxygen bond in the chain to break and as this process is independent of the nature of the end-group the glycols diacetates and dimethyl ethers behave similarly. Step-wise degradation is thought to occur during the thermal depoly- merisation of polyoxymethylene polymers ; it is believed that the first bond to break is one very near the end of the polymer chain since the stability of a polymer depends to a considerable extent upon the nature of the end- groups. If traces of hydrogen chloride or boron trifluoride are added to formalde- hyde polymers in a vacuum a t about loo" the depolymerisation is acceler- ated.l4 It is thought that the polymer chains are split into reactive frag- ments by the same type of process as that suggested for the branching process during polymerisation and that the increase in the rate of depoly- merisation results from the increase in the number of points from which monomer molecules can be shed.The mechanism of the depolymerisation of poly(methy1 methacrylate) has been elucidated by determining the molecular weight of the polymer at stages during the reaction but a similar method cannot be used for formaldehyde polymer owing to the impossibility of determining molecular weights. Polyoxymethylene glycols have definite dissociation pressures of forrnal- BEVINGTON THE POLYMERISATION OF ALDEHYDES 149 dehyde and these pressures have been measured a t various temperatures.These results have been used l5 to calculate the changes in heat content and entropy accompanying the reaction at 25". and - 32.5 & 4 cal./deg. mole respectively; from the equation it is calculated that AG is equal to zero at 137". According to these cal- culations gaseous formaldehyde a t l atmosphere should not polymerise at temperatures above 137" ; this predicted " ceiling temperature " is in fair agreement with observed values. Properties of Linear Polyoxymethylene Polymers.-The lower poly- oxymethylene glycols are soluble in water acetone and ether but the solubility decreases as the molecular weight rises and the compound HO*[CH,*O],,*H is only very sparingly soluble in acetone.Separation of the lower members of this series of polymers has been effected by fractional precipitation from acetone solution by light petroleum. The higher glycols appear to dissolve slowly in water but the process involves partial depoly- merisation and so is not true solution. The lower diacetate and dimethyl ether polymers are soluble in a number of liquids but the solubility falls off wit!h increasing chain-lengths. The melting points of some of the lower members of the series of polymers have been measured and as expected the melting point rises with chain-length. The linear polymers of formaldehyde are shown by X-ray diagrams to be fairly crysta1line.l The degree of crystallinity of a polymer prepared a t low temperatures can be increased by warming it to the softening point and then cooling it.The magnetic susceptibilities of various high polymers of formaldehyde have been measured. Farquharson l6 reasoned that for a linear polymer with repeating unit *CH,*O* the magnetic susceptibility should approach the value of - 0-496 x A marked dis- crepancy was observed for a material known as 6-polyoxymethylene formed by prolonged treatment of polyoxymethylene dimethyl ether of high mole- cular weight with boiling water. It appears that this material is not a simple polyoxymethylene ; from the nature of its degradation products Staudinger had concluded previoiisly that 6-polyoxymethylene contains a few *CH(OH)* groups in its chain formed by rearrangement of *CH,*O* groups. High Polymers of Acetaldehyde CHsOgas at 1 atm. -+ solid polymer The calculated values for AH and AS are - 13.3 & 1.5 kcal./mole AG = A H - T .A S as the chain-length increases. Preparation and Structure.-A high polymer of acetaldehyde was first dmcribed by Letort l 7 and Travers l8 independently. If a sample of liquid aldehyde is frozen or if the vapour is condensed on a glass surface cooled in liquid air and then the solid is warmed to room temperature a viscous 1 5 Dainton and Ivin Trans. Faraday SOC. 1950 46 331. l7 Compt. rend. 1936 202 767. l6 Ibid. 1937 33 824. 1* Trans. Faraday Xoc. 1936 32 246. 150 QUARTERLY REVIEWS liquid or even a gel is sometimes produced. Unchanged aldehyde can be removed under reduced pressure leaving a rubber-like solid which can also be recovered by precipitation with water. Osmotic measurements on solutions in ethyl methyl ketone gave a value of 510,000 for the average molecular weight of a sample of the polymer ; l9 the material had undergone partial depolymerisation before the measurements were made however.The course of the polymerisation is influenced profoundly by many factors for example the purity of the aldehyde the state of the surface upon which condensation occurs and the rate of solidification or condensation. Staudinger 2o suggested that the polymer has a long-chain structure with repeating unit *CHMe*O* ; this structure has been confirmed by considera- tion of the properties of the polymer,21 X-ray diffraction data,22 and infra- red spectra.Z3 A technique in which acetaldehyde vapour is condensed on a cold surface under carefully controlled conditions has been used in a detailed study of the polymerisation.24 The variations in the yield of polymer with changing conditions can be explained readily if centres for polymerisation are assumed to be formed on the cold surface as the aldehyde condenses and if an aldehyde molecule can either be incorporated in it growing polymer molecule or condense on the cold surface without reaction.It appears that the active centres have fairly long lives and that they can be deactivated in three ways vix. by a process involving a centre and a molecule of aldehyde by one involving a centre and a molecule of a foreign substance and spon- taneously. Since condensation without reaction is possible yields of polymer less than l0Oyo are to be expected ; reductions in the yield may be due either to decreases in the number of primary centres or to premature termination of the growing chains.Although there are no effective inhibitors for the low-temperature poly- merisation of formaldehyde small amounts of certain substances e.g. water ethyl alcohol and paraldehyde suppress the polymerisation of acetal- dehyde a t its freezing point. Acetic acid has a marked effect upon the reaction increasing the yield of polymer but decreasing its molecular weight ; it also catalyses the depolymerisation. The effects of acetic acid are believed to be similar to those of hydrogen chloride on the polymerisation of formaldehyde ; the acid is thought to cause the rupture of carbon- oxygen bonds to give reactive points which either take up or shed monomer molecules according to the prevailing conditions. If the break occurs during poly- merisation the chance of an aldehyde molecule condensing without reacting is reduced and consequently the yield of polymer is increased and a t the same time the average molecular weight is reduced.There have been conflicting reports on the necessity for some substance l9 Muthana and Mark J. Polymer S c i . 1949 4 91. 2o Trans. Faraday SOC. 1936 32 249. 21 Rigby Danby and Hinshelwood J. 1948 234. 2 2 Powell quoted in ref. (21). 23 Sutherland Philpotts and Twigg Nature 1946 157 267. 24 Bevington and Norrish Proc. Roy. SOC. 1949 A 196 363. BEVINGTON THE POLYMERISATION OF ALDEHYDES 151 other than acetaldehyde to be present in order to initiate the polymerisation. Recent careful work by Letort and P&ry 25 indicates that traces of a peroxide in the aldehyde are essential for polymerisation.It has been shown conclusively 21 24 26 that solidification is essential if polymerisation is to occur ; if liquid aldehyde is supercooled below its freez- ing point no reaction takes place. Letort and Pktry 25 have shown that there is a connection between the number of centres for the crystallisation of liquid acetaldehyde and the yield of polymer. The polymer is formed while solidification is proceeding and the yield is unaffected by the length of time for which the aldehyde is frozen; inhibitors can be added to frozen aldehyde without influencing the amount of polymer which can be recovered. There have been suggestions 21y 26 that the latent heat of fusion can in some way be used for activation. An alternative explanation for the need for solidification is that the growing centres are thereby fixed and ring-closure is made impossible; further in liquid systems a reactive point might be effectively screened by a " cage " of molecules.No information about the crystal structure of acetaldehyde is available but it would be interesting to know whether in the crystal the carbon and oxygen atoms of the carbonyl groups lie in positions corresponding approximately to their positions in the polymer chains if this were so and a polymer once began to grow propaga- fion of the chain during growth of the crystal might proceed quite easily. An ionic mechanism for the polymerisation would be in accord with the following facts (i) reaction occurs readily a t low temperatures ; (ii) the formation of cyclic polymers of acetaldehyde is catalysed by acidic sub- stances ; (iii) the carbonyl bond is strongly polar.Letort and Pktry z 5 favour a mechanism involving free radicals in view of their conclusion that a peroxide is an essential catalyst for the reaction. High polymers of acetaldehyde have been the subject of a patent.27 It was claimed that boron trifluoride or other active halide catalyst converts acetaldehyde into the polymer CH,*CH( OH)*[CH,*CH( OH)],*CHO n having values up to about 5000. The suggested formula is probably incorrect since it corresponds essentially to polyvinyl alcohol which is insoluble in acetone and other liquids claimed to dissolve the polymer. Formaldehyde propaldehyde and methyl ketones were reported to polymerise similarly but in all cases the conditions for reaction were not defined closely. The polymerisation of formaldehyde can certainly be catalysed by boron tri- fluoride and similar substances but attempts to make polymers of other aldehydes and of acetone with these catalysts have failed.Comparison of the Properties of the High Polymers of Formaldehyde and Acetaldehyde.-The high polymer of acetaldehyde is more soluble than those of formaldehyde dissolving in acetone diethyl ether carbon tetrachloride butyl acetate and other liquids. At room temperature the methyl-substi- tuted polyoxymethylene is more rubbery than the unsubstituted material. The differences in solubility and mechanical properties can be related to 25 Compt. rend. 1950 231 519 545. 26 Letort Duval and Rollin ibid. 1947 224 50. 2'U.S.P. 2,274,749 March 3 1942. 152 QUARTERLY REVIEWS differences in crystallinity; the presence of the methyl groups in the acetaldehyde polymer must make it more difficult to fit the chains into a lattice and consequently the acetaldehyde polymer would be less crystalline than the formaldehyde polymer under similar conditions.The replacement of hydrogen atoms by methyl groups affects the properties of other polymers in a similar way ; for example the polymers of acrylic acid and acrylonitrile are much less soluble than those of methacrylic acid and methacrylonitrile and polyethylene is much less rubbery than polyisobutene. The Stability of the Acetaldehyde Polymer.-The high polymer of acetal- dehyde is less stable than the linear polymers of formaldehyde the rate of depolymerisation of the former being appreciable a t room temperature ; as stated already the cyclic trimer of acetaldehyde is less stable than the corresponding polymer of formaldehyde.The differences between the stabilities of the polymers of the two aldehydes is ascribed to a weakening of the carbon-oxygen bonds caused by the methyl groups. The replace- ment of hydrogen atoms by methyl groups affects similarly the stabilities of polymers of certain olefinic monomers ; for example polyisobutene is considerably less stable thermally than polyethylene. As will be shown later replacement of the methyl groups in polyacetaldehyde by larger alkyl groups reduces the stability of the polymer still further. Polymerisation under Other Conditions.-Acetaldehyde polymerises to a certain extent when subjected to very high pressures; 28 for example treatment for 30 mins. a t 300" under 5000 atmospheres gives a viscous liquid from which a rubbery material can be isolated.Water is formed during the treatment and crotonaldehyde can be detected easily. It appears that the polymer is derived from the unsaturated aldehyde and not directly from acetaldehyde and there is no evidence that the polymer is a polyoxy- methylene derivative. Ultrasonic waves have been used to induce polymerisation in acet,alde- h ~ d e ~ ~ but it is clear that the polymer was of the aldol type. The action of high-speed cathode rays on acetaldehyde vapour ' produces solid com- pounds which have been referred to rather loosely as polymers.30 They cannot be truc polymers of the aldehyde since their production is accom- panied by the appearance of decomposition products and acetaldehyde and acetone seem to give the same solid material.High Polymers of Butaldehyde Formation of the Polymer at High Pressures.-Butaldehyde has been converted into a high polymer by the application of a pressure of 12,000 atmospheres for 24 hours at 30".31 It was shown that very careful purifica- tion of the aldehyde reduced the tendency for reaction and that oxidation products of the aldehyde are probably essential catalysts. It was concluded Le$unski; and Rienov Compt. rend. Acad. Sci. U.R.S.S. 1941 30 624. 29 Demann and Asbach Tech. Mitt. Krupp Forschurqsber. 1940 3 12 ; see Chem. 30 McLennan and Patrick Canadian J . Res. 1931 5 470. 31 J . Amer. Chem. Xoc. 1932 54 628. A h . 1940 34 3972. BEVINGTON THE POLYMERISATION OF ALDEHYDES 153 that a peroxide derived from the aldehyde and not butyric acid was the active substance.The peroxide decomposes slowly under normal condi- tions and it was suggested that the initiation of polymerisation is associated with this decomposition. Benzoyl peroxide could be used as a catalyst for the polymerisation and certain substances such as ethyl alcohol acted as inhibitors. Depolymerisation of the Polymer.-The polymer of butaldehyde reverted to the monomer quite readily the rate rising with temperature. The process resembled the depolymerisation of other aldehyde polymers both linear and cyclic in that it was accelerated by acids ; alkalis also acted catalytically. The ease of depolymerisation and the high yield of monomeric aldehyde indicate that the polymer was most probably of the polyoxymethylene type with repeating unit *CH(C,H,)*O*.The polymer was suspended in 1 4-dioxan and the freezing point of the liquid gradually fell ; on the assumption that the depression was due entirely to liberated monomer the rate of depolymerisation was calculated. From the rates of depolymerisation a t a series of temperatures it is estimated that the overall activation energy for the process is very roughly 15 kcal. per mole of monomer. It appears that the polymer of butaldehyde is less stable than the high polymers of formaldehyde and acetaldehyde. Linear polymers of formaldehyde can be stabilised to a certain extent by replacement of the terminal hydroxyl groups by other groups such as *O*CH and *O*CO*CH,. Attempts to stabilise the butaldehyde polymer similarly were unsuccessful for the necessary reagents themselves acceler- ated the depolymerisation.The Mechanism of Polymerisation.4onant and Peterson 31 pointed out that the decomposition of the peroxide believed to be the essential catalyst for the polymerisation occurs at normal pressures and that the need for high pressures during the polymerisation is not associated with that decomposition. The high pressures cause the molecules in the liquid to be packed tightly and perhaps also to be orientated in a way favourable for the propagation of a reaction chain; the polymer is denser than the monomer and so high pressures should favour the production of polymer. It was suggested that polymerisation proceeds by an energy chain reaction but this seems improbable. Peroxide-catalysed reactions involve free radicals and since this polymerisation is catalysed by peroxides there is evidence for a radical mechanism.The overall activation energy for the polymerisation can be estimated by comparing the activation energy for depolymerisation and the heat of polymerisation. As indicated already the activation energy for depolymerisation is roughly 15 kcal. per mole ; the heat of polymerisation of butaldehyde is probably a little less than that of formaldehyde which is in the range of 12-15 kcal. per mole. If the polymerisation and depolymerisation proceed via the same reactive inter- media,tes it is probable that the polymerisation requires an activation energy of roughly 3 kcal. per mole. A direct comparison of the polymerisations of formaldehyde acetalde- An attempt was made to study the depolymerisation more closely. 154 QUARTERLY REVIEWS hyde and butaldehyde is not possible because the high polymers are formed under such widely different conditions.Qualitatively it is clear that the ease with which polymerisation occurs and the stability of the polymer both decrease as the size of the alkyl group is increased. Certain observa- tions on the polymerisation of thioaldehydes confirm that the size of the group attached to the carbonyl group may influence the polymerisation. Thioformaldehyde and other thioaldehydes of low molecular weight poly- merise extremely readily and only thioaldehydes with high molecular weights are at all stable in the monomeric form. Most thioaldehydes form cyclic polymers but a few having very large molecular weights tend to form linear polymers. 32 Polymerisation of Other Aldehydes at High Pressures.-isoButaldehyde n-valeraldehyde and n-heptanaldehyde resemble n-butaldehyde in giving substituted polyoxymethylenes when subjected to very high pressures.31 Evidence was obtained that the two butaldehydes can form co-polymers Polymers of Ketones Apart from the patent claim in respect of methyl ketones (see p. 151) there are no records of ketones polymerising to compounds with repeating unit *C(R,R,)*O*. Thioketones readily form cyclic trimers of general formula (VIII). This difference between ketones and their sulphur analogues may occur because the sulphur atom is larger than the oxygen atom. The reason for acetone for example not forming a cyclic trimer may be that the methyl groups on one carbon atom interfere with those on another ; in thioacetone the carbon atoms are further apart and so the chance of interference between the methyl groups is reduced.O*CR ,R ,.O I I CRIR,*O.CR,R (VIII) The Polymerisation of Dialdehydes Pure glyoxal polymerises only slowly and the process is accelerated by traces of water but very little is known about the polymerisation of com- pounds of this type. Glyoxal might give rise to two distinct types of polymers with repeating units -O*CH CH*O= and *O*CH(CH 0). and so be compared with butadiene which also contains conjugated double bonds and for which both 1 2- and 1 4-addition are possible. It should be noted that a polymer with the first of the repeating units above would not be a polyoxymethylene derivative since the chain would consist of alternate pairs of carbon and oxygen atoms and also that if such a polymer existed it could be produced only by a reaction of the free-radical type and not by one of the ionic type.A polymer having the second repeating unit could have a branched or cross-linked structure. The polymerisation of methylglyoxal CH,*CO*CHO was studied by 32 See e.g. " Organic Chemistry " edited by Gilman Wiley New York 2nd edn. 1942 p. 923. BEVINGTON THE POLYMERISATION OF ALDEHYDES 155 Moulds and Riley,33 who found that the rate was affected by the purity of the material and that traces of water catalysed the reaction. The polymer was sometimes produced as a hard glass a d it could be decomposed to monomer quite easily indicating that it was probably of the polyoxymethyl- ene type. It was suggested that the first stage in the reaction is the addition CH,*CO*CHO + H20 -+ CH,*CO*CH(OH) and that then two molecules of t,he product reacted thus /OH /OH CH,CO.CH/ CH,.CO.CH/ \OH --+ >O +H,O /OH CH ,-CO CH' CH -C 0 *C d \OH \OH Further similar condensations would lead to the building of large molecules.This mechanism is essentially the same as those discussed already in con- nection with the polymerisation of aldehydes to cyclic trimers and t,he polymerisation of formaldehyde in aqueous solution. It was argued that the polymerisation of methylglyoxal d.id not involve the carbonyl group having a methyl group attached to it since diacetyl CH,*CO*CO*CH does not polymerise under these conditions. The second carbonyl group of methylglyoxal is essentially ketonic and t,herefore would not be expected to enter into polymerisation reactions.Polymers of Unsaturated Aldehydes Keten CH C 0 polymerises readily and the reaction shows character- istics of an ionic process ; for example the speed of reaction varies with the dielectric constant of the medium and it may be appreciable a t tempera- tures as low as - 80°.34 A dimer of uncertain structure 35 is produced but it is clearly not of the polyoxymethylene type. Certain substituted ketens polymerise more readily than keten itself. Solutions of dimethylketen C(CH3) C 0 in ether undergo vigorous polymerisation if treated with traces of trimethylamine a t - 80°.36 Hard glassy products of quite high molecular weight are produced. These poly- mers revert to the monomer a t this ready depolymerisation may *c.o* II CMe (IX) temperatures in the range 100-200" and indicate that the main chain of the polymer CMe ,*C* A (XI consists of alternate carbon and oxygen atoms i.e.the repeating unit is (IX) and not (X). There is also some chemical evidence that the polymer 33 J. 1938 621. 34 Rice and Greenberg J. Amer. Chem. SOC. 1934 56 2132. 35 Whiffen and Thompson J. 1946 1005. a6 Staudinger Helv. Chim. Acta 1925 8 306. 156 QUARTERLY REVIEWS is a substituted polyoxymethylene wix. (i) the polymer readily absorbs bromine indicating the presence of olefinic bonds (ii) ozonolysis yields acetone and non-volatile products suggesting that the polymer contains CMe groups. On the other hand when the polymer is treated with alcoholic potash at loo" the chief product is CHMe,*CO*CHMe indicating that it may contain the unit *CMe,*CO*CMe,* which would be the case if the olefinic and not the carbonyl bond opened during polymerisation. Infra- red measurements would most probably distinguish between the two possible repeating units since the one contains the C-C bond and the other the C=O bond. Acraldehyde and a-methylacraldehyde give products of high molecular weight in the presence of dilute aqueous alkali,37 but they are not poly- oxymethylene derivatives. 3' Gilbert and Donleavey J. Amer. Chem. SOC. 1938 60 1737 1911.

ISSN:0009-2681    

DOI:10.1039/QR9520600141

出版商:RSC    

年代:1952

数据来源: RSC

摘要:

By L. L. BIRCUMSHAW M.A. D.Sc. (THE UNIVERSITY BIRMINGHAM) and A. C. RIDDIFORD PH.D. (THE UNIVERSITY SOUTHAMPTON) I. Introduction THE study of the kinetics of heterogeneous reactions covers a wide field. In one system the problem may be purely chemical iii nature but in another the characteristics which are of interest to the chemist may be entirely determined by non-chemical factors. In this Review an account will be given of systems in which the observed rate of reaction is either wholly or partly determined by the rate of a transport process i.e. by the rate of mass-transfer of a species by convection and diffusion. In general atten- tion will be confined to systems subject to forced convection although many of the considerations apply also to mass-transfer under natural convection.The subject has been reviewed as a whole or in part by Centnerszwer,l Taylor,2 Bowden and Agar,3 T o ~ b i n ~ and by Moelwyn- hug he^.^ Hixson and Crowell,6 and Hixson,’ have reviewed the topic from the standpoint of the study of agitation. The general case of a reaction between a solid and a solution resulting in soluble products will be taken as a model. The overall process may comprise as many as five primary steps (a) Transport of solute molecules to the interface. ( b ) Adsorption a t the surface. (c) Reaction a t the surface. (d) Desorption of the products. ( e ) Recession of the products from the interface. Of these steps ( b ) ( c ) and ( d ) are characterised by an interaction between the solid and solute and quite generally will be termed chemical processes. Complications may arise in other systems ; for instance the products may be gaseous or may form an insoluble layer on the solid surface or may undergo further reaction in the bulk of the solution.For a liquid-liquid system in which reaction takes place between two solute species one in each phase step (a) will comprise two transport processes etc. By con- 1 2 . physikal. Chem. 1929 141 A 297. “ A Treatise on Physical Chemistry ” 2nd edn. Macmillan 1931 Chap. XV. Ann. Reports 1938 35 90 ; see also Agar ibid. 1947 44 6 and Hickling Quart. Bull. Sci. Univ. Kiev 1939 No. 4 155. “The Kinetics of Reactions in Solution ” 2nd edn. Clarendon Press 1947 I n d . Eng. Chem. 1931 23 923. Reviews 1949 111 95. p. 357. ’ Ibid. 1944 36 488. 157 158 QUARTERLY REVIEWS fining attention mainly to clear-cut examples however the modifications necessary for other systems will be obvious.According to Nernst,8 it is highly probable that the chemical processes a t the interface are always very much faster than one or other of the trans- port processes so that unless there is a slow process occurring within the bulk of one of the phases the observed rate is transport-controlled. This hypothesis caused considerable controversy. His theory outlined in the next section was supported by several workers who conducted investiga- tions designed to test its validity e.g. B r ~ n n e r ~ Jablczynski and St. Jablonski,lo and Van Name and his co-workers.ll On the other hand Ericson-Aurh and Palmaer,12 Wildermann,13 Gapon,'* Miyamoto,15 and Roller l6 are but a few of the workers who have criticised the theory some- times to the extent of denying that transport control ever occurs.11. The Nernst Theory5 Noyes and Whitney studied the dissolution of cylinders of benzoic acid and of lead chloride in water. The cylinders were rotated and the process of dissolution was followed by analysing the solution at intervals. They expressed their results by the first-order equation dc/dt = E(c - c ) . * ( 1 ) where c is the concentration of the solute at time t and c is the solubility of the compound in water at the experimental temperature. They sug- gested that a saturated layer is rapidly formed a t the interface and that the observed velocity is simply the rate a t which solute molecules diffuse from this layer into the bulk of the solution. Their results were confirmed by Bruner and St.Tolloczko,ls who showed that the rate of dissolution is proportional to the apparent surface area A . Nernst,s in extending these views to heterogeneous reactions assumed that equilibrium is established almost instantaneously a t the interface between two phases. Thus for the dissolution of magnesium in acid for exa,mple the equilibrium concentration of hydrogen ion in contact with the metal is vanishingly small and the dissolution velocity will be deter- mined solely by the rate a t which hydrogen ions arrive a t the interface. Provided the solution is well stirred the concentration in the bulk of the solution may be regarded as uniform and the hydrogen ions reach the surface by diffusing through a thin layer of solution of thickness 6 adhering to the solid surface. 2. physikal. Ghem.1904 47 52 ; Macmillan 1923 p. 669. Ibid. 1904 47 56. " Theoretical Chemistry " 5th English edn. lo Ibid. 1911 75 503. 11 (a) Van Name and Edgar Amer. J. Sci. 1910 29 237; 8. physikal. Chem. 1910 73 97 ; (b) Van Name and Bosworth Amer. J. Sci. 1911 32 207 ; (c) Van Name and Hill ibid. 1913 36 543; ( d ) Van Name ibid. 1917 43 449. l2 8. physikal. Chem. 1906 56 689. l4 2. Elektrochem. 1928 34 803. l6 J . Phys. Chem. 1935 39 221. Ibid. 1909 66 445; Phil. Nag. 1909 18 538. l5 Trans. Farahy SOC. 1933 29 789. l7 2. physikal. Chem. 1897 23 689. Ibid. 1900 35 283. BIRCTJMSHAW SND RIDDIFORD * TRANSPORT CONTROL 159 Consider the solid of surface area A in contact with a volume V of solution of concentration c. - dc/dt = (DA/V)dc/dy - * (2) where dc/dy is the concentration gradient normal to the surface and D is the coefficient of diffusion of the solute.Nernst assumed that the con- centration gradient can be expressed by (c - c,)/6 where ci is the con- centration a t the surface ; by substitution in (2) - dc/dt = DA(c - Ci)/VvG . * (3) and the first-order constant is given by k = DA/VvG * (4) It has the dimensions of a frequency. Equation (3) may be compared with equation (1). For the dissolution of a metal in acids and for similar systems he further assumed that the concentration of solute a t the surface is practically zero. - dc/dt = cDA/V6 * ( 5 ) Tests of the Theory.-If the coefficient of diffusion of the solute is known the thickness of the diffusion layer can be calculated from the observed velocity constant per unit area a t unit volume kT,* since kT = k V / A = D/6 .* (6) This unit constant has the dimensions of a velocity. Thus Brunner gave the values listed in Table 1 (see also Table 3). He observed that these Then by Pick's law Equation (3) then reduces to TABLE 1. Values of 6 at 20" System Dissolution of benzoic acid in water . . . . . I , magnesia in benzoic acid . . . . , , magnesia in acetic acid . . . . . 9 9 , marble in HC1 + MgC1 . . . . 7 9 , magnesium in benzoic acid . . . . Y7 , silver acetate in water . . . . . S. mm. 0.02-0*03 0.029 0.028 0.032-0.036 0.022 0.031-0.039 values are all of the same order of magnitude and claimed that they are not unlikely values. Moelwyn-Hughes,5 whilst pointing out that these values are physically improbable draws attention to the significant fact that 6 has roughly the same value for many reactions of quite different chemical character consistent with the view that the rates are determined by a diffusion process.An increase in the rate of stirring should be accompanied by an increase in the observed velocity since the thickness of the liquid layer adhering to the surface will decrease. For a given rate of stirring moreover the efficiency of the stirrer will be a function of the dimensions and shape of the system. These effects have been observed for some systems. For example Nernst and A second test of the theory concerns the effect of stirring. * Throughout this Review quantities relating to a transport process and t o a chemical process are denoted by the subscripts T and C respectively. 160 QUARTERLY REVIEWS Merriam l9 studied the electrolysis of acidified KI solutions between plati- num electrodes one being rotated a t constant known speeds.They found that the limiting current increases with increasing rate of stirring. Many investigators have confirmed this effect usually expressing their results in a power relation of the form ET cc (r.p.m.)" . * (7) where the power a < 1. Nernst and Merriam found a = 0.6 ; some of the values reported by other workers are given in Table 2 together with Dissoln. of Na cylinders in liq. NH . . . . Dissoln. of Zn cylinders in dil. acetic acid solns. . TABLE 2. Values of a the stirring coeflcient 116-834 450-26 400 1 System Electrolysis of KI solns. . . . . . . . Redn. of 0 to H,O a t Cu cathode . . . . Dissoln. of Zn in aq. iodine s o h . . . . . . Electrolysis of HC1 + KC1 solns. . .. . . Dissoln. of magnesia in benzoic acid . . . . Reaction of Na amalgam with dil. aq. acid . . Dissoln. of Hg in aq. iodine s o h . . . . . . Electrolytic oxidri. of chromous chloride . . . Dissoln. of Mg in MeOH-H,O solns. of HCl . . Concn. polarisation at a silver cathode . . . Dissoln. of Mg cylinders in HC1 solns. . . . H .p.m. ~ ~~ . 120-400 30-3600 50-200 50-400 135-182 500-2450 170-300 300-350 350-700 700-2730 < 300 300-1 500 0-1000 1000-5600 a 0.42 0.5 0.56 0.6 0.66 0.7 0.8 0.85 < 1 1 < 1 1 < 1 - 1 1 1 Ret . 20 21 22 23 9 24 1 l a 25 20 27 28 29 30 the range of stirring speeds investigated. Hixson and Baum,31 studying the rate of dissolution of benzoic acid pellets in sodium hydroxide solutions calculated the value of 6 for various rates of stirring (Table 3).* Over the range studied 6 is inversely proportional to the rate of stirring i.e.a = 1. Rather less attention has been paid to the influence of the Is 2. physikal Chern. 1905 53 235. 2O Sackur ibid. 1906 54 641. z1 Siver and Kabanov J . Phys. Chem. U.S.S.R. 1948 22 53 ; 1949 23 428. 22 Ridtliford and Bircumshaw J . 1952 698. 2 Eucken 2. physikal. Chem. 1907 59 72. 24 Dunning and Kilpatrick J . Phys. Chenz. 1938 42 215. 2 5 Jablczynski 2. physika2. Chem. 1908 64 748. 26 Garrett and Cooper J . Phys. Colloid Chem. 1950 54 437. 27 Nagel and Renner 2. Elektrochem. 1950 54 547. 28 King and Braverman J . Amer. Chem. Soc. 1932 54 1744. Z9 Johnson and McDonald ibid. 1950 72 666. 30 King and Schack ibid. 1935 57 1212. 31 Ind. Eng. Chem. 1944 36 528. * For this system the diffusion layer is composed of two regions a layer adjacent to the solid surface wherein the concentration of benzoic acid decreases from the saturation concentration at the surface to zero at points distant 6 from the surface and a second layer in which the concentration of hydroxyl ions increases from zero at sa to the bulk concentration at points distant 6 from the surface.The values recorded in Table 3 are for the total layer thickness. BIRCUMSHAW AND RIDDIFORD TRANSPORT CONTROL 161 Values of 6 for the dissolution of benxoic acid in TABLE 3. sodium hydroxide solutions - - R.p.m. . . . . 200 250 300 ~ 350 450 I geometry of the system on the observed rate. In most investigations the effect is eliminated by maintaining the same geometry for all experiments. Some studies of the effect have been made however ; these will be referred to in Section V.Nernst considered that if the thickness of the adhering layer is deter- mined by experiment for one system with a given type and rate of stirring the velocity constant k can be calculated for other reactions taking place under the same flow conditions. Thus Brunner determined 6 from measurements of the rate of dissolution of benzoic acid in water and then calculated the rate of dissolution of magnesia in various acids using the known coefficients of diffusion. The calculated values were in reasonable agreement with the experimentally determined rates. In this considera- tion Nernst has introduced a further assumption namely that 6 is a function of the type and rate of stirring only. This can be shown to be incorrect (see below).Without making this assumption however it is evident that the dissolution of a number of solids in the same solution under the same experimental conditions should proceed at the same rate provided that in each case the observed rate is determined solely by the rate of transport of solute to the solid surface the solids being insoluble in the pure solvent. and Van Name and Bosworth lib found that under the same conditions mercury cadmium zinc copper silver iron nickel and cobalt dissolve in aqueous iodine solutions a t the same rate within the limits of experimental error. Their results have been confirmed.22 Again from Nernst’s assumption that 6 is a function of the rate and type of stirring only k should be proportional to the coefficient of diffusion of the solute and hence inversely proportional to the viscosity of the solu- tion.Results in qualitative agreement with these views were soon obtained e.g. Van Name and Edgar lla have shown that bromine attacks metals more rapidly than does iodine,32 Jablczynski 25 found that an increase in the viscosity of the solution decreased the rate of oxidation of chromous chloride Jablczynski and St. Jabloqki lo obtained similar results for the dissolution of magnesium and marble in hydrochloric acid Van Name and Hill llC demonstrated that the addition of sucrose or alcohol to the solution decreased the rate of solution of cadmium in aqueous iodine solutions whilst in their earlier papers King and his co-workers 33 published similar results for other systems. Finally if 6 is a function of the rate and type of stirring only it should 32 See also Trotman-Dickenson and James J .1947 736. 33 (a) King and Liu J . Amer. Chem. SOC. 1933 55 1928 ; (b) King ibid. 1935 Thus Van Name and Edgar 57 828; see also ref. 28. 162 QUARTERLY REVIEWS be temperature invariant from eqn. (6) then E and D should have the same temperature coefficient. At 25" the value of the energy of activation for diffusion ED ranges between about 2800 and 6500 ~al./mole,~~ depend- ing on the solute and on the solvent. For many electrolytes in aqueous solution ED - 40004500 cal./mole. Many heterogeneous reactions do have values of E, the observed critical increment of this order e.g. for the electrolysis of 0.66~-potassium chloride E = 4300 cal./mole ; 9 for the dissolution of cadmium in aqueous iodine solutions,lld EA = 4000 cal./mole.It must be observed however that chemical reactions having a critical increment of this order or less are Defects of the Nernst The~ry.~~-As detailed above Nernst in formu- lating his theory assumed (1) that the chemical processes at the surface always proceed very much faster than one of the two transport processes ; (2) that in a well-stirred system the concentration gradient is confined to a thin layer of solution adhering to the solid surface ; (3) that within this layer the concentration varies linearly with distance measured normal to the surface ; and (4) that the thickness of this layer whilst being a function of the rate of stirring and of the geometry of the system is independent of the coefficient of diffusion of the solute of the viscosity of the solution and of the temperature.None of these assumptions is entirely correct. recognised the possibility of chemically-controlled reactions. That the chemical effect predominates in some systems is shown by the non-dependence of the observed rate on the rate of stirring as in the case of the dissolution of certain metals in acid 36 and of glass in alkali.37 Marc 38 found that the rate of crystallisation from aqueous solutions is independent of the rate of stirring when this is sufficiently intensive.* Again the large tempera- ture coefficients of certain reactions rule out any possibility of transport- control. Moelwyn-Hughes has discussed the kinetics of the decomposi- tion of sodium hypochlorite in aqueous solutions catalysed by a cobalt peroxide suspension studied by Howell.39 For this reaction E is 16,600 cal./mole and the kinetics are satisfactorily explained on the assumption that the rate is controlled by activated collisions between the hypochlorite ions and the catalyst surface.Other cases of heterogeneous catalysis of this type and of the type where the rate is governed by an adsorption process are too well known to require mention. Lastly if the 34 Taylor J . Chem. Phys. 1938 6 331. 3 6 E.g. Centiierszwer and Zablocki 2. physikal. Chem. 1926 122 455 ; Centner- 37 Schmidt and Durau ibid. 1923 108 128 ; but contrast Muller and Weinstein 3 8 2. physikal. Chern. 1908 61 385 ; 1909 67 470 ; 1909 68 104 ; 1910 73 685. 39 Proc. Roy. Soc. 1923 A 104 134. * At 0" the process is approximately of the second order whereas at 25" the process is first-order. Since the r61e of diffusion in the mechanism of the crystallisation pro- cess in certain circumstances has been established beyond doubt (see e.g.tho recent discussion of the Faraday Society on Crystal Growth ; No. 5 1949) one must con- clude that for some systems a change in the experimental conditions may effect a change of control. Despite his support of the Nernst theory Brunner 3 5 See also King ref. 33b. szwer ibid. 1928 137 352. Acta Physicochim. U.R.S.S. 1935 3 465. This point) is considered further in Section VII. BIRCUMSHAW AND RIDDIFORD TRANSPORT CONTROL 163 observed rate is transport-controlled it should be independent of the particular crystal face exposed to attack. This has been shown to be the case for systems which on other grounds are known to be transport- ~ontrolled.~~ On the other hand cases are known where the observed rate varies according to the face exposed to attack,41 the effect sometimes being shown by the preferential development of certain crystal faces during the course of reaction.42 The assumption that the diffusion layer is stationary with respect to the surface must now be considered.As previously mentioned concordant values of 6 of the order of 0.03 mm. a t 20" have been obtained for many reactions of very different types. Now such a value corresponds to a layer some 50,000 molecules thick which on general grounds would appear to be improbably high. This conclusion is supported by experimental evi- dence. Roller,lG for example cites seemingly overwhelming evidence in favour of the view that fluid motion persists down to very short distances from the solid surface if not to the surface itself.The most striking evi- dence is afforded by the work of Fage and T ~ w n e n d ~ ~ who using the ultramicroscope studied turbulent flow in pipes. Minute particles present in the tap-water used in the investigation were intensely illuminated and these particles used to follow the motion of the fluid. In this way they were enabled to measure the maximum values ul vl and wl of the three components u w and w of the velocity disturbance a t any point. The radial dependence of the mean velocity U was obtained by the same means. At the centre of the pipe ul vl and w1 are approximately equal. As the wall is approached the ratio vl/U obtained from the velocity disturbance normal to the wall decreases to zero whilst the corresponding ratios ul/U and wr/U increase.At the wall itself it was found that the flow tended to the laminar type (see next section),.motions of the particles in the laminae being observed to within a distance of 0.6 x lor4 em. from the wall. If fluid motion persists up to points very close to the solid surface the assumption that the concentration is a linear function of the distance y (measured normal to the surface) over the range 0 < y < 6 can be but an approximation. For the case of a plane disc electrode rotating about an axis perpendicular to the plane Levich 44 has calculated values of cy the concentration a t any point y. Fig. 1 shows the ratio cy/c plotted against y/6 where c is the concentration in the bulk of the solution.* The full line shows the calculated dependence of cy on y ; the broken line (a) King and Appleton Trans.Electrochem. Soc. 1940 77 219 ; (b) Glauner Chem. Zentr. 1934 11 2129; see also ref. 22. 41 E.g. (a) Spring 2. physikal. Chem. 1888 2 13 ; (b) n7ildermann ibid. 1910 '71 401 ; (c) Glauner ibid. 1929 142 A 67 ; ( d ) Gwathmey and Benton J. Chem Phys. 1940 8 431 569; see also ref. 40a. 4 2 E.g. Tovbin and Baram J. Phys. Chem. U.S.S.R. 1949 23 406. 4 3 Proc. Roy. SOC. 1932 A 135 656. 44 Actu Physicochim. U.R.S.S. 1942 17 257; J . Phys. Chem. U.S.S.R. 1944 * Fig. 1 has been adapted from Fig. 2 in Levich's paper (ref. 44). See also Levich 18 336. Discuss. Faraday SOC. 1947 1 37. M 164 QUARTERLY REVIEWS indicates the concentration gradient in the hypothetical Nernst layer. The calculated concentration gradient occurs over a distance a’ where 6’ is related to 6 by the expression (see Section IV) 6 = 0.8936’ * (8) Finally there is the question of the dependence of 6 on the coefficient of diffusion of the solute on the viscosity of the medium and on the temperature.According to Nernst 6 is independent of the diffusion coeffi- cient D and hence E should be proportional to D. Experimentally this is not the case. King and Cathcart 45 find the relation k T ot . * (9) for the dissolution of magnesium cylinders in acids whilst King 46 suggests that the power on D should be 0-75 for dissolution under turbulent flow conditions. Eucken 47 found the power on D to be 0.66 for laminar flow. These results indicate that 6 is a function of D and this together with the fluid flow within the Nernst layer discussed above suggests that 6 is probably also it function of the viscosity of the medium and of the temperature.Having regard to the similarity between the observed criti- cal increment and the energy of activa- tion for the diffusion process in certain transport-controlled systems it is to be expected that the dependence of 6 on temperature will be slight (see Section VII). Thus whilst the experiments of Page and T ~ w n e n d ~ ~ and of others indicate that the concept of a station- ary layer of liquid adjacent to the solid surface is wrong there appear to be excellent grounds for the belief that in systems for which the observed rate is a function of the rate of stirring and of the diffusion coefficient of the solute etc. there is a con- centration gradient existing between the solid surface and points in the liquid distant 6 from the surface.Indeed the existence of this region can be demonstrated very simply in certain cases,48 and Antweiler 49 has photo- graphed the diffusion layer at the dropping-mercury cathode. The pro- perties and extent of this region in any given system will be determined 4 6 J . Amer. Chem. SOC. 1937 59 63. 40 Trans. N . Y . Acad. Sci. 1948 11 10 262. 4 7 2. Elektrockem. 1932 38 341. 48 E.g. King and Brodie J . Amer. Chem. ~Soc. 1937 59 1375 ; see also Hixson 49 2. Elektrochem. 1938 44 719 ; some of the photographs have been reproduced and Baum (ref. 31). by Kolthoff and Liiigane “ Polarography ” Interscience 1941 p. 128. BIRCUMSHAW AND RIDDIFORD TRANSPORT CONTROL 165 by the value of the diffusion coefficient of the solute by the viscosity of the solution and by the nature of the flow of fluid in the system.Before dealing with the connection between the diffusion layer and the properties of the solute and solvent it is instructive to consider certain qualitative aspects of the dynamics of fluid flow. III. Qualitative Aspects of the Dynamics of Fluid Flow50 Consider the steady flow of fluid through a pipe. The nature of the flow may be investigated qualitatively by ejecting a coloured fluid from a capillary tube into the main body of fluid. Study of the forms assumed by such " flow-indicators " affords valuable evidence. For instance for low velocities of fluid flow the coloured filament flows parallel to the axis of the pipe and remains sharply defined although its width increases very slowly owing to diffusive and convective transfer normal to the axis.For this type of flow termed laminar (streamline or viscous) flow u the velocity component parallel to the axis depends only on the radial distance from the axis. The velocity component v normal to the axis is very small ; (a> FIG. 2 that it is not zero follows from the fact that at the solid surface there is no slip between the surface and the fluid in contact with it.* This stationary layer retards the fluid in contact with it which in turn retards the next layer and so on the thickness of the layer of retarded fluid increas- ing with distance along the solid surface. At a certain distance from the pipe inlet the retarded layer fills the whole cross-section of the pipe ; beyond this point the velocity distribution over any cross-section is para- bolic as indicated in Fig.2a. As the rate of flow is increased a velocity is reached at which the nature of the fluid flow alters abruptly; the filament no longer flows parallel to the axis but instead mixes rapidly with the main fluid body. On close examination it is seen that the mixing.is due to very irregular fluid motions across the pipe. For this type of flow termed turbulent flow the velocity components u and v are now to be regarded as mean values taken over a sufficiently long interval of time. Their actual values at any given point fluctuate with time owing to the movement of turbulent eddies through the fluid body. The mean velocity component parallel to See Goldstein " Modern Developments in Fluid Dynamics " Clarendon Press 1938. * This postulate is of fundamental importance in the study of fluid dynamics (see ref.50 p. 676). It must be noted however that it applies only to solid-fluid boun- daries. Levich (Discuss. Paraday SOC. 1947 1 37) has pointed out that the tangential component of the velocity remains continuous at fluid-fluid boundaries resulting in very much more favourable conditions for convective transfer. 166 QUARTERLY REVIEWS the axis is zero a t the wall again owing to the absence of slip at the solid surface but the distribution curve is now very much steeper near the wall and flatter near the axis (Fig. 2b). The transition from laminar flow to turbulence is governed largely by the value of the dimensionless group UpZ/q where U is the mean rate of flow p and q are respectively the density and viscosity of the fluid and I is a length which characterises the system (in the present case for example the diameter of the pipe).This group is termed the Reynolds number Re and if the fluid may be regarded as incompressible and 7 as independent of the rate of shear is directly related to the mean velocity of the fluid. If the fluid entering the pipe is in laminar flow the flow is laminar through- out the pipe for all Reynolds numbers below ca. 2100. Again if the flow is turbulent a t the inlet the fluid reverts to laminar flow beyond a certain critical distance from the inlet for numbers below 2100; should the pipe be shorter than this critical length which is a function of Re and of the diameter of the pipe the flow remains turbulent. On the other hand for Reynolds numbers greater than 2100 turbulence may develop.The onset of turbulence is favoured by sharp angles roughness of the walls con- strictions in the pipe etc. Once the character of the flow changes the degree of turbulence increases with increasing Re until maximum turbu- lence results. Ideally maximum turbulence should mean that the whole body of fluid is in turbulent flow. It appears however that this state is not realised in practice ; experiment (see e.g. Fage and Townend 43) indicates that between the main fluid body in turbulent flow and the wall there is always a region of laminar flow. The thickness of this laminar boundary layer depends upon the mean velocity of the fluid and upon the degree of smoothness of the wall. The transition between the laminar boundary layer in which the viscous forces predominate and the turbulent main body where the effect of the viscous forces is negligible is continuous.For kinetic studies of heterogeneous reactions the system is usually of a different type. The liquid is contained in a vessel and is stirred either with a stirrer (the propeller and paddle being the most common types) or with the solid itself often in the form of a cylinder.* The same general considerations apply subject to an important qualification. Unless the dimensions of the system are so large (or those of the solid so small) that the volume of fluid may be regarded as infinite there are now other interfaces to consider i.e. between the stirrer and the fluid and at the walls of the containing vessel. This necessarily must render the treatment more complex. Paddle-stirred systems are further complicated by the phenomenon of cavitation behind the trailing edges of the stirrer at high speeds of rotation.Hixson and his co-workers 51 have made a detailed h1 (a) Hixson and Crowell Ind. Eng. Chem. 1931 23 1002 1160 ; (b) Hixson and Luedeke ibid. 1937 29 927 ; (c) Hixson and Baum ibid. 1941 33 478 1433 ; ( d ) idem ibid. 1942 34 120 194; see also refs. 6 and 31. * Levich 44 points out that the cylinder appears to have been chosen on grounds of simplicity of flow although the system is in fact extremely complicated from a hydrodynamical point of view. BIRCUMSHAW AND RIDDIFORD TRANSPORT CONTROL 167 study of the characteristics of stirred systems. In particular they have shown that there is a lower critical Reynolds number governing the transi- tion from non-turbulent to turbulent flow analogous to the case of pipe flow.Although for most systems the quantitative treatment of fluid flow is difficult if not impossible the qualitative treatment is of significance in the study of mass-transfer rates. For laminar fluid flow parallel to the solid surface the transport of matter must be governed partly by convec- tion and partly by diffusion the role of diffusion becoming more important as the surface is approached. Moreover for turbulent flow the state more often encountered in systems of chemical interest the transport of matter to the surface can be broadly resolved into two processes ( a ) the transport by forced convection from the body of the fluid to the boundary layer and ( b ) the transport by diffusion and convection across the boundary layer to the solid surface.The important feature is that both for turbulent and for non-turbulent flow there is a region adjacent to the solid surface wherein the diffusive transport of matter is important. Thus we have a new albeit approximate interpretation of the physical nature of the layer of thickness 6 determined from the simple Nernst theory. For certain systems a quantitative treatment of mass-transfer in terms of the fluid flow is possible (see next section) ; in general however the difficulties involved in the quantitative treatment are very great and one must resort to semi-empirical methods. Quite apart from the complex geometry of many systems of practical importance the assumption that the entire resistance to mass-transfer lies within the boundary layer is sometimes incorrect.Thus Sherwood and Woertz s2 have shown that as much as 57% of the total resistance to mass-transfer may be due to the turbulent main body of the fluid. IV. The Quantitative Treatment of Mass-transfer If a t a given point in a solution the velocity components parallel to the x y and z axes are 'u v and w respectively the rate of change of concentration at that point is given by the expression ac D-+-+- - u - + v - + w - at = (ii a y 2 a2c a z 2 ( i ay ac aZ i e . &/at is the sum of the rate of change due to diffusion and of that due to convection. The velocity components are determined by the Navier-Stokes equations and by the equation of continuity together with the appropriate boundary conditions. I n general a solution is possible only for those cases where the equations of fluid motion can be reduced to a simpler form.53 Solutions have been obtained for certain systems and in one case Levich 54 5 2 Trans.Amer. Inst. Chem. Eng. 1939 35 517. 53 For a short account of the equations of fluid motion and of mass-transfer see Agar Discuss. Faraday SOC. 1947 1 26. 54 Ref. 44. See also the further papers by Levich (a) Acta Physicochim. U.R.S.S. 1944 19 117 133 ; ( b ) J . Phys. Chern. U.S.S.R. 1948 22 575 711 721 ; (c) D ~ C U S S . Faraduy SOC. 1947 1 37. 168 QUARTERLY REVIEWS has given a detailed treatment of mass-transfer under non-turbulent condi- tions. It is convenient to use cylindrical polar co-ordinates in which equa- tion (10) becomes A brief account of bhe system is given below. vr vd and vy are the velocity components a t a point ( r + y).The system consists of a flat disc of very large area the plane of which is taken as y = 0 rotating with a constant angular velocity co about an axis ( r = 0 ) perpendicular to the plane of the disc in an infinite volume of solution. Under these conditions the interface between the disc and the fluid is the lY F I G . 3 only boundary surface in the system i.e. the shape of the containing vessel is with- out effect upon either the fluid flow or the rate of mass-transfer to the disc. The very large diameter of the disc means that any effects connected with the edge of the disc may be neglected. The solution of the equa- tions of fluid motion due to von KBr~n&n,~~ leads to the following picture of the flow of fluid in the system where we consider the side of the plane for which y is positive.At large values of y vy has a constant value (- O.S862/v% where v is the kinematic vis- cosity of t'he fluid) whilst the other velocity components are zero ; i . e . a t points far dis- tant from the disc the fluid moves towards the disc with a constant velocity. On the other hand a thin layer of fluid at the surface is dragged by the disc and acquires rota- tional and radial motion Finally a t y = 0 vr and vy are zero and the rotational velocity a t any point on the disc distant r from the axis iscor. The flow is illustrated in Fig. 3 (after Levich 44). Thus there is a transition from flow essentially normal to the surface to flow parallel to the surface pointing to the existence of a viscous boundary layer. At a distance y N 2.S(v/cr))) the value of the component vd is one- tenth of the corresponding value a t t,he solid surface whereas the normal component vy is about 80% of its maximum value.This distance then gives the approximate thickness of the layer of fluid dragged by the rotating disc. For water a t room temperature and for an angular velocity of 25 radians per second the layer thickness is ca. 0-05 em. All three com- ponents are functions of v co and y ; v and v4 are also dependent upon the radial distance from the axis whereas the normal component is inde- pendent of r. Levich solved equation (11) for this system under non-turbulent flow s6 See ref. 50 p. 110. BIRCUMSHAW AND RIDDIFORD TRANSPORT CONTROL 169 conditions.* A detailed account of his treatment is beyond the scope of this Review ; it is of interest however to see how equation (11) was simplified and to examine the solution since Levich’s conclusions stress the analogy between the fluid flow in the system (momentum-transfer) and mass-transfer.The solution of the equations of fluid motion described above refers to t’he st’eady motion of the viscous fluid. For convective transfer the condition for t,he steady state may be expressed &/at = 0. Moreover from t’he axial symmetry of the system it is evident that the concenbration c must be independent of the angle $ ; the terms in q5 then disappear from (11). Finally Levich assumed that c is independent of r i.e. that c = c(y) so that equation (11) reduces to and we are dealing with one-dimensional transport. The form of equa- tion (12) suggests the nature of the solution.At large values of y vy is constant and provided this constant fluid velocity is sufficiently great (i.e. reasonably high values of m) the contribution of diffusion to the transport process will be small. This indicates a uniform concentration in the bulk of the solution. Very close to the disc however wy decreases in magnitude and the rate of mass-transfer is increasingly governed by diffusion. Pro- vided the volume of solution V is large enough t o reduce the effects of t)he vessel walls to a minimum and the diameter of tlhe disc is very much larger than the thickness of the boundary layer the solution may be applied to a systcni of finite size (13) where 6’ is the thickness of the region over which the concentration differs significantly from the bulk concentration. From the integration of (12) 6’ = ~-SOS(D/Y)*(Y/LI))).Comparison of equations (3) and (13) shows that the thickness of the Nernst layer is given by the expression 6 = 0.8936’ = 1*612D*d~0-) . * (14) I- Levich’s theory has been tested by Siver and Kabanov,21 who studied t’he limiting currents (ci = 0) a t an amalgamated copper disc cathode The solution of equation (12) has already been given (see Fig. 1). dc DA (C - Ci) dt - V ’ 0.8936’ * For this system the Reynolds number Re = a2w/v whore a is the radius of the Experiment (see e.g. ref. 50 p. 368) indicates that the lower critical R e is t [Added in proof :] Wagner ( J . AppZ. Phys. 1948 19 837) has solved the equation corresponding to eqn. (1 1) for heat-transfer in this system. In a personal communica- tion to King he points out that his treatment when applied to mass-transfer in the system leads to the expression 6 = 1.78DbQw-8.Wagner however has used the approximate solutions of the Navier-Stokes equations given by von KarmBn (2. angew. Math. Mech. 1921 1 244) whereas Levich employed the more exact solutions obtained by Cochran (Proc. Camb. Phil. SOC. 1934 30 365). The Reviewers are grateful to Professor C. V. King for drawing their attention to Wagner’s paper and for informing them of Wagner’s communication t o him. disc. - 105. 170 QUARTERLY REVIEWS rotating in various solutions ; in each case the limiting (diffusion) current was proportional to cd. A more important test however consists in the comparison of values of 6 calculated from equation (14) with those deter- mined by experiment using the simple Nernst theory.Their results for the discharge of hydrogen ions from a solution 0.002N in hydrochloric acid and 0 . 1 ~ in potassium chloride (added to eliminate ionic migration) are shown in Table 4. TABLE 4 w radians/sec. . . . . . 9.42 19.92 37.70 56-53 75.40 lo3& cm. calc. from (6) 10.16 5-24 4.20 lo3& cm. calc. from (14) 1 10.31 1 ;::! 1 5.16 1 4.22 1 ::: 1 It appears that equation (14) also applies to a rotating paddle-type specimen.22 In Table 5 the calculated values of 6 for the dissolution of zinc specimens in aqueous iodine solutions are compared with tihe values derived from the Nernst theory. From Tables 4 and 5 it will be seen TABLE 5 w radians/sec. . . . . 5.23 7-85 10.5 20.9 lo3& em. calc. from (6) . ~ 7.36 1 6-01 1 5.20 1 '::;5 1 3-68 lo3& cm. calc.from (14) . . 7.98 6.40 5.61 '4.37 3-72 that the agreement between theory and experiment is satisfactory particu- larly a t the higher rates of stirring as required by Levich's treatment. Thus for these two systems the thickness of the conventional diffusion layer (and hence of the rate of mass-transfer) can be calculated without reference to the experimentally observed rate. From this and the qualitative considerations advanced in Section 111 it is reasonable to suppose that for any system under turbulent or non- turbulent conditions the thickness of the fictitious Nernst layer (and of 6' also) may be expressed as a function of D Y and the characteristic velocity of the system U. For non-turbulent flow the form of the dependence will be similar to equation (14) ; thus Levich has shown that for laminar flow along a flat plate (or a surface having a very large radius of curvature) 6 = 3xaD*dU-i * (15) provided the dimensions of the system are such that other wall and edge effects are again negligible.For this case 6 increases as the square root of the distance x from the edge of the plate.* Since D is an inverse function of Y the small values of the powers on D and Y explain why early workers assumed 6 to be independent of these quantities and why similar values of 6 are often obtained for different reactions under the same flow condi- * Eucken's treatment for mass-transfer to a plane surface from fluid in laminar flow parallel to the surface (ref. 47) has been criticised by Levich. Eucken's solution [S = 1.24(0x/U1)4 where U is the fluid velocity a t unit distance from the surface] may be compared with (15).Experimentally Triimpler and Zeller (Helv. Chim. Acta 1951 34 952) find that S cc U-**43. BIRCUMSHAW AND RIDDIFORD TRANSPORT CONTROL 171 tions (see Table 1). Except for systems of very simple form the constant of proportionality which depends only upon the geometry of the system must be determined by experiment. The quantitative treatment of mass-transfer under conditions of turbu- lent fluid flow is very much more difficult. In addition to the problem of solving the equations of fluid motion it is uncertain to what extent the analogy between mass- and momentum-transfer is valid in this region. For the allied problem of heat-transfer Goldstein s6 regards the assumption of similarity in the velocity and temperature distributions as doubtful whilst Levi~h,~' from a t,heoretical study of mass-transfer to a smooth plate asserts that the analogy definitely breaks down.On the other hand Sherwood 58 has recently discussed the question of mass- heat- and momentum-transfer under turbulent flow conditions and concludes that there is a good measure of experimental support for the analogy. The empirical approach described in the next section suffices for most practical purposes and as will be seen shows that the dependence of 6 on D and Y is again such that one would expect 6 to have approximately the same value for quite different reactions under the same flow conditions. It is desirable that more extensive investigations be carried out on systems of simple geometric form such as those described above and pipe systems where a metal pipe reacts with a fluid flowing through it.* In practice however the geometry of the system is usually governed by other factors with the result that a solution of the equations of convective transfer is not possible.For such systems the use of dimensional analysis has proved to be of value. V. Application of Dimensional Analysis to Mass-transfer In the preceding sections attention has been confined to cases of forced convection. Although natural convection arising from variations in density due to concentration changes and from temperature gradients within the system must play some part in such cases the effect is negligible for most systems. The following discussion will also be limited to cases of forced convection although the method is applicable with modification in detail to natural convection.t Let j represent the mean transfer rate per unit surface area being defined by the relation 56 Op.cit. Chap. XV. 6* Ind. Eng. Chem. 1950 42 2077 ; see also Bedingfield and Drew ibid. p. 1164. * E.g. the dissolution of a copper pipe in ammonia solutions studied by Uchida and Nakayama ( J . SOC. Chem. Ind. Japan 1933 B 36 635) and in nitric acid and ferric chloride solutions studied by Buben and Frank-Kamenetzkii ( J . Phys. Chem. U.S.S.R. 1946 20 225). t For a more detailed account of both natural (free) and forced convection and references to the literature see Agar 5 3 and the recent study of mass-transfer under natural convection by Wagner ( J . Phys. Colloid Chern. 1949 53 1030). Ref. 54b p. 711. 172 QUARTERLY REVIEWS where A is the surface area and dm/dt is the actual rate of mass-transfer per unit area at any point on the surface.Then for transport under forced convection in systems where the solid surface is tlhe only boundary surface we require to know the dependence of j upon the characteristic velocity of the fluid U the kinematic viscosity Y (= q/p) the diffusion coefficient of the solute D the concentration difference (c - ci) = Ac and upon the characteristic length of the system 1. The dimensions and units of these quantities are given in Table 6 together with the corresponding details for the quantities involved in heat -transfer problems. 59 The dependence may be expressed as j = @(U,E,v,D,Ac) . * (17) where the form of t'he function depends only on the shape of the solid surface. For many purposes the assumption that @ is a power function of t'he form 3 = BUPI/P~VP~DP~ACP~ * (18) is satisfactory,* whence by expressing the condition that the product of the dimensions of the quantities on the right-hand side ( B being it number) must be the same as the dimensions of j the powers may be expressed in terms of any two of them e.g.in terms of p and p4 j = R u P J P 1 - l)v(l -Pl-Pa)DP,AC * (19) This can be rearranged to form a relation between three dimensionless groups. Of the possible groups the following are usually chosen since i t is convenient to have the velocity appearing in but one group TABLE 6 J Ac D V 1 U Mass- txansfer Mean transfer rate per unit area (mole/cm. s-sec.) Concn. diff. (mole/~m.~) Coeff. of diffusion (ern. z/sec.) Kinematic viscositj (em. z/sec.) Characteristic length (em.) Characteristic velocity (cm./sec. ML- zT-l ML-3 L2T-1 L2T-l L LT-1 Heat-transfer Mean transfer rate per unit area (eal./cm. 2-sec.) Heat difference (cal./cm. 3 Thermal diffusivity (cm. z/sec.) Kinematic viscosity (em. 2/sec. ) Characteristic length (em.) Characteristic velocity (cm./sec. 1 Heat. k 2 T - ' Heat.L-3 L2T-1 L2T-l L LT-1 where for heat-transfer s p A0 is the specific heat at constant pressure (cal./g.-deg.) is the density of the fluid ( g . / ~ m . ~ ) is the temperature difference (deg.) and A is the thermal conductivity of the fluid (cal./cm. -sec . -deg. ). Mass-transfer Heat-transfer Tho Nusselt number 7 . . Nu = jl/DAc NU = @!/AAO Tho Prandtlnumber t . . Pr = v/D Pr = vsp/A Tho Reynolds number . . Re = UZ/v Re = Ul/v BIRCUMSHAW AND RIDDIFORD TRANSPORT CONTROL 173 Equation (19) may then be written Nu = B(Re)a(Pr)b .* (20) where p1 and (1 - p 4 ) have been replaced by a and b respectively. Since U appears only in Re the power on Re is evidently the stirring coefficient (see Table 2). Thus a may be determined by studying t'he variation of Nu (or j ) with change of fluid velocity all other factors being held constant. If the dependence of D on Y is known 6 may similarly be determined from a study of tlhe variation of Nu wit>h change of fluid viscosity and B can be evaluated. The advantage of the method lies in the fact that comparatively few measurements are required to establish the dependence of j on the other factors. Moreover once the relationship has been established it may be used to calculate the rate of mass-transfer for values of the individual variables lying outside the ranges covered by actual experiment provided only that tJhe values of the groups fall within the ranges studied.60 The physical significance of the dimensionless groups is of interest.For mass-transfer the Nusselt number represents t,he ratio of the actual rate of mass-transfer per unit area j to the rate of mass-transfer by diffusion in the stationary fluid AcD/Z. The Prandtl number is the ratio of the fluid property governing t'he transfer of momentum by viscous effects due to a velocity gradient to the fluid property governing mass- transfer by molecular diffusion due to a concentration gradient. The Reynolds number may be regarded as the ratio of fluid momentum per unit area per unit time pU2 to the viscous drag force per unit area qU/Z against which it is balanced.Evidently Re is the group which determines the velocity distribution (see Section 111). I n practice however it is not always possibleGto design the system so t'hat the solid reactant surface is the only effective boundary surface. If this cannot be done J will also be a function of other lengthr such as e.g. in stirred systems the diameter of the containing vessel the depth of the solid reactant below the liquid-air interface etc. Provided all measurements are carried out upon the same system or upon a series of geometrically similar systems this dependence can be incorporated in the number B but where for example it is desired to compare mass-transfer to the same reactant surface in a series of vessels of similar shape but differing dimensions further dimensionless groups of the form (Z'/Z) must be included in equation (20).Comparatively little work has been done in this connection. The papers by Hixson and his co-workers 51 have 69 See e.g. Fishenden and Saunders '' An Introduction to Heat Transfer " Clarendon Press 1950; Goldstein op. cit. Chapters XIV and XV; Sherwood ref. 58; Wicke Chem. Ing. Tech. 1951 23 5. Fishend'en and Saunders op. cit. ref. 59. * Equations of another type have been proposed for presenting heat-t,ransfer results over wide ranges (see e.g. Eckert " Introduction to the Transfer of Heat and Mass " McGraw-Hill 1950 p. 140). p For mass-transfer the Nusselt and the Prandtl number are sometimes termed the Sherwood (Sh) and the Schmidt (Sc) number respectively. 174 QUARTERLY REVIEWS already been cited and Pratt has published an extensive study of the resolution of B for heat -transfer in a somewhat complicated paddle-stirred system.Comparison of equations (16) and (20) shows that the relation between 6 and the other variables can be expressed by 6 = (Z/B)(Re)-”(Pr)-b . * (21) from which since Ac does not appear in the equation follows the important fact that the thickness of the Nernst diffusion layer must be independent of the concentration difference ; 53 the same holds true for 6’ the thickness of the actual diffusion layer. It is of interest to consider the values of a found for mass-transfer under non-turbulent and turbulent conditions since the type of flow is largely governed by Re. Although very few investigations of mass-transfer in the non-turbulent region have been reported it is known that a = 0.5 for laminar flow along a plate and for the rotating disc system (see Section IV) ; the same value is found for heat-transfer to a flat plate.On the other hand a lower power (a = 0.33) expresses the dependence for heat-transfer in pipes.60 For turbulent flow the dependence of 6 on Re increases with the degree of turbulence until u reaches a maximum value corresponding to fully developed turbulence. It appears that for most systems amax. - 1 (see Table 2) although for both heat- and mass- transfer in pipes the value is again somewhat lower (amax - 0-8). The dependence of 6 on Pr illustrates an important difference between heat-transfer in liquids and gases mass-transfer in gases and mass-transfer in liquids. The Prandtl numbers for heat-transfer and for mass-transfer in gases are usually of the order of unity the highest numbers (Pr - 300) being observed in the study of heat-transfer in viscous oils.For these liquids however v is markedly dependent on the temperature with the result that the velocity distribution differs considerably from that obtain- ing under isothermal conditions. In particular the distribution may vary according to the direction of heat flow so that the theoretical interpreta- tion of the results is rendered very difficult. This difficulty does not arise for mass-transfer in liquids however and the process may be studied a t Prandtl numbers of several thousands (e.g. for aqueous solutions at room temperature Pr - 1000).62 Several workers have postulated that the value of the power on the Prandtl group should be the same both for heat- and mass-transfer under comparable condition^.^^ The theoretical value of b is 0.33 for mass-transfer in the non-turbulent region whilst the same value is suggested 64 for heat- transfer through a laminar boundary layer even when Pr is as low as 0.7.The postulate seems to be in accord with experiment although the experi- mental determinations of b for mass-transfer are relatively few. Some of 61 Trans. Inst. Chem. Eng. 1947 25 163. 83 See e.g. (a) Colburn Tram. Amer. Inst. Chern. Eng. 1933 29 174 ; ( b ) Chilton Buben and Frank-Kamenetzkii J . Phys. Chem. U.S.S.R. 1946 20 225. and Colburn Ind. Eng. Chem. 1934 26 1183. Lighthill Proc. Roy. SOC. 1950 A 202 359. BIRCUMSHAW AND RIDDIFORD TRANSPORT CONTROL 175 the recorded values for heat- and mass-transfer are given in Table 7 ; the type of flow (turbulent T or non-turbulent N.T.) is specified where known.With the exception of the first value the values fall within the range 0.3-0-5. So far as can be judged b is the same for both turbulent and non-turbulent flow and for most purposes can be regarded as having the constant value 0.33. TABLE 7. Values of the Prandtl power System Dissolution of metal cylinders in acids etc. . . . Dissolution of Zn foil in aq. iodine solutions . . . Dissolution of Zn Cd in dil. HCl and HOAc etc. . Heat-transfer in the same apparatus . . . . . Dissolution of Mg cylinders in acids . . . . . Heat-transfer in paddle-stirred system . . . . Reaction between wool and NaOCl solutions . . Dissolution of Ph*CO,H in aq.NaOH and NH . . Dissolution of Ph*CO,H in water flowing through Heat-transfer in conduits . . . . . . . . Vaporisation of liquids from a wetted wall column Dissolution of Ph*CO,H in various solvents . . . Heat-transfer in same apparatus . . . . . . Dissolution of Ph*CO,H in water etc. . . . . - - _ _ _ _ _ _ ~ _ _ ~ ~ _ _ _ ~ _ _ _ _ _ _ _ _ ~ _ ~ Dissolution of Ph*CO,H in water etc. . . pipe etc. into air Type of flow ___- b .____ 0.17* 0.27 0.3* 7 0.3 0-3* 0.3 0.3-0.5 0.3 3-0.4 $ 0.33 0.33 0.4 0.44 0-5 0.5 0.5 Ref. 65 22 66 66 45 61 67 68a 68b 69 70 71 51c 51c 61d . * King and his co-workers determine b from the relation velocity constant K Donst.. Provided v is constant (which will be approximately the case for the dilute solutions used) const. = (1 - b). 7 King and Howard's results have been recalculated by Hixson and Baum,slc who find the expression Nu = B(Re)0'41(Pr)0'4.$ The system consisted of granules of the solid packed into a tube with the fluid flowing through it. For continuous fluid flow b = 0.4 whereas for film flow b = 0.33. VI. A Classiiication of Heterogeneous Reactions Since the publication of Nernst's theory many experimental investiga- tions of heterogeneous systems have been made. Some of the processes studied are properly to be interpreted in the light of the modified form of his theory but there are numerous examples of reactions for which the observed rate is determined by the rate of chemical reaction at the inter- face. Van Name and Hill 7 2 were the first t o propose a classification of heterogeneous reactions of the type considered in this Review.On the s5 Ref. 33b. 6 6 King and Howard Ind. Eng. Chem. 1937 29 75. 67 Alexander Gough and Hudson Trans. Paraday Soc. 1949 45 1058. e * Van Krevelen and Krekels Rec. Trav. chim. (a) 1948,67,512 ; (b) 1950,69,1519. g9 Linton and Sherwood Chem. Eng. Progress 1950 46 258. '" McAdams " Heat Transmission " 2nd edn. McGraw-Hill 1942. 71 Gilliland and Sherwood Ind. Eng. Chem. 1934 26 516. 7 2 Amer. J . Sci. 1916 42 301. Results recalculated by King and Howard ref. 66. 176 QUARTERLY REVIEWS basis of the rate-determining step they divided heterogeneous reactions into three types (i) The rate of chemical reaction a t the interface is very much faster than the rate of transport of reactant to or of products from the surface. The observed rate is therefore determined solely by the rate of the slowest transport process.It seems preferable to term these “ transport-con- trolled ” processes rather than “ diffusion-controlled ” since mass-transfer occurs both by diffusion and by convection. (ii) The rate of chemical reaction at the interface is much slower than the rate of either of the transport processes and hence determines the observed rate. These may be termed “ chemically-controlled ” processes although the term ‘( activation-control ” is sometimes used. (Since we are concerned with the r81e of transport in heterogeneous processes it is con- venient to term all steps in which transport plays no part e.g. the actual processes of solution or deposition a t the solid surface adsorption etc. chemical processes.) (iii) Both rates are of the same order of magnitude and the observed rate is determined by some function of the two.These have been termed reactions of intermediate type. From general kinetic principles the distinction between the possible types of system will not be sharp rather would one expect heterogeneous reactions to show a gradation between the limiting cases of chemical and transport control. Van Name and Hill’s third class is evidently the general case. It is noteworthy that whereas the true rates of chemical reaction a t interfaces vary over a very wide range the overall rates range from extreme slowness to an upper limit determined solely by the rate of a transport process; 33b the same is true of course for homogeneous reactions in the liquid phase.73 Chemically-controlled Reactions.-Since we are only incidentally con- cerned with reactions of this class a brief treatment will suffice for our present purpose.For chemically-controlled reactions we have the condition that the concentration of solute is uniform throughout the fluid body.* The solute molecules must collide with the surface before reaction can occur hence the simplest case takes the form of a bimolecular process in which the (‘concentration ” of the solid is represented by the surface area. For a given area moreover the rate will again be inversely proportional to the volume of solution so that the rate may be expressed - dc/dt = k ~ A c n / V * (22) where kc is the chemical velocity constant per unit area a t unit volume 73 Christiansen J . Colloid Sci. 1951 6 213. * This is not strictly true when adsorption occurs at the surface but adsorbed layers are of negligible thickness compared with the usual thickness of the diffusion layer.Adsorption processes will not be considered however except for cases where the reaction products are adsorbed on the surface so reducing the area of surface available for attack. BIRCUMSHAW AND RIDDIFORD TRANSPORT CONTROL 177 and n is a power expressing the order of the reaction. For many reactions of this type n = 1 reactions of higher order being uncommon. Centner- szwer claimed that the reaction between tin and hydrochloric acid is of the fourth order but the evidence on which the claim is based seems doubtful. 5 The General Case. Reactions of Intermediate Type.-When the con- centration of solute at the interface is neither the same as that in the bulk of the solution nor equal to the equilibrium concentration (where the latter implies the solute concentration after an infinite time of reaction excess solid being present) the observed rate is a function both of the rate of chemical reaction a t the interface and of the rate of a transport process.Many workers have proposed extensions of t3he Nernst theory to reactions of this type.74 In essentials these treatments are very similar. The rate of the chemical proceas may be expressed as - dci/dt = kcA~y/V . * (23) where ci is the concentration of the solute at the interface. For the trans- port process - dc/dt = k ~ A ( c - Ci)/V . * (3') Then for the steady state kC'cin = E,(c - ci) from which ci may be expressed in terms of the remaining quantities. For first-order chemical reactions ci = ckT/(kC + kyl) and substituting in equation (3') we have Then k = kCkT/(k 4- k T ) .* (25) where k is the observed velocity constant per unit area a t unit volume so that if the chemical process is of the first order with respect to the solute other criteria must be used to diagnose the rate-determining step as was pointed out by Van Name and Hill.72 It will be evident that when E > E, the observed rate is determined solely by the chemical process at the interface. Similarly the condition k < k represents the case of transport control. The reported cases of "pure" systems are those in 74 E.g. (a) Heymann 2. physikal. Chem. 1913 81 204 ; (b) Tu Davis and Hottel Ind. Eng. Chem. 1934 26 749 ; (c) Damkohler in "Der Chemie-Ingenieur " Akademische Verlagsgesellschaft Leipzig 1937 Vol.111 p. 413 ; ( d ) Frank-Kame- netzkii J . Phys. Chem. U.S.S.R. 1939 13 756; Acta Physicochim. U.R.S.S. 1940 12 9 ; ( e ) Kimball J . Chem. Phys. 1940,8 199 ; (f) Tovbin J . Phys. Chem. U.S.S.R. 1946 20 1435; (g) Zdanovskii ibid. p. 869; (h) Tanaka and Tamamushi Bull. Chem. SOC. Japan 1949 22 187 ; (i) Hochberg and King J . Electrochem. SOC. 1950 97 191 ; see also refs. 16 and 72. The treatment of fast electrode processes when the current is subjected to a small alternating current relative to the solution (Randles Discuss. Faraday SOC. 1947 1 11 ; Ershler ibid. p. 269) is essentially similar. The original suggestion that the concentration a t the interface may not be the same as the equilibrium concentration was made by Berthoud J . Chim. phys. 1912 10 624. * When the chemical reaction at the interface is of higher order the expressione are more complex ; for second-order reactions - * = ""c( 1 + p - d#3 + :) dt V where fi = (rC~/kcc).'4~ 178 QUARTERLY REVIEWS which one or other of these conditions is satisfied under the particular experimental conditions employed.In certain circumstances a change in the experimental conditions effects a change of control ; this is considered in the following section. There are however two further points to be considered in connection with the simple treatment given above. It is assumed that the areas appearing in equations (23) and (3') are the same but in general this will not be true. As mentioned in Section 11 the rate of a transport-controlled reaction is proportional to A the apparent surface area.75 For chemically-controlled reactions however the rate is a function of the true surface area A,. If the fraction of this area avail- able for attack is o the rate is proportional to oA,. With these modi- fications equation (25) becomes k = kCkT/[kC + kT(A/gAC)] - (26) where lc is now the observed unit constant based upon the apparent area. The conditions for the limiting cases are now (i) Ic,oAc > ETA for transport control and (ii) k,oAc < ETA for chemical control i.e. the conditions are based upon the overall rate constants.* The distinction is of importance since although the apparent area A will usually change so slowly during the course of reaction (particularly for the case of the massive solid) that it may be regarded as constant yet the available surface oA may increase owing to progressive roughening of the surface.Secondly since the thickness of the Nernst diffusion layer is independent of the concentration difference at the surface (Section V) the modified Nernst theory may be applied to k for reactions of intermediate type. For the simple systems discussed in Section IV k may be calculated and Ic then determined from the observed rate constant by means of equation (25). For other systems lc must be determined experimentally. I 0-88 0.82 0.82 $ 0-86 0*86$ 0.244 TABLE 8. The dissolution of metal cylinders in acid solutions (3.5 x 10-3~-p-Benzoquinone ; metal cylinders 2 X 2-54 em. ; 25" ; 3200 r.p.m. ; 250 ml. of solution. Experiments conducted under nitrogen.) Acid or buffer ~ Metal I k (cm.min.-l) O ~ ~ M - H C ~ O.O5~-glycine . . . O.~M-HOAC,~ 0.lM-NaOAct .. O-IM-HOAC 0-1M-NaOAc . . . O - ~ M - H C ~ . . . . . . . - O-~M-HOAC 0-liu-NaOAc . . . 0.03M-HC1 0*07~-KHPh" . . . Cd Cd Cd Pb Sn c u 1- ._ * Phthalate. t Acetate. $ Estimated from initial rate. 7 5 See ref. 18. Laitinen and Kolthoff ( J . Phys. Chem. 1941 45 1061) find this also to hold for mass-transfer under natural convection. * Equation (24) may be written - dc/dt = KcK,c/V(Kc + KT) where KT = ETA and Kc = ECUA c are the overall transport and chemical constants respectively. The use of the unit rate constants (being independent of area) seems preferable however although the separate evaluation of kc E T and A/aAc may not be possible for a given system. BIRCUMSHAW AND RIDDIFORD TRANSPORT CONTROL 179 As an example of the latter we may consider the dissolution of metal cylinders rotating in acid solutions in the presence of p-benzoquinone as a depolariser studied by Hochberg and King.74i The experimental con- ditions and results are detailed in Table 8.The first five results suggest that the observed rate is determined solely by the rate of a transport process hence k1 = k = 0.85 cm.min.-l. For copper however the rate constant is smaller (k = 0-244 cm.min.-l) ; from equation (25) the differ- ence between A and oA being neglected E = 0.34 cm.min.-l. The authors give further examples in their paper. VII. Factors affecting the Observed Rate To conclude this Review the factors which affect the observed rate must be considered. With the exception of temperature and of the initial concentration of solute which are considered separately at the end of this section the factors may broadly be divided into those affecting the chemical rate and those affecting the rate of the transport process.King,46 Kress- man and K i t ~ h e n e r ~ ~ and Zimmerman 77 have discussed the question of suitable criteria for diagnosing reaction type. A change in the rate-determining step will be described as a change of control this may be from transport control to the intermediate type of control for example or from one form of transport control to another.78 More generally an alteration in the dependence of the observed rate con- stant on the rate constants of the primary processes an alteration which may or may not result in a change of control will be termed a shift of control. Factors aecting the Rate of the Chemical Process.-If the true surface area A increases during the course of reaction the overall chemical rate constant Kc will increase with shift towards transport control.In fact varying the method of preparation of the solid surface provides one test of whether the chemical rate exerts any control on the observed rate,* provided the critical value of Re corresponding to the surface roughness (see below) is not exceeded. It is known moreover that E is often depen- dent on the particular crystal face exposed to attack (see Section 11). The preferential development of certain crystal faces during the course of reaction may result in an increase in the mean value of kc reinforcing the effect of the increase in A,. As we have seen E is independent of the crystd face under attack. For cer'Bn systems the observed rate is decreased by the addition of small qur itities of other substances.This effect in some instances is due 76 DiscuSs. Paraduy SOC. 1949 7 90. J . Phys. Colloid Chem. 1949 53 562. 78 E.g. (a) Boyd Adamson and Myers J . Amer. Chem. SOC. 1947 69 2836; ( b ) Alexander Gough and Hudson Trans. Faraduy SOC. 1949 45 1058 1109 ; Alex- ander and Hudson J . Phys. Colloid Chem. 1949 53 733. * This test is particularly useful in the case of metal surfaces (see e.g. Salzberg and King J . Ebectrochem. SOC. 1950 97 290 ; Bircumshaw and Riddiford J . 1951 698) ; for non-metallic solids submicronic disintegration of the surface (see Traube and v. Behren 2. physikal. Chem. 1928 138 A 85) when this occurs is an added complication. N 180 QUARTERLY REVIEWS to complex formation between the solute and the foreign substance as in the case of the depression of the polarographic diffusion current of cadmium by bovine serum albumin studied by Tanf01-d.'~ The increased size of the solute species results in a decrease of the diffusion coefficient and hence of ZC (see below).In other cases the effect has been traced to the adsorp- tion of the foreign substance on the surface with consequent decrease in a the fraction of the area available for attack and hence in Kc. Thus Marc 38 found that the addition of small quantities of certain dyes e.g. quinoline-yellow reduces the rate of crystallisation of salts ; Paine and Prance 80 have reported similar effects for the growth of alum crystals from solutions containing diamine sky-blue. The inhibition of corrosion of metals has long been known and Jenckel and Braucker 81 have shown that 16-naphthaquinoline effects a reduction in the rate of dissolution of aluminium in hydrochloric acid solutions.The reduction in Kc will result in a shift towards chemical control provided that a t the surface the rate of transport of solute parallel to the surface is too fast to influence the observed rate i.e. provided the apparent surface area is unaltered. Owing to the fact that fluid motion persists up to points very near to the surface (Sections I1 and 111) such that convection parallel to the surface is still marked although convection normal to the surface is negligible this assumption would appear reason- able for systems subject to forced convection. There seems little doubt that the assumption is generally valid.Thus Volmer 82 demonstrated the existence of an adsorbed layer on crystals grown from the vapour or from the melt and showed that the molecules in this layer are very much more mobile than in the melt. Moreover a lateral flow of solute along the surface of a crystal must be assumed to explain the two-dimensional growth of crystals from aqueous s0lutions,8~ for whereas the normal concentration gradient a t the surface of the growing crystal varies from a maximum at the face centre to a minimum a t each edge the crystal face has generally the same linear rate of growth all over the face. It should be noted that the influence of such added substances in itself offers no proof that the undisturbed system is subject to any degree of chemical control although this has sometimes been assumed ; as we have seen the effect is to be interpreted as a shift of control.The magnitude of the shift may of course be increased by coupling the use of such foreign matter with a suitable alteration of the experimental conditions e.g. Jenckel and Braucker 81 found that the reduction in the rate of dissolution of aluminium in the presence of /?-naphthaquinoline was enhanced by an increase in the concentration of hydrochloric acid and by a decrease in temperature (see below). 79 J . Arner. Chern. Xoc. 1951 73 2066. *O J . Phys. Chem. 1935 39 425. slZ. anwg. Chern. 1935 221 249. 82 Trans. B'araday SOC. 1932 28 359. 8s (a) Bunn Discuss. Paraday Xoc. 1949 5 132 ; (b) Berg Proc. Roy. SOC. 1938 A 164 79 ; (c) Humphreys-Owen ibid. 1949 A 197 218 ; Discuss. B'aradccy SOC. 1949 5 144.In this discussion recent work on the effect of impurities dyes etc. on crystal growth is reported. BIRCUMSHAW AND RIDDIFORD TRANSPORT CONTROL 181 Closely related to this effect is the case where accumulation and/or adsorption of the products affects the observed velocity. The adsorption of a soluble product may be considered f i r ~ t . 8 ~ It is assumed that adsorp- tion and desorption are rapid compared with the rate of reaction. Then the available fraction of the surface may be expressed by 0 = 1/(1 + k’x) . - (27) where k f is a constant and x is the concentration of product a t time t . The rate of reaction is then (28) where El = IcckT/(kc + E,A/Ac) and k” = k‘/(l + Ic,Ac/E,A). If x = 0 a t t = 0 El has the same significance as in equation (26). For strong adsorption of the product CJ will decrease very rapidly and the observed effect will be a shift towards chemical control as in the case of added foreign substance.On the other hand if the product is less strongly adsorbed CT will decrease more slowly and the observed effect will be a progressive fall-off in the overall rate ie. an apparent change of order. Salzberg and Kings5 have considered the more complex case when both reactant and product are adsorbed on the surface adsorption equilibrium not being maintained whilst dissolution is proceeding. Secondly there is the case when the product accumulates as an insoluble layer on the surface. If this layer is impermeable to the solvent the mechanism will undergo a marked change; diffusion of reactants either way through the film may or may not now influence the observed rate.This case differs so markedly from the type of system under review that it will not be further considered.*6 For the case when the coating is per- meable to the solvent it seems probable that both the chemical and the transport rates will be affected. There is no doubt that the film will modify the characteristics of the diffusion layer owing to immobilisation of the solvent within the layer of product possibly to the extent that 6 becomes independent of the fluid vel0city.~7 Certainly it appears that for several systems of this type the film offers additional resistance to mass-transfer.22 Factors affecting the Rate of Transport.-As mentioned above a change in the roughness of the solid surface will be expected to affect Ac and hence the overall chemical rate constant rather than the corresponding quantities A and ETA for the transport process.Nevertheless the magni- tude of Ic may be affected in certain circumstances. From the standpoint of dimensional analysis the roughness of the surface may be expressed as a dimensionless group termed the relative roughness defined as the ratio of the mean height h of the surface irregu- 84 Cf. Prutton and Day J . Phys. Colloid Chem. 1949 53 1101 and ref. 74i. 86 For a review of solid-gas systems of this type see Evans J. Corrosion 1948 5 No. 4 16. For an interesting study of the change from transport control to control by diffusion through a compact layer of product see Jaenicke 2. Elektrochern. 1951 55 186. J. Electrochem. SOC. 1950 97 290. *’ E.g. Bircumshaw and Everdell J. 1942 598.N* 182 QUARTERLY REVIEWS larities to the characteristic length of the system. Equation (20) (Section V) then becomes Nu = B(Re)"(Pr)b(h/Z)" * (29) For fluid flow through pipes the characteristic length being taken to be the radius of the pipe Nikuradse88 has shown that relative roughness over the range 1/507 to 1/15 is without effect on the friction factor in the region of laminar flow. For turbulent flow however the relative roughness effects a considerable increase in the friction factor the effect becoming noticeable a t a particular critical Reynolds number as the relative roughness increases so the critical Re value becomes smaller whilst the increase in the friction factor becomes larger. Similar effects are t o be expected in stirred systems. Thus we would expect the surface roughness and hence the method of preparation of the solid surface to be without effect on k in the region of non-turbulent flow (i.e.m = 0). In the turbulent region however k may well depend on the method of preparation of the surface may be unusually sensitive to the rate of fluid flow over a certain range of velocities and further may alter during the course of reaction owing to progressive roughening of the surface. An increase in the friction factor may be regarded most simply as an increase in the characteristic velocity of the system ; hence the effect of an increase in surface roughness where apparent will be to increase the value of k with consequent shift towards chemical control. Since many of the pub- lished investigations describe systems in turbulent flow this effect clearly merits attention.From the discussion in Sections TV and V the dependence of k on the coefficient of diffusion of the solute D tlhe kinematic viscosity of the fluid v and the fluid velocity U may be expressed by k T = F( U,D,l/y) . - (30) where D and v are not independent variables.89 Hence an increase in D (or decrease in v) will increase E with a corresponding shift towards chemical control. The dissolution of magnesium in ethanol-water solutions of acetic acid,g0 and that of silver in mineral oil solutions of sulphur,91 appear to be examples of this effect. Similarly an increase in the fluid velocity will cause a shift towards chemical control. For some 7G it is observed that as the rate of stirring is increased the observed velocity constant reaches a limiting value and thereafter is independent of the fluid velocity.This may be due to a change to chemical control or to the stirrer's having reached maximum efficiency; 7G for any given case other criteria must be applied to distinguish between these possibilities. In particular the use of baffles in stirred systems is a convenient way of promoting turbulence and so increasing the efficiency of the stirrer. 1933 4B July/August. The remaining factors need little discussion. 8 8 Forschungsheft No. 361 Suppl. to Forschung auj dem Qebiete des Ingenieurwesens 89 Bircumshaw and Riddiford J. 1951 1490. go Roehl King and Kipness J . Amer. Chem. SOC. 1941 63 284. 91 Foley Morrill and Winslow J . Phys. Colloid Chem. 1950 54 1281. BIRCUMSHAW AND RIDDIB’ORD TRANSPORT CONTROL 183 Finally since k may depend upon the geometry of the apparatus (Section V) it will sometimes be possible to vary k+ by using a different type of stirring or by altering the dimensions of the apparatus.The Initial Concentration of Reactat.-Since 6 is independent of the concentration difference at the surface and D in comparison with the other variables is but slightly dependent on concentration an increase in the initial concentration of reactant will be accompanied by an increase in the maximum rate (corresponding to ci = 0) a t which reactant can be transported to the surface. Then if for a transport-controlled reaction the initial concentration is increased a point may be reached a t which the maximum transport rate exceeds the rate a t which chemical reaction at the surface can deal with the supply.A change in the initial con- centration might thus result in a change of control. A clear example of this effect is provided by the exchange adsorption of cations from aque- ous solutions by phenol-formaldehyde resins ; 78n for 0-001M-solutions the observed rate is determined by the rate of transport of cations to the resin surface whereas a t 0-lM-concentration diffusion through the resin is the rate-determining step. The reaction between wool and potassium per- manganate solutions between wool and chlorine solutions 78b and the exchange of 64Cu between a single-phase copper amalgam and an aqueous solution of cupric The Temperature C~efficient.~~-An increase in temperature will increase both the rate of chemical reaction a t the interface and the rate of the transport process.It is convenient to begin therefore by considering the effect on the unit rate constants for each step and then to consider the temperature coefficient of the observed unit rate constant. For the chemical reaction a t the interface the dependence of Ec on temperature may be expressed by means of the Arrhenius equation Jcc = ZCe-Ec/RT . * (31) where the pre-exponential factor is of the order of magnitude of the collision frequency per unit area for molecules colliding with the surface E is the apparent critical increment for the chemical process,g4 and the remaining symbols have their usual significance. In practice it is found that E varies from a lower limit of zero. Similarly it is found that the dependence of ET on temperature may be expressed as kT = ZTe-%lRT . * (32) On the simple Nernst theory E = ED the activation energy for diffusion this conclusion as we have seen is based on the assumption that 6 is temperature-invariant.For solid-liquid systems however we may write ij cc U-flv(a - h ) D b . * (33) are probably further examples of this effect. 92 Kayas Conapt. rend. 1948 226 2144. 93 Riddiford J. Phys. Chem. in the press. s4 Cf. Hinshelwood “ The Kinetics of Chemical Change ” Clarendon Press 1945 p. 214. 184 QUARTERLY REVIEWS (see Section V) i.e. for constant fluid velocity d(lns)/dT = ED - (a - b)E,]/RP where E is the activation energy for kinematic viscous b are temperature-invariant 6 will vary exponentially the dependence however will be slight in comparison of the other quantities.* From equation (34) ET = (1 - b)ED + (a - b)Ey .- (34) flow. Since a and with temperature with the variation since k = D/S. from 0.5 for non-turbulent flow to 1 for strongly turbulent flow. ( 3 5 ) may then be written From the discussion in Section V b - 1/3 and a varies Equation ' ( 3 6 ) and since ED and E are of the same order of magnitude E will be some- what smaller than ED for non-turbulent or mildly turbulent flow. On the other hand E may be larger than ED under strongly turbulent condition^.^' As a rough guide ET -ED when a = 2/3. From equation (25) the dependence of the observed unit constant on temperature may be expressed by ET = ED + E,,)/6 for non-turbulent flow = ED + E1,)/3 for strongly turbulent flow * If an experimental energy of activation E, is defined by the relation d(lnk,)/dT = E A / R T ~ E may be determined at any value of T from the tangent to the curve of Ink plotted against 1/T ; then from equation (37) and in general the variation of E with temperature is governed by the variation of the ratio kc/kT.Since in practice Zc>ZT we may distin- guish three cases (i) Ec = ET = E, for which case Ic varies exponentially with T ; (ii) E < E, which corresponds to the condition k > ET a t all temperatures i e . E = E at all temperatures ; (iii) E > ET when at low temperatures the observed rate will be chemically-controlled whereas a t high temperatures the process will be transport-controlled. For case (iii) then an alteration of temperature will result in a shift of control; if a sufficiently wide temperature range is practicable a change of control will be observed. The combustion of carbon spheres in a flowing air stream 74b the catalytic oxidation of sulphur dioxide on platinum pellets,95 and the dissolution of very pure iron in are examples of systems conforming to case (iii).9 6 Olson Schuler and Smith Chem. Eng. Progress 1950 46 614. s6 Abramson and King J . Amer. Chem. SOC. 1939 61 2290. * No detailed study of the temperature coefficient of 6 under forced convection appears to have been made. Laitinen and Kolthoff (J. P h p . Chem. 1941 45 1079) conclude that the effective thickness of the diffusion layer a t a rotating platinum microelectrode remains practically constant for small temperature changes (see also Jaenicke 2. Elelitrochem. 1951 55 648). BIRCUMSHAW AND RIDDIFORD TRANSPORT CONTROL 185 The Reviewers wish to thank all the many authors who have so kindly sent them reprints of their publications and they have pleasure in acknow- ledging their debt to Miss E.Koutaissoff for her kindness in translating several Russian papers. Figs. 1 and 3 are reproduced by kind permission of the Academy of Sciences of the U.S.S.R.

ISSN:0009-2681    

DOI:10.1039/QR9520600157

出版商:RSC    

年代:1952

数据来源: RSC

摘要:

!WE ENERGETICS OF REACTIONS INVOLVING HYDROGEN PEROXIDE ITS RADICALS AND ITS IONS By M. G . EVANS N. S. HUSH and N. URI IN discussions of free-radical reactions in solution it is necessary to know the energetic and the thermodynamic quantities involved in the primary steps. In aqueous solution these reactions frequently involve ions and ionised species of atoms and free radicals. In recent years great attention has been focused on the free radicals and ions arising from water and hydrogen peroxide. These species are H OH HO, O, Hf OH- HO, and 0,. To understand the reactions of these entities in aqueous solution with each other and with metal ions we need fundamental quantities such as bond-dissociation energies electron affinities ionisation potentials sol- vation energies and the corresponding entropy changes.In this Review we present values of the above quantities which we consider to be most firmly based and show how these quantities are derived from experimental data. We also give tables of AGO AH" and Axo for most of the reactions of the above entities amongst themselves and with metal ions in different oxidation states which have been studied directly or discussed as steps in an overall reaction scheme. I. The Bond-dissociation Energies in Hydrogen Peroxide.-(a) The HO.. ..OH Bond-dissociation Energy. This has been obtained from the following cycle Qf -.Ha/ HZ + O2 / - D o . - "f2.,.,,, 2 H + 2 0 - 20H DOH whence where the values of DH, Do, and DOH used are those given in Table 1 ; DHO....HO is evaluated as 55.6 kcal. This value refers to the gas-phase reaction.Assuming that the heat of solution of OH is similar to the heat of condensation of water we obtain for the solution phase a bond-dissociation energy of 46.7 kcal. (the heat of hydration of liquid H,O being taken as 0.7 kcal. ; Evans Baxendale and Uri I). ( b ) The HOO .... H Dissociation Energy. This may be evaluated from the following cycle DHO .... OH = Qf + DH + Do,-~DoH . * (1) -DHo .... H "i'"" - H +O*H - ( 2 + S)H,O ( - 1 H + s H + ) 1 i ( E + S h o z (H2Oz)aq. - Hi&. + HOzaq. Q Trans. Faraduy Soc. 1949 45 236. 186 EVANS HUSH AND URI REACTIONS INVOLVING HYDROGEN PEROXIDE 187 whence DHO .... H = - Q - ( A + S)H~O + ( - IH + SH+) + WHO + SHO,-) (2) The value of Q has been measured by Evans and Uri; the value of ( A + S)H,O is taken as 11.6 + 0.7 = 12-3 kcal.An idea of this value may be obtained from the energy of the absorption maximum in the spectrum of the Fe3+H02- ion-pair complex in solution. The absorption spectra of ion-pairs of the type Fe3+X- have been described as electron-transfer spectra and the energy of the absorption maximum identified with the energy change of the reaction Fe3+X- % Pe2+X. Under these conditions because of the Franck-Condon principle the position of the nuclei both in the ion-pair itself and in the solvation shell will be unchanged. The terms entering into the absorption energy can be seen from the following cycle The unknown quantity remaining in this equation is ( E + A!Y)~~,. --D8Fer+X Fea+X ___j Fez+ + X - N h V / Jm/ 1- (E +SIX in which DSFe3+X- is the heat of formation of the normal ion-pair from the separate ions in solution.I)8Fea+X is the heat of formation of Fe2+X in a configuration in which the centres and the water molecules have exactly the same positions as in Fe3+X-. .DFe*+X will therefore differ from the normal heat of formation of this entity by energy terms arising from the fact that neither the centres nor the water molecules are in their normal equilibrium configuration (3) From direct measurements 130 of heats of formation of ion- pairs we know the value of DFe3+X- and over the range I4O of ions F- Cl- OH- H02- Br- this value does not vary 2 150 by more than & 5 kcal. 5 That the energy terms aris- - 7 160 ing from the non-equilibrium 3 configuration of Fe2+X are considerable is seen from 170 - a comparison of Nhv and in cases where J - ( E + S) no no loo 90 80 Rl is known with some cer- A,,,.(kcal.) o f Fe3+X" tainty.However in spite of this it is an experimental fact that the value of ( E + S) is approxi- mately linear with Nhv (see fig.). This experimental observation encourages 2 Trans. Paraday SOC. 1949 46 220. - DsFea++x- Fe3+X- - Fez+ + X- - Nhv = - D'FeS+X- + DFea+X + J2-3 - ( E + S)X . X - C-DFeS+-X + J - ( E + S)XI 1880 188 QUARTERLY REVIEWS us to interpolate a value of (E + AS)^^ from the known energies of the absorption maxima. This method leads to a value of ( E + X)HO = 136 kcal. which used in equation (2) leads to a value of 102 kcal. for the bond- dissociation energy HO ,....€€. This value seems to us to be-as we shall discuss later-in conformity with chemical evidence ; although it has been derived through an argument involving values of (E + X) for various ions the value DH....O,H is independent of the particular way in which the overall energy of the reaction Fe2+ + $X -+ Fe3+ + X- is split up between J2-3 and ( E + X),.From the cycle (2) it is seen that the value of DHO,....H = 102 kcal. applies to the gas phase and in obtaining the corresponding value for bond dissociation in solution we have identified the heat of solution of HO with that of hydrogen peroxide. This is obtained from the following cycle ( c ) The DH.e..o Bond-dissociation Energy. -D H....o*H - H+HO th .... 0 3 + HifTg -DHp H + 0 - H + H + O DH .... 0 = - DH .... O,H + Qf + DH . * (4) Using the values of bond dissociation obtained above together with the data in Table 1 (a) we have evaluated the bond-dissociation energy DH....02 as 3578 kcal.As we have indicated in the above discussion when we obtain heats of dissociation or reaction in the gas phase from measurements in solution heats of solution of free radicals are required. There are no direct measurements of such heats and we have had to make the assumption shown in Table l(b). TABLE 1 * ( a ) Dissociation processes whence 104.1 IlHOH-+H + O H . . o,-,o+o. 118.2 IIHO - H + O . . Reaction 1 D (kcal.) Reaction 1 D (kcal.) 120.7 100-1 Radical or molecule . . . . . . . . . H O . O H . . H O . . . . . . H,O . . . . . ( b ) Heats of solution of radicals and molecules Q (kcal.) I Radical or molecule Q (kcal.) H,. . . . . . . 1.4 H . . . . . . . 1 ::: 1 12.3 0 . . . . . . 3.9 12.3 11 0 . . . . . . . 2 EVANS HUSH AND URI REACTIONS INVOLVING HYDROGEN PEROXIDE 189 Reaction ( 1 ) Reactions in gas p b e OH+H,O -+H,O+HO .OH+HO +H,O+O . . H02+ H,O + H,O + OH+ 0 H02+H0 +O,+H,O . . OH+O +HO,+O . . HO,+H +H,O,+H . . (2) Reactions in solution phase OH+H,O +H20+H0 . OH+HO +H,O+O . . H0,+H202 -+ H,O + OH+ 0 H02+H0 -+H,O,+O . . TABLE 2 Reaction 19 85 30 67 - 65 -2 19 77 30 58 HO S H + O . . . H,O -+H+HO . . . OH+OH-+H,O . . . OH+OH+H,O+O . . . OH+H2 +H,O+H . . . OH+OH-+H,O . . . OH+OH+H,O+O . . . OH+H -+H,O+H . . . Heat (kcal.) - 36 - 102 66 21 17 47 12 18 II. Heat Free-Energy and Entropy Changes of Ionic Reactions involving Hydrogen Peroxide in Aqueous Solution.-( a) Xtartdard Entropies of Gaseous Radicals and Molecules. We have taken the standard entropies * ( S O ) at 1 atmosphere pressure and 298" K. of OH fg) H (g) H,O (g) and 0 (g) to be 43.9 31.2 45.1 and 49.0 e.u./mole respectively and have estimated a value of 50 e.u./mole for HO (g) and H,O (g).( b ) Heats Entropies and Free Energies of Gaseous Molecules and Radicals and Related Data. Where Go is the standard free energy of the solute gas at one atmosphere pressure and @' its standard partial molal free energy in aqueous solution then a t concentrations such that = H" we have for the free energy of hydration AG(hyd.) = Go - Go = - RT In a where a is the activity of the species in solution. The entropy of hydration is calculated either from the relationship AS(hyd.) = - [AG(hyd.) - AH(hyd.)]/T or where the temperature coefficient of solubility is known from the relationship - AS(hyd.) = d[AG(hyd.)]/dT. The entropy of hydration of gaseous molecules and radicals is due almost entirely to the difference in translational entropy of the species in the gas and in the solution phase and we have estimated the entropies of hydration of HO TABLE 3 - H2O (g) *H,O (1).. . . . . Ha08 ( g ) +H2O (1) . . . . . HZO (1) + HZ02 (as-) - H,O (aq.) -+ H+ (aq.) + HO,- iaq:) H,O (aq.) + H+ (aq.) + OH- (aq.) . H (g) - + H 2 (aq.) . . . . . 0 (g) -+ 0 (aq.). . . . . . HO(g) +HO(aq.) . . . . . O,H (g) + 0,H (aq.) . . . . . AH kcal./mole - 10.5 - 11.6 - 0.74 8-2 13.3 - 1.42 - 3-85 - 10 - 12 AS e.u./mole - 28.4 - 27.2 8.4 - 25.7 - 27.3 - 18.8 - 26.0 - 25 - 25 AO kcal./mole - 2.05 - 3-50 - 3.24 15.8 21.6 4.18 3.90 - 2.5 - 4.5 4 Latimer " The Oxidation States of the Elements and Their Potentials in Aqueous Solution " (Prentice-Hall 1938). 190 QUARTERLY REVIEWS and OH by analogy with the inert gases (He - 19 ; Ne - 22 ; A - 23 ; Kr - 26 ; Xe - 27 ; Rn - 29 e.u.) as - 25 e.u.The heats of solution of OH and HO have here been taken t o be approximately - 10 and - 12 kcal./mole respectively (cf. Table 1). The heats and entropies of solution of the monatomic halogen atoms have been assumed to approxi- mate to those of OH. The more important values are listed in Table 3. Ionisation and vaporisation data are included here for reference. The standard entropies go (9 = x - R In a) are referred to So(H+ aq.) = 0 at 298" K. The entropies of OH-(aq.) and 0,H-(aq.) may be obtained from the entropy cycles ( c ) Xtundard Partial Molal Entropies of Aqueous Ions. A S W A s h (H+) HZ02 (1) + A s h (HO o-) - ASdf HzOz (aq-) - H+ (aq.) + HO2- (aq.) ASt (aq.) whence where AS, AS, ASi(,) AS (H+) AXh (OH-) AS (H0,-) and AS (aq.) are the entropy changes respectively on dilution vaporisation gaseous ionisation hydration of H+ hydration of OH- hydration of H0,- and aqueous ionisation.- A& + AS + + ASh (H+) + Ash (OH- or H0,-) = A& (aq.) For OH- we have Ash (OH-) = - A&(,) - Ash (H+) - AS + ASd + ASi (84.) = - 22 + 26.0 - 28.4 + 8.0 - 27.3 = - 43.7 whence L!?(OH- aq.) = - 2.5 B.U. we have assumed that the difference between the sum of the vibrational and rotational entropies of H,O and H0,- is small and is approximately the same as the difference for H,O and OH-. In order to calculate P(HO,- aq.) we require the value of We then have ASh (HO2-) = - A&g) - A s h (H') - AS + ASa + A& (rtq.) = - 22 + 26.0 - 27.2 + 8.4 - 25.7 = - 40.6 Eley Trans.Faraday SOC. 1939 86 1283. EVANS HUSH AND URI REACTIONS INVOLVING HYDROGEN PEROXIDE 191 whence P(HO,- aq.) = 7 0.u. It is assumed that the value of % for 0,- (aq.) is close to that for H0,- (aq.). Values of AH AG and A 8 for reactions involving the bi- and ter-valent Fe Co Cr and V ions are given in Table 4. M(aq.)'+ + H+(aq.) + M (aq.)3+ + &Hz (g) - * ( 5 ) which are involved have been estimated in different ways. For Fe AH(5) is taken to be 11.3 kcal./mole * and AG(5) to be 17.8 k~al./mole.~ The corresponding entropy change is - 21.8 e.u./mole which is in agreement with Latimer's ionic entropies (Fe2+ - 25.9 ; Fe3+ - 63.3 e.u.). The thermochemical values of AH(5) for Co Cr and V are less accurately known than are the free energies consequently for these ions the experimental free energies t have been employed in the calculations and the heats and entropies estimated.$ The entropies of the aqueous ions from Zn to V may be expected to vary only by relatively small amounts for instance the'entropies of Zn2+ Cu2+ Fe2+ and Mn2+ are - 25.9,' - 26.5 - 25.9 and - 25.9 ex.respectively. We consider it a good approximation to take AS(5) to be the same for the reactions involving Co Cr and V ions as for those involving Fe ions. (d) Subsidiary Thermochemicul Quantities. The corresponding values for the reaction For the reaction Cu+ +H+ -+ Cua+ + +H,. * (6) we have used Fenwick's value 8 for AG (3.85 kcal./mole). From the simi- larity of ionic radii of Cu+ and Na+ it is estimated that ASh,,. of Cu+ is - 21 e.u. ; hence x"[Cu+ (as.)] = 17 e.u.Together with Latimer's value for @"Cu2+ (aq.)] AS(6) is calculated to be - 28 e.u. and AH(6) - 4.4 kcal./mole. For the reaction Fe(CN),4- + H+ -+ Fe(CN),3- + &Hs we assume Kolthoff and Tomsicek's value 9 for A# (8.3 kcal./mole) and Bichowsky and Rossini's value for AH (25.1 kcal./mole). The correspond- ing entropy change is 56-3 e.u. The data for the halogens with the exception of fluorine are taken from Bichowsky and Rossini and from Latimer (opp. cit.). In accordance with Evans Warhurst and Whittle's suggestion,1° we have taken the a J . Amer. Chem. SOC. 1944 66 1573. Stokes and Stokes Trans. Paraclay SOC. 1945 41 688. J . AWT. Ohem. Soc. 1926,48,860. J . Phya. Ghem. 1035,39,945. lo J . 1950,1524. * Calculated from the heat of the reaction Fe*+ (aq.) + &Hga+ (aq.) + Fe3+ (aq.) + Hg (1) and the heats of formation and solution of FeCA and Fe(NO,), together with the heats of formation of Fe*+ (aq.) and of the anions.Data from Bichowsky and Ro~sini.~ t Data from Latimer (op. cit.). Value for V from Jones and C01vin.~ $ The heats of formation of Cra+ and Crs+ given by Bichowsky and Rossini lead with the above free energies to entropies of - 37 and - 93.6 e.u./g.-ion respectively it is assumed that the free-energy data are more reliable. Jones and Colvin ( b c . cit.) find a difference of 0.01 v between the normal potentials at 273" and 298" for V*+ -+ Va+ + e corresponding to M ( 6 ) = - 9.3 e.u. but the method is not accurate enough to establiah a significant difference from the entropy change in the Fe reaction. 192 OH- and metal ions Co3+ + OH- +Co2+ + OH .. . . . Fe3+ + OH- - + F e z + + OH . . . . . 1 V3+ + OH- +Vz++OH . . . . . Cr3+ + OH- +Cr2+ + OH . . . . . Cu2+ + OH- +Cu+ + OH . . . Fe(CN) .. + OH- + Fe(CN) .. + O H . . . 0,. and metal ions Co3+ + 0,. +Co2+ + 0. . . . . . . Fe3+ + 0.. -+Fez+ + 0. . . . . . . Cr3+ + 0.. +Cra+ + 0. . . . . . . Fe(CN). .. + 0.. --+ Fe(CN). .. + 0. . . . . Co3+ + HO.. +Co2+ + HO. . . . . Fe3+ + HO.. +Fez+ +HO. . . . . V3+ + HO.. -+V2++H0. . . . . . Cr3+ + HO.. +Cr2+ + HO. . . . . Cu2+ + HO.. --+ Cu+ + HO. . . . . Fe(CN). .. + HO.. + Fe(CN). .. + HO. . . . O H + F - + O H - + F . . . . . . . OH + C1- + OH- + C1 . . . . . . . OH + Br- + OH- + Br O H + I - +OH.+I . . . . . . . . OH + SH- -+ OH- + SH . . . . . . . 0.. and halogen atoms v3+ + 0.. +v.++o.. . . . . . CuZ+ + 0.. +cut- + 0. . . . . . HO.. and metal ions OH- and halogen atoms . . . . . . . 0.+ F. -0.. + F . . . . . . . . 0. + c1- + 0.. + c1 . . . . . . . . 0. + I- + 0.. + I . . . . . . . . 0. + SH- + 0.. + SH . . . . . . . . HO. + F- +HO.. + F . . . . . . . HO. + C1- + HO.. + C1 . . . . . . . HO. + Br- + HO.. + Br . . . . . . . HO. + I- -+ HO.. + I . . . . . . . HO.+SH.+HO.. +SH . . . . . . HO..+OH-+HO.+O H. . . . . . . 0.. + HO. -+ 0. + HO.. . . . . . . . 0.. + OH -+ 0. + OH- . . . . . . . 0. + Br- -+ 0.. + Br . . . . . . . . HO.. and halogen atoms Radical-ion transfer QUARTERLY REVIEWS TABLE 4 . Ionic Reactions in Aqueous Solution (298" K.) (a) Non-bond breaking Reaction 1 A H - 1 (kcal./mole) 21 44 68 72 60 30 . 42 . 18 6 10 - 2 . 32 7 31 54 58 47 18 34 3 . 10 .27 . 30 I 96 66 52 36 32 46 17 3 . 14 . 17 . 13 . 60 . 62 I I 59 59 59 59 65 . 19 53 53 53 53 59 . 25 55 65 55 55 62 . 23 - 6 . 21 . 24 . 29 . 10 - 1 . 14 . 19 . 23 - 8 - 3 . 17 . 21 . 25 . 10 - 3 - 2 - 5 3 27 51 54 41 36 .58 -34! .10 - 6 . 20 .24 - 9 15 38 42 29 25 36 10 - 3 .18 .27 96 71 58 43 35 47 22 9 - 6 .14 .12 .49 .61 I EVANS HUSH AND URI REACTIONS INVOLVLNG HYDROGEN PEROXIDE 193 TABLE 4 (cont.) ( b ) Bond-breaking Reaction +H,+HO +HO,-+H+ . . . . . HO +H++O,- . . . . . . +H,+OH + O H - + H + . . . . . +H2+0 +O,-+H+ . . . . . H0,- + H,O,+HO + OH + OH- . . . 0,- + H,O + O + OH + OH- . . . H,O,++H -+OH-+OH+H+ . . . Co3+ + H,O +Co2+ + OH + H+ . . Fe3+ + H,O +Fez+ + OH +H+ . . V3+ + H,O - + V 2 + + O H + H + . . Cr3+ + H,O +Cr2+ + OH + H+ . . Cu2+ + H,O Fe(CN)B3- + H,O -+ Fe(CN),4- + OH + H+' Co3+ + HO *Co2+ + H+ + 0 .. Fe3+ +HO +Fe2+ + I€+ + 0 . . V8+ + HO * V 2 + + H + + 0 2 . . . Cr3+ + HO --+Cr2+ + H+ + 0 . . Cu2+ + HO Fe(CN),3- + HO + Fe(cN)64- + H+ + 0 H,O + Co3+ . H,O + Fe3+ . H,O + V3+ +V2+ + HO + H+ . . H,O + Cr3+ . H,O + Cua+ + Cu+ + HO + H+ . . Fe(CN),3- + H,O + HO + H+ + Fe(CN)64- F+H,02 + H O + H + + F - . . . . C1 +H,O +HO,+ H+ + C1- . . . . Br +H,O +HO,+ H+ + Br- . . . . I + H,O +HO,+H+ + I- . . . . SH + H,O,-+HO,+ H+ + SH- . . . . . Co2+ + H,0,-,Co3+ + OH + OH- . . . Fe2+ + H,O + Fe3+ + OH + OH- . . . V2+ + H,O -+V3+ + OH + OH- . . . Cr2+ + H,O -+ Cr3+ + OH + OH- . . . Cu+ + H,O -+ Cua+ + OH + OH- . . . H,O,+F- + O H + O H - + F . . . . H,O,+Cl- + O H + OH- +C1 . . . . H,O,+Br- -+OH+ O H - + B r . . . . H,O,+I- + O H + O H - + I .. . . H,O,+SH-+OH+OH-+SH . . . Oxidation of H,O by metal ions -+ Cu+ + OH + H+ Oxidation of HO by metal ions -+ Cu+ + H+ + 0 Oxidation of H,O by metal ions -+ Co2+ + HO + H+ . -+ Fe2+ + HO + H+ . --+ Cr2+ + HO + H+ . Oxidation of H,O by halogen atoms Reduction of H,O by metal ions Reduction of H,O by halide ions AH kcal ./mole) - 42 - 3 - 65 7 36 - 13 - 7 34 57 81 85 74 43 - 45 - 21 '3 7 - 5 - 35 16 39 63 66 55 26 - 38 - 9 6 22 25 - 28 5 - 19 - 23 - 11 81 62 38 23 15 AS (e.u.) - 24 - 18 - 28 - 22 5 3 - 21 31 31 31 31 38 - 46 35 35 36 35 41 - 43 28 28 28 28 35 - 49 - 23 - 9 - 5 0 - 16 - 61 - 51 - 61 - 61 - 67 2 - 12 - 16 - 21 - 15 AG kcal./mole) - 36 3 - 47 14 35 - 14 1 25 48 72 76 62 67 - 56 - 31 - 7 - 3 - 17 - 22 . 7 31 65 68 46 41 - 31 - 6 7 22 30 43 220 - 4 - 8 6 80 66 43 29 20 194 QUARTERLY REVIEWS TABLE 4 (cont.) (c) Further ionW reactions Reaction OH and metal ions Ag+ + OH +Age++ OH- .. . Ce3+ + OH +Ce4+ + OH- . . . Mn2+ + OH *Mn3++OH- . . . Fe(Cl,H,N,)32+ + OH + Fe(C,,H8N,)3+ + OH- . Mo(CN),"- $. OH + Mo(CN),~- + OH- . . MnO, + OH +MnO,-+OH- . . . Mn(CN),,- + OH +Mn(CN),3- + OH- . . FeO,,- + OH +FeO,- + OH- . . C0(CN)e4- + OH + CO(CN)63- + OH- . . HO and metal ions Ag+ + HO -+Ag2+ + H0,- . . . *Ce4+ + H0,- . . . +Mn3+ + H0,- . Ce3+ + HO Mn2+ + HO Fe(Cl,H8N,)2+ + HO -+ Fe(Cl,H8N,)3+ + H0,-' Mo(CN),~- + HO + Mo(CN),~- + H0,- . MnO,,- + HO +MnO,- + H0,-. . . MII(CN),~- + HO -+ Mn(CN),3- + H0,-. . FeO,,- + HO -+ Fe0,- + H0,- . . . . CO(CN),4- + HO + CO(CN),3- + H0,- . . Agf + 0 +Ag2+ + 0,- . . . . Ce3+ + 0 +Ce4+ + 0,- . . . . Mna+ + 0 +Mn3+ + 0,- .. . . Mo(cN),,- + 0 +Mo(CN),~- + 0,- . . MnO,,- + 0 +MnO,-+ 0,-. . . . MXI(CN),~- + 0 --+ MII(CN),~- + 0,- . . FeO,*- + 0 +FeO,- + 0,- . . . . 0 and metal ions Fe(C12H8N2)32+ + 0 + Fe(C12H8N2)33+ + 0 2 - CO(CN)64- + 0 -+CO(CN)63- + 0 . . . ( d ) Reactions invoking ozone Reaction 0 + 20H- 0 3 + H,O 0 + OH- H 2 0 2 + 0 3 H0,- + 0 3 HO + 0 3 OH + 0 3 Cot+ + 0 + H,O 0 3 + H,O (HO + Co3+ + OH- +20,-+H,O . . . . + 2H0 . . . . . + 0,- + Ho . . . +OH+HO,+'O . . + O H + O - + O . . . + O H + 2 0 . . . . 4 H 0 + 0 . + c O ~ + + OH- 4 0 +'OH +H,O + O2 . . . . +Co2+ + H,O + 0 . . AH :kcal./mole; - 19 - 24 - 28 - 22 - 44 - 58 - 6 - 10 - 14 - 9 - 31 - 46 44 39 35 40 18 4 AS (ex.) - 66 - 62 - 59 19 19 19 - 62 - 48 - 55 23 23 23 - 60 - 46 - 53 25 26 26 A@ ,kcal./mole 1 - 8 - 10 - 19 - 28 - 32 - 50 - 60 - 64 13 4 2 - 7 - 16 - 20 - 38 - 49 - 62 62 63 61 42 33 29 11 0 - 3 kcal./mole) AH I I 16 49 33 19 8 - 42 - 31 16 - 10 - 68 AS (e.u.) 6 - 5 0 6 11 9 - 2 - 62 - 1 61 AG (kcal./mole) 14 51 33 18 5 - 45 - 30 31 - 10 - 76) EVANS HUSH AND URI REACTIONS INVOLVING HYDROGEN PEROXIDE 195 dissociation energy of flubrine to be 42 kcal./mole (mean value).Values calculated for reactions involving SH are very approximate these employ West’s estimatell of the electron afkity of SHY and it is assumed that D(S,) = 83 kcal./mole.* (e) Heat and Free Energy of Formution of 02-. We have estimated the heat of hydration of 0,- to be 63 kcal./g.-ion relative to the value 282 kcal. for H+. is due to the fact that in the present calculations we have following the method of Eley and Evans,12 taken account of small effects on the reorienta- tion energy of the co-ordinated water molecules in the fist hydration shell and on the energy of Born charging caused by the asymmetry of the ion.Together with the electron affinity of 0 calculated from the lattice energies of peroxides (16 kcal.) this yields The small difference between this and a previous estimate AH A# AG 0 (aq.) + +H (g) -+ 0,- (aq.) + H+ (aq.) 7 - 31 16 The value of AQ for this reaction in conjunction with our data for HO, correspond to a free energy of ionisation of HO of - 3 kcal./mole. This agrees well with the results in the iron catalysis of hydrogen peroxide decomposition obtained by Barb Baxendale George and Hargrave,13 and by Weiss and Humphrey.14 With the above value HO would be dis- sociated to the extent of approximately 50% at pH 2 whereas at pH 0 the ratio 0,- KO would be approximately 1 100.The hydrogen-ion dependence of the consumption ratio AH202/AFe2+ in the catalytic burst at the start of the reaction can thus be well accounted f0r.t (f) Reactions involving Ozone. A number of schemes have been sug- gested for reactions of ozone involving radicals and ions of the type we have been considering (cf. Taube and Bray le). Coupling the heat and free-energy change of 0 -+ 0 + 0 with the data already provided we obtain the thermodynamic quantities in Table 4(d) for several of the re- actions which have been discussed. These reactions are certainly composite and do not necessarily represent the primary steps. The thermodynamic quantities may be helpful in the elucidation of the reaction mechanisms.Hill l7 has postulated that the Co2+-catalysed decomposition of ozone involves the [Co(0H)l2+ ion-pair complex. Some approximate quantities involving this complex and 0 are included in Table 4(d) it having been assumed that the energetics of the reaction Co3+ + OH- -+ [Co(0H)l2+ are the same as those of the corresponding Fe3+ reaction. A number of estimates have been made of the oxidation-reduction potentials of the O,/O,- (9) Standard Oxidation-Reduction Potentials. l1 J . Phys. Chem. 1935 39 493. l3 Nature 1949 163 692. l5 J . Amer. Chem. SOC. 1942 64 2468. Ibid. 1948 70 1306. * In calculating the values listed in Table 4(c) we have used the free-energy data given in Latimer’s compilation,4 and in some cases the tonic entropies have been estimated.for the dissociation constant of HO has been independently postulated by Taube.15 l2 Trans. Faraday SOC. 1938 34 1093. l4 Ibid. p. 691. l6 Ibid. 1940 62 3357. t A value of 196 QUARTERLY REVIEWS OH/OH- and HO,/HO,- systems the best-known of which are those contained in Latimer's monograph.4 Since the publication of the latter (1938) a number of thermodynamic quantities relating to these potentials have been revised.* On the basis of data here proposed the standard potentials (to the nearest 0.1 v) of these systems at; 25" c. are as follows E (volts againat N.H.E.) OH(aq.) + e -+ OH-(aq.) . . 2.0 HO (aq.) + e + H0,- (aq.) . . 1.5 0 (as.) + e -+ 0,- (aq.) . . - 0.6 18 Present communication. * Our value for the H....O bond-dissociation energy is within the limits of accuracy of a recent estimate by RobertsonlB based on the appearance potential.

ISSN:0009-2681    

DOI:10.1039/QR9520600186

出版商:RSC    

年代:1952

数据来源: RSC