GENERAL AND PHYSICAL CHEMISTRY.1. INTRODUCTION.THE chemistry of deuterium and its compounds continues to provea fruitful field of investigation, and since the last Report ondeuterium, published two years ago, nearly 300 papers have beenpublished dealing with various aspects of the subject. It is not aneasy matter to review all this work in one article, especially as manyof the studies have been made in connexion with spectroscopicinvestigations and with the object of elucidating the mechanisms ofthermal and photochemical reactions. The general and physicalchemistry of deuterium is, therefore, described in one section ofthe Report and references are given to other articles, in this andprevious Reports, where applications relating to particularproblems are mentioned.Attention should be called here to animportant symposium on “ Deuterium and its Compounds ” heldunder the auspices of the Bunsen Gesellschaft in September,1937 ; the papers contributed are important, not so much for thenew work reported, as for the fact that they are generally verycomplete surveys of the field with full references to the originalpapers. Mention may be made of the doubt which has been caston the existence in Nature of a third isotope of hydrogen, withmass number 3, called tritium. It will be of interest to see if theclaim2 to have obtained water enriched in this isotope can besubstantiated. As forecast in a previous Report, supplies of watercontaining appreciable proportions, vix., 2-3 yo, of H,180 are nowbecoming available both in England and in the U.S.A., andprogress in the study of isotopic exchange reactions with oxygenis to be expected.The most satisfactory method of preparing thiswater is by fractional distillation. Developments are aIsomentioned in this Report of the partial separation of the isotopesof lithium, carbon, nitrogen, and argon by physical and chemicalmethods.Once again a report on chemical kinetics is inevitable : variousaspects of the subject have been discussed at meetings of theChemical Society and the Faraday S~ciety.~ Although noimportant advance is to be recorded in the “ transition-state ” or‘‘ activated-complex ” theory of reaction velocity, the fundamental1 2. Elektrochem., January 1938.3 For references, see this vol., p.44.See Ann. Reports, 1935, 32, 50GLASSTONE : INTRODUCTION. 31assumptions of the theory have been examined, with the resultthat there has been a clarification of the essential postulates.Progress continues to be made in the study of the combustion ofgaseous hydrocarbons and of reactions in solution.It is five years since the subject of photochemistry was treatedin a comprehensive manner in these Reports: in the intervalthere have been many advances, both in the study of photochemicalprocesses and in the parallel field of light absorption. Particularmention may be made of progress in the investigation of thephotolysis of carbonyl compounds and of the hydrogen-chlorinereaction. Many of the complexities of the latter process have nowbeen straightened out, and it is probably not an exaggeration tosay that the problem is well on the way to a solution as completeas is compatible with the limitations of the present state ofscientific knowledge.One of the most striking developments of this decade is theattention paid to the study of intermolecular, i.e., van der Waals,forces and to the theory of the liquid state.The time isundoubtedly ripe for an account of the present status of these andrelated subjects. Owing to the diversity of interests involved inthe study of liquids, it has been necessary to leave some aspects tobe considered in the section of this Report dealing with Crystal-lography. During 1936 the Faraday Society held a discussion on“ Structure and Molecular Forces in Liquids,” and in 1937 thecentenary of the birth of van der Waals was celebrited by asymposium, held in Amsterdam, on the properties of liquids andhighly-compressed gases.The papers have been publi~hed,~ andthe whole forms a valuable contribution to our knowledge of theliquid state and of solubility. It is appropriate to record theappearance of a second edition of J. H. Hildebrand’s well-knownmonograph on solubility : the revision has been so thoroughthat it is virtually a completely new work.Although there have been no spectacular developments,interesting results have been obtained in various fields of theelectrochemistry of solutions : a few only of these have been chosenfor discussion in this Report. Special attention may be called tothe method for determining dissociation constants of acids whichappears to be, from the theoretical standpoint, the most reliableprocedure yet devised.In conclusion, attention may be directedto the publication of a new edition of it familiar text-book,g and of4 For references, see this vol., p. 75.5 “ Solubility of Non-Electrolytes,” Reinhold Publishing Corporation, New6 S. Glasstone, “ The Electrochemistry of Solutions,” Methuen and Co.,York, U.S.A., 1936.Ltd., 193732 GENERAL AND PHYSICAL CHEMISTRY.a, monograph in which an attempt is made to interpret theproperties of ions in solution in terms of energy levels.'The writer wishes to place on record his sincere appreciation ofthe work of those who have so generously collaborated with himin the compilation of the General and Physical Chemistry sectionof these Reports during the past three years.S. G.2. NON-RADIOACTIVE ISOTOPES.(Continued from Ann. Reports, 1935, 32, 40.)Deuterium.-Separation. The normal value of the separationfactor * in electrolysis is about 6-43, but in the previous Report 1mention was made of the possibility of an exceptionally highfactor, about 100, in water containing relatively little deuterium.This result, which was of great theoretical importance since itappeared to be a striking instance of " quantum-mechanicalleakage," was based on the value for the H/D ratio in normalwater concerning which there was some uncertainty, One of thechief causes of error in the latter connexion has been the failure tomake allowance for the change in the isotopic composition ofoxygen in the preparation of deuterium-free water as a standard ofcomparison; this has been done in recent work and the followingH/D ratios in ordinary water have been recorded : 6400 & 200(Lake Mendota, Wisconsin),2 > 5600 (Osaka, Japan),3 and 6900(Lake Michigan)? From new experimental work it appears thatif the last of these ratios is accepted the separation factor in theelectrolysis of dilute alkali has an average value of about 7 at anickel cathode, both in normal water (0-014% D,O) and in watercontaining 0.25y0 of D,O.If the normal water has a higherdeuterium content than that implied by the H/D ratio of 6900, theseparation factor would be even lower.It is suggested, therefore,that the abnormally high factor mentioned above is in error.5From time to time reports have been published of the presence ofa higher concentration than normal of deuterium in substances ofnatural origin6 It is probable that some of these results are in7 R. W. Gurney, " Ions in Solution," Cambridge Univ. Press, 1936.1 Ann,. Reports, 1935, 32, 41.2 N. F. Hall and T. 0. Jones, J . Amer. Chem. Xoc., 1936, 58, 1916.3 N. Morita and T. Titani, BulE. Chem. SOC. Japan, 1936, 11, 403.4 J. L. Gabbard and M. Dole, J. Amer. Chem. Xoc., 1937, 59, 181.5 H. F. Walton and J. H. Wolfenden, J., 1937, 1677.6 For summaries, see M. Dole, J. Amer. Chem. SOC., 1936, 58, 580; N.Morita and T. Titani, Bull. Chem. SOC. Japaw, 1936, 11, 419; see also J.S.Anderson, H. V. A. Briscoe et al., J., 1937, 1492.separation factor is the H/D ratio in the gas divided by the H/Dratio in the liquid phase.ULASSTONE : NON-RADIOACTIVE ISOTOPES. 33error because, when the substances are burnt, the oxygen in theresulting water has not the same isotopic composition as theoxygen in ordinary water (cf. p. 42) : the observed differences indensity may be due to this cause and not to differences in theisotopic ratio of the hydrogen. After making all the necessarycorrections, it has been found that the H/D ratio in benzeneJ8presumably originating from coal, in cholesterol, and in halibut-liveroil9 is approximately the same, within the limits of experimentalerror, as in normal water. In view of the possibility that watersfrom different sources may have different isotopic ratios of bothhydrogen and oxygen, it might be desirable to define the litre andthe Centigrade degree in terms of lH216Q; because of thedifliculty of obtaining this substance it is recommended that thepresent definition be maintained, but that the source and treatmentof the water should be specified.10 The problem of isotopiccomposition as related to the determination of atomic weights hasbeen already considered.l1The deuterium content of the occluded gas obtained from apalladium cathode after electrolysis is greater than that evolvedduring electrolysis : l2 this result is explained in terms of thetheory of Halpern and Gross,13 that the electrolytic separation ofhydrogen and deuterium is due to the different rates of combinationof the atoms on the surface of the electrode.The same view issaid to be supported by theoretical considerations based oncalculations of absolute reaction velocities by the transition-statemeth0d.1~ On the other hand, arguments have been put forwarc!in support of the suggestion that the electrolytic separation canbe explained by the different rates of discharge of hydrogen anddeuterium ions, resulting from the difference in the zero-pointenergies of the hydrated protons (H30+) and deuterons (D,O+).15It is of interest in this connexion that wnaphthaquinoline, whichmight be expected to inhibit any catalytic process in the separationof hydrogen and deuterium, actually lowers the Separation factorsat mercury and silver cathodes, and also reverses the sign of thetemperature coefficient .168 M.Dole, loc. cit.M. Dole and R. B. Gibney, J . Amr. Chem. SOC., 1936,58, 2552.10 E. H. Riesenfeld and T. L. Chang, Physilcal. Z., 1936, 37, 690.11 Ann. Reports, 1936, 33, 141-142.12 A. Farkas, Trans. Paraday Soc., 1937, 33, 552.13 Ann. Reporta, 1935, 32, 41.14 G. Okamoto, J. Horiuti, and K. Hirota, Xci. Papers Inst. Phys. Chem.16 J. A. V. Butler, 2. Elektrochem., 1938, 44, 55.16 H. F. Walton and J. H. Wolfenden, Nature, 1936, 138, 468.Res. Tokyo, 1936, 29, 223.REP.-VOL. X X W . 34 QENERAL AND PHYSICAL CHEMISTRY.In addition to work on the electrolytic method, research has beencontinued on other methods of separation of the isotopes ofhydrogen: the efficiency of the diffusion method, using Hertzpumps, has been co&med,17 and another procedure, in which arapidly moving stream of mercury vapour acts as the membrane,has been described.18 Preliminary experiments have shown thatfractional desorption of hydrogen and deuterium gases fromactivated charcoal in a vacuum is an efiicient method of separation,l9and the separation of light and heavy waters by distillation hasbeen considered further.20 The failure to bring about enrichmentof deuterium by crystallisation of salt hydrates has beenalthough there appears to be a difference of opinion as to whetherthe isotopic ratio in the first four molecules of water of crystal-lisation of copper sulphate is the same as in the fifth molecule.22Analysis.Improvements have been recorded in the methodsfor determining the isotopic composition of hydrogen and itscompounds. The procedure involving measurement of the densityof small quantities of water has received some attention,23 and anumber of workers have used the thermal-conductivity methodwhich is applicable to minute quantities of hydrogen gas, providedthey contain a fair proportion of deuterium.24 The latter has alsobeen modified for the analysis of mixtures of light and heavy ~ a t e r . ~ 5The interferometric method, depending on the different refractiveindices of these two types of waters, has also been modified,26 anda gas-density method involving the use of tt quartz-fibre micro-17 R. Sherr and W.Bleakney, Physical Rev., 1936, 49, 882.19 K. Peters and W. L o b a r , 2. physikal. Chern., 1937, A, 189, 51.2O J. Horiuti and G. Olzamoto, Bull. Chem. SOC. Japan, 1935, 10, 503;G. B. Pegram, H. C. Urey, and J. Huffman, Physical Rev., 1936, 49, 883;E. R. Smith and M. Wojciechowski, J . Res. Nut. Bur. S t a d . , 1936, 17, 841.2 1 E. H. Riesenfeld, Ber., 1935, 68, 1962; E. H. Riesenfeld and T. L.Chang, ibid., 1936, 69, 1302.22 J. S. Anderson, H. V. A. Briscoe et al., J., 1937, 1492 (1499); H.Perperot and F. Schacherl, J. Chim. phydque, 1937, 34, 257.23 Idem, J . Phys. Radium, 1935, 6, 319; K. Fenger-Eriksen, A. Krogh,and H. Ussing, Biochem. J., 1936, 30, 1264; W. H. Hamill, J. Amer. Chem.SOC., 1937, 59, 1152; J. S. Anderson, H. V. A. Briscoe, et al., J., 1937, 1492;H. Fromherz, R.Sonderhd, and H. Thomas, Ber., 1937, 70, 1219.24 A. Farkas, L. FarkBs, and E. K. Rideal, Nature, 1936, 137, 315; K.Wirtz, 2. physikal. Chem., 1936, B, 32, 334; D. D. Eley and J. L. Tuck,Trans. Faraday SOC., 1936, 32, 1425; J. L. Bolland and H. W. Melville, ibid.,1937,33, 1316; G. H. Twigg, ibid., p. 1329; N. R. Trenner, J . Ckm. Physics,1937, 5, 382, 761.D. MecGillavry, Rec. trav. chim., 1937, 56, 330.2s A. Farkas, Trans. Paraday SOC., 1936, 32, 413.26 N. S. Filipova and M. M. Sluckaja, Acta Phy&ochim. U.R.S.S., 1936,51, 131QLASSTONE : NON-RADIOACTIVE ISOTOPES. 35balance has been described.27 The change in the consolutetemperature of the phenol-water system with varying isotopiccomposition of the water has been suggested as a means fordetermining the H/D ratio in liquid water.28 In certain instancesthe relative absorptions at various wave-lengths in the infra-redcan be applied to estimate the isotopic composition of hydrogencorn pound^.^^Properties. Further measurements have been made of thethermal conductivity 30 and viscosity of deuterium gas ; 31 fromthe latter it appears that the effective molecular diameters of thetwo isotopes do not differ by more than 2%.The refraction anddispersion,32 and the compressibility of the gas a t varioustemperatures33 have also been studied. The difference in thevapour pressure of the equilibrium 'mixture of ortho- and para-deuterium and the normal mixture has been determined attemperatures be5ween 15" and 20.4" K.,34 and measurements havebeen made of c, of the two isotopes in the liquid and solid states.35The density, compressibility, and thermal expansion of solidhydrogen and deuterium have been determined at 4.2" 1c.36Observations have been made of the solubility37 of the gases inpalladium, and of their rates of diffusion through heated palladiumand platinum and through an iron The rates ofadsorption on carbon, copper, platinum, and nickel surfaces havebeen ~tudied.3~ The theory of the interaction of atoms and27 N.R. Trenner, J. Amer. Chem. SOC., 1937, 59, 1391 ; see also M. Calvin,Trans. Paraday SOC., 1936, 32, 1428; P. Holemann and K. Clusius, 8.physikal. Chem., 1937, B, 35, 261.28 W. H. Patterson, J., 1937, 1745.2a W.S . Benedict, H. S. Taylor, et al., J . Chem. Physics, 1937, 5, 1.so C. T. Archer, Nature, 1936, 138, 286; G. W. Kannuluik, ibid., 1936,137, 741 ; W. Nothdurft, Ann. Physik, 1937, 28, 157.s1 H. C. Torrey, Physical Rev., 1935, 47, 644; A. B. Van Cleave and 0.Maass, Canadian J. Bee., 1935, 13, B, 384.92 T. Lamen, 2. Physik, 1936,100, 543 ; W. J. C. Om, Tram. Paraday SOC.,1936, 32, 1556.s3 E. Bartholom6, 2. physikal. Chem., 1936, B, 33, 387; K. Schiifer, ibid.,1937, B, 36, 85.34 F. G. Brickwedde, R. B. Scott, and H. S. Taylor, J. Chem. Physics,1935, 3, 653.36 E. Bartholom6 and A. Euckon, 2. Elektrochem., 1936, 42, 547.57 A. Sieverts and G. Zapf, 2. physikal. Chem., 1935, A, 174, 359; A.Sieverts and W. Dam, ibid., 1936, B, 34, 158.3t3 A.Farkas, Trans. Faraday SOC., 1936, 32, 1667; R. Jouan, J . Phys.Radium, 1936, 7, 101 ; P. C. Blokker, Rec. trav. chim., 1936, 55, 979.3Q R. M. Barrer, Trans. Paraday SOC., 1936, 32, 481; R. A. Beebe et al.,J . Amer. Chem. SOC., 1935, 57, 2527; E. B. Maxted and C. H. Moon, J . ,1936, 1542 ; A. Magnus and G. Sartori, 2. physikal. Chem., 1936, A , 175, 329.H. D. Megaw and F. Simon, Nature, 1936,138, 24436 GENERAL AND PHYSICAL CHEMISTRY.molecules with solid surfaces has been worked out and the resultsillustrated by reference to H,, HD, and D, molecules.40Properties* of Deuterium Oxide.-The diamagnetic susceptibility ofliquid deuterium oxide is 0.64 x 10-6 at 20", compared with0.72 x for water,41 but the surface tensions are identicalwithin O*05~0,42 contrary to results reported previously.Atordinary temperatures the vapour pressure of heavy water is lessthan that of light water-hence the possibility of separation bydistillation-but the difference diminishes as the temperature israised : it is estimated that at 224" the values would be identical.The latent heat of vaporisation of D,O exceeds that of H,O, buthere also increase of temperature decreases the differen~e.4~ Thelatent heats of fusion are in the opposite direction, the values inmixtures of light and heavy water being expressed by L 5 79.67 -4.38n2 cals. per g., where 12, is the mole-fraction of D,0.44 It hadbeen stated that the isotopic composition of ice which separatesfrom water is the same as that of the water itself ,45 but this has beenshown to be incorrect : 46 the solidus and liquidus curves are,however, not far apart, the maximum separation being 0.02" for aliquid containing 42% of D20.The specific heat of liquid heavywater is appreciably greater than that of ordinary water, thedifference being largest at low temperatures. The mean specificheat of D,O between 4" and 26" is 1.018 cals.47 New determinationshave been made of the viscosity of liquid D,O, which is found tobe greater than that of H20,48 and of the densities of mixtures oflight and heavy water. It appears that there is no volume changeon mixing, and that the system is ideal, as is to be expected; from40 5. E. Lennard-Jones and A. F. Devonshire, Proc. Roy. SOC., 1936, A,156, 16.4 1 F.E. Hoare, Nature, 1936, 137, 497; (Miss) V. C. G. Trew and J. F.Spencer, ibid., pp. 706, 998; V. Nehra and M. Qureski, Current Sci., 1937,5, 533 ; H. P. Iskenderian, Physical Rev., 1937, 51, 1092.42 H. Flood and L. Tronstad, 2. physikal. Chern., 1936, A , 175, 347; H.Lachs and I. Minkow, Rocz. Chem., 1937, 17, 363 ; G. Jones and W. A. Ray,J . Chern. Physics, 1937, 5, 505; J. Timmermans and H. Bodson, Cornpt.rend., 1937, 204, 1804.43 F. T. Miles and A. W. C. Menzies, J. Amer. Chem. SOC., 1936, 58, 1067;E. H. Riesenfeld and T. L. Chang, 2. physilal. Chern., 1936, B, 33, 120,127.44 0. Redlich and J. Zentner, Monatsh., 1936, 68, 407; E. A. Long andJ. D. Kemp, J . Arner. Chem. SOC., 1936, 58, 1829.45 M. Deielih, 2. anorg. Chem., 1935, 225, 173.46 A.Eucken and K. Schafer, ibid., p. 319.4 7 R. S. Brown, W. H. Barnes, and 0. Maass, Canadian J . Res., 1935,13, B,18 G. Jones and H. J. Fornwalt, J. Chem. Physics, 1936, 4, 30; W. N.167; A. Ferguson and A. H. Cockett, Nature, 1936, 138, 842.Baker, ibid., p. 294GLASSTONE : NON-RADIOACTIVE ISOTOPES. 37the known densities, dzP, of pure H20 (0.99705) and of pure D,O(1.10466), it is shown that n,,, = 9.235AS/(l - 0.0309AS), wherenDeo is the mo1.-fraction of D20 in a mixture and AX is equal to- d-.)/dH20, the d values being for 25" compared withordinary water a t 4'?9 The density of liquid deuterium oxide atvarious temperatures and the volume change on freezing havebeen studied; the influence of pressure on the freezing point ofdeuterium oxide is 0.00705" per kg./cm.z, which is approximatelythe same as for ~ater.~O The apparent molar volume of deuteriumoxide dissolved in dioxan is about 0.5% less than that for water,in agreement with theoretical e~pectation.~l The refractive indexof 99.2y0 D20 vapour for light of A 5462-23 A.is 1.0002501 comparedwith 1.0002527 for H,O; the dispersion has also been studied.52The difference in the ionic products of the two forms of water,mentioned in the last report, has been confirmed, the ratioKHso/KDt0 being about 5-5 at 25".53 In agreement with previouswork, it is found that replacement of water by deuterium oxideraises the upper and depresses the lower consolute temperatures ofa binary liquid system.54 The adsorption of the vapour of heavywater on active charcoal has been investigated.55Deuterates.-The dissociation pressures of the deuterates of anumber of salts, vix., CuSO,, 5D20, Na,SO,, 10D20, MgSO,, 7D20,SrCl,, 6D20, CoCl,, 6D,O, NiCl,, 6D20 and NaBrz, 2D,O, have beenmeasured over a range of temperatures.The deuterates havelower vapour pressures than the corresponding hydrates, and theheat changes involved are also lower ; as far as has been examined,the transition points of deuterates, with the exception ofNa,S04, 10D20, appear to be lower than for the correspondinghydrates.56 The colour of CuS04, 5D,O is considerably lighterthan that of CuSO,, 5H20, and a similar difference is observed insolutions in heavy and light water, respectively, a t equal concen-tration~.~' Deuterates of the inert gases, vix., Kr, 6D20 and49 L.G. Longsworth, J. Amer. Chem. SOC., 1937, 59, 1483.50 J. Timmermans et al., Conzpt. rend., 1936, 202, 1061.51 R. A. Robinson and R. P. Bell, Trans. Paraday Soc., 1937, 33, 650.62 C. Cuthbertson and M. Cuthbertson, Proc. Roy. SOC., 1936, A , 155,213.53 W. F. K. Wynne-Jones, Trans. Faraday Soc., 1936, 32, 1397; H.Erlenmeyer and A. Epprecht, Helw. Chim. Ada, 1936, 19, 677, 1292.64 J. Timmermans and G. Poppe, Compt. rend., 1935,201, 608.55 A. King and C. G. Lawson, Trans. Paraday SOC., 1936, 32, 478; K. Arii,Bull. Inst. Phys. Chem. Res. Japan, 1937, 16, 749.58 H. Perperot and F. Schacherl, J. Phys. Radium, 1935, 6, 439; J. R.Partington and (Mrs.) K. Stratton, Nature, 1936, 137, 1075; F.Schacherland 0. BBhounek, ibid., 1936, 138, 406; J. Bell, J . , 1937, 459.5 7 Idem, Nature, 1936, 137, 53438 GENERAL AND PIIYSICAL CHEMISTRY.Xe, 6D20, have been prepared; they are said to resemble thecorresponding unstable hydrates.58Solutions in Heavy Water.-The solubilities of various salts havebeen measured in heavy water and compared with the values inordinary water : in general, the former are the lower when referredto the same number of mols. of s0lvent.5~ A number of studiesare recorded of heats of dissolution and of dilution of electrolytesin light and in heavy water. The heat absorbed on dissolution inthe latter is greater than for the same salt in ordinary water, thedifferences being attributed mainly to differences in the heats ofsolvation of the ions.In dilute solution the integral heats ofdilution are independent of the isotopic composition of the water,but at higher concentrations the values become greater, numerically,in passing from ordinary to heavy water.60 The quinhydroneelectrede has been applied to the measurement of acidity in heavywater, and the dissociation constants of salicylic and acetic acidsand of quinol have been determined; the values, like the ionicproduct, are less than in normal water.61 The whole subject ofionic equilibria in H20-D,O mixtures has been considered fromthe theoretical standpoint, especially with reference to thequinhydrone and the hydrogen electrode, and the dissociation ofacids.62 The dissociation constant of acetic acid has also beendetermined from conductance measurements with potassiumacetate in deuterium oxide ; the product of the equivalentconductance a t infinite dilution and the viscosity, i.e., the Waldenconstant, changes by nearly 2% in passing from normal to heavywater.Similar results have been obtained with potassiumchloride solution^.^^ Determinations have been made of thetransport numbers and conductances of lithium, sodium, potassium,and hydrogen chlorides in heavy water, and the ionic conductancesat infinite dilution have been estimated. The value for thechlorine ion varies in a linear manner with the mo1.-fraction of D,O58 M. Godchot, (Mlle.) G. Cauquil, and R. Galas, Compt. rend., 1936, 202,759.59 F. T. Miles, R. W. Shearman, and A.W. C. Menzies, Nature, 1936, 138,121; R. W. Shearman and A. W.. C . Menzies, J . Amer. Chm. Soc., 1937,59, 185; F. T. Miles and A. W. C. Menzies, ibid., p. 2392.60 E. Lange and W. Martin, 8. Elektrochem., 1936, 42, 662; 2. physikal.Chem., 1937, A , 178, 214; 179, 427; 180, 233; W. Birnthaler and E. Lange,2. Elekt.rochem., 1937, 43, 643; for mmmary, see E. Lange, ibid., 1938, 44,31.61 IT. K. LaMer, and S. Korman, Science, 1936, 83, 624; J. Amer. Chm.SOL, 1936, 58, 1396; J. P. Chittum and V. K. LaMer, ibid:, 1937, 59, 2425.62 W. H. Hamill, ibid., p. 1492; G. Schwarzenbach, 8. Elektrochem., 1938,44, 46.63 V. K. LaMer and 5. P. Chittum, J. Amer. Chem. Soc., 1936, 58, 1642ClLASSTONE : NON-RADIOACTIVE ISOTOPES. 39in the water, but for the hydrogen ion there are considerabledeviations indicating a complex conductance me~hanism.~~ Potas-sium acetate and acetic acid have been electrolysed in heavy water,but the resulting ethane contained very little deuterium.65 Thehigher overvoltage of hydrogen in deuterium oxide at a mercurycathode has been confirmed; 66 the importance of the results inconnexion with the theory of overvoltage is considered elsewhereOther Deuterium Compounds.-The refractive indices of gaseousD,Se,67 ND,, and DC168 have been measured and compared withthe values for the analogous hydrogen compounds; the former aredways slightly less than the latter.The melting points andtransition points of CD,, D,S, and D,Se have been observed,Gg andtransformations in the solid state studied optically and by X-raymethods.' 0 Solid tetradeuteroammonium chloride, like its hydrogenanalogue, undergoes changes in its lattice dimensions at lowtemperatures, the respective transition points being - 24" and - 30°.71 The critical temperatures of various simple deuteriumcompounds have been determined and found to be less than for thecorresponding hydrogen co~npounds.~~ A number of organicdeuterium compounds have been prepared andHomogeneous Exchange Reactions .-Investigations have beenmade of exchange reactions involving isotopes of hydrogen and thehalogen acids; heats of reaction and equilibrium constants havebeen e~aluated,'~ and the results found to be in agreement with(p. 109).64 L. G. Longsworth and D. A. MacTnnes, J .Amer. Chem. BOG., 1937, 59,1666; J. P. Chittum and V. K. LaMer, Zoc. cit., ref. (61) ; see also, V . K. LaMer,Chem. Reviews, 1936, 19, 363.66 K. Erlenmeyer and W . Schoenauer, Helv. Chim. Acta, 1937, 20, 222;P. Holemann and K. Clusius, loc. cit., ref. (27)-66 J. Nov&k, Coll. Czech. Chem. Comm., 1937, 9, 237.67 0. E. Frivold, 0. Hassel, and T. Skjulstad, Physikal. Z., 1936, 37, 134.68 0. E. Frivold, 0. Hamel, and S. Rustsd, ibid., 1937, 38, 191.6Q A. Kruis, L. Popp, and K. Clusius, 2. Elektrochem., 1937, 43, 664. '* A. Kruis and K. Clusius, Z . physikal. Chem., 1937, B, 38, 156; PhysiEal.Z., 1937, 38, 510; E. Justi and H. Nitka, ibid., p. 514.71 A. Smits, G. J. Muller, and F. A. Kroger, 8. physilcal. Chem., 1937, B,38, 177 ; I. Nittrt and K.Suenaga, Sci. Papers Inst. Phys. Chem. Res. Tokyo,1937, 32, 83; J. Weigle and H. SaYni, Arch. Sci. phys. nut., 1937, 19, Suppl.28-29.72 H . Kopper, 8. physikal. Chem., 1936, A, 175, 469.78 For summary, see H. Erlenmeyer, 2. Elektrochem., 1938, 44, 8 (9); seealso Arm. Repwta, 1936, 33, 228, 291.74 P. C. Cross and P. R. Leighton, J . Chem. Physics, 1936, 4, 28; A. F.Kapustinsky, J . Amer. Chem. Soc., 1936, 58, 460; K. Wirtz, 2. physikal.Chem., 1936, B, 31, 309; Physikal. Z., 1936, 37, 165; J. R. Partington andR. P. Towndrow, Nature, 1937,140, 15640 GENERAL AND PHYSICAL CHEMISTRY.those calculated from spectroscopic data.75 The exchange reactionsbetween the isotopic forms of hydrogen and water, hydrogensulphide and water, and hydrogen and ammonia have also beenthe subject of theoretical and experimental studies.76 Theequilibrium constant and velocities of direct and reverse reactionsin the exchange between ethyl alcohol and HDO have beenmeasured, and from the results it is concluded that the reactionproceeds ionically.77 Atomic deuterium reacts with saturatedaliphatic hydrocarbons, e.g., methane, ethane, etc., hydrogenatoms being replaced by those of deuterium; the energies ofactivation of such processes have been measured, and the resultsapplied to some problems of reaction mechanism and energies oflinkage~.7~ It has been confirmed that in aqueous solutions thereis complete exchange between the hydrogen of the ammino- andaquo-groups of complex cobaltammines and the deuterium of heavywater; 79 the mechanism of the exchange has been discussed.**The hydrogen in hypophosphorous acid exchanges almost com-pletely with deuterium in deuterium oxide solution, but thedeuterium in the salt Ba(D,PO,), does not exchange with thehydrogen in ordinary water after several hours.81 Some workersclaim that the hydrogen atoms in alkali acetates can undergoexchange to a small extentYs1 whereas others 82 say that noapparent exchange occurs.Investigation of the reaction betweenacetylene and water has been continued; heats of reaction andequilibrium constants have been measured, and the latter havealso been calculated from spectroscopic data.s3 A number of75 See Ann. Reports, 1935, 32, 66.7 6 T. Jones and A. Sherman, J. Chem.Physics, 1937, 5, 375; K. Wirtzand K. F. Bonhoeffer, 2. physikal. Chem., 1936, A, 177, 1; K. Wirtz, ibid.,1935, B, 30, 289; A. Farkas, J., 1936, 26; P. A. Small, Trans. Paraday SOL,1937, 33, 820.7 7 W. 5. C. Orr, ibid., 1936, 32, 1033; see, however, J. C. Jungers andK. F. Bonhoeffer, 2. physikal. Chem., 1936, A , 177, 460,7 8 E. W. R. Steacie and N. W. F. Phillips, J . Chem. Physics, 1936, 4, 461;E. W. R. Stcacie, Canadian J . Res., 1937, 15, B, 264; N. R. Trenner, K.Morikawa, and H. S. Taylor, J. Chem. Physics, 1937, 5, 203; cf. this vol.,70 F. W. James, J. S. Anderson, and H. V. A. Hriscoe, Nature, 1937, 139,109; J. Horiuti and G. Okamoto, Sci. Papers Inst. Phys. Chem. Res. Tokyo,1937, 31, 205; G. Qkamoto, J. Pac. Sci. Hokkaido Imp. Univ., 1937, 111, 2,818.80 J. S.Anderson, N. L. Spoor, and H. V. A. Briscoe, Nature, 1937, 139, 508.81 H. Erlenmeyer, W. Schoenauer and G. Schwarzenbach, Helv. Chim.82 S . Liotta and V. K. LaMer, J . Amer. Chem. SOC., 1937, 59, 946.83 L. H. Reyerson and B. Gillespie, ibid., 1936, 58, 282; 1937, 59, 900;p. 49.Actcc, 1937, 20, 726.K. Hirota and G. Okamoto, Bull. Chem. Soc. Japan, 1936, 11, 349GLASSTONE : NON-RADIOACTNE ISOTOPES. 41papers dealing with exchange reactions involving organic com-pounds have appeared,84 some of which are of physicochemicalinterest .85Heterogeneous Exchange Reactions.-Experiments on the catalysedexchange reactions between the isotopes of hydrogen in the gaseousform and liquid water, e.g., D, + H,O =+ HD + HDO andHD -+ H,O + H, + HDO, have been continued, and the subjecthas also been considered from the theoretical standpoint, as it hassome bearing on the problem of the electrolytic separation factor.It appears that the rate of atomisation of molecular hydrogen isan important factor in determining the rate of the process, butother factors may also be operative.The heterogeneous reactionsbetween gaseous hydrogen and methyl and ethyl alcohol, and alsoprobably with other organic compounds, appear to involve similarmechanisms. 86 Deuterium and ammonia gas undergo exchangeon iron catalysts ; the reaction has been investigated between 160"and 230", and shown to proceed in the adsorption layer as follows :NH,+ NH, + H, D,+ ZD, D + H + HD and D + NH,--+ NH,D, the last stage being the one which determines theover-all rate of the process.87 Studies are being made of theexchange between deuterium and gaseous paraffin hydrocarbonson nickel catalysts ; from the results, information is being obtainedconcerning the activation of particular linkages of the hydrocarbonon the catalyst surface.88Other Properties of Deuterium Compounds.-The spectroscopy ofdeuterium compounds continues to be a subject of investigation ;this has been already considered in these Reports,s9 and a furtherdiscussion must be left until a comprehensive treatment is possible.The photochemistry and kinetics of reactions involving deuteriumand its compounds are considered on pp.68 and 49; referencemay also be made to a previous and to valuablesummaries .9184 For complete review, see C.K. Ingold and C. L. Wilson, 8. Elektrochem.,85 A. E. Brodski, Trans. Faraday Soc., 1937, 33, 1180.86 A. Farkas, ibid., 1936, 32, 922; D. D. Eley and M. Polanyi, ibid., p.1388; M. Calvin, ibid., p. 1428; J. Horiuti and G. Okamoto, ibid., p. 1492;A. Farkas and L. Farkas, ibid., 1937, 33, 678, 827; G. Okamoto, K. Hirota,and J. I-Ioriuti, Sci. Papers Inst. Phys. Chem. Res., 1936, 29, 223; 30, 151 ;1937, 31, 211.1938, 44, 62 ; see also Ann. Reports, 1936, 33, ?91.87 A. Farkas, Trans. Paraday Soc., 1936, 32, 416.88 K. Marikawa, H. S. Taylor et at., J. Amer. Chem. Soc., 1936, 58, 1445,89 Ann. Reports, 1935, 32, 64; 1936, 33, 53.01 0. Reitz, 2. Elektrochem., 1938, 44, 72; K. H. Geib, ibid., p.81.1795; 1937, 59, 1103.Ibid., pp. 92, 9842 GENERAL AND PHYSICAL CHEMISTRY.Tritiwm-Since the claim to have concentrated tritium, i.e., thehydrogen isotope of mass 3, in water was made over two years ago,there have been no further developments in this field; from acritical review of the whole problem, however, it appears that theevidence for the existence of tritium in Nature is open to seriousdoubt?2Oxygen. Isotopes.-The important discovery has been made 93and confirmedg4 that oxygen from the air is richer in the heavierisotopes than is the oxygen in ordinary water; the difference inatom weights is 0-00011 unit. Water obtained from atmosphericoxygen is thus about 7 p.p.m. heavier than lake, sea, or river waterhaving the same isotopic ratio of hydrogen.The differenceappears to be due to the exchange reaction 2H21s0(Z) + 1602(g)1802(g) + 2H2l6O(Z), for which the equilibrium constant is 1.012 a t25°,95 so that the ratio 1sO/160 is greater in the gas phase, i.e., theatmosphere, than in the liquid, i.e., water. Normally, a relativelyhigh temperature and a catalyst are required for this equilibriumto be e~tablished,~~ but it is possible that in Nature it may beattained relatively rapidly in the stratosphere under the influenceof ultra-violet radiation from the ~ ~ 1 1 . ~ 7 Another possibility isthat the excess l80 in the atmosphere may result from the exchangebetween carbon dioxide and water, leading to excess of the heavyoxygen isotope in the and this is subsequently liberatedas molecular oxygen by the photosynthetic action of plants.99The two isotopic forms of water, H,160 and H21S0, have a slightdifference of vapour pressure, and this has been utilised to bringabout a partial separation in a specially designed fractionatingcolumn consisting of seven 5-ft.lengths, each fitted with 87 fixedand 87 rotating conical plates. A steady supply of watercontaining 2.5% of H2lsO is expected.1 The ready availability of92 (Lord) Ruthsrford, Nature, 1937, 140, 303.93 M. Dole, J. Amr. Chem. Soc., 1935, 57, 2731; J . Chem. Physics, 1936,4, 268, 778; N. Morita and T. Titmi, Bull. Chem. SOC. Japan, 1936, 11, 36,414.94 C. H. Greene and R. J. Voskuyl, J. Amer. Chem. Soc., 1936, 58, 693;W. H. Hall and M. L. Johnston, ibid., p.1920; T. 0. Jones and N. F. Hall,ibid., 1937, 59, 259; E. R. Smith and H. Matheson, J. Res. Nu). Bur. Xtand.,1936, 17, 625.O6 Ann. Reports, 1935, 32, 52.g6 N. Morita and T. Titani, Bull. Chm. SOC. Japan, 1937, 12, 104; T. 0.97 H. C. Urey, quoted by M. Dole, J. Chem. Phy&s, 1936, 4, 268 (272).Q8 Ann. Reports, 1935, 32, 62.Jones and N. F. Hall, loc. cit., ref. (94).C. H. Greene and R. J. Voskuyl, loo. cit., ref. (94).J. R. Huffman and H. C. Urey, Id. Eng. Chem., 1937,29, 531 ; see alsaS. C. Datta, J. N. E. Day, and C. K. Ingold, J., 1937, 1968 (1969)WYNNE-JONES : CH.EMICAL HINE'J!IUS. 43water enriched with respect to I80 will encourage further studiesof oxygen-interchange reaction in solution.2Other Isotopes.-Considerable enrichment of the heavier isotopeof carbon (13C) in methane, of nitrogen (15N) in nitrogen gas, andof the lighter isotopes of argon (36A and 38A) has been obtainedby use of the Hertz type of diffusion apparat~s.~ By allowing finedrops of lithium amalgam to fall repeatedly through a long columncontaining lithium chloride dissolved in ethyl alcohol, or lithiumbromide in alcohol-dioxan, exchange of the lithium between thetwo phases occurs, with the result that the isotopic ratio 'Li/6Li ischanged from 11.6 to 5.1 at the bottom of the column where thelithium in the amalgam is extracted by the solution.4 A similarenrichment has been observed in the direct electrolysis of aqueouslithium hydroxide,5 the 7Li isotope remaining in the electrolyte,although failure is recorded with a lithium chloride solution and acathode of circulating mercury.6 Passage of lithium and ammoniumchlorides through a column of zeolite results in an increase in theproportion of the heavier isotope of lithium and of the lighterisotope of nitrogen, respectively, in the solutions ; this is probablythe result of preferential adsorption of the other isotope by thezeolite.5 By means of chemical exchange reactions, e.g., betweencarbon dioxide and aqueous potassium hydrogen carbonate in thepresence of carbonic anhydrase, and between ammonia and aqueousammonium sulphate or nitrate, enrichment of the isotopes 1sC and15N, respectively, has been obtained.' The best results have beenrecorded with ammonium sulphate, a 6.5-fold increase in theproportion of 15N being achieved.S. G.3. CHEMICAL KINETICS.The two main aspects of the study of chemical reactions arethe elucidation of their kinetics and the interpretation of theirmechanisms in terms of the properties of the reacting molecules.These problems may be treated separately, but the interpretationof mechanisms is dependent upon a correct analysis of the kinetics ofCf. S. C. Datta, J. N. E. Day, and C. K. Ingold, J., 1937, 1968 (1969);E. Blumenthal and J. B. M. Herbert, TTans. Farday Soc., 1937, 33, 849.D. E. Wooldridge and W. R. Smythe, Phy&al Rev., 1936, 50, 233;D. E. Wooldridge and F. A. Jenkins, ibid., 1936, 49, 404, 704; H. Barwickand W. Schiitze, 2. Physik, 1937,105, 395.4 G. N. Lewis and R.T. Macdonald, J. Amer. Chem. Soc., 1936,58, 2519.6 T. I. Taylor and H. C. Urey, J. Chem. Phyaics, 1937, 5, 597.6 G. Champetier and P. Regnault, Bull. SOC. chim., 1937, 4, 592.7 H. C. Urey et al., J . Chem. Physics, 1936, 4, 622; 1937, 5, 856; J. Amer.Chem. Soc., 1937, 59, 140744 GENERAL AND PHYSICAL CHEMISTRY.reactions, and also upon the use of an adequate theory for theformulation of reaction rates. The general theory of the rates ofelementary reactions, which has been reported in previous years,lhas been the subject of recent discussions held by the ChemicalSociety 21 3 and the Paraday Society,4 in which various diflicultieshave been cleared away. The theory, which can be most simplystated in the formk* = K K ~ . kT/hwhere k* is the specific reaction rate constant, K$ the equilibriumconstant between the activated complex and the reactants, k theBoltzmann constant, T the absolute temperature, h is Planck’sconstant and K is the transmission coefficient, involves twoassumptions, vix., (1) that the equilibrium number of activatedcomplexes is maintained throughout the course of the reaction, and(2) that for the motion of the nuclei quantum effects may be neglectedor regarded as corrections to the classical treatment. I f K is takenas unity, this is a third assumption which must be considered.Attention has been called to the first assumption by variousauthors 5s 6 who doubt its validity : the difficulty was first raised byR.H. Fowler 7 with regard to kinetic-theory calculations of reactionrates, but it was shown by L.S. Kassel that, for approximatelyequal molecular masses of the reactants, the disturbance of theMaxwell distribution by reaction would usually be of the order of0.1%. A more general argument is that k* for elementarybimolecular reactions is constant throughout the reaction andindependent of the pressure ; further, for reactions reaching anequilibrium, where it has been possible to study both the forwardand the back reactions, the ratio of the two rate constants hasbeen found to agree with the equilibrium constant. It is extremelyunlikely that in ordinary liquids, where the solvent molecules willhelp t o maintain the Maxwell distribution, there will be any greaterdivergence than in gases, and we may therefore regard the firstassumption as valid for all ordinary reaction^.^The second assumption is generally accepted, and it seems probablethat, except for hydrogen atoms, deviations from classical behaviourAnn.Reports, 1935, 32, 89; 1936, 33, 86.2 M. Polanyi, J . , 1937, 629.4 Trans. Paraday SOC., 1938, 34, 1.5 E. A. Guggenheim and J. Weiss, ibid., p, 57.6 R. H. Fowler, ibid., p. 24.7 “ Statistical Mechanics,” 1929, 461.C. N. Hinshelwood, ibid., p. 635.Physical Rev., 1930, 35, 261.This is shown in yet another way by J. A. Christiansen (Trans. ParadaySOC., 1938, 34, 73) from his theory of intramolecular diffusion, 2. physikal.Chem., 1936, B, 33, 145; 1937, B, 37, 374WYNNE-JONES : CHEMICAL KINETICS. 45are not significant.lO* l1 The evaluation of the transmissioncoefficient is not so easy : a t sufficiently low temperatures we mayregard it as unity for all reactions, but, as E.Wigner 11 has shown,if the potential surfaces are very complicated, the value of K mayfall much below unity. If the reaction is not “ adiabatic,” i.e.,if during the motion of the nuclei the electrons do not remain inthe lowest quantum state, there will also be a decreased rate whichmay be treated as a small transmission coefficient.12The general theory of reaction rates can be applied in variousways and the particular mode adopted depends upon the informationavailable with regard to molecular properties and potential-energysurfaces. For certain gaseous reactions, it has been possible to carryout a detailed analysis which leads to the calculation of the rate ofthe reaction from the properties of the reacting m0lecules.1~ Forother reactions, in particular those occurring in solution, it is moreconvenient to employ a thermodynamic formulation * and to assignto the activated complex an activity coefficient which may becompared with those of other molecular species and hence enablereaction mechanisms to be deduced.This is analogous to theprocedure which has proved so useful in the study of the equilibriumproperties of solutions, and has the merit of emphasising that anunderstanding of equilibria is essential for an interpretation ofreaction rates.A striking feature of the work under review is the increasingrealisation of the importance of equilibrium studies for the interpret -ation of the rates of reactions.For chain reactions and otherreactions showing complex kinetic behaviour it is necessary to useelaborate kinetic formulations, but for the majority of reactionsfollowing simple kinetic laws any abnormality in the rate is reflectedin the equilibrium data, and this makes it in the highest degreeimprobable that explanations involving specifically kinetic factorshave any significance.A. E. Stearn and H. Eyring l4 have discussed the mechanism ofthe Menschutkin reaction from this point of view, and have shown10 R. P. Bell, Proc. Roy. SOC., 1933, A, 139, 466.11 Trans. Faraday Xoc., 1938, 34, 29.12 A. E. Stearn and H. Eyring, J . Chem. Physics, 1935, 3, 778.13 Ann.Reports, 1936, 33, 89-91.14 J . Chem. Physics, 1937, 5, 113.* This has given rise to the unfortunate expression “the quasi-thermo-dynamic theory of reactions.’’ There is no qualification to be attached tothe thermodynamic formulation of the general theory of reaction rates; ifcompetently done, the thermodynamic expressions are precise and can beapplied with complete confidence. The theory itself is not, of course, atheorem in thermodynamics, and should not be so described. In any casethe prefix “ quasi- ” is meaningless when applied to thermodynamics46 GENERAL AND PHYSICAL CHEMISTRY.that the activity coefticient of the activated complex varies with thedielectric constant of the medium in the way to be expected for amolecule with a very high dipole moment ; they have also estimatedthe free energy of proton transfer from one water molecule to anotherfrom the data for the mobility of the hydrogen ion and for thedispersion of the dielectric constant of ice.E. A. Guggenheim l6has written a valuable account of the thermodynamic relations foractivated complexes, and the use of the correct forms of theserelations should go far to remove misunderstandings.E. Wigner 16 has considered the association of atoms by three-body collisions, and has pointed out that in such a case there is noactivation energy and that the transition state is not so muchthe passage through a surface in space as the passing of the relativeenergy of the atoms through the value zero. He is able to fixupper limits for the rates of such reactions, and these values agreefairly well with the results of E.Rabinovitsch and W. C. Wood 1 7on the recombination of iodine atoms. H. Gershinowitz 18 has usedthe energy surfaces constructed for the interaction of three hydrogenatoms in a, consideration of the transfer of energy in molecularsystems : the probability of energy transfer is dependent upon thestate of excitation of the reagents, and we thus have quite differentrelative efficiencies of gases for energy transfer in the dispersion ofsound and in unimolecular reactions.Gas Reuctiorns.-F'resh examples of unimolecular reactions havebeen found with features similar to those already investigated:amongst those investigated are the cis-tr~~s-isomerisation ofp-cyanostyrene,l9 the decomposition of tert.-butyl halides,2* thedecomposition of methylene diacetate, dipropionste, and dibutyrateand of allied 22 and the decomposition of benzylidene-azine and of a-a~otoluene,~~~ 24 of ethyl and n-propyl nitrites 25 andof tetramethylsilicon .26The status of many unholecular reactions was questioned someyears ago by F.0. Rice and K. F. Her~feld,~' who suggested thatthey are chain reactions involving free radicals. The presence of1 5 Trans. Faraday SOC., 1937, 33, 607. l6 J . Chem. Physics, 1937, 5, 72Q.17 Ibid., 1936, 4, 497. Is Ibid., 1937, 5, 54.19 G. B. Kistiakowsky and W. R. Smith, J . Amer. Chem. SOC., 1936, 58,20 G. B. Kistiakowsky and C. H. Stauffer, ibid., 1937, 59, 165.21 C. C. Cof€in and W.B. Beazley, Canadian J . Res., 1937, 15, B, 229.22 C. C. Coffin, J. R. Dacey, and N. A. D. Parlee, ibid., pp. 247, 254, 260.23 A. Williams and A. S. C. Lawrence, Proc. Roy. SOC., 1936, A , 156, 444.24 Idem, ibid., p. 455.25 E. W. R. Steacie and S. Katz, J . Chem. Phy&x, 1937, 5, 125.26 D. F. Helm and E. Mack, J . Amer. Ohern. Soc., 1937,59, 60.27 Ibid., 1934, 56, 284.2428WYNNE-JONES CHEMICAL KINETIUS. 47free radicals in these reacting systems has been shown by Pease 28* 29and by C. J. M. Fletcher and G. K. Rollef~on,~O and L. A. K. Staveleyand C. N. Hinshelwood 31 have made a detailed study of the effectof nitric oxide on the kinetics of these reactions. Nitric oxide issupposed to act by removal of free radicals, aiid the inhibitedreaction is regarded as the simple decomposition : for the ethers,aldehydes, and ketones, the mean chain length, defined as the ratioof the uninhibited and inhibited reactions, is small, and this isregarded as indicating that only a small proportion of the primaryacts yields radicals but these give rise to long chains.Since theinhibited reaction shows the same kinetic behaviour as the un-inhibited, conclusions previously drawn as to the nature of thesereactions are substantially On the other hand, M. W.Travers and his collaborators 33 have carried out an extensive seriesof detailed analyses of the reacting systems and conclude that thewhole course of these reactions is far too complex for the manometricmethod of following the reaction rate to yield results of significance.For the decomposition of dimethyl ether, P.3’. Gay and M. W.Travers 34 find that the effect of increasing additions of nitric oxideis to suppress completely the normal decomposition and to set upanother process of oxidation. It is evident that much more detailedstudy is required before our knowledge of these processes can beregarded as satisfactory.Reactions between nitric oxide and oxygen, chlorine, or bromineare the best known examples of third-order reactions in the gaseousstate, but there has always been discussion about their mechanism.R. H. Crist and G. M. Calhoun35 have studied the oxidation ofcarbon monoxide in presence of nitric oxide, and claim from theirresults at low oxygen pressures that the rate-determining step isthe formation of nitrogen trioxide from nitric oxide and oxygen;other authors have regarded the molecule (NO), as the reactingspecies.W. Krauss and M. continuing the work ofM. Bodenstein’and W. have shown that the reaction2N0 + C1, = 2NOC1 is accurately of the third order, and theirresults are of great importance, because, when combined with thedata of G. Waddington and It. C. Tolman3s for the bimolecular28 R. N. Pease, J . Arner. Chern. SOC., 1937, 59, 425.29 L. S. Echols and R. N. Pease, ibid., p. 766.30 Ibid., 1936, 58, 2129.31 Proc. Roy. SOC., 1937, A , 159, 192.33 Trans. Faraday Soc., 1937, 33, 1342.3 5 J . Chem. Physics, 1937, 5, 301.36 2. physikal. Chem., 1937, A , 178, 245.37 Ibid., 1936, A, 175, 294.3 8 J .Arner. Chem. Soc., 1935, 57, 689.32 J., 1937, 1568.s p Ibid., p. 75648 GENERAL AND PHYSICAL CHEMISTRY.decomposition of nitrosyl chloride, they give with high accuracy-J. K. Dixon’s 39 values for the equilibrium constants.Combustion of Gases.-The phenomenon of combustion is charac-terised by considerable kinetic complexity and marked dependenceupon the nature and dimensions of the containing vessel; never-theless, considerable progress has been made as the result of carefulexperimentation and the detailed working out of reaction schemes,particularly following the work of N. Semen~ff.~O R. G. W. Norrishand S. G. Foord 41 have examined the combustion of methane andshown that the marked induction period followed by a gradualincrease in the reaction velocity, the almost complete inhibition ofthe reaction by packing the vessel, and the variation of the ignitiontemperature with pressure can all be accounted for on the atom-chain hypothesis. It is assumed that oxygen atoms are producedat the surface by some such reaction asCH4 + 0, = HCHO + H,O; H*CHO + 0, = H*C02H + 0I J.H20 + COthe formaldehyde concentration then steadily increases by a straightchain of the type0 + CH, = CH, + H,O; CH, + 0, = H*CHO + 0Formaldehyde molecules may be occasionally removed by oxidationby molecular oxygen, thus causing chain-branching and allowingan exponential growth of the formaldehyde concentration. Theultimate removal of formaldehyde from the system is assumed tooccur by the reactionHCHO + 0 = CO + H2Owhich causes a decrease in the effective chain-branching factor andmakes the rate of combustion reach a steady value. As the pressureis increased, the rate increases until the system is unable to dissipatethe heat of reaction and ignition occurs.This scheme is simple and seems to account for most of the facts,but G.von Elbe and B. Lewis42 consider modification necessaryin order to explain the maximum in the rate which is observed atabout 350” for all paraffin hydrocarbons except methane, also thephenomenon of “cool” flames, and the production of methylalcohol in large amounts during the oxidation of propane. The39 2. physikal. Clz,em., 1931, Bodenstein Festband, p. 679.40 “ Chemical Kinetics and Chain Reactions,” Oxford, 1934.41 Proc.Roy. SOC., 1936, A, 157, 503.42 J . Arner. Chern. SOC., 1937, 59, 976WYNNE-JONES : CHEMICAL EINETICS. 49scheme suggested by these authors involves hydroxyl-radical chains,and for the oxidation of methane they suggestOH + CH, = CH, + H2OCH, + 0 2 = HGHO + OHfollowed byOH + HCHO = HCO + H,OHO, + H*CHO = CO + H,O + OHThis leads to much the same results as the treatment of Norrishand Foord, but in addition affords an explanation of the formationof methyl alcohol, since with higher hydrocarbons CH,R radicalswill be formed and these may react with oxygen giving first peroxideradicals and then methyl alcohol. The luminescence is ascribedto the breakdown of radicals, yielding formaldehyde with very largeexcess of energy.Numerous other investigations have been carried out on thecombustion of hydrocarbons 43 and the hydrogen-oxygen reaction.44Interesting results have been obtained from studies of the exchangereactions between hydrogen and deuterium atoms and variousmolecules containing hydrogen.A. Farkas and H. W. Melville45examined the mercury-sensitised exchange reactions of deuteriumwith ammonia, methane, and water, and compared the rates withthe absolute intensity of the light and with the rate of the ortho-para-hydrogen conversion in the reacting mixture which enabled acalculation of the concentration of hydrogen or deuterium atoms.The energies of activation for the reactions of deuterium atoms withmethane, ammonia, and water were found to be 13,11, and 7 kg.-cals.respectively.The value for methane is confirmed by the work ofE. W. R. S t e a ~ i e , ~ ~ but the measurements of H. S. Taylor and hiscollaborators 47 on the methane reaction, using the mercury reson-ance radiation and also a Wood's discharge tube for producinghydrogen atoms, suggest that the reaction is complex and that theHCO + 0, = HO, + CO43 B. V. Aivazov and M. B. Neumann, J . Phys. Chem. Russia, 1936, 8, 543;1937, 9, 231; E. A. Andreev, Acta Physicochirn. U.R.S.S., 1937, 6, 57; M.Rivin and A. Sokolik, ibid., p. 105; P. Sadovnikov, ibid., p. 419; G. P. Kane,E. A. C. Chamberlain, and D. T. A. Townend, J., 1937, 436.44 G. von Elbe and B. Lewis, J . Amer. Chern. Soc., 1937, 59, 2025; H.Kondrateeva and V. Kondrateev, Acta Physicochim. U.R.S.S., 1937, 6,625; G.von Elbe and B. Lewis, J . Amer. Chem. Soc., 1937, 59, 656; N.Semenova, Acta Physicochim. U.R.S.S., 1937, 6, 25; A. Biron and A. Nal-bandjan, ibid., p. 43.45 Proc. Roy. Soc., 1936, A, 157, 625.46 Canadian. J . Res., 1937, 15, B, 264.47 N. R. Trenner, K. Morikawa, and H. S. Taylor, J . Chem. Physics, 1937,5, 203; K. Morikawa, W. S. Benedict, and H. S. Taylor, ibid., p. 21250 GENERAL AND PHYSICAL CHEMISTRY.minimum activation energy for the exchange between deuteriumatonis and methane is 15-6 kg.-cals.Reactions in Solution.-Reactions catalysed by acids and bases areof considerable value in the study of the various factors influencingthe rates of reactions in solution, and afford the best known andmost carefully investigated example of the relation between ratesof reaction and chemical equilibria, vix., the Brernsted equationk = G , K". Recent developments in the study of these reactionshave been summarised in papers by R. P. Bell, K. J. Pedersen,W. F. K. Wynne-Jones, and K. F. Bonhoeffer, presented a t thediscussion held by the Faraday Society.48 Particular reference mustbe made to Pedersen's, whose analysis of protofropic reactions isthe most important contribution to our understanding of thesechanges since Brsnsted first showed what is the true function of anacid or basic catalyst.The decomposition of nitroamide, which was the first reaction forwhich the Brarnsted conception was shown to be applicable,49 hasbeen re-examined by E.C. Baughan and R. P. Bell," who find thatthe Brransted relation hold6 accurately for catalytic effects of anumber of anions a t different temperatures. V. K. LaMer andJ. Greenspan 61 have studied this reaction in heavy water, in whichthe rate of the solvent-catalysed reaction is about one-fifth of thatin ordinary water : at the discussion, LaMer stated that subsequentmeasurements have shown that the rate of decomposition ofdideuteronitroamide is the same as that of nitroamide, proving thatthe two hydrogen atoms in nitroamide exchange rapidly with thoseof water.Another classical reaction, the mutarotation of glucose, has beenexamined by G. F. Smith,52 who has determined values of thecatalytic coefficients of the hydroxyl, glucosate, acetate, and otherions and also of the molecules of water, acetic, and chloroaceticacids a t different temperatures. The energies of activation areabout 17,000--18,000 cals.for all the catalysed reactions, but the'' steric factors " vary enormously and can be approximatelycalculated from the equation P = 0-36 x where K is thedissociation constant of the acid. The data also show clearly thatthe energy of activation for the water-catalysed reaction is far frombeing constant over the temperature range 0-35".Tb e depolymerisation of paraldehyde in benzene, nitrobenzene,48 Tram. GaTaday Soc., 1938, 34, 229, 237, 245, 252.4Q J. N. Br~nsted and K. J. Pedersen, 2. physikal. Chem., 1924,108, 185.50 Proc. Roy. SOC., 1937, A , 158, 464.51 Trans. Faraday SOC., 1937, 33, 1266.S2 J ., 1936, 1824; G. F. Smith and (Miss) IT. C. Smith, J., 1937, 1413WYNNE-JONES : CHEWCAL KINETICS. 51amyl acetate, and anisole has been found to be catalysed by acids,and the order of the reaction suggests that all three oxygen atomsof the paraldehyde molecule are involved.53 An attempt to studythe same reaction in the gaseous phase showed that it is pre-dominantly heterogeneous.54The interesting problem of the difference between specific effectsof hydrogen and hydroxyl ions and general acid and basic catalysis,and also the question of pre-equilibria in these reactions, have beenconsiderably clarified by experiments with heavy water. Avaluable method of investigating equilibria involving hydrogenions was proposed by P.Gross, H. Suess, and H. Steine~,~~ whoshowed that the equilibrium constant for a reaction of this typewill vary with the isotopic composition of the water according tothe equation -Bin = KO . cp + K1 . cD, where Kn is the equilibriumconstant for association with hydrogen ions in a solvent with amo1.-fraction n of deuterium oxide, KO and Kl are the correspondingconstants for pure water and deuterium oxide, and cp and CD are theeffective proton and deuteron concentrations, If the rate of areaction is governed by a similar equilibrium, we have the analogousequation for the specific rate constantsIn this way it has been possible to relate the decomposition ofdiazoacetic ester,56 the hydrolysis of a ~ e t a l , ~ ~ and the inversion ofsucrose 56 to the data for the equilibrium constants of p i ~ r i c , ~ ~and acetic acids.60 These results show that the hydrogenion is a specific catalyst for these reactions.For reactions whichrequire both an acid and a bme we may have general acid andbasic catalysis in spite of the existence of pre-equilibria between thesubstrate and hydrogen ions: an important discussion of thisproblem, based on work on the bromination of acetone and ofnitromethane as well as on the mutarotation of glucose, has beengiven by K. P. Bonhoeffer and 0. Reitz.61I n last year’s Report reference was made to a determination ofthe rate of the exchange reaction C,H,-OH + HOD = C,H,*OD +H20; if this were a slow reaction, the neutralisation of acids andbases in alcohol would be slow, which is contrary to experience;53 R.P. Bell, 0. M. Lidwell, and M. W. Vaughan-Jackson, J., 1936, 1792.54 R. P. Bell, and R. le G. Burnett, Trans. Paraday SOC., 1937, 33, 355.55 Ibid., 1936, 32, 883.58 P. Gross, H. Steiner, and F. Gauss, ibid., p. 877.67 W. J. C. Orr and J. A. V. Butler, J., 1937, 330.68 P. Gross, H. Steiner, and H. Suess, Trans. Paraday Soc., 1936, 32, 883.6% p. Gross and A. Wischin, ibid., p. 879.60 V. K. LaMer and J. P. Chittum, J. Amer. Chem. Soc., 1936,58, 1642.61 Z. physikal. Chem., 1937, A, 179, 13652 GENERAL AND PHYSICAL CHEMISTRY,it is therefore satisfactory that a repetition of the work has shownthat the reaction is too rapid to be measured.62The dissolution of metals in acids, which used to be regarded as itreaction controlled largely by the rate of diffusion of hydrogen ionsto the metal surface, was shown by M.Kilpatrick and J. H.RushtonG3 and also by J. N. Brransted and N. L. Ross Kane 64 to beeffected by acid molecules, The work of C. V. King and M. M.Braverman65 made it appear that diffusion plays an importantpart in controlling the reaction rate, but these authors were unableto formulate a mechanism which would restore the unique positionof the hydrogen ion. Additional evidence has now been obtainedof the action of molecules other than the hydrogen ion,66 but ithas also been shown that the diffusion coefficients of different acidsrun roughly parallel to the rates at which they react with metals.q7Relations similar to the Bronsted equation have been observedfor overvoltage (the Tafel equation) and for oxidation-reductionreactions (Dimroth’s relation), but a generalisation of an unexpectedtype was made by L.P. Hammett and H. L. Pfluger 68 when theyround a correlation between the rates of reaction of triethylaminewith certain esters and the dissociation constants of the acids fromwhich the esters were derived. A summary of this work is givenby Hammett,6g and other recent contributions have been madeby him and Roberts 7O and also by G. N. Burkhardt, C. Horrex, andD. I. Jenkins.71These linear relationships between the free energies of equilibriumprocesses and the allied rates have been considered by Evans andP ~ l a n y i , ~ ~ who have shown that they may be fitted into a con-sistent framework * on the basis of the general theory of reactionrates. J.A. Christiansen, in developing his theory of a chemicalreaction considered as it phenomenon of intramoleeular diff~sion,~G2 J. C. Jungers and K. F. Bonhoeffer, 2. physikal. Chem., 1936, A, 1’97, 460.G3 J . Physical Chem., 1930, 34, 2180; 1934, 38, 269.64 J . Amer. Chem. Soc., 1931, 53, 3624.66 Ibid., 1932, 54, 1744.66 M. Sclar and M. Kilpatrick, ibid., 1937, 59, 584; A. Quartaroli, AttiV Cortgr. Naz. Chim., 1936, 2, 466; M. Centnerszwer and W. Heller, J . Chirn.physique, 1937, 34, 217.6 7 C. V. King and W. H. Cathcart, J . Amer. Chem. SOC., 1937, 59, 63.6 8 Ibid., 1933, 55, 4079.70 L. P. Hammett, J . Amer. Chem. SOC., 1937, 59, 96; I. Roberts and L.P.71 J., 1936, 1649, 1654.72 Trans. Faraday SOC., 1936, 32, 1333; M. Polanyi, ibid., 1938, 34, 11;M. G. Evans, ibid., p. 49. * The criticism of the “ derivation ” of these relations is really only astatement that Evans and Polanyi have made a consistent formulation.Tram. Paraday SOC., 1938, 34, 156.Hammett, ibid., p. 1063WYNNE-JONES : CHEMICAL KINETICS. 53is led to the same type of relationship and also to the well-knowncorrelation of the frequency factor and energy of activation forrelated reactions.Polymerisation reactions are being extensively studied , and someof them are of great value because they can be examined in thegaseous phase as well as in a, number of solvents. A, Wassermann 73has given an account of the results obtained by himself and hiscollaborators for a number of reversible diene syntheses.Theassociation reactions, which both in the gas and in solution appearto be simple and bimolecular, have a '' steric factor '' of aboutin all phases; this corresponds to a small entropy of the activatedstate and is in agreement with considerations of the equilibria.Wassermann has shown that if the " steric factor " is assumed tobe the same in all phases, the collision number in a liquid is aboutten times the value in the gas,J. B. Harkness, G. B. Kistiakowsky, and W. H. Mears '* haveexamined some gaseous polymerisations and found that, although thedimerisations of cyclopentadiene, AaY-pentadiene, and py-dimethyl-hay-butadiene proceed smoothly and follow a simple kinetic course,yet styrene, vinyl acetate, and methylacetylene do not appear topolymerise.Other authors 75 claim to have measured the rates ofpolymerisation of the latter substances, but their results are ascribedby A. C. Cuthbertson, G. Gee, and E. K. Rideal 76 to catalysis byperoxides.Polymerisation and other reactions have been examined at thesurfaces of solutions by following the change in the interfacialpotential difference : the results show the marked effect oforientation of the molecules upon the course of theAn investigation which merits particular notice is the work ofJ. Steinhardt 7* on the inactivation of pepsin, a reaction which isprobably typical of protein denaturation processes. By a carefuland systematic examination of salt effects, Steinhardt was able toshow that the reaction rate is inversely proportional to the fifthpower of the hydrogen-ion concentration and that it is thereforethe pepsinate ion which reacts ; this interpretation changes theactivation energy from the apparent value of 63,500 cals.for the73 Trans. Paraday Soc., 1938, 34, 128.74 J . Chem. Physics, 1937, 5, 682.7 6 G. V. Schulz and E. Husemann, 2. physikal. Chern., 1937, B, 36, 194;7 G Nature, 1937, 140, 889.77 0. Gee and E. K. Rideal, J., 1937, 772; E. K. Rideal and J. S. Mitchell,Proc. Roy. SOC., 1937, A , 159, 206; A. E. Alexander and J. H. Schulman, ibid.,161, 115; A. E. Alexander and E. K. Rideal, ibid., 163, 70.H. Suess, K. Pilch, and H. Rudorfer, ibid., 1937, A , 179, 361.7 8 Kgl.Danske Videnskab. Selskab, 1937, 14, 1164 GENERAL AND PHYSICAL CHEZVIISTRY.reaction at constant pH to a value of 18,300 cala. for the reaction ofthe ion itself, and enables a much simpler mechanism t o beformulated.Amongst other important reactions which have been studied insolution may be mentioned the formation and decomposition ofquaternary ammonium salts,79 for which valuable equilibrium datawere also obtained; the hydrolysis of esters and the formation ofmethylpyridinium iodide in various solvents ; 80 esterification inbenzene solution ; *1 the rearrangement of N-chloroacehnilide,for which the energy of activation has been shown to vary with thetemperatureYs2 and the kinetics of which have been examined bythe use of radioactive hydrochloric acid ; g3 and the reaction betweenarsenic acid and iodine, which has been followed at equilibrium bythe exchange of radioactive arsenic.84 W.I?. K. W.-J.4. PHOTOCHEMISTRY.Since the last Report on photochemistry appeared1 a veryconsiderable amount of work on photo-reactions has been carriedout, and publications of the last few years in particular have led toa greatly increased understanding of the subject in relation to thebroader principles of general chemical kinetics. As a basis for suchinvestigation, it may be stated that the Einstein law is applied onlyto the elementary act as an absorption of light by a quantumprocess involving one quantum per absorbing molecule; the totalphotochemical yield is determined by the secondary thermalreactions of the system produced by absorption.It does notfollow that every absorbing molecule becomes activated ordissociated with a consequent primary quantum yield of unity ;internal redistribution or degradation of energy is possible, moreespecially in polyatomic molecules, while collisional deactivationmust be considered. Further advance has been made in theelucidation and comparison of all such processes in both gas-phaseand solution. Considerable progress is recorded in the fullerinvestigation of the photolysis of the simpler organic compounds,the results of which, in addition to their intrinsic interest, againyield valuable information and guidance in the examination of thefundamental processes of absorption ; in this group, carbonyl79 W.C. Davies and R. G. Cox, J . , 1937, 614.80 R. A. Fairclough and C. N. Hinshelwood, ibid., p. 538.81 M. M. Davies, Trans. Paraday SOC., 1937, 33, 331.82 J. 0. Percival and V. K. LaMer, J . Amer. Chem. SOC., 1936, 58, 2413.83 A. R. Olson, C. W. Porter, F. A. Long and R. S. Halford, ibid., p. 2467.84 J. N. Wilson and R. G. Dickinson, ibid., p. 1368.1 Ann. Reports, 1932, 29, 46RITCHIE : PHOTOCHEMISTRY. 55compounds have been subjected to further detailed investigation,and it is to them that attention is first directed.Carbonyl Compounds.4tudies in this series are of particularinterest in view of recent conclusions regarding the thermaldecomposition of organic molecules in general: e.g., L. A. K.Staveley and C. N.Hinshelwood conclude that the primary act inthe thermal decomposition of an organic molecule may involveinternal rearrangement to stable products or the production offree radicals or a combination of both these processes. Norrishand his co-workers 3* * have previously recognised the existence oftwo types of photo-decomposition, type I being predominant withshort-chain compounds and expressed byROCOR'-+ CO + (RR + RR' + R'R'),while type I1 is characteristic of long-chain compounds andinvolves the cracking of the long hydrocarbon chains in a positionC Z ~ to the absorbing carbonyl group, giving a lower carbonylcompound and an olefin. Investigation of these modes ofdecomposition has been extended to the liquid state and tosolution. Liquid acetone or acetone in benzene solution shows noreaction, but in cyclohexane some reaction occurs with thesolvent, the molecules being deactivated by collision, or removedby chemical reaction, with the solvent molecules.With di-n-propylketone in cyclohexane a t 20°, practically pure ethylene is evolved(by type I1 mechanism), whereas in the gas phase both types ofdecomposition occur; hence, in solution at room temperature thetype I mechanism of decomposition is suppressed. In a similarway, methyl ethyl and diethyl ketones show no appreciablephotolysis at 20"; but at loo", the type I mechanism is againobserved, with the important difference that radicals formed byphotolysis do not readily combine to give ethane, etc.; instead,RH is formed in an amount which is more than double theequivalent amount of evolved carbon monoxide, the general resultssuggesting that the reaction CH,*CO + CnH2%+ a = CH,*CHO +C,H,,I is possible in the solution.In the aldehydes acetaldehyde,butaldehyde, and isovaleraldehyde, on the other hand, decom-position at room temperature is the same as in the gas phase, nounsaturation in the solvent being found, and less than 2% ofhydrogen in the products; accordingly, Norrish and Bamford areof the opinion that under such conditions the type I mechanism inJ., 1937, 1668.C. H. Bamford and R. G. W. Norrish, J., 1935, 1504.E. J. Bowen and A. T. Horton, J., 1936, 1685.3 R. G. W. Norrish and (Miss) M. E. S. Appleyard, J., 1934, 874.6 R. G. W. Norrish and C. H. Bamford, Natave, 1937,140, 19556 GENERAL AND PHYSICAL CHEMISTRY.aldehydes is strikingly differentiated from that in ketones, theprimary act being represented by RCHO + hv = RH + CO.Notemperature coefficient is thus observed, in contrast to the effectwith the free-radical mechanism of the ketones. Type I1mechanism is unaffected by solution ; for instance, methyl n-butylketone decomposes in cyclohexane as in the gas phase, the productsbeing independent of temperature. The mechanism of type I isthus regarded as entirely separate from that of type 11, thisdifference indicating that such molecules may be activated in twoways.No hydrogen is obtained by illumination of acetone vapour bylight corresponding roughly to 1995-1 820 A. , the absorptionspectrum in this region consisting of discrete bands overlaid witha faint continuum.The overall reaction is expressed byCO(CH,), = C,H, + CO. The independence of quantum efficiency(y) with the energy absorbed (labs.) indicates no generation ofacetone from decomposition products, while the addition of carbondioxide and nitrogen increases the quantum efficiency and indicatesdissociation of activated acetone molecules at every collision ;ethane is less efficient. y does not appear to approach zero at lowpressure; no fluorescence was detected and if it does not exist, anactivated but stable state of the molecule must be postulated,corresponding to the idea of degradation to heateg The results arecorrelated by the use of potenfial-energy surface diagrams.Theseresults for the banded region of absorption are generally confirmedby R. Spence and W. Wild,lO who conclude that the excitedmolecule may dissociate directly into ethane and carbon monoxidewithout the intervention of radicals. On the other hand, diacetylcan be isolated from acetone illuminated at room temperature bylight of the continuous region of absorption,ll this indicatingCO(CH3), + hv = CH,*CO + CH, as the primary process. Theratio C,H,/CO is no longer unity but approaches a maximum valueof about 2.5 with increasing light intensity, one of the contributingreactions being CH, + CH,*CO = C,H, + CO. The acetyl radicalis concluded to be fairly stable at room temperature, but thestability rapidly decreases as the temperature is raised.Increaseof acetone pressure or decrease in light intensity increases theproportion of methane in the products at 60°, the reactionsuggested being CH, + CO(CH,), = CH4 + CH,*CO*CH,.7 R. 0. W. Norrish and C. H. Bamford, Nature, 1936, 138, 1016.8 J. P. Howe and W. A. Noyes, jr., J. Arner. Chem. Soc., 1936, 58, 1404.9 R. G. W. Norrish, H. G. Crone, and (Miss) 0. D. Saltmarsh, J., 1933, 1533.10 J., 1937, 352.11 See also M. Barak and D. W. G. Style, Nature, 1935,135, 307RITCHIE PHOTOCHEMISTRY. 57Various types of evidence may be adduced in general support ofthe above schemes. The quantum efficiency of decomposition ofmethyl n-butyl ketone (2480-2770 A.) is given l2 as 0.03 for type Imechanism, and 0.27 for type 11, condensation or polymerisationaccounting for 0.04.No fluorescence was detected, and since theabsorption spectrum is completely diffuse, the 65 % of moleculeslosing their excitation energy are regarded as deactivated withoutdecomposition either by internal processes or by collision. Bymetallic-mirror methods, methyl and ethyl radicals have beenisolated in the photo-dissociation of dimethyl and diethyl ketones,l3and n-propyl radicals result similarly from illumination of di-n-propyl ketone l* and from diisopropyl ketone, although in thelatter case, the isopropyl radical is first formed and immediatelytautomerises.15 The interconversion of ortho- and para-hydrogenhas been employed to show the production of the paramagneticmethyl radical from illuminated acetone and methyl iodide, but nofree radicals from propaldehyde.The polymerisation of acetyleneand ethylene photo-sensitised by acetone is regarded as due to thepresence of free radi~a1s.l~ These were not isolated fromacetaldehyde, however,18 and no free hydrogen atoms were obtainedfrom f~rmaldehyde,~~ either in the region of fine structure or inthat of predissociation, but at high temperatures the chaincharacter of the photolysis is explained in terms of free hydrogenor methyl radicals;20 when light in the banded region of thespectrum is absorbed by molecules in high vibrational levels, theprobability of transition from the excited state to an unstablestate yielding free radicals is greatly increasedl0 In acetone, thequantum measurements show that at high temperatures no chainis involved, but that chain formation occurs with acetaldehyde; 21these results have been confirmed 22 for formaldehyde, propaldehyde,and acetone, and are kinetically in agreement with the idea ofvirtually simultaneous rupture of the two bonds attaching theradicals to the carbonyl grouping.Chains in the photochemicaldecomposition of acetaldehyde . and propaldehyde at 300" are12 B. M. Bloch and R. G. W. Norrish, J., 1935, 1638.13 T. G. Pearson and R, H. Yurcell, J., 1934, 1718; 1935, 1151; N.14 T. G. Pearson and R. H. Purcell, J., 1936, 253.1 5 H. H. Glazebrook and T. G. Pearson, ibid., p. 1777.16 W. West, J. Amer. Chem. SOC., 1935, 57, 1931.17 H. S. Taylor and J. C. Jungers, Trans. Faraday SOC., 1937, 33, 1353.1 8 T.G. Pearson and R. H. Purcell, J . , 1934, 1718.19 F. Patat, 2. physilcal. Chem., 1934, B, 25, 208.20 J. A. Leermakers, J. Amer. Chem. SOC., 1934, 56, 1537.21 Idem, ibid. ; C. A. Winkler, Trans. Faraday SOC., 1935, 31, 761.22 E. I. Akeroyd and R. G. W. Norrish, J., 1936, 890.Prileshajeva and A. Terenin, Trans. Paraday SOC., 1935, 31, 148358 GENERAL AND PHYSICAL CHEMISTRY.suppressed by the addition of small amounts of nitric oxide,23 onemolecule of the latter being required for each light quantumabsorbed by acetaldehyde and one for each five absorbed bypropaldehyde. Chains are regarded as broken by the reactionbetween nitric oxide and methyl, the initial step in the mechanismbeing the production, by light absorption, of methyl and CHOradicals followed by the collision of CH, with acetaldehyde to givecarbon monoxide, methane, and methyl.F.E. Blacet and J. G. Roof 24 have shown the presence ofhydrogen in the products of photolysis of acetaldehyde, and foundthat the percentage of hydrogen increases as the quantum absorbedincreases (3130-2537 A.). This is regarded as due to the differencein kinetic energy imparted to the products of dissociation, theauthors preferring to regard the initial step as CORH + hv =R + HGO, the hydrogen being derived from the recombination ofHCO radicals, which are here not regarded as unstable.25Crotonaldehyde, contrary to expectations based on analogy withthe saturated aldehydes from a comparison of their absorptionspectra,26 shows little decomposition on illuminati~n.~~ Nofluorescence was observed even at low pressures.Although theenergy transfer without decomposition may be admitted forspectral regions where the absorption is obviously banded, theauthors 27 prefer the concept of a predominating reverse reactionwith the above-mentioned initial dissociation into R and HCO.The low quantum efficiency of decomposition of acraldehyde 28 mayalso be regarded from the internal-energy degradation point ofview. A review by G. K. Rollefson29 of work on aldehydes ingeneral and on acetaldehyde in particular indicates that suchphoto-activated molecules can react in several ways a t comparablerates, e.g., emission of radiation, production of free radicals,rearrangement to give carbon monoxide and methane, polymeris-ation; the procedure favoured is a function of the energy of theexciting wa~e-length,~O considered in relation to the molecularstructure.Hydrogen atoms but no free methyl radicals are found by the23 J.W. Mitchell and C. N. Hinshelwood, Proc. Roy. SOC., 1937, A , 159,32; 600 H. W. Thompson and J. W. Linnett, Trans. Paraday SOC., 1937, 33,874; H. W. Thompson and M. Meissner, Nature, 1937, 139, 1018.24 J . Amer. Chem. SOC., 1936, 58, 278.25 See also M. Burton, ibid., p. 1655.2 6 See also P. A. Leighton, Chem. Reviews, 1935, 17, 393.27 F. E. Blecet and J. G. Roof, J. Arner.;Chem. SOC., 1936, 58, 73.28 H. W. Thompson and J. W. Linnett, J., 1935, 1452.29 J . Physical Chem., 1937, 41, 259.30 See P. A. Leighton, L.D. Levanas, F. E. Blacet, and R. D. Rowe, J..Amer. Chem. SOC., 1937, 59, 1843RITCHIE : PHOTOCHEMISTRY. 59metallic-mirror method during the photolysis of acetic acid,sl afree-radical mechanism being postulated. No hydrogen atoms arefound in the corresponding decomposition of formic acid, photolysisbeing regarded in this case as the formation of stable molecules inone primary act.32 Acetyl peroxide33 is readily decomposed byultra-violet light, the main products being carbon dioxide, ethane,and methane; the process is a complex one, although it is perhapsnoteworthy that the ratio C,H,/CH4 in the products is highestwhen the solid itself is illuminated, and falls off rapidly for theliquid and for solutions in cycluhexane and alcohol.Absorption by Halogens.--The extinction coefficients of iodinevapour a t room temperature have been measured 34 between 4300and 6200 A,, and found to be considerably greater than previouslyaccepted values; it is pointed out that the correct mean extinctioncurves in the band region can be obtained by the addition of inertgas, such addition having no effect on the continuous part of thespectrum. The “limiting” extinction curve so obtained foriodine vapour resembles that of the extinction curve of iodinedissolved in carbon tetrachloride. Extinction coefficients areincreased by solution, especially in the cases of chlorine andbromine, while the general shape or position of the maximaremains unchanged ; the electronic transitions involved are of theintercombination type lC- ---+ 311, prohibited normally in verylight atoms, but occurring under the influence of the electric fieldsof the surrounding molecules.The decrease in extinction coefficientobserved35 in strongly illuminated solutions of iodine in carbontetrachloride and benzene is due to the dissociation of iodinemolecules into atoms, where the observed interdependence of theenergy absorbed and the iodine concentration is in agreement withtheory; recombination of iodine atoms in solution is about 1000times greater than that of bromine atoms in helium a t atmosphericpressure, in accordance with the fact that triple collisions arenecessary for recombination and that each collision is a recombiningone in solution. The quantum yield of dissociation of iodinemolecules in solution is the same in the spectral region below 5000 A.in the continuum and above 5000 A.in the band region; loss ofenergy by collisions of the second kind between activated iodinemolecules and solvent molecules does not OCCUF to any appreciableextent. The quantum yield of iodine dissociation may be unity.3t M. Burton, J. Amer. Chem. Soc., 1936, 58, 1645.32 Idem, ibid., p. 1656; see E. Gorin and H. S. Taylor, &id,, 1934, 58,33 0. J. Walker and G. L. E. Wild, J., 1937, 1132.84 E. Rabinovitsch and W. C. Wood, Trans. PaTaday Soc., 1936, 32, 540.86 Idem, ibid., p. 547.2042; A. Terenin, Acto Physicochim. U.R.S.S., 1935, 3, 18160 GENERAL AND PHYSICAL CHEMISTRY.and the average lifetime of free atoms equal to the average collisioninterval in a monatomic iodine gas; or the quantum yield may bel l n (1 due to ‘‘ primary recombination ” 36 and the lifetime of theatoms n times larger than the collision interval in the gas.Thestationary concentration of radicals in such an illuminated solutionis independent of the occurrence of primary recombination providedthat the state of the atoms or radicals formed by light absorptionis not essentially different from that of the same particles meetingaccidentally in the solution.37 This last point is also discussed byG. K. Rollefson and W. F. Libby,3* who find it necessary to assumethat the primary action of light on a molecule must be ofcomparable efficiency in solution and in the gas phase, if there is aclose correlation between the two systems.Review has also beengiven of photo-reactions in non-ideal solutions,39 and a generalcorrelation of photo-reactions in gases with those in solution .40In illuminated iodine solutions in benzene and other solvents,41 thekinetics are somewhat obscured by the reaction of iodine atomswith the solvents.In iodine vapour 42 practically every excited iodine molecule(4990-6200 A.) dissociates into atoms by collision with othermolecules, the quenching of iodine fluorescence being then due todissociation and occurring at the first kinetic collision for gasesexcept helium. The transition from the stable excited state 31TOfto the unstable repulsive state 31T0-, normally prohibited in a freeiodine molecule by the symmetries involved, is made possible bythe effect of a magnetic or non-homogeneous electric field.Sincethe transition is the more difficult the further apart the potential-energy curves of the two states, the fluorescence is less sensitive(to added gas molecules) in the red than in the green. In theiodine-sensitised decomposition of ethylene iodide in carbontetrachloride,43 light corresponding to the region of continuousabsorption is only slightly more effective in bringing aboutdissociation of iodine in the solution than light corresponding tothe discontinuous absorption; in this paper attention is also givento the intensity distribution throughout the solution. Similarly,study of the photochemical reaction of bromine atoms with finely36 J.Franck and E. Rabinovitsch, Trans. Faraday SOC., 1934, 30, 120.37 E. Rabinovitsch and W. C. Wood, ibid., 1936, 32, 1381.38 J . Chem. Physics, 1937, 5, 569.39 G. K. Rollefson, Chem. Reviews, 1935, 17, 425.40 R. G. Dickinson, ibid., p. 413.4 1 E. Rabinovitsch and W. C. Wood, Trans. Paraday SOC., 1936, 33,42 Idem, J . Chem. Physics, 1936, 4, 358.43 R. G. Dickinson and N. P. Nies, J . Amor. Chem. SOC., 1935, 57, 2382.816RITCHIE : PHOTOCHEMISTRY. 61divided platinum** leads to the conclusion that the initial step inboth yellow and blue light reactions is the same.In general, the efficiency coefficient for a process involvingcollision between an excited and an unexcited atom to form adiatomic molecule with emission of radiation is very ~ma11,~5 andthere is no doubt that the general recombination of iodine andbromine atoms requires the presence of a third body as energyacceptor and stabiliser.Absolute values for the recombination ofiodine atoms with various inert gases as third bodies have beengiven by E. Rabinovitsch and W. C. Wood; 46 calculation on theordinary basis shows, for example, that one collision in 530 is arecombining one in helium, and one in 50 in carbon dioxide, atatmospheric pressure. The recombination velocity of bromineatoms is found 47, 48 to be approximately one-third of the iodinevalues; absolute figures have also been given by K. Hilferding andW. S t e k ~ e r . ~ ~ Relative values in the case of bromine have beenestimated from the rate of hydrogen bromide photo~ynthesis,~O andfrom studies of the Budde effect; 51 in general, the efficiencies ofdifferent gases in such a stabilising process may be expressed forthese gases which act as simple energy acceptors by a commonseries CO, > 0, > N, > A > Ne > He.This seems also to applyto the formation of other complexes, e.g., H02,52 NC14,s3 theefficiency being thus higher for diatomic and polyatomic moleculesthan for monatomic ones, and increasing with growing molecularsize and the intensity of the molecular fields of the collidingparticles. The recombination of chlorine atoms does not, in general,take place by a similar mechanism; G . K. Rollefson and H.Eyring 54 have discussed the formation of Cl, from atomic andmolecular chlorine, and use has been made in other connections 55of this conception, which is considered in greater detail later.Observation in such cases is frequently complicated by surface44 J.Urmston and R. M. Badger, J. Amer. Chem. SOC., 1934, 56, 343.45 R. Rompe, Physikal. Z., 1936, 37, 807.46 J . Chern. Physics, 1936, 4, 497.4 7 E. Rabinovitsch and W. C. Wood, Trans. Faraday Soc., 1936, 32, 907.4 8 E. Rabinovitsch and H. L. Lehmann, ibid., 1935, 31, 689.49 2. physikal. Chern., 1935, B, 30, 399.50 M. Ritchie, Proc. Roy. SOC., 1934, A , 146, 828.51 W. Smith, M. Ritchie, and E. B. Ludlam, J., 1937, 1680.62 M. Ritchie, ibid., p. 867.53 J. G. A. GrBiths and R. G. W. Norrish, Trans. Faraday SOC., 1931, 27,451 ; Proc. Roy. Soc., 1934, A , 147, 140.54 J .Amer. Chem. Soc., 1932, 54, 170.55 H. C. Craggs and A. J. Allmand, J., 1936, 241; H. C. Craggs, G. V. V.Squire, and A. J. Allmand, J . , 1937, 1878; C. F. Fisk and W. A. Noyes, jr.,J . Amer. Chem. SOC., 1936, 58, 170762 GENERAL AND PHYSICAL CHEMISTRY.action. The wall of the vessel may act as third body, or removeatoms from the gas phase by adsorption; diffusion and convectionmust be adequately considered.51~ 56 Various efficiencies of atomremoval have been recorded and disc~ssed.~' The reaction ofatomic iodine with quartz has been inve~tigated.~~The triple collision has been discussed by various authors :H. Senftleben and W. Hein 59 concluded that the effective collisiondiameter in the hydrogen-atom triple collision is not essentiallygreater than the gas-kinetic diameter.E, Rabinovitsch 6o dis-tinguishes between the collision mechanisms (A' + A") + X and(A' + X) $- A", assuming that recombination takes place atcomparatively large distances between the three particles, avoidingthe idea of quasi-molecules, and considering all three partners asindependent. The number of triple collisions then depends on27&g#8 + T ~ , ~ , where T ~ , ~ , . . is the period of the collision A' + A" andT ~ . ~ that of the collision A' + X ; application to recombinationdata shows that for nitrogen, oxygen, methane, and carbon dioxideas third bodies in the recombination of hydrogen, bromine, andiodine atoms, every gas-kinetic contact leads to recombination, butwith argon the efficiency is slightly smaller, and those of hydrogenand helium are about one-tenth of that of the heavier gases.Reactions involving the Halogens.-Recent studies of thehydrogen-chlorine photocombination lend support to the view thatthe various aspects of this complicated reaction may now bepresented in one comprehensive and correlated system.Ingeneral, the kinetics may be considered in three sections : (a)oxygen-free mixtures where the rate of homogeneous gas-phasereaction is proportional to the square root of the absorbedenergy,61* 62* 63 ( b ) oxygen-rich mixtures where the rate is pro-portional to the first power of the absorbed energy, and (c) atransition region where the homogeneous gas-phase rate isproportional to a power of the energy absorbed lying between 0.5and 1, and depending on the amount of oxygen present.64 In allsections, the production of chlorine atoms by the original light66 E.Rabinovitsch and W. C. Wood, Trans. Faraday SOC., 1936, 32, 917;E. Rabinovitsch, 2. physikal. Chem., 1936, B, 33, 275.67 G. Kornfeld and S . Khodschaian, ibid., 1937, B, &j, 403; W, Steiner,Trans. Faraday SOC., 1935, 31, 962; G. M. Schwab (with H. Friess), 2.phgsikal. Chem., 1936, 178, 123.5 8 G. Brauer, ibid., 1935, 174, 435.b0 Ann. Physik, 1936, 22, 1.60 Trans. Paraday SOG., 1937, 33, 283.61 J. C. Potts and G. K. Rollefson, J . Amer. Chem. Soc., 1935, 57, 1027.62 H. C. Craggs and A. J. Allmmd, J., 1936, 241.63 Craggs, Squire, and Nlmand, Eoc. cit., ref. (55).64 R. G. W.Norrish and M. Ritchie, Prm. Roy. SOC., 1933, A, 140, 713RITCHIE : PEOTOCHEMISTRY. 63absorption is followed by the propagation of chains by the Nernstmechanism.In oxygen-free mixtures, the intensity relationship indicates thatchains are brought to an end by recombination of chain carriers.In the experiments of Craggs, Squire, and Allmand 6s an importantfraction of the chains was broken by such a, recombination processunder conditions where triple collisions could be of very infrequentoccurrence; 65 a termolecular mechanism analogous to theabove-mentioned recombination of bromine and iodine atoms ishere out of the question. Recombination of chlorine atoms mustthen take place by the reversible formation of C1, from chlorineatoms and chlorine molecules, followed by the bimolecular collisionof two GI3 molecules, the reaction between Cl, and hydrogen beingtaken as negligible in comparison with that between C1 andhydrogen.Treatment of this mechanism in the usual way leads toan expression where the quantum efficiency is proportional to thepressure of hydrogen, in agreement with the results of earlierworkers, but inversely proportional to the pressure of chlorine, aneffect not previously. observed but now confirmed.66 It is alsopossible that chlorine atoms may be removed by reaction with GI,molecules, but study of the kinetic " order '' for chlorine indicatesthat this process is insignificant a t higher pressures of chlorine,although it, may play an increasing part as the chlorine pressurefalls.At low pressures of chlorine, the rate tends to become directlyproportiond to the energy absorbed, indicating that, oxygen beingabsent, reaction cliains are broken by adsorption and recombinationof chain carriers on the walls of the reaction vessel.The relativevalues of the wall collision efficiencies for chlorine atoms and GI,molecules are then of importance, that of the latter being taken asconsiderably greater.63 If the efficiency of chlorine-atom removalis high, addition of chlorine may raise the overall reaction rate by'' poisoning " the walls; but if the chlorine-atom efficiency is low,then addition of chlorine may retard the reaction rate by facilitatingthe production of the more easily adsorbed €3,. In general, the ratio[Cl,]/[Cl] must be considered in relation to pressure of chlorine,light intensity, and the nature of the walls.Similaz effects arefound with hydrogen chloride. A retardation 67 is regarded as dueto the stabilisation of the more easily adsorbed GI3 moleeules, anacceleration as due to the poisoning of the catalysed recombination of65 See also M. Tamura, Rev. Phys. Chem. Japan, 1937, 11, 1.66 G. V. V. Squire and A. J. Allmsnd, J., 1937, 1869.67 See M. Ritchie and R. G. W. Narrish, Proc. Rog. Soc., 1933, A, 140,.112; and ref. (61)64 GENERAL AND PHYSICAL CHEMISTRY.chlorine atoms on the walls, the latter state of affairs being favouredby reactive walls and low intensity.These general results of Allmand and his co-workers arecompatible with those recorded by earlier workers.Under certainconditions 63 " abnormal " intensity and chlorine-pressure effectsare obtained and related to diffusion of chlorine atoms (e.g., fromthe light beam to the dark zones).In oxygen-rich mixtures the rate of hydrogen chloride photo-synthesis is proportional to the first power of the absorbed energy,and mutual recombination of chain carriers does not occur to anyappreciable extent. When the ratio [H,]/[CI,] is large, chains arebroken predominantly by the removal of hydrogen atoms byreaction with oxygen and a third molecule; 68i G9* 62 the quantumefficiency may then be reduced by the addition of inert gas,52 ofhydrogen chloride and of hydrogen.'O In the cases of inert gas andhydrogen chloride, the process is taken to be the stabilised formationof the HO, complex, which reacts with hydrogen to give water;the quantum efficiency of water formation thus rises with increasingpressures of hydrogen chloride, carbon dioxide, nitrogen, oxygen,and hydrogen.52 The maximum efficiency (yHzo) appears to be2.52*68~ 71, 72 Hydrogen peroxide, which can be isolated undercertain conditions from illuminated mixtures of hydrogen, chlorine,and oxygen, is rapidly destroyed by illuminated chlorine,68 theperoxide molecule being able to break two chains in the hydrogen-chlorine rea~tion.'~ When the ratio [H2]/[C12] is small, chlorineatoms may be removed by reaction with oxygen, presumably bytriple collision processes also, where KO, may be produced by thereaction between chlorine atoms, oxygen, and hydrogen chloridemolecules. 74Consumption of oxygen without inhibition of hydrogen chlorideformation has been recorded and discussed.62Craggs and Allmand 62 confirm the earlier conclusion thatphoto-combination of hydrogen and chlorine will take place inlight of 5460 A., no reaction being observed at 5790 A.At 5460 A.energy chains are improbable; the primary process in the bandedregion generally is optical dissociation to chlorine atoms, the processbeing connected with the overlying continuum. These authors68 M. Bodenstein and P. W. Schenk, 2. physikal. Chem., 1933, B, 20, 420.69 K. B. Krauskopf and G. K. Rollefson, J. Amer. Chem. SOC., 1934, 56,70 R. G. W. Norrish and M. Ritchie, Proc. Roy.SOC., 1933, A , 140, 713.7 1 K. B. Krauskopf and G. K. Rollefson, Zoc. cit., ref. (69).72 D. L. Chapman and J. S. Watkins, J., 1933, 743.78 G. Kornfeld, 2. physilcal. Chew., 1937, B, 35, 236.74 G. Kornfeld and S. Khodschaian, ibid., p. 403.327RITCHIE : PHOTOCHEMISTRY. 65oalculate that the quantum efficiency rises from 5460 to 4 9 0 0 ~ .and suggest that normal chlorine molecules with v” I: 3, whenabsorbing light of these wave-lengths, dissociate into normal atomsby means of a transition from the excited state to one of thestates In, or 3xlU; according to N. S. Bayliss 75 this assumption isnot necessary, R. G. Aickin and N. S. Bayliss 76 having shown thatthe continuous absorption in this region is due a t least partly totransitions to the 3nIu state dissociating into normal atoms.Further, if it be assumed that the reaction in this region is causedby only that part of the absorption which is continuous, thequantum yields calculated by Craggs and Allmand are increased toa value practically the same as that at wave-lengths less than4 7 8 5 ~ . These authors also report that the quantum efficiency ispractically constant in the region 47854000 A., but decreases atstill shorter wave-lengths when the chlorine pressure is low andsurface action predominant ; differences may then be attributedto the more rapid absorption of Cl,.The maximum quantumefficiency is related 75 to the region in which photo-dissociationproduces both normal and excited chlorine atoms ; a6 wave-lengthsless than 4000 A., practically only normal atoms are pr0duced.7~Recalculation of Craggs and AZlmand’s earlier results by Baylissshows that the temperature coefficient of the reaction is the sameat 5460 as a t 4360a., the energy of activation being thusindependent of wave-length.It now seems definitely established that intense drying does notretard the photo-combination of hydrogen and chlorine, thequantum efficiency of the gas-phase reaction being independentof small amounts of water v a p ~ u r .~ ~ ~ 78 Where surface action isappreciable, an adsorbed film of water is found to facilitatesomewhat the catalytic action of the walls on chlorine atoms,77with a, corresponding effect on the overall rate of hydrogen chlorideformation. These results are analogous to those found for theBudde effect in bromine,51 where the rise in pressure on illuminationis practically independent of the presence of small amounts ofwater vapour, although again there is an indication that anadsorbed film of water vapour increases slightly the surface removalof bromine atoms.Other reactions involving the halogens continue to be the object75 Tram.Paraday SOL, 1937, 33, 1339.7 7 H. C. Craggs and A. J. Allmand, J., 1937, 1889.7 8 W. H. Rodebush and W. C. Klingelhoefer, jr., J. Arne?. Chem. SOC.,1933, 55, 130; F. Bernreuther and M. Bodenstein, Sitzungsber. peuss. Akad.Wis~., 1933, 333; G. K. Rollefson and J. C. Potts, J. Amw. Chern. Soc., 1933,55, 860.Ibid., p. 1333.REP.-VOL. XXXIV. 66 GENERAL AXD PHYSICAL CHEMISTRY.of invatigation and review.79 The radical C,H,Cl, is postulated inthe photochlorination of gaseous ethylene,s0 which a t constantlight intensity proceeds a t a rate proportional to the pressure ofchlorine, but is independent of the ethylene concentration ;practically no hydrogen reacts in a mixture of hydrogen, chlorine,and ethylene, while the ethylene reaction goes to completion.The photoformations of tetrachloroethane from trans- 81 and fromcis-dichloroethylene 82 are long chain reactions, retarded by oxygenand probably involving the radical C2M2C13; of similar nature isthe formation of pentachloroethane from trichloroethylene.82 Ofchain character also is the photochemical chlorination of dichloro-benzenes 83 and the reaction of chlorine with formic a ~ i d , ~ 4 thelatter involving chloroformic acid as a probable intermediate, andsurface removal of chlorine atoms being observed under certainconditions.Chlorination of tert.-butyl chloride and related com-pounds 85 and of liquid pentane 86 has been described.Photo-brominations include those of tetrachloroethylene andchloroform,87 of acetylene 88 and other unsaturated compounds, 89and photo-iodination of the ethylenic bond has also beenin~estigated.~~~ 90 The reactions of bromine with water 91 and ofiodine with sodium formate in aqueous solution92 have beendiscussed. The velocity of combination of hydrogen and fluorinein magnesium and platinum vessels is not appreciably increased byill~mination,~~ surface removal of atoms being regarded aspredominant.Halogen Compounds.-The quantum efficiency of decompositionof iodoform and related iodides is of the order unity,e* and increases79 H.J. Schumacher, Angew. Chem., 1936, 49, 613; 1937, 50, 483; Trans.80 T. D. Stewart and B. Weidenbaum, J. Amer. Chern. Soc., 1935,57, 2036.81 K. L. Miiller and H. J. Schumacher, Z. physikal. Chem., 1937, B, 35,82 Idem, ibid., p . 455.88 C. F. Fisk and W. A. Noyes, jr., J . Amer. Chem. Soc., 1936, 58, 1707.84 (Miss) H. L. West and G. K. Rollefson, ibid., p. 2140.85 A. 0. Rogers and R. E. Nelson, ibid., p. 1027.86 T. D. Stewart and B. Weidenbaum, ibid., 1935, 57, 1702.87 J. Willard and F. Daniels, ibid., p. 2240.88 W. Franke and H. J. Schumacher, 2. phgsikal. Chem., 1936, B, 34, 181.89 A.Berthoud and M. Mosset, J. Chim. physique, 1936, 33, 272.90 G. S. Forbes and A. F. Nelson, J. Amer. Chem. SOC., 1936, 58, 182;1937,59, 693; R. E. De Right and E. 0. Wiig, ibid., 1936,58, 693.91 H. A. Pagel and W. W. Carlson, 6. Physical Chem., 1936, 40, 613.Oa N. R. Dher and P. N. Bhargava, ibid., 1935,89, 1231.*a N. Bodenstein, H. Jockusch (with S. H. Chong), t. a w g . Chem., 1937,O4 K. E. Gibson and T. Iredale, Trans. Paraday Soc., 1936, 32, 571.Electrochem. Soc., 1937, 71 (26), 297.285.231, 24RITCEUE : PHOTOCHEMISTRY. 67with the length of the carbon chain, replacement of hydrogen byhalogen, and change of the central carbon atom from primary totertiary: these effects recall the reactivity of the sodium atomwith the chlorides,95 diminishing activation energies correspondingto increased quantum yields.Other investigations reported are ofethylene i0didQ6 iodof 0rm,~7 methyl iodide, and hydrogeniodide,g8 the initial light absorption being taken to form a radicalby the breaking-off of the halogen atom. In the case of oxalylchloride,9g results are in agreement with the assumption that themolecule breaks at the C-C bond in light of short wave-length(2537 A,), but with light of 3650 A. fission at, one oE the GC1 bondspredominates.Relatively large amounts of dichlorine hexoxide are formed inthe photo-decomposition of chlorine dioxide in carbon tetrachloridesolution,l where the mechanism probably resembles that operativein the gas system.2 The decomposition of chlorine monoxide bylight of the continuous absorption region of the visible spectrum isaccelerated by the presence of hydrogen, some hydrogen chlorideand water beiig formed; the reaction between (310 and hydrogenis regarded as mainly responsible.3 The quantum yield ofdecomposition of dichlorine monoxide is approximately the same at3650 A.as in the bromine-sensitised decomposition at 5460 A.,*sensitisation by bromine molecules being favoured ; some dichlorinehexoxide is formed and is taken to be due to the bromine-sensitiseddecomposition of chlorine dioxide.2Reactions of Hydrides.-In the direct decomposition by ultra-violet light of methane, hydrogen is the main product, togetherwith acetylene, some ethane and ethylene, and other hydrocarbons :the quantum efficiency is o€ the order 1.3.5 The decomposition ofethylene is also reported.6There is little doubt that the photo-decomposition of ammonia,involves the production of NH, and H in the initial stages; final96 H.von Hartel, N. Mew, and M. Polanyi, 2. physikal. Chem., 1932, B06 R. E. De Right and E. 0. Wiig, J . Amer. Chem. SOC., 1935, 57, 2411.9 7 R. Spence and W. Wild, Proc. Leede Phil. SOC., 1936-38, 3, 141.9 8 T. Iredale and (Miss) D. Stephan, Trans. Faraday SOC., 1937, 33, 800.99 I<. B. Krauskopf and G. K. Rollefson, J . Amer. Chem. SOC., 1936,58,443.1 J . W. T. Spinks and H. Taube, ibid., 1937, 59, 1155; see also B. Luther2 J. W. T. Spinks and J. I. Porter, J. Amer. Chem. Soc., 1034, 56, 264.3 T. Iredale and T.G. Edkards, ibid., 1937, 59, 761.4 A. G. Brown and J. W. T. Spinks, Canadian. J . Res., 1937, 15, B, 213.5 P. A. Leighton and A. B. Steinsr, J . Amer. Chem. ~ o c . , 1936, 58, 1823;6 R,. D. McDonald and R. G. W. Norrish, Proc. Roy. Soar, 1936, A, 157, 480.19, 139.and R. Hoffmann, 2. physikal. Chem., 1936,177, 17.W. Groth and H. Laudenklos, Natumuiss., 1936, 24, 79668 GENERAL AND PHYSICAL CHEMISTRY.products may markedly depend on the experimental conditions,e.g., size and material of the reaction vessel, total pressure, etc. Inthe initial stages of the illumination the proportion of hydrogen inthe non-condensable products may approach loo%,' the quantumefficiency then approaching unity and indicating little regenerationof ammonia from NH, and H.Triple or wall collisions accountfor the formation of hydrogen and hydrazine. As decompositiongoes on, the proportion of hydrogen falls to 75% as given by earlierworkers: a suggested reaction is H + N,H, = NH, + NH,. Athigher pressures of non-condensable products, the quantum yieldis much less than unity, and may depend markedly on the pressureof ammonia : decreased pressure decreases the number ofefficient triple collisions, e.g., in the regeneration of ammonia fromNH, and H and y rises; on the other hand, further reduction ofpressure causes the reaction to become heterogeneous, regenerationof ammonia being taken to occur at the walls, the quantumefficiency then decreasing. Under the experimental conditions,the rate was found to be proportional to the energy absorbed.The mechanism has been further discussed el~ewhere.~The direct photo-decomposition of trideuteroammonia 10 isapproximately 1.4 times slower than that of ammonia in thepredissociation bands at 2100 A.The same kind of dependence ofy upon pressure is found to hold as for ammonia under similarconditions : the mechanisms are similar, and the differences in ratedue to slower recombination of ND, radicals (+ N, + 2D,).The quantum efficiency at 2138 A. is greater than at 2100 A., thisbeing perhaps due to direct production of nitrogen and deuteriumfrom collision of activated trideuteroammonia molecules with anormal molecule without the production of radicals. The quantumefficiency of the mercury -sensitised decomposition of ammonia a tvarious temperatures 1l is found to be approximately the same asthat for the direct reaction; once the molecule is dissociated, thesecondary reactions are the same in both cases.Earlier work12points out the errors which may arise by taking the effective meanlife of the mercury atom as a constant and neglecting thereabsorption of resonance radiation at higher mercury pressures ; 137 H. J. WeIge and A. 0. Beckman, J. Amer. Chem. SOC., 1936, 58, 2462.8 E. 0. Wiig, ibid., 1935, 5'9, 1559; 1937, 59, 827.9 W. Mund, G. Brenard, and L. Kaertkemeyer, Bull. SOC. chim. Belg., 1937,46, 211 ; W. Mund and A. van Tiggelen, ibid., pp. 104, 227.10 E. 0. Wiig, J . Amer. Chem. SOC., 1937, 59, 955.11 H. W. Melville, Proc. Roy.SOC., 1936, A, 157, 621.l2 Idem, ibid., 1935, A, 152, 325.1s See also K. Morikawa, W. S. Benedict, and H. S. Taylor, J . Ciaem.Phyaics, 1937, 6, 21669 RITCHIE PHOTOCHEMISTRY.absolute and even relative values may become ~ncertain.1~ Theapplication l2 of such correction leads to the conclusion that withammonia and the trideutero-analogue, dissociation occurs oncollision with metastable mercury (3P0) atoms, derived in turn byquenching collision from mercury (3P1) atoms. The former collisionsare equally efficient for ammonia and the deutero-compound; andthe mechanism accounts for the abnormally high inhibiting effect ofhydrogen, which is 2 0 4 0 times greater than that computed fromthe relative quenching radii of ammonia and hydrogen, andtrideuteroammonia and deuterium. The efficiency of hydrogenand deuterium atoms in inhibiting the reaction of ammonia andthe trideutero-compound is equal ; the more rapid decompositionof ammonia is then due to the more rapid removal of NH, radicalsto give hydrogen and nitrogen.Quantum efficiencies in the mercury-sensitised decompositionof phosphine and trideuterophosphine l5 may vary with thepressure; i.e., the secondary reactions are not all influenced bychange in pressure to the same degree.Calculation indicates that,in contradistinction to the case of ammonia, phosphine quenchesthe initially excited mercury atom to the ground state directly.Otherwise, the mechanism resembles that of the ammoniadecomposition ; again, direct and photo-sensitised decompositionsinvolve the same secondary reactions.The temperature coefficientof the mercury-sensitised phosphine decomposition is unity,16 as isfound with the direct decomposition. The decomposition of PH,and PD, gives P, units, appearing as a deposit of red phosphorus.17The quantum yield in the mercury-sensitised photo-decompositionof arsine is found 18 to be approximately unity, the decompositionbeing only very slightly inhibited by hydrogen. The correspondingdecomposition of silane 19 yields hydrogen and a polymerisedhydride [SiHJn(x <0.9), or a mixture of silicon and such a polymer :the radicals SiH, or SiH, probably play a part in the mechanism.Photo-sensitised exchange reactions have also been furtherinvestigated. In the mercury-sensitised exchange with deuteriumand phosphine,16 the quantum efficiency of exchange exceeds unityat higher temperatures (>500"), and indicates the chain mechanism1 4 See also N.R. Trenner, K. Morikasva, and H. S. Taylor, J . Chem. Physics,15 H. W. Melville, J. L. Bolland, and H. L. Roxburgh, Proc. Roy. Soc.,1 6 H. W. Melville and J. L. Bolland, ibid., p. 384.17 See H. W. Melville and S. C. Gray, Trans. Faraday SOC., 1936, 32, 271;18 N. L. Simmons and A. 0. Beckman, J. Amer. Chem. SOC., 1936,58, 454.10 H. J. Emel6us and K. Stewart, Trans. Faraday Soc., 1936, 32, 1577.1937, 5, 208.1937, A , 160, 406.G. Rathenau, Physsica, 1937, 4, 50370 GENERAL AND PHYSICAL CHEMISTRY.(400-500") D + PH, = PH,D + H, H + D, = HD + D. ThePD,-H, exchange proceeds similarly , both reactions being charac-terised by a high temperature coefficient.At room temperatures,the temperature coefficient is unity, and the exchange proceeds bythe recombination of the radicals produced by the mercurysensitisation : PH2 -+ D = PH,D, E€ + 1), = HD + D. Mercury-sensitised exchange reactions of hydrogen and deuterium have beeninvestigated for ammonia, methane, and water, hydrogen- ordeuterium-atom concentrations being determined by the ortho-pareconversion of the element.20 For methane 2*i 21 and ammonia,20the exchange above 300" takes place by the chain D + XH =DX + H, H + D, = HD + D : the quantum efficiency is greaterthan unity. The possible reactions in the mercury-sensitiseddecomposition of methane, deizteromethanes, and the hydrogenisotopes have been extensively discussed ; 21 at high temperatures(>300") exchange owurs by the chain process as above, but atlower temperatures exchange is mainly by recombination of themethyl radical and atomic deuterium.Deuterisation of alkylradicals in presence of atomic deuterium is a more rapid processthan that of the saturated hydrocarbons.22 Approximate energiesof activation of these reactions are discussed: the quenchingefficiency of the excited mercury atom by methane to give methyland a hydrogen atom is given a temperature coefficient correspondingto an activation energy of approximately 4.5 kg.-cals. Withwater,lfr20 reaction is slow and the exact mechanism is not socertain. The exchange efficiencies in this case are much largerthan those of the mercury-sensitised decomposition itself , and muchlarger than values calculated on the assumption that in thedeuterium-water mixtures, deuterium acts simply as quenching theexcited mercury atoms; l1 exchange does not therefore take placeappreciably by the dissociation and subsequent re-formation of thewater molecule.Photo-oxidation.-Photo-oxidations may be classified into threemain groups.23 I n the first, oxidation is due to excited oxygenmolecules or atoms, e.g., the oxidation of carbon monoxide,2*nitrogen and carbon monoxide.25 Monosilane itself is not decom-posed by light from an aluminium spark or mercury arc, but may20 A.Farkas and H. W. Melville, Proc. Roy. SOC., 1937, A , 157, 625.z1 K.Morikawa, W. S. Benedict, and H. S. Taylor, J . Chem. Physics, 1937,22 N. R. Trenner, K. Morikawa, and 33. S. Taylor, ibid., p. 203.2s H. J. Schumacher, 2. Elelitrochem., 1936, 42, 522.24 B. Popov, Acta Physicochim. U.R.S.S., 1935, 3, 223; M. SisIrin, V.26 V. Kondrateev, Acta Phyaicochim. U.R.S.S., 1935, 3, 247.5, 212.Kondrateev, and T. Suschkevitsch, J . Phys. Chern. Russia, 1936, 8, 281IuI[TCJHlX : PHOTOCHEMISTRY. 71be made to explode with oxygen under these conditions a6temperatures below the normal range for thermal igniti0n.1~Mention may also be made of the effect of altered temperature onthe photo-formation of ozone by an aluminium spark;26 inagreement with theory, [O,] = K[O2l2 at equilibrium, andcalculation gives the energy of activation of the reaction betweenoxygen atoms and ozone as 6160 & 100 cals.per mole.The second group is that where the quantum is absorbed by themolecule which is oxidised, e.g., hydrogen iodide,27 hydrogen iodideand deuterium iodide,28 involving the collision-stabilised formationof HO, and DO,, where the zero-point energies of these complexesare concluded to be of the same magnitude as those of the halidemolecules. With phosphine and trideuteroph~sphine,~~ the corre-sponding reactions involved in the chain propagation, branching,and termination at t.he walls proceed with similar velocities in thetwo cases, the positions of both lower and upper explosion limitscoincide, and the chain lengths above and below these limits arerespectively identical.In the case of methylem iodide, CH, andthe peroxidic CH202 and (CH,O), are regarded as primary products,final products being hydrogen, carbon monoxide, formaldehyde,formic acid, and ethylene glycol.30 The oxidation of gaseousformaldehyde takes place by way of formic acid, no peroxide orperacid being found; 31 with acetaldehyds the occurrence ofdiacetyl peroxide is confirmed. Both oxidations are short-chainprocesses ; in aldehydes generally peroxides may be derived fromthe radical RC0.32 Peroxidic formation has been studied 33 formethyl and other alcohols, glycerol, acetone, fructose, formic andother acids, glycol, and paracetaldehyde : in each case the activatedoxidisable molecule is supposed to form with molecular oxygen aperoxidic compound, which in turn reacts with the original activatedmolecule. In the case of r~brene,,~ the initially activated moleculeis unable to form B, stable oxygenated compound by coltision withoxygen, unless the complex is stabilised by subsequent collision orthe rubrene molecule partially deactivated to a second state where28 A.Eucken and F. Patat, 2. phyaikal. Chem., 1936, B, 33, 459.27 V. Kondrateev, E. Kondrateeve, and A. Lauris, J. Phys. Chem. Russia,28 G. A. Cook and J. R. Bates, J. Amer. Chern. Soc., 1935, 57, 1775.29 H. W. Melville, J. L. Bolland, and H. L. Roxburgh, Proc. Roy. Soc.,30 R. A. Gregory a,nd D. W. G. Style, Trans. Faradccy SOC., 1936,52, 724.31 J. E. Carruthers and R. G. W. Norrish, J., 1936, 1036.32 H.L. J. BackstrBm, 2. physikal. Chem., 1934, B, 25, 99.83 R. Cantioni, Ber., 1936, 69, 1101, 1386, 1796, 2282, 2286; Helv. Chhn.34 W. Koblitz and H. J. Schumacher, 8. ph@at. Chem., 1937, B, 35, 11.1934, 5, 1411.1937, A , 160, 417.Acta, 1936, 19, 1153; 2. wiss. Phot., 1937, 38, 90, 116, 11972 GENERAL AND PHYSICAL CHEMISTRY.8 stable complex results directly. In this group also may be citedthe reaction of mercury vapour with oxygen by light of theresonance line 2537 A., ozone formation being recorded ; 35 selectiveoxidation of mercury isotopes is claimed36 by irradiation of amercury-oxygen mixture with either the I11 or the IV componentof the 2537 line.Halogen-sensitised reactions play an important part in the thirdgroup of photo-sensitised oxidations.Radicals are formed by theaction of the halogen atom, which again form peroxide chains. Onthis basis may be considered the production of carbon dioxidefrom the monoxide, chlorine, and oxygen (radical COCl),37 and ofcarbonyl chloride from carbon tetrachloride, chlorine, and oxygen(radical CCl,) .38 Intermittent illumination of carbon monoxide-chlorine mixtures with a small oxygen content 39 indicates theformation of an active oxygen-containing intermediate, of longerlife than C1 and COCl, and able t o initiate new chains in the dark.The formation of carbonyl chloride from chloroform, chlorine, andoxygen may be almost completely suppressed by added methyl orethyl alcohol or ammonia, due to the removal of chlorine atoms,an induction period being observed .40 The chlorine-sensitisedoxidations of methane, methyl chloride, ethylene dichloride, andformaldehyde are discussed on similar lines to the above ; z3quantum efficiencies (referred to carbon monoxide formation) risefrom about 80 to 800 in the order CH, < CH3Cl < CH,Cl,, aperoxide of short life being formed from the CHCl, radical in eachcase.41 The bromine-sensitised oxidation of mandelic acid has beenfurther investigated under various ~onditions.~~ Mention mayalso be made of the mercury-sensitised oxidation of mon~silane,~~the oxidation of chloroacetic acid by potassium permanganatewith uranyl salts as photo-sen~itisers,~~ and that of various organicsubstances by hydrogen peroxide with inorganic sols as catalysts.4435 I.M. Frank, J . Phys. Chem. RussZ'a, 1934, 5, 1013; Acta Physicochim.38 K. Zuber, HeZv. Physica Acta, 1935, 8, 481 ; Chem. Zentr., 1936, 1, 494.37 M. Bodenstein, W. Brenschede, and H. J. Schumacher, Z . physikal.H. J. Schumacher and K. Wolff, ibid., 1934, B, 25, 131; A. T.as M. Bodenstein, W. Brenschede, and H. J. Schumacher, 2. physikaz.40 H. J. Schumacher and D. Sundhoff, ibid., 1936, B, 34, 300.41 W. Brenschede and H. J. Schumacher, ibid., 1936, 177, 245.42 J. C. Ghosh and S . K. Bhattacharyya, ibid., 1936, B, 31, 420; J. C.Ghosh and B. B. Roy, ibid., 1936, B, 32, 158.45 Idem, J . Indian Chem. Soc., 1936, 13, 1.44 T. Banerjee, ibid., 1937, 14, 69.U.R.S.S., 1934, 1, 833.Chm., 1935, B, 28, 81.Chapman, J . Amer.Chem. SOC., 1935, 57, 419.Chem., 1937, B, 35, 382RITCHIE : PHOTOCHEMISTRY. 73A photo-stationary state in an oxidation-reduction system isdescribed,45 and the photochemical reductions of ferric salts byorganic acids 46 and of ceric ions by water.47General.-Photolysis of tetramethyl- and tetraphenyLlead48gives metallic lead and other products which are mainly thehydrocarbons, ethane and triphenyl respectively. Use of aradioactive indicator method gave y values of the order unity forthe tetramethyl compound as vapour, though short chains,retarded by oxygen, are indicated. In solution, the value is of theorder 0.3 for both compounds, these lower values in solution beingregarded as due to deactivation or secondary recombinahion. Theformation of free radicals has been demonstrated during thephotolysis of tetramethyl-lead vapour ; the decomposition ofdimethylmercury and tetraethyl-lead, as well as of acetone, hasbeen studied similarly,49 the maximum production of radicals fromthe first being at 2200a., in the region of diffuse predissociationbands.With light of wave-length 2 5 3 7 ~ . , which is near thelong-wave limit of the overlapping continuum, the quantumefficiency of decomposition of dimethylmercury is unity at roomtemperature, but at higher temperatures rises to greater values bya chain process which is suppressed by the addition of nitricoxide.50 The primary act on light absorption is regarded as theproduction of CH, and Hg*CH,; 51 results suggest that one methylradical reacts with each molecule of nitric oxide.The absorptionspectra of polyatomic molecules containing methyl and ethylradicals have been discussed 52 and the emission spectra of freeradicals produced by photodissociation of polyatomic molecules inthe Schumann ultra-violet studied.53The quantum yield of photo-decomposition of azomethaneapproaches unity as its upper limit for initial decomposition a t lowpressures.54 Change of wave-length from 2080 to 2540 A. approx-imately halves the efficiency, indicating a decreased probability of46 G. Holst, Nature, 1937, 139, 285; 2. physikal. Chem., 1937, 179, 172.** P. La1 and P. B. Ganguly, 2. anorg. Chem., 1936, 229, 16.47 J. Weiss and D. Porret, Nature, 1937, 139, 1019.48 P. A. Leighton and R. A.Mortensen, J . Amer. Chem. SOC., 1936. 58,4% N. Prileshajeva and A. Terenin, J . Phys. Ghem. Rwsia, 1936, 8, 111.50 H. W. Thompson and J. W. Linnett, Trans. Paraday Soc., 1937, 33,51 See N. Prilesliajeva and A. Terenin, ibid., 1935, 31, 1483.62 K. W. Thompson and J. W. Linnett, PTOC. Roy. SOC., 1936, A , 156,58 H. Neujmin and A. Terenin, Acta Physicochim. U.R.S.S., 1936, 5, 465.54 G. S. Forbes, L. J. Heidt, and D. V. Sickman, J . Amer. Chem. Soc.,448.874.108; 160, 539.1935, 57, 193574 GENERAL AND PHYSICAL CEEMISTRY.dissociation before deactivation. With increasing pressure: y fallsoff rapidly, apparently due to increased collisional deactivation ;argon, nitrogen, and methane also reduce the rate of decomposition.65Detailed analysis of the products of photolysis shows that, ingeneral, the amount of nitrogen produced exceeds the amount ofhydrocarbon gas, and indicates again that the molecule does notdecompose exclusively by rupture but that it may also decomposeby a rearrangement to form stable molecules.560 ther reactions involving nitrogen compounds are : the primaryprocess in the light arrangement of o-nitrobenzaldehyde,57 thedecomposition of nitromethane and nitroethane, 58 of some aliphaticnitroso-compounds,59 of nitrates, nitrites, and nitro-compounds,Mreduction of nitrates,6i deamination of amino-acids.62In relation to the changes in chemical activity with moleoularorientation in a monolayer, the photo-decomposition of stearanilidecan be suppressed by compressing the layer so that the benzenenuclei no longer absorb in the ultra-violet region.63 Illuminationof carbon suboxide by light from a, mercury-vapour lamp leadschiefly to a polymerisation with little decomposition; 64 fromsulphur dioxide, sulphur and the trioxide may be obtainedY65 andwith nitrous oxide a t least part of the primary dissociation(1850-2300 A.) is into nitrogen molecules and oxygen atoms,66although it is not possible to state definitely that none of theprimary dissociation is into nitric oxide molecules and nitrogenatoms. Mention may be made also of reactions of thiol compoundsin solution,67 photo-dissociation of gallium halides 68 and of leadhalides,Gg action of light on zinc ~ulphide,~O lead oxides,'l and6s G.Goldfhger, Compt.rend., 1936, 282, 1502.66 M. Burton, T. W. Davis, and H. A. Taylor, J. Amer. Chem. SOC., 1937,5 7 L. Kuchler and F. Patat, 2. Elektrochem., 1936, 42, 529.68 E. Hirschlaff and R. G. W. Norrish, J., 1936, 1580.59 ,D. L1. Hammick and M. W. Lister, J., 1937, 489.60 C. H. Purkis and H. W. Thompson, Trans. Faraday SOC., 1936, 32, 1466.63 C. Weizmann, E. Bergmann, and Y. Hirshberg, J . Amer. Chem. SOC.,63 E. K. Rideal and J. S. Mitchell, Proc. Roy. SOC., 1937, A, 159, 206; see64 H. W. Thompson and N. Healey, Proc. Roy. Soc., 1936, A, 157, 331.65 M. Konstantinova-Schlesinger, J . Phys. Chem. Rzcssia, 1935, 6, 601.6 6 W. A. Noyes, jr., J . Chem. Physics, 1937, 5, 807.67 J. Weiss and H. FishgoId, Nature, 1936, 187, 71.68 A. Petrova, Acta Physicochim.U.R.S.S., 1936, 4, 559.6B B. Popov and H. Neujmin, J . Phys. Chem. Russia, 1934,5, 863.70 H. Platz and P. W. Schenk, Angew. Chem., 1936, 49, 822.7l J. Eoffmann, 2. anorg. Chem., 1936,228, 160.59, 1038, 1989.R. Cultrera, Gazxetta, 1936, 88, 440.1936, 58, 1675.J. S. Mitchell, J . Chem. Physics, 1936, 4, 725BUTLER : INTERMOLECULAR FORCES. 75white pigments.72 Measurements of photochemical yield incrystals may be made by optical and electrical means.73J. Franck and K. F. Herzfeld 74 give a theory of photosyntheticproduction of oxygen covering quantitatively many observationsdescribed in the literature, the scheme involving four photochemicalsteps and two dark reactions. Plant acids are regarded assynthesised by the same mechanism; these can be photo-oxidisedin a reaction sensitised by chlorophyll.Descriptions have been given of improved sources of continuousultra-violet radiation,75* 76 and of the 2537,77* 78 1285, and1469 A .' ~ resonance radiations. M. R.5. INTERMOLECULAR FORCES AND THE PROPERTIES OF LIQLTDS.The remarkable revival of interest in the nature and propertiesof the liquid state which has taken place in the last few years isshown by a great increase in the number of papers on this subject,which has also been reviewed in two general discussions.In September 1936 the Paraday Society held a general discussionon the Structure and Intermolecular Forces in Liquids andSolutions,l and the centenary of the birth of J. D. van der Waalswas marked by a commemoration in Amsterdam in November1937, a t which papers on the properties of dense gases and liquidswere read by many of the leading workers.2 Reference should bemade to these symposia for points of view not mentiomd here.The question of structure in liquids as revealed by X-ray diffractionis discussed elsewhere in these report^.^Intermoleculur Forces.-The general state of our knowledge of theforces between molecules, which are responsible for deviations fromthe perfect-gas law, cohesion in liquids and solids, capillarity, etc.,has been summed up by F.London in an article which provides asuitable basis for this Report. The first suggestion of the origin ofthese forces was that of W. H. KeesomY5 who presumed the genera172 C. F. Goodeve, Trans. Paraday Soc., 1937, 33, 340,74 J .Chem. Physics, 1937, 5, 237.7 5 R. H. Munch, J . Amer. Chem. SOC., 1935, 57, 1863.713 G. Jacobi, Physikal. Z., 1936, 37, 808.77 G. Kornfeld and F. Muller-Skjold, 2. physikal. Chem., 1936, B, 31, 223.'@ W. Groth, 2. Elektrochem., 1936, 42, 533.1 Trans. Paraday SOC., 1937, 33, 1.3 P. 169.1922, 23, 225.R. W. Pohl, Proc. Physical SOC., 1937, 49, Extra part, 3.H. VV. Melville, Trans. Paraday SOC., 1936, 32, 1525.Physica, 1937, 4, 915.Trans. Faraday SOC., 1937, 33, 8.Leiden Camm. Suppl., 1912, 24, 25, 26; Ph.ysiJca!. Z., 1921, 22, 12976 GENERAL AND PHYSICAL CHEMISTRY.existence of permanent dipoles in molecules. These will give risesometimes to an attractive and sometimes to a repulsive forcebetween two molecules, according to their mutual orientation, butaveraging over all possible orientations, Keesom obtained a generalorientational attraction, the mean energy of which is given bywhere p is the dipole moment and r the distance between the dipoles.that dipolar molecules would also setup induced moments in molecules in their vicinity, the magnitudeof which is proportional to their polarisability, 01.This induction$fleet also gives rise to an attractive force, the energy of which isP. Debye pointed outNeither of these effects can account for the attraction betweencompletely non-polar atoms such as the inert gases, and they alsolack the additivity which is required to account for the generalcohesion.In 1930 London showed 7 that, according to the description ofsymmetrical molecules given by wave mechanics, although averagedover a period of time the charge distribution is perfectly sym-metrical, yet a t any instant there will be various configurations ofnuclei and electrons showing instantaneous dipole moments.Thesequickly varying dipoles will act on the polarisability of othermolecules and produce in them induced dipoles, giving rise to anattractive force, having to the first a.pproximation the averageenergyU = - 2 . hvoa2/r6where vo is a frequency characteristic of the molecule.This dispersion eflect * has the characteristic of additivity, i.e., thetotal interaction between a number of molecules is the sum of theinteractions of all the pairs of molecules, and therefore is of the typerequired to account for the general cohesive or van der Waals forcesbetween molecules.The magnitude of the three effects for somesimple molecules is shown in the following table, from which it canbe seen that the inductive effect is always practically negligibleand that the orientation effect is small except with highly polarmolecules.-6 Physsikal. Z., 1920, 21, 178; 1921, 22, 302.7 2. Physik, 1930, 60, 491; 63, 245; Z. physikal. Chem., 1930, B, 11,* So called because it depends on the characteristic frequencies of the223.moleculesBUTLER : INTERMOLECULAR FORCES.Relative magnitudes of molecular interactions.p . 1Ol8. a . loz4. hv,,.co ......... 0.12 1.99 14.3HI ............ 0.38 5.4 12HBr ......... 0.78 3.58 13.3HCl .........1.03 2-63 13.7NH, ......... 1.5 2.21 16H,O ......... 1.84 1-48 18He ............ - 0.20 24.5A ............ - 1.63 15.4Xe ............ - 4.00 11.5Orient -ation Inductioneffect, effect,Q . p4/kT.* 2p2a.*0.0034 0-0570.35 1.686.2 4.0518.6 5.484 10190 10 - -- -- -77Disper-sioneffect,%a2hvo. *67.53821761059347522171.2* In erg.-cm.s xMore accurate calculations of the dispersion effect have been madeby various authors,8 and the effect of including terms of a higherorder, such as that due to the interaction of a dipole in one moleculewith a quadrupole in another, or between two quadrupoles has beenin~estigated.~ The former gives rise to an r8 term which mayamount in halide lattices to 20y, of the total interaction; but thelatter, an T-10 term, is apparently negligible.At small distances it is obvious that repulsive forces must comeinto play to prevent molecules collapsing into each other.Theseare due to the facts that : (1) at short distances the electron cloudsof two molecules no longer screen the nuclear charges completelyand the latter repel each other by their electrostatic forces; (2) theelectron clouds themselves interact and produce a repulsion. Therepulsive potential can be represented empirically by a power law ,loe.g., hr-12, but calculations by wave mechanics l1 appear to favouran exponential law R = be-rlp, where b and p are constants whichmust usually be determined empirically.12London hasshown13 that the heats of sublimation of the inert gases and anumber of simple diatomic molecules are in good agreement withthe values calculated by his dispersion formula, and an approximatecalculation of the van der Waals constant a also gave reasonableagreement.A. Muller found l4 that the formula could be success-J. C. Slater and J. G. Kirkwood, PhysicaZ Rev., 1931, 37, 686; J. G.Kirkwood, Physikal. Z., 1932, 33, 57; H. Hellmann, Acta Physicochirn.U.R.S.S., 1935, 2, 273; R. A. Buckingham, Proc. Roy. Soc., 1937, A, 160, 94,113.Various tests of these equations have been made.H. llargenau, Physical Rev., 1931, 38, 747 ; R. A. Buckingham, Zoc. cit.lo J. E. Lennard-Jones, Proc. Physical SOC., 1931, 43, 461.l1 M. Born and J. E. Mayer, 2. Physilc, 1932, 75, 1 ; F.London, Zoc. cit.l2 See J . E. Mayer, J . Cham. Physics, 1933, 1, 270, 329; W. E. Bleick andl a LOC. cit., ref. (1).J . E. Mayer, ibid., 1934, 2, 252.l4 Proc. Roy. SOC., 1936, A, 154, 62478 GENZRAL AND PHYSICAL CHEMISTRY.fully applied to the sublimation energies of the long-chain paraffins.Helium and hydrogen are, however, exceptions, because the zero-point energy in condensed phases is nearly as large as the van derWaals energy. Helium is of great interest in that the solid phaseis not stable even at absolute zero, except at pressures exceeding25 atrn0s~heres.I~ There is a transition point a t 2.19" K., at whichthe heat capacity curve has a discontinuity and the viscositydecreases by one-tenth. Calculations by F. London,lG taking thezero-point motion into account, indicate that below this temperaturethe stable liquid has a tetrahedral structure.A more direct estimate of the law of force between two moleculeshas been made by H.Kuhn and P. London,17 from the effect of thepressure of an inert gas in broadening a spectral line. In thecase of the mercury resonance line, argon produces a broadeningwhich is directly proportional to the pressure of argon and ofmercury and is therefore produced by single transits. The resultsindicate that the c r 6 law is valid between 3.4 and 4-8 xEquations of State and Partition Functions of Liquids.-Theclassical approach to the liquid state, from imperfect gases, culmin-ated in various equations of state, of which van der Waals's equationis the best known. The most useful empirical equation of thiskind is, however, that suggested by Kamerlingh Onnes and Keesom,vix.cm.PV = RT(1 + B/V + C/V2 .. .)where B, C , etc., are the first, second, etc., virial coefficients.Hitherto it has only been possible to derive the first virial coefficientfrom a given law of molecular interaction with any accuracy, andthis is only sufficient for very moderate compressions. Animportant contribution to the solution of the general problem ofdetermining all the coefficients for any law of force has been madeby J. E. Mayer,Is and the method employed has been discussed andextended by M.The introduction of the intermolecular forces into equations ofequilibrium is most conveniently made by means of the partitionfunction of R.H. Fowler and C. G. Darwin. For a perfect gas thisfunction isfg = ( 2 n d ~ T ) ~ ' ~ V b ( T ) / N h ~ . . . . (1)15 W. H. Keesom, Physica, 1934, 1, 128, 161; F. Simon, Nature, 1934, 138,16 Proc. Roy. SOC., 1936, A , 153, 576.17 H. Kuhn and F. London, Phil. Mag., 1934, 18, 983, 987; H. Kuhn,Proc. Roy. Soc., 1937, A , 158, 212, 230.18 J . Chem. Phgsics, 1937, 5, 67, 75.460, 529.Physica, 1937, 4, 1034BUTLER : INTERMOLECULAR FORCES. 79where b(T) is that part of the whole partition function which dependson the vibrational and rotational energy of the molecule.* Mayerintroduces into this function coefficients expressing the interactionin clusters of 2, 3, 4, etc., molecules. The calculation of theseCoefficients from the law of force is, however, very laborious, andit is unlikely to become an effective method for the study of gasesand liquids, at least for some t h e .A less fundamental method which avoids many of these difficultiesis to introduce the average potential field of the neighbours of agiven molecule.Thus, on the assumption that the vibrationaland rotational states are the same in the liquid as the gas, Eyring 2oexpresses the partition function of a liquid aswhere AE is the energy of vaporisation t and VIN is replaced by Vf,the free volume of the liquid. The essential part of the problem isthen the determination of the free volume of the liquid. For simplecubical packing of spherical molecules, this is S( V11'3 - whereV , is the molecular volume of the liquid and d the incompressiblediameter of the molecules, or more generally Vf = b3( V11'3 - d)3.These simple expressions are in reasonable agreement with many ofthe properties of simple liquids.Mercury is, however, an exceptionand this case has been examined in detail.21When the partition function of a liquid is completely known, theheat capacity can be calculated. If Eyring's function (2) is valid,and it is assumed that the free volume of the liquid does not varywith the temperature when the total volume is constant and thatthe rotational and vibrational terms are the same in the liquid as inthe gas, it is found that the heat capacity of the liquid at constantvolume (0,) is the same as that of the gas. This is true within& 2 cals.for many polyatomic liquids,22 except associated liquids.Thus, although these assumptions may be adequate for a broadtreatment, they fail to account for the finer features. For simpleliquids Cv follows a very uniform course between the melting pointand the critioal point.23 With monatomic liquids, such as neon,2o R. F. Newton end H. Eyrhg, Trans. Paraday SOC., 1937, 33, 73;21 J. F. Kencaid and H. Eyring, J. Chem. Physics, 1937, 5, 687 ; see also22 LOC. cit., ref. (20).23 E. Bauer, M. Magat, and M. Surdin, J. Phys. Radium, 1936, 7, 441.* The Helmholtz maximum work function is A = KT logf per molecule, andt More correctly, AE a t 0' K. should be used.H. Eyring and J. Hirschfelder, J. Physical Chem., 1937, 41, 249.ref.(40).the pressure is obtained as (dA/du)F = -p80 GENERAL AND PHYSICAL CHEMISTRY.argon, mercury, and czesium, C, has near the melting point thevalue 3R = 6 cals., which is characteristic of the solid state. Asthe temperature rises, the value falls and approaches 2R = 4 cals.at the critical point. L. Brillouin 24 has suggested that the twodegrees of freedom of transverse vibrations in the solid become a,rotational motion in the liquid, which makes a contribution R(instead of 2R) to C,. The higher values a t lower temperaturesare ascribed to some sort of quasi-crystallinity. E. Bartholomi5 andA. EuckenZ5 are able to account for the large values of C,, byassuming a particular shape for the potential field of a molecule inthe liquid, but to explain the full decrease they are obliged toassume an association between molecules which decreases withrising temperature.J. D. Bernal 26 ascribes the high values of C,near the melting point to the energy required to bring about thenecessary changes in the equilibrium configuration of the liquid.The theory has not, however, been developed far enough fornumerical calculations. Liquid hydrogen and helium have muchlower Cv’s, rising with temperature, because a t these low tempera-tures the classical equipartition has not beenJ. E. Lennard-Jones and A. F. Devonshire 28 start from the samebasis as Eyring, but they go further and evaluate the potential fieldproduced by the neighbours of a given molecule from the knowninteraction constants. A molecule in a dense gas is regarded asconfined for most of its time in a cell or box, the walls of which arecomposed of its nearest neighbours. When this box is large, thepotential inside is fairly uniform except near the boundary, wherethere is a region of low potential.As the density of the gas increasesand the size of the box diminishes, the boundary fields begin tooverlap and the potential in the middle of the box falls, reachinga minimum and rising again a t high compressions when therepulsive forces become important. By using constants in theequation +(r) = Am - Br-m, derived from the deviations fromthe perfect-gas law at low densities, the potential of a molecule inthe box can thus be expressed as a function of the volume of theform x(v) = uIvv - p/vp.When this is inserted in place of AEin (2), a partition function is obtained which is capable ofreproducing the critical phenomenon? and the critical temperaturesof hydrogen, neon, nitrogen, and argon calculated from it are ingood agreement with the observed values.24 J . Phys. Radium, 1936, 7, 153; Trans. Paraday Soc., 1937, 33, 54.=5 Ibid., p. 45.27 E. Bartholorn6 and A. Eucken, 2. Elektrochem., 1936, 42, 547.28 Proc. Roy. SOC., 1937, A , 163, 63; also J. E. Lennard-Jones, Physica,26 Ibid., p. 27.1937, 4, 941BUTLER : INTERMOLECULAR FORCES. 81The entropy of vaporisation of a liquid according to (1) and (2)is approximatelyIt has long been known that AX = L/T is nearly constant for normalliquids at their boiling point (Trouton's rule).J. H. Hildebrand 29found that a better constancy was obtained when liquids are com-pared at temperatures at which they give rise to equal vapourconcentrations. Under this condition Vs is constant and (3) thenimplies that the free space in the liquid is also constant. Eyring 3Ohas pointed out that, since the energy required to make a hole ina liquid is the same as that required to remove a molecule from theliquid, the concentration of holes in the liquid must be equal to theconcentration of molecules in the vapour. At equal vapourconcentrations liquids will therefore have equal " concentrationsof holes "; in conjunction with (3) this gives a simple explanationof Trouton's rule.* When the molecular rotations are not fullydeveloped in the liquid, as in associated liquids in which directedbonds exist, the entropy of vaporisation will be greater than (3),as is found to be the case.It is evident that in the theory of liquids, so far as it has beendeveloped, the free space is of the first importance and is the maindistinction between a solid and a liquid.The existence of free spacein the liquid permits the diffusion of molecules and the movementsnecessary for viscous flow. The concept of holes of molecular sizealone is, however, much too simple. No doubt holes of all sizesexist, and the distribution of such sizes has been considered byW. Altar.31The Viscosity of Liquids.-An exponential relation between theviscosity and temperature was apparently first suggested by J.deGuzmBn Carran~io,~~ wiz., r ) = AeBIRT. Although many variantsof this equation have been suggested,33 e.g., Andrade's equation,34q = A/vl'3 . ec'wT, which frequently gives somewhat better agree-ment with the measurements, the simplicity of the original equationAB=Rlog(V,/Vj) . . . . ' (3)29 J . Amer. Chem. SOC., 1915, 37, 970.30 J . Chem. Phyaiw, 1936, 4, 283; see also 0. I(. Rice, ibid., 1937,5, 353.31 Ibid., p. 577.33 For bibliography, see A. G. Ward, Trans. Paraday SOC., 1937, 33, 88.34 Phil. Mag., 1934, 7, 17, 698.* If V , is the volume of 1 mol. of vapour, the concentration of holes in theliquid is equal to the concentration of molecules in the vapour, i.e., NIT',,where N = Avogadro's number. The free space in 1 mol.of liquid is therefore7, = (N/ V,)v,,,V,,,, where v, is the volume of the molecule and V, the rnoIe-cular volume of the liquid. Thus V,/V, is approximately V,Z/V,*. Thevalue of this for benzene a t its b. p. is 8.9 X lo4, which gives Ah'= 22.8 cds.per degree, in quite good agreement with the Trouton's law constant.32 Anal. Pis. Quim., 1913, 11, 35382 GENERAL AND PHYs1CA.L CREMISTRY.makes it more suitable as a starting point of theories of viscosity.Recent discussions have been mainly concerned with the natureof B, which can be regarded as the activation energy of the viscosityprocess. Guzm&n showed that in some cases there is a roughcorrespondence between B and the latent heat of fusion. A. G .Ward,35 discussing viscosity from the point of view of 9.D. Bernal'stheory 36 of the configuration of liquids, suggests that such a relationis to be expected when the liquid is merely a disordered version ofthe solid, but when the configuration in the liquid is essentiallydifferent, B and L will not be connected.H. Eyring has discussed the theory of viscosity in a series ofpapers.57 The theory has developed progressively and all theideas brought forward are not easily summarised, but broadly twomechanisms are considered : (1) A " unimolecular " mechanism inwhich each molecule moves independently ; this movement can onlytake place if a hole of suitable size is wailable and the activationenergy is the energy required to make the hole. (2) A " bimole-cular " process in which two molecules in adjacent layers, in relativemotion, rotate round each other through 90".The free spacenecessary for a movement of this kind is considerably smaller thanfor the first. Since the energy required to make a hole of molecularsize is equal to the energy of vaporisation, the activation energyof the viscosity process should be related to the energy ofvaporisation.Some very interesting facts emerge from the comparison of thesequantities. Liquids fall into well-defined groups. For carbontetrachloride, benzene, cydohexane, methane, argon, nitrogen, andcarbon monoxide, which have presumably practically sphericallysymmetrical fields of force, the ratio ( AE)vap./(AE)dsc. is 2-3-24, butfor unsyrnmetrical molecules such as pentane, chloroform, ethyliodide, carbon disulphide, and diethyl ether the ratio is 3-5-4.0.These ratios fit in with the bimolecular hypothesis. The metalshave much larger ratios, varying from 8 to 30, and the interestingsuggestion is made that while the unit for vaporisation is theatom, the unit of flow is the much smaller metal ion.If thedifference of size of the atom and ion is allowed for by using theratio (AEvap./AEvisc,) (riOn l ~ a t o m ) ~ , normal values of 2 4 are obtained.The normal hydrocarbons all have ratios of ca. 4, which suggeststhat the chains are curled up in the liquid state,With water, the ratio decreases as the temperature rises, i.e.,AEd, decreases more rapidly than AEv,.. Eyring suggests that36 LOC. cit., ref. (33).s7 J. Chem. Physics, 1936, 4, 283; R.H. Ewell and H. Erying, &bid., 1937,36 Trans. Farday Soc., 1937, 33, 27.5, 726; R. H. Ewell, ibid., p. 571BUTLER : INTERMOLECULAR FORCES. 83when there are directed forces, such as hydrogen bonds, the activ-ation energy of the viscosity process will consist of the usual fraction(ca. l/4) of that part of the energy of vaporisation which is due toundirected attractive forces, but all the energy of directed bondswhich must be broken in the process of flow. The rapid decrease ofAE~s, is then due to the decrease in the number of hydrogenbonds with rising temperature. The high visoosity of glycerol,sugar solutions, etc., is due to the preponderence of this structuralactivation energy.MeEting.-One of the most diEcult points to understand aboutliquids is the existence of a sharp melting point.If a liquid differsfrom a solid merely in a greater degree of thermal agitation anddisorder, it is not clear why the process of melting is so sharp.The nature of melting has been considered by a number of authors.J. Frerike13* has argued that the crystalline and the liquid state,like the gaseous and the liquid state, are separated by a continuousseries of states of increasing disorder, the intermediate stages,however, being unstable. This idea has also been supported byF. Simon,39 who suggests that, since the properties of the liquidand the crystalline state of a substance approach each other at highpressures, in the limit a critical point may be reached. J. D. Bernal,however, regards the liquid state as characterised by a fundamentalirregularity, so that the difference between solid and liquid is one ofkind rather than degree.At the melting point the free energy of the solid and liquid arethe same, and the greater energy of the liquid is therefore com-pensated by a greater entropy.One factor in this is the communaluse of the free volume of the liquid by all its rn0lecules.~0 In thesolid each molecule can be regarded as confined to a region in theneighbourhood of the lattice point, but in the liquid it is reasonableto suppose that the free volume is shared by all the molecules.This gives rise to a term R in the entropy of the liquid, and thus,when the rotational and vibrational terms are the same in the solidas in the liquid, the entropy of fusion may be expected to be 2 units.This is approximately the case with metals, but with most othersubstances rotation does not occur in the solid, and the entropy offusion is greater on account of changes of the rotations andvibrations.With many organic compounds, the entropy of fusionis about l2,41 and when a much smaller value than this is found,8 8 Trans. Paraday Sac., 1937, 33, 58; Acta Physicochim. U.R.S.S., 1935,39 Trans. Paraday Soc., 1937, 33, 65.40 J. Hirschfelder, D. Stevenson, and H. Eyring, J. Chem. Physics, 1937,4 1 P. WaIden, 2. Elektrachem., 1908, 14, 713.3, 633, 913; Bull. Awd. Sci. U.R.S.S., 1936, 371.6, 89684 GENERAL AND PHYSICAL CHEMISTRY.it is indicated that rotational degrees of freedom already exist inthe solid below the melting point.The appearance of such rotations in solids has now been observedin a considerable number of cases.42 Usually there is not a sharptransition, marked by a latent heat, when the molecules begin torotate, but this occurs gradually over a range of temperature inwhich the heat capacity rises to a maximum and falls again andanomalous dielectric constants are often observed.The theory ofthis transition has been worked out by R. H. Fowler.43Miscelbneous Relations.-E. Bauer, M. Magat, and &I. Surdin 44have found that many of the properties of a large variety of liquidsfall OE identical curves when plotted against a reduced temperature,8 = (7’ - Tf)/(Tc - Tf). This function varies from 0 a t thefreezing point (T,) to 1 at the critical point (TJ, but except at lowtemperatures it is not very sensitive to Tf, and M.Surdin45 has sinceshown that in many cases the van der Waals reduced temperatureT/Tc is equally effective. The effect of temperature on normalliquids at least is thus remarkably uniform, and it is clear that asingle model of the liquid state may be expected to apply to allnormal liquids.R. H. Fowler 46 has considered the surface tension of liquids in theneighbourhood of the critical point, and derived the relationQ = const. x (1 - T/Tc)2 for this region. Since van der Waals’sor Dieterici’s equations of state give (1 - Tc/T) K (dii,. - dVapJ2,it follows that o = K(dli,, - dvaPj4. This is Macleod’s equation?the basis of Sugden’s parachor, P = MK1/4, which has hithertobeen quite empirical.The derivation cannot, however, be extendedto lower temperatures at which Macleod’s relation still holds. Theabsolute calculation of the parachor by similar methods gavepromising results.Dielectric Properties of Polar Liquids.Dielectric Constants.-When an electric field is applied to a vapourof dipolar molecules a certain amount of orientation in the directionof the field is produced, and according to P. Debye’s well-knowncalculation the average value of the orientational polarisation permol. is related to the dipole moment p byPo = (4x/3)Np2/3kT . . . . * (1)42 For review, see C . P . Smyth, Chem. Reviews, 1936, 19, 329.43 Proc. Roy. SOC., 1935, A , 149, 1; A , 151, 1; cf.0. K. Rice, J. Chem.44 J. Phys. Radium, 1936, 7, 441 ; Trans. Paraday SOC., 1937, 33, 81.45 J. Phys. Radium, 1937, 8, 294.4 6 Proc. Roy. Soc., 1937, A, 159, 227.Physics, 1937, 5, 492 ; T. S. Chang, Proc. Camb. Phil. Soc., 1937, 33, 524BUTLER : INTERMOLECULAR FORCES. 85The total molecular polarisation is related to the dielectric constantE by the Clausius-Mosotti formula,where (4x/3)Nao is the non-orientational part of the polarisation.If the dipolar molecules were quite free to rotate in liquids, thesame orientational polarisation should be produced in solution innon-polar solvents or in the pure liquid. Actually, it is well knownthat the polarisation in the pure liquid is much smaller than that inthe vapour, or in dilute solutions in non-polar solvents; e.g., fornitrobenzene in dilute carbon tetrachloride solution a t 20", Po =369 c.c., whereas in the pure liquid Po = 95 C.C.When the molarfraction of the dipole molecules is increased from 0 to 1, two kindsof polarisation curves are encountered : (1) the polarisation of thedipolar constituent falls continuously, (2) it passes through amaximum (e.g., alcohols in non-polar solvents). This behaviourhas hitherto been ascribed to the formation of groups of two ormore molecules, the dipole moment of which may be greater orless than that of the single dipoles, according to their disposition,but no clear quantitative treatment of this conception has beenp~ssible.~'P. Debye has now extended 48 his theory of orientational polaris-ation to include the possibility of coupling of dipoles.If it besupposed that the rotation of the molecule may be hindered by apotential field which can be represented by - E cos 8 , where 8is the angle between the axis of the dipole and an axis in which themolecule is held a t any instant by the fields of neighbouringmolecules, the orientational polarisation is found to bewhere L(p) is Langevin's function of p = E[kT. When B > kT,the factor [l - L2(p)], by which the polarisation is reduced by thecoupling energy, is 2kTIE. The values of EIkT required to accountfor the polarisations in pure water and nitrobenzene are 11 and 10.It is also shown that when E is large, dielectric saturation occurs atmuch larger field strengths than is to be expected from the simpletheory, which is in accordance with experimental results.In solutions E might be expected to be a function of the con-centration of dipoles, and F.H. Muller has discussed49 the nature4 7 See C. P. Smyth, " Dielectric Constant and Molecular Structure,'' 1931,48 Physikal. Z . , 1935, 36, 100, 193; Chem. Rewiews, 1936, 19, 171.49 Physikal. Z., 1937, 38, 498.Chap. IX86 GENERAL AND PHYSICAL CHEMISTRY.of this function, comparing Debye's formula with A. E. van Arkeland J. L. Snoek's formula 50which accounts for the variation of Po with concentration over avery wide range.51 When the degree of coupling is great (cnv2 >3kT; n is the number of dipoles per c.c., and c a constant), thisbecomes Po = (4x/3)N(l /cn), and hence a comparison with Debye'sformula suggests that E = 2cnp2/3.As an explanation of thisrelation, Miiller suggests that the internal field acting on a dipole ina dielectric liquid is F = Cnp, and the Debye coupling energy, whichis the interaction of a given dipole with this field is E = @' =Cnp2. The constant C is almost the same for a wide variety ofmolecules (excluding hydroxy-compounds) and is independent ofT to near the critical temperature. In dilute solutions, where thecoupling energy is small, this does not, of course, apply, and altern-ative expressions are given.This theory gives a continuous fall in the polarisation as theconcentration of dipoles is increased, and gives no explanation of themaximum in the polarisation-concentration curves which occurswith alcohols, which are also anomalous in that the pure liquidshave dielectric constants that decrease with rising temperat~re.5~The anomalous behaviour is no doubt connected with the specialcharacteristics of hydrogen bonds.W. D. Kumler has shown 53that the orientational part of the dielectric constant, when correctedso that it applies to equal numbers of molecules, is a linear functionof the dipole moment in many liquids, but when hydrogen bond8are present the value is considerably greater.In a very important paper, L. Onsager 54 has discussed liquids ofhigh dielectric constant from a radically novel standpoint. Histreatment agrees with Debye's original value of the orientationalpolarisation, but rejects the Clausius-Mosotti formula (2) connectingthe polarisation with the dielectric constant.In the latter theinternal field is represented by F = E + 4x1, where E is theexternally applied field, and I the electric moment per unit volume.In Onsager's treatment the field which acts on a molecule in apolarised dielectric is divided into a cavity field G, proportional toE,' and a reaction field R, which is proportional to the total electricmoment and depends on the instantaneous orientation of themolecule. In Mosotti's formula it is assumed that the total internal80, 707.Physikal. Z., 1932, 33, 662; 1934, 35, 187; Trans. Paraday SOC., 1934,61 See Ann. Reports, 1936, 33, 132.52 Cf. P. Girard, Tram. Puraday SOC., 1934, 30, 763.ss J.Amer. Chem. Soc., 1935, 57, 600. 64 Ibid., 1936, 58, 1486BUTLER : INTERMOLECULAX FORCES. 87field is effective in orientating the molecule, but according foOnsager, R can never exert a torque on the molecule and theorientating couple is produced entirely by Q. The reaction field Rcauses an enhancement of the dipole moment of the molecule andalso increases the moment induced by R. Hence the dipole momentis a function of the dielectric constant of the medium. As the resultthe polarisation is represented by(E - r2)(2c - r2)/c(r2 + 2)2 = 4ni~p,2/9kTwhere r is the refractive index; or, when E is large,2&/(r2 + 2)2 - 41tN. p,-,2/9kT . . . . (5)J. Wyman 55 had found that the Clausius-Mosotti formula failedt o give an adequate interpretation of the polarisation of liquids andsolutions of high dielectric constant.In aqueous and alcoholicsolutions of amino-acids, the dielectric constant increases linearlywith the concentration (n); dzldn, is practically independent of thesolvent and is closely correlated with the relative dipole momentsof the solute molecules, so that in these solutions the dielectricconstant appears to be an additive property of the moments present.These facts find no simple interpretation on the Mosotti formula,and Wyrnan therefore suggested that for large values of E, thedielectric constant is a linear function of the polarisation. He hassince shown 56 that in many pure liquids the dielectric constant i srelated to the polarisation, calculated from the dipole moment byDebye's formula, by P = ( E + l)/q where x varies from 6 to 10 andis usually close to 8.5.Onsager's expression (5) gives this valuewhen r = 1-46.Dispersion of the Dielectric Constant .-When the frequency ofelectric oscillations is increased, a point is eventually reached atwhich the molecular rotations of the dipolar molecules can no longerkeep pace with the applied field, and the dielectric constant thenfalls from its low-frequency value E, to that produced by theelectronic part of its polarisation alone, E ~ . In Debye's theory ofthe effect, the orientational polarisation for the frequency v is givenby the complex quantityP o = T ' - 4xN p2 1.3!cT * 1 + 2XiVT 'where T is the relaxation time. Fkom the real part of this, thedielectric constant is obtained asE = EO + (Em- E0)/(1 + V2/VS2)J .Amer. Chem. SOC., 1934, 56, 636; Chem. Review, 1936, 19, 213,J . Amer. Chem. Soc., 1936, 58, 1482$8 GENERAL AND PHYSICAL CHEMISTRY.where v, is the critical frequency at which E is half-way betweenthe two limits E, and E ~ , and is related to 7 byvs = 1/2X+O + 2)/(E,+ 2)Connected with this are two other effects, which are more con-veniently measured : (1) a high-frequency conductivity whichincreases as E decreases, (2) a dielectric loss or absorption of energyin the diele~tric,~' which shows itself in an easily measured heatingeffect, This is negligible in non-polar liquids, and in dilute solutionsis proportional to the number of dipolar molecules ; it is a maximumat the frequency vs* and at considerably lower frequencies isproportional to v ~ .~ ~On the assumption that the frictional resistance to rotation of thedipoles may be represented by Stokes's law, Debye calculates therelaxation time as T = 4~ya,~/kT, where 3 is the viscosity of theliquid and a the radius of the dipolar molecule. The relaxationtimes obtained by either of the above methods are of the order of10-11 sec., and for dilute solutions in non-polar solvents are in quitegood qualitative agreement with the calculated values. 59 Quantit-atively there are discrepancies ; e.g., W. Holzmiiller has investig-ated solutions of a series of ketones and finds a progressive increasein z with the size of the molecule, but the molecular radii requiredare about half the actual values, and the ratio of the relaxationtimes in hexane and benzene is not constant and is usually smallerthan the ratio of the viscosities.Recent observations by L. D'Orand J. Henrion suggest that the relaxation time may apply tothe rotation of parts of the molecule, for z for pdi(chloromethy1)-benzene, which has rotating dipoles, is considerably less than thatfor o-dichlorobenzene.In pure liquids there are considerable discrepancies, particularlywith hydroxylic compounds, which have been extensively investig-ated with wave-lengths between 1 and 100 cm.G2 P. Girard andP. Abadie G3 found that in a series of polyhydric alcohols, a decreasesF. Harms, Ann. Physik, 1901, 5, 564; J. Malsch, Physikal. Z., 1932, 33,19; 1936, 37, 849; Ann.Physik, 1932, 12, 865; 1934, 20, 33; M. Wien,Physikal. Z., 1936, 37, 155.68 P. Debye, Trans. Paraday SOG., 1934,30,679 ; Physikal. Z., 1934,35, 102.69 G. Martin, ibid., 1936, 37, 164, 665.6o Ibid., 1937, 38, 574.Ibid., p. 635; Compt. rend., 1936, 202, 398.e2 M. von Ardenne, 0. Gross, and G. Otterbein, Ph,ysikal. Z., 1936,37,533 ;M. Wien, ibid., p. 869; J. Malsch and E. Keutner, ibid., 1936, 36, 288; C.Schreck, ibid., 1936, 37, 157, 549; C. Schmelzer, ibid., p. 162; Ann. Physik,1937, 28, 35; A. Esau and G. Boz, Physikal. Z., 1937, 38, 774; G. Hettner,ibid., p. 771 ; D. Elle, Ann. Physik, 1937, 30, 354.6s J . Phys. Radium, 1935,6, 296; 1936,7,211; Compt. rend., 1936,202,308BUTLER : MTERMOLECTJLAR FORCES. 89as the number of hydroxyl groups is increased, and with glycerol,complex curves are obtained indicating more than one relaxationtime. In solutions of these compounds, T is a maximum at thesame concentration as the maximum of the polarisation.The discrepancies from the simple theory might be due to thenot unlikely incorrectness of Stokes's law when applied to rotations.Debye has attempted to improve the theory by taking account ofhindrance of dipole rotation by the potential field of surroundingmolecules (p.85), and finds that when the coupling energy E ismuch greater than kT, (6) is replaced by4xN ,p2 1 P o = - -3 ' 3kT 0*5p( 1 + 2xiu~/O*5p)The meaning of this is that P, is divided by the factor p/2 as before(P = EJkT), and the effective relaxation time is obtained bydividing T by the same factor, i.e., T' = ~/0-5p.It is too early to judge the success of this development. Againstit are observations of W.Hackel,65 who has determined the relax-ation times of the lower alcohols over a range of temperature andfinds that for each alcohol the experimental value 7' is a linearfunction of T and at certain temperatures T i / 7 is greater than 1.Hydrogew Bonds in Associated Liquids.-The association ofmolecules in liquids through the formation of hydrogen bonds hasbeen discussed in a prcvious Report 66 and it is only necessary toreview recent developments. The whole subject is covered by alengthy article by M. L. Huggins.67 J. D. Bernal and H. D.Megaw 68 suggested that two kinds of bonds should be distinguished,for which they proposed the names, hydrogen and hydroxyl bonds,severally.In the former, the hydrogen atom is attached with equalfirmness to both oxygens, i.e., -0----33----0-, the energy of thebond is about 8000 cals./mol., and the distance between the oxygens2-5-4.65 A. In the latter, the hydrogen remains unsymmetricallyattached to one of the oxygens, i.e., -0----H-0-, giving a bondenergy of ca. 5000 cals./mol. and an oxygen distance of 2.7-2.9 A .Since it is admitted-that probably an almost continuous series ofcases between these extremes exists, the necessity of the distinctionhas not been everywhere accepted.@ The nature of the hydrogen64 Physikal. Z., 1935,36, 100, 193; P. Debye and W. Ramm, Ann. Physik,1937, 28, 28.65 Physikal.Z., 1937, 38, 195; W. Hackel and M. Wien, ibicl., p. 767;cf. E. Keutner and G. Potakempo, ibid., p. 635.6g Ann. Reports, 1934, 31, 37.6 7 J . Org. Chem., 1936, 1, 407; cf. E. N. Lassettre, Chem. Reeriews, 1937,g8 Proc. Roy. Soc., 1935, A , 151, 384.139 See, e.g., W. H. Rodebush, Chem. Review8, 1936, 19, 69.20, 25990 GENERAL AND PHYSICAL CHEMISTRY.bond has also been discussed by A. Sherman and R. H. Gillette,'Owho suggest that since its exact nature is a t present unknown, it isbest defined by the thermochemical binding energy.Hydrogen atoms attached to oxygen give rise to well-markedabsorption bands in the near infra-red, which are usually easilydistinguished from the bands due to CH, and, being characteristicallyinfluenced by changes of the binding energy, they promise to givemuch information about the nature of the hydrogen bonds, and havebeen the subject of a large number of investigations.E.Ganz 71has observed the effect of changes of temperature and the additionof salts for a wide range of wave-lengths. The two bands at 0.77 pand 0.85 p, for example, sharpen considerably as the temperature israised from 12" to go", and in concentrated salt solutions theabsorption resembles that in pure water a t a higher temperature,ie., the salts raise the " structural temperature " of water, the effectbeing greater the larger the anion. The bands at 3 p, 4.7 p, and 6 phave also been studied,72 and the effects of temperature on themaxima are recorded. J. R. Collins and C.Moran 73 have alsostudied the effect of salts on the infra-red bands of water, andfind that small ions, such as those of lithium and magnesium,decrease, while large ions increase, the structural temperature, whichagrees with J. D. Bernal and R. H. Fowler's conclusions as to theeffect of these salts on the temperature of maximum density.74G. Bosschieter and J. Errera 75 studied dilute solutions of water incarbon disulphide and carbon tetrachloride and found bands between3500 and 3700 cm.-l, which they attribute to the valency vibrationsof the single molecule. In liquid water and in more concentratedsolutions in various solvents 76 another band appears a t 3300 cm.-l,which arises from interactions between the molecules.The Raman spectrum of liquid water has been examined byM.Magat 77 and J. H. Hibben,T8 who interpret the low-frequencyshifts in terms of the Bernal-Fowler model. I. R. Rao and P.K0teswaram,7~ however, suggest that some of the lines on whichthese investigators rely are excited by weak secondary lines of theThe situation with water itself is not very clear.70 J . Arner. Chem. Soc., 1936, 58, 1135; J . Physical Chem., 1937,41, 117.2. physikal. Chem., 1936, B, 33, 163; Ann. Physik, 1936, 26, 331; E.Ganz and W. Gerlach, Physikal. Z., 1936,37, 358.72 E. Ganz, Ann. Physik, 1937, 28, 445.73 Physical Rev., 1936, 49, 869; 1937, 52, 88; cf. J. R. Collins, ibid., 1925,74 J. Chem. Physics, 1933, 1, 515.76 J . Phys. Radium, 1937, 8, 229.7 7 Ann. Physique, 1936, 6, 108; Tram.Paraday SOC., 1937, 38, 114.7 8 J . Chern. Physics, 1937, 5, 166.26, 771.76 Compt. rend., 1937, 204, 1719.70 Ibid., p. 677BUTLER : INTERMOLECULAR FORCES. 91mercury arc. P. C. Gross, J. Burnham, and P. A. LeightonS0have made a more elaborate study, and conclude that water israther more than 2-co-ordinated between 25" and go", while in icea t O", 4-co-ordination predominates.The first systematic study of the infra-red spectra of hydroxy-and amino-compounds was that of 0. R. Wulf and U. Liddel,81who, working with a glass spectrograph, examined the absorptionat about 1-5 p ( k e . , the first overtone of the fundamental band) ofmany of these compounds in dilute solution. They found aremarkable constancy of wave-length and intensity, except incompounds containing intramolecular bonds, where this bandappeared to be completely absent.(Recent investigation suggeststhat it has probably shifted to wave-lengths beyond the regionexamined.) Certain o-substituted groups in phenol cause a splittingof the band into two, ascribed by L. Pauling 82 to the possibility ofcis-trans-isomerism.Many hydroxyl compounds have, besides the sharp fundamentalband at cn. 2.75 p, another much wider band near 3 p. P. Molletand J. Errera s3 showed that in the case of ethyl alcohol the latteris due to association and disappears on dilution of the alcohol with anon-polar solvent, while the former becomes more pronounced.Similar observations were made by R. Freymarqs4 who found thatthe former band is strong in the vapour, but weak in the liquid a troom temperature, and is intensified by dilution and by raising thetemperature, while the latter predominates in the solid and liquid atlow temperatures.Similar observations have been made by A. M.Buswell, V. Dietz, and W. H. Rodebu~h.~~ J. J. Fox and A. E.Martins6 have made a careful examination of the absorption inthis region of phenol and various aliphatic and aromatic alcohols.The wave-length of the sharp band varies slightly in the differentcompounds. With phenol the intensity of the longer-wave" association band " is proportional to the number of moleculesnot contributing to the short-wave band, and it is possible todetermine the numbers of single and associated molecules. Theassociation band is complex, and its maximum moves to shorterwave-lengths as its intensity falls.The results suggest that,particularly with the aliphatic alcohols, the complex molecule may80 J. Amer. Chem. SOC., 1937, 59, 1134.81 Iblbid., 1933, 55, 3574; 1935, 57, 1464; 0. R. Wulf, U. Liddel, and S. B.82 Ibid., p. 94.8s Compt. rend., 1937, 204, 259; Trans. Furaday SOC., 1937, 33, 120.84 Compt. rend., 1937, 204, 261, 1063 ; Bull. SOC. chim., 1937, 4, 944.8 5 J. Chem. Physics, 1937, 5, 501.8 6 Proc. Roy. SOC., 1937, A , 162, 419.Hendricks, ibid., 1936, 58, 228792 QENERAL AND PHYSICAL CREMISTRY.contain more than two single molecules, in agreement with thethermoohemical results of K. L. Wolf .8'have made very similar observ-ations on the second harmonic at about 1 p.In the first place theystudied acetic acid in the vapour state and found that the intensityof the sharp band at 0.97 p was proportional to the concentration ofsingle molecules. In solutions of alcohols they found 89 two broadbands, one on each side of the narrow hydroxyl band, whichincreased in intensity as the concentration increased. Thefrequencies of the primary bands of various molecules are shiftedin the liquid state or in solutions by amounts which are proportionalto the interaction energy of the hydroxyl group with the surroundingatoms. A. Naherniac 90 has also made an extensive study of theshifts of these bands in passage from the vapour to the liquid stateand finds a correlation between the shifts and the magnitude of thevan der Waals forces.It is evident that the infra-red spectra promise to provide a verypowerful method of investigating the interactions of hydroxyl andsimilar groups with each other and with other molecules in pureliquids and solutions, but so far the work has been exploratory andthe precise origin of the broad secondary bands has not yet beenelucidated.Other investigations, which cannot be described indetail, are listed bel~w.~lR. M. Badger and S. H. BauerJ. A. V. B.6. ELECTROCHEMISTRY.Thermal Data for E.M.F. Measurements.--It has been indicated inprevious Eeports 1 that the study of heats of dilution and of otherthermal properties of solutions of strong electrolytes has attractedinterest in recent years. In view of the difficulties connected withprecise calorimetric measurements, especially with dilute solutions,it is important to record the development of methods based ondeterminations of the E.M.F.'s of reversible cells.The E.M.F. ofsuch a cell is a measure of the free-energy change in the process taking87 Trans. lraraday SOC., 1937, 33, 179; 2. physilcal. Chem., 1935, B, 28, 1.88 J . Chem. Physics, 1937, 5, 369, 605.s1 E. L. Kinsey and J. W. Ellis, J . Chem. Physics, 1937, 5, 399; P. Bar-chiewitz, Compt. rend., 1937, 204, 1184; s. Mizushima, Y. Uehara, and Y.Morino, Bull. Chem. SOC. Japan, 1937,12, 132 ; D. Williams and E. K. Plyker,J. Chem. Physics, 1936, 4, 460; J . Opt. SOC. Amer., 1936, 26, 149; M. V.Volkenstein, Acta Physicochim. U.R.S.S., 1936, 4, 357; W.Gordy et al.,J . Amer. Chem. Soc., 1937, 59, 464; Physical Rev., 1936, 50, 1151; 1937,51, 564; J . Cham. Physics, 1935, 3, 664; 1936, 4, 749; 1937, 5, 284; J .Physical Chem., 1937, 41, 645.89 Ibid., p. 399.Ann. Physique, 1937, 7, 528.Ann. Reports, 1932, 29, 29; 1934, 31, 58GLASSTONE : ELECTROCHEMISTRY. 93place therein, and by means of the Gibbs-Helmholtz equation it ispossible to calculate the decrease in heat content for the sameprocess. This method, as applied to “ chemical cells,” has been inuse for several years for the determination of the heat changesassociated with chemical reactions, and the results have been recog-nised as more accurate than those obtained by direct thermo-chemioal measurement .2 It should be possible, theoretically, byapplying an analogous method to ‘‘ concentration cells ” to acquireinformation related to heats of dilution, but it is only in the pastfew years that the technique of the study of these cells has attainedthe standard necessary for the measurements to be of value. Theappropriate thermodynamic equations have been known for sometime and various attempts were made to apply them, but the moderndevelopments may be considered as having originated four yearsago.It is probably not yet true to say that the results obtainedfrom concentration cells are more accurate than those derivedfrom the best calorimetric studies, but the former procedure has theadvantage of being simpler and more rapid.Consider a concentration cell without transport, ie., one in whichtwo solutions of the same electrolyte at different concentrations,c and c’, are separated by a suitable electrode reversible with respectto one of the ions of the electrolyte ; for the passage of x faradays ofelectricity, where x is the valency of the other ion, the quantityof electrolyte in the more concentrated solution ( c ) is decreased by1 mol., whereas that in the other (c’) is increased by the same amount.At constant pressure the change in heat content of the first solutionis equal to the partial molar heat content * of the solute (R2) at theconcentration c, whereas for the second solution ( c ‘ ) the correspond-ing change is Hz’.The free-energy change is represented by xP . AE,where AE, the E.N.F. of the concentration cell, is generally expressedas E - E’; this corresponds to the fact that in cells of the typeunder discussion it is the practice to measure the E.M.F.’s of thetwo halves separately.By the Gibbs-Helmholtz equation itfollows thatH2 - B2’ = zP[(E - E’) - T(6(E - E’)/W>] . . (1)and hence if E and E’ are measured at a number of tem-peratures, so that the temperature coefficient at constant pressure,i.e., 6(E - E’)/6T, is known, the change in heat content can bedetermined.From the thermodynamic standpoint the important quantity is2 Cf. H. S. Taylor and G. St. J. Perrott, J . Amer. Chem. SOC., 1921, 43, 45.* This is defined as the heat change on the addition of 1 mol. of solute to aninfinitely large amount of solution, i.e., 6H/6n, the pressure being constant94 GENERAL AND PHYSICAL CHEMISTRY.gz - g2, i.e., the value when one of the solutions is infinitelydilute : this is known as the relative partial molar heat content, z2,and represents the change of heat content relative to an idealsolution.Two methods have, in general, been used to determineL, from the experimental quantity R2 - H i . First, the E.M.F.’sfor a series of different concentrations are extrapolated to give Eo.Graphical extrapolation to zero concentration. is at present insuffi-ciently accurate for the purpose and the following procedure hasbeen adopted. The E.M.F. of the half-cell containing the solutionc may be represented by the equationE = EO + (vRT/xP) log, cf . . . . . (2)where Eo is the E.M.F. of the cell when c is zero, f is the meanactivity coefficient of the ions a t the concentration c, and v is thenumber of ions produced by one molecule of electrolyte.Theactivity coefficient can then be expressed in terms of the concen-tration by means of an extension of the Debye-Huckel equation,e.g., that of Gronwall, LaMer, and Sandved, and by using theE.M.P.’s for a number of cells with different concentrations of solutethe best value for Eo may be found.3 The substitution of Eo forE’ in equation (1) then gives the corresponding value of B2 - R2O.Instead of using the actual E.M.P. of the cell ( E ) for purposes ofcalculation it has been proposed to replace it by a quantity E,,where E is equal to E, - (vRT/xE’) log, c ; substituting this for Ein equation (1) and replacing E’ by EO, it is readily seen that theequation becomesRz - B,o = Lz = zE’[(Ec - EO) - T{S(Ec - EO)/ST>] .(3)so that z, is obtained directly.4 In order to evaluate the temper-ature coefficient the E.M.P.’s are expressed as a function of thetemperature, e.g., E = a + bt + ct2 + . . ., where t is the Centi-grade temperature, and then differentiated, giving b $- 2ct + . . .The second method involves the determination of the relativepa.rtia1 molar heat content of any convenient solution, say c’,i.e., H,‘ - H: is determined; hence by adding this value to thatof H , - H i , obtained from equation (l), the quantity E, for thesolution c is obtained. The relative partial molar free energy P - %Oof the solute in a solution is, by definition,F - Fi-J = (VRT) logef + (vR17) log, cwhere the symbols have the same significance as before; from this,See I.A. Couprthwaite and V. K. LaMer, J . Amer. Chem. Soc., 1931, 53,4333.4 V. K. LaMer and I. A. Cowperthwaite, ibid., 1933, 65, 1004GLASSTONE : ELEOTROCHEMISTRY. 95by direct application of the Gibbs-Helmholtz equation, it followthatBy means of an extended form of the Debye-Huckel equation,e.g., the form usually known as the Huckel equationY5 it is possibleto express logJ in terms of the concentration : then by usingequation (4) an expression for z2 may be obtained involving concen-trations, the temperature coefficient of the dielectric constant of thesolvent, and other quantities which are either universal constants ormay be calculated from the experimental E.M.F.data. By meansof the resulting equation, a,l - &O may be evaluated for theconcentration c’, and so L2 for other concentrations may be obtained.In order that the results should be as little dependent as possibleon the method used for determining z2’ - B20, it is evident thatthe reference solution c’ should be as dilute as possible; concen-trations as low as 0.001 molal have been used. For a range of nottoo concentrated solutions, for which the Huckel equation applies,the z2 values for the whole series may be calculated by means of theequation relating z2 to the concentration : the results obtained inthis way are in good agreement with those given by the othermethod6 This is, of course, to be expected provided the adjustableconstants in the Huckel equation have been correctly chosen.The familiar Kirchhoff equation states that AC, = 8(AH)/8TY andapplied t o the case under discussion this leads toH - = E2 = - v ~ T 2 ( 6 logef/sT) .. . . (4)where Zp2 is the partial molar heat mpacity of the solute * in the givensolution and is the corresponding value at infinite dilution; thedifference is called the relative partial molar heat capacity, and thismay be obtained from the temperature coefficient of z2 at constantpressure. If z2, obtained as described above, is expressed as afunction of temperature, then 6z2/6T is readily obtained by dif-ferentiation. It will be observed from equations (1) and (2) thatthe determination of z2 involves differentiation of the E.M.F.withrespect to temperature; hence zap - zi9 is dependent on the seconddifferential and a high order of accuracy is not to be expected unlessthe E.M.F. measurements are very exact. Nevertheless, wheredirect calorimetric studies have been made the results are generallyin satisfactory agreement with those obtained from E.M.F. data.6- 76 E. Hiiukel, Physikal. Z., 1925, 26, 93.* H. 8. Harned and R. W. Ehlers, J . Amer. Chem. Soc., 1933, 55, 2179.7 H. S. Harned and J. C. Hecker, ibid., p. 4838; H. S. Marned and M. A.* This is defked as the change in heat capacity of a large amount of solutionCook, ibid., 1937, 59, 496, 1290.on addition of 1 mol. of solute96 GENERAL AND PHYSIUAL CHEMISTRY.The methods described above for the determination of thermaldata from measurements on reversible cells have been applied tothe study of solutions of hydrochloric acid in water and in methylalcohol-water mixtures 8 ~ 6 ; sodium and potassium hydroxides ; '* *cadmium,10 thallium,ll b&rium,l2 and potassium 13 chlorides ;sodium,14 zinc 39 15 and cadmium 16 sulphates ; sulphuric acid ; 1'hydrobromic acid in water and in lithium chloride solutions; l8and sodium bromide.19Thermal Properties and CmLcentration.-According to the limitingDebye-Huckel equation, applicable to very dilute solutions, thelogarithm of the activity coefficient (f) is a linear function of dz;it - follows, - therefore, from equations (4) and (5) that both z2 andcpa - cgp should also vary in a linear manner with 6 .2 0 It isdoubtful whether the limiting slopes obtained by the electrochemicalprocedure for very dilute solutions have any significance, since someform of the Debye-Huckel equation is always used for the purpose ofextrapolation to infinite dilution. Until an independent method ofextrapolation is used, therefore, the data are only of theoreticalvalue in connexion with the study of relatively concentrated solu-tions. The thermal data obtained from E.M.P. measurements forsuch solutions, like those obtained calorimetrically, show a linearvariation with dz, although the concentrations are too high for thelimiting equation to be applicable; the slopes of the lines are,however, not in agreement with those to be expected at high dilu-tions, and each electrolyte shows individual behaviour.An attempthas been made to determine whether this individuality could beeliminated by expressing thermal properties a t constant volumeinstead of at constant pressure, thus avoiding any disturbing effectsthat might arise from the thermal expansibility and the compressi-bility of the solution; the results show that these factors are not* H. S. Harned and H. C. Thomas, J. Amer. Chem. SOC., 1936, 58, 761 ;G. Wkerlof and J. W. Teare, ibid., 1937, 59, 1855.9 H. S. Harned and M. A. Cook, ibid., p. 496.lo H. S. Harned and (Miss) M. E. Fitzgerald, ibid., 1936, 68, 2624.l1 I. A. Cowperthwaite, V. K. LaMer, and J. Barksdale, ibid., 1934, 56, 54.l2 E. A. Tippetts and R.F. Newton, ibid., 1934, 56, 1676.la H. S. Harned and M. A. Cook, ibid., 1937, 59, 1290.lo H. S. Harned and J. C. Hecker, ibid., 1934, 56, 650.l6 H. S. Harned, ibid., 1937, 59, 360.l6 V. K. LaMer and W. G. Parks, ibid., 1933, 55, 4343; V. I(. LaMer andE. L. Carpenter, J. Physical Chem., 1936, 40, 287.17 I. A. Cowperthwaite and J. Shrawder, J. Amer. Chem. BOG., 1934, 56,2345; H. S. Harned and W. J. Hmer, ibid., 1935, 57, 27.l8 H. S. Harned, A. S. Keston, and J. G. Donelson, ibid., 1936, 58, 989;H. S. Harned and J. G. Donelson, ibid., 1937, 59, 1280.lo H. S. Harned and C. C. Crawford, ibid., p. 1903.2o Anw. Reports, 1934, 31, 59GLASSTONE : ELECTROCHEMISTRY. 97responsible for the specific properties of differentIt appeared possible that the square-root relationship, at least asapplied to concentrated solutions, might have little theoreticalsignificance, as the non-electrolytes urea and mannitol were reportedto show a similar variation of thermal properties with concentration.22It must be pointed out, however, that the measurements extendedover a relatively small range, the heat capacities varying by onlysmall amounts, so that the conclusions cannot be regarded as estab-lished. In fact, measurements with sucrose show that the apparentmolar heat capacity is more nearly a linear function of the first powerof c than of 2/23 whereas for urea, over a considerable concentrationrange, neither relationship is appli~able.~*Since the last report on the subject,2* relatively little calorimetricwork has been done on the thermal properties of dilute solutions ofelectrolytes; some accurate measurements have been made of theintegral heat of dilution * of sodium chloride solution^,^^ and of theheat capacities of sodium and barium chloride solutions.26 Thesequantities can be related to one another and also to the relativepartial molar heat contents and heat capacities discussed above?’The Debye-Huckel limiting law is undoubtedly applicable in verydilute solutions, and the slope of the line showing the variation ofthe integral heat of dilution with the square root of the concen-tration is in better agreement with the requirements of this law 28than was a t first r e a l i ~ e d .~ ~ The difficulties lie, not only in themeasurements, but also in the extrapolation of the results obtained.Theory of Concentrated Solutions of Electrolytes.-The problem ofdilute solutions of strong electrolytes can be regarded, a t least forthe present, as solved, and consequently attention is being turnedto the theory of more concentrated solutions and of mixtures, Itappears that in both these connexions the ((principle of specificionic interaction,” originally postulated by J.N. Brmsted in 1922 2921 F. T. Gucker and T. R. Rubin, J . Amer. Chern. SOC., 1936,67, 78.22 C. M. White, ibid., 1936, 58, 1620.23 F. T. Gucker and F. D. Ayres, ibid., 1937, 59, 447.26 E. A. Gulbransen and A. L, Robinson, ibid., 1934,56,2637.26 T. F. Young and J. S. Machin, ibid., 1936, 58, 2254; C.M. White, ibid.,27 See M. Randall and F. D. Rossini, ibid., 1929,51, 323.28 T. F. Young and W. L. Groenier, ibid., 1936, 58, 187.20 See Ann. Reports 1933, 30, 24.* The integral heat of dilution per mol. of solute is the heat change ondiluting a solution containing 1 mol. of solute at a given concentration toinfinite dilution; it is related to z2 by the expression Ljmz2. dm, where WI isIdem, ibid., p. 2962.p. 1615.m athe molality of the solution.REP.-VOL. XXXIIV. 98 GENERLL BND PHYSICAL CHEMISTRY.and subsequently developed more explicitly,30 is likely to play animportant part. According to this principle the interaction betweenions of the same sign is determined entirely by electrostatic (“ long-range ”) forces, since as a result of their mutual repulsion they spendrelatively little time in close proximity, but for ions of opposite signspecific (“ short-range ”) forces, dependent on the nature of theiom, also make themselves felt.The original Debye-Huckeltreatment considered only the electrostatic forces between ions of afinite size in a medium of dielectric constant equal to that of thesolvent; later, Huckel applied a correction for the effect of theions on the dielectric constant, regarded in the nature of a ‘‘ salting-out ” effect, which was of importance for relatively concentratedsolutions. To make the picture more complete, however, it isnecessary to add a term for the interaction between ions andundissociated molecules, equivalent to the salting-out of non-electro-lytes, and a term for molecule-molecule interaction, representingthe departure of solutions of non-electrolytes from ideal behaviour.The exact trwtment is too complicated to be carried further atpresent, but it can be simplified by introducing the principle ofspecific ionic interaction: it is assumed that as far as the short-range forces are concerned only ions of opposite sign need be con-sidered, In this way G.Scatchard31 has obtained an expressionfor the “non-ideal” free energy, and using the simple thermo-dynamic relationship between this and the activity and osmoticcoefficients,* he obtained for these quantities expressions the firstterm of which corresponds to the Debye-Huckel equation. Thezeis, however, an important difference : it has been argued that thequantity K which appears in the Debye theory, and represents thesquare of the reciprocal of the thickness of the ionic atmosphere,should be proportional to the number of ions in unit volume ofsolvent, rather than of solution.This means that the quantityZcx2, where c is the volume concentration, should be replaced byXmz2, where m is the molality, i.e., mols. per 1000 g. of solvent, atconstant temperature and pressure. In order to test Scatchard’sequations, there are only two quantities which are not known withsufficient accuracy and must be obtained empirically from measure-ments on two solutions : these are (a) the volume occupied by theions in solution, and (b) the coefficient for molecule-molecule inter-action.The former is found to be 3.15 times the actual volume, as80 E, A. Guggenheim, PhiE. Mag., 1935,19, 586; 1936, 22, 322.81 Chern. Reviews, 1936,19, 309.* The osmotic coefficient is obtained by dividing by actual osmotic effect,e.g., depression of the freezing point, by the value for an ideal solution,allowance being made for ionisation assumed to be completeGLASSTONE : ELECTROCHERIISTRY. 99compared with four times to be expected from the van der Waalstheory, and the latter is one-third the value for an aliphatic hydro-carbon in water. By using these data, the calculated values for t h eosmotic coefficient have been compared with those observed for the15 alkali chlorides, bromides, and iodides at concentrations up toand generally exceeding 4 ~ .The agreement, although not perfect,a t least shows that the new development is a step in the right direc-tion, It i s of interest to record that numerous valuable experimentaldata concerning osmotic and activity coefficients, covering a consider-able range of have recently become available as aresult of an improvement in the " isopiestic '' method of determiningvapour pressures of solutions 33 first attempted, without greatsuccess, by W. R. Bousfield in 1917.Mixtures of Electrolytes.-The application of the theory outlinedabove to mixtures of electrolytes is a matter of difKculty, butprogress has been made by means of a simplified treatment. Startingfrom a general equation, based on statistical considerations, of thefree energy of a quantity of fiuid containing a number of com-ponents, an equation has been deduced3* which, for the osmoticcoefficient (+) of a particular ion in a mixture of electrolytes, can bewritten in the form4 - 1 = + Bm + Cm1'5 + Dm2 + + .. . . (6)where m is related to the total ionic concentration of the solutionand A is the appropriate Debye-Hiickel constant, and representsthe interaction between the ion and its atmosphere. The quantitiesBy C, D, E , ete., depend on the various forces in the solution and onthe respective ionic concentrations ; B is the equivalent of the short-range interaction between a pair of ions, C is that for the pair ofions and their ionic atmospheres, D for three ions, E for three ionsand their atmospheres, and so on.It has been found experi-mentally35 that B and C are linear functions of the ionic concen-trations, whereas D and E axe quadratic functions. These relation-ships are exactly those derivable from the principle of specificinteraction which would make B and C , for two ions of the samesign, and D and E, for three ions of the same sign, so small as to benegligible. In the original application of the principle Brmnstedconsidered an equation equivalent to the A and B terms only of32 R. A. Robinson et cd., J . Amer. Chem. Soc., 1934,50,1839; 1935, b7, 116133 D. A. Sinelair, J , Physical Chem., 1933, 87, 395.34 G. Scatchard and S. 8. Prentb, J . Amer. Chem. SOC., 1934,66,1486,2314.85 Idem, ibid., p. 2320; me abo B. B. Owen and T. F. Cooke, ibid., 1937,gQ,1165; 1936, 58, 959; 1937,59, 84.2273, 2277100 GENERAL AND PHYSICAL CHEMISTRY.equation (6) ; hence the modern development represents the exten-sion of the concept to the coefficients of higher powers of the concen-tration of the solution, which take further inter-ionic forces intoconsideration. It may be mentioned that a similar relationshipfor B and D, although not quite as simple for C and E, applies inconnexion with the activity coefficient of an ion in a mixture ofelectrolytes.A simpler treatment of concentrated solutions and of mixedelectrolytes, which appears, however, to be less promising, has beendeveloped by G.Akerlof : for two different strong electrolytes of thesame valency type a t the same concentration and with equal ap-parent mean ionic diameters, the Huckel equation gives the relation-ship for the two activity coefficients logfi/’f2 = Ec, where c is theconcentration.This has led to the formulation of the empiricalrule logy - log yR = Elm, where is the activity coefficicnt, interms of molality, of any electrolyte and yR is the value for a referencesubstance at the same molality m, and El is a constant for the formerelectrolyte. Activity data for a number of substances appear to bein good agreement with this rule36 for relatively concentratedsolutions. In further development of the argument, it was postu-lated that in a series of solutions of two electrolytes, at constant ionicstrength, the logarithm of the activity coefficient of either is a linearfunction of its concentration, and is independent of the total con-centration, Combining these rules with the fact that the activityof any solute must be constant in its saturated solutions, a t a definitetemperature, equations have been deduced whereby the compositionsof saturated solutions of two 36 or more 37 soluble strong electrolytescan be calculated, provided certain activity data are available. Theprocedure can also be reversed, and the activity coefficients of hydro-chloric acid a t high concentrations have been determined frommeasurements of the solubility of sodium chloride in these acidsolutions.38The postulates described above have, however, been subjected tosome criticism : the results of activity-coefficient measurements bythe freezing-point method show that the first rule is certainly notcorrect in solutions more dilute than l ~ ., although at higher concen-trations it may be a satisfactory approximati~n.~~ The rule mayalso be tested by means of calorimetric measurements ; by combiningAkerlof’s equation with equation (4), it can be shown that ananalogous expression L2 - Lz(R) = k2m should apply : this has beena e G. Akerlof and H. C. Thomas, J. Amer. Chem SOC., 1934, 56, 593.3 7 G. Akerlof, {bid., p. 1439; G. Akerl6f and 0. Short, ibid., 1937, 59, 1912.38 G. Akerlof andH. E. Turek, ibid., 1934, 56, 1876.39 G. Scatchard and S. S. Prentiss, Zoc. cit., ref. (36)GLASSTONE : ELECTROCHEMISTRY. 101found to be only approximately true, there being marked deviationsin dilute solutions.40The extensive work of H.S. Harned and others, on activity co-efficients in mixtures of halogen acids and their salts, shows that theresults may be expressed by the equationsandlogyl = aim, + logy10 . . . . . (7)log y2 = a2m2 + log yZ0 . . . . . (8)where y, and y2 are the activity coefficients of acid and salt, respec-tively, in the mixture; ylo is the value for the acid a t zero concen-tration in the salt solution, and yZ0 that for the salt at zero concen-tration in the acid; ml and m2 are the molalities of acid and saltrespectively, the sum of which is constant and equal to the ionicstrength. It can be shown that according to Akerlijf’s rules a1 anda2 should be constant and independent of the total concentration :this is not the case in dilute solution, although the values appear tobecome more constant a t higher con~entrations.~1 The rules can,therefore, only be regarded as approximate, but they may proveuseful in the study of concentrated solutions.Harned41 has shown that the application of the principle ofspecific ionic interaction in its simple form requires al and a2 inequations (7) and (8) to be equal numerically but of opposite sign,i.e., al + or2 = 0.The actual sum appears to tend towards zero invery dilute solutions, but appreciable deviations occur at higherconcentrations : it is possible that Scatchard’s extension of theBrransted principle, described above, to higher powers of the con-centration may remove the discrepancy.Dissociation Constants.-The potentiometric method in commonuse for the determination of the dissociation constants of acids andbases involves the measurement of the hydrogen-electrode potentialin mixtures of the acid, or base, and its salt, a calomel electrode beingused as standard. These measurements not only involve the un-certainty of a liquid junction, but in addition there is a divergenceof opinion concerning the potential of the calomel electrode on thenormal hydrogen scale.Both these difficulties have been avoidedin a method for the determination of the thermodynamic dissociationconstants of weak acids developed by H. S. Harned and his colla-b o r a t o r ~ . ~ ~ For an acid HA a cell without liquid junction of thetypeAglAgCl ( s ) NaCl (m3) NaA (mz) HA (m1)lH2 (1 atm.)is set up, where m,, m2, and m3 are the concentrations of acid, its40 A.L. Robinson and H. S. Frank, J. Amer. Chem. Soc., 1934,56, 2312.4 1 H. S. Harned, ibid., 1935, 5’9, 1865.42 H. 8. Harned and R. W. Ehlers, ibid., 1932, 54, 1350102 GENERAL AND PHYSICAL CHEMISTRY.salt and sodium chloride, respectively.cell is given bywhere the a terms represent the activities of the ions indicated in thesubscripts, and E, is the E.N.P. of the cell consisting of AgIAgCland the hydrogen electrode, H, (1 atni.), in a solution containinghydrochloric acid having a mean activity of unity. This is known,from measurements on cells with the acid at different concentrationsand appropriate extrapolation to infinite dilution, with an error ofless than 0.1 mv.The thermodynamic dissociation constant ( K )of the acid is equal to &.uA,/aEA, and making use of the fact thatthe activity (a;) may be replaced by the product of the activitycoefficient (y) and the molality (m), it can be readily shown thatequation (9) can be mitten in the formThe E.M.F. (E) of thisE = E, - (RT/P)log,a,.a,,, . . . . (9)and henceThe right-hand side of this equation may be put equal to - (RT/P)log K‘, where K’ becomes identical with K a t infinite dilution,for then the various activity coefficients are unity and the corre-sponding term is zero. Since E, is known and E can be measured,the left-hand side of the equation can be evaluated for various con-centrations of acid, salt, and sodium chloride; the value of m,,,may be put equal to m3 in dilute solution, mHA is given by m, -[H’], and mAt by m2 + [H’], a sufficiently accurate value of [H’],the hydrogen-ion concentration, being obtained from an approximatevalue of the dissociation constant.If the resulting quantities,which are equal to - (RT/P) log, K’, are plotted against the ionicstrength of the mixture in the cell and extrapolated to zero concen-tration, the intercept gives - (RT/P) loge K , from which the thermo-dynamic constant K may be obtained. Alternatively K‘ itself,obtained by multiplying the value of the left-hand side by PIRTand taking antilogarithms, may be plotted against the ionic strength,whence the intercept for infinite dilution gives K directly. Thisprocedure involves more calculations, but the extrapolation ispresumably more accurate.The general method has been used todetermine the dissociation constants of the following acids : aceticin water and in methyl alcohol-water and dioxan-water,43 pro-43 H. s. Warned and R. w. Ehlers, J. Amer. Ghern. Xoc., 1933, 55, 652;H. S. Harned and N. D. Embree, ibid., 1935, 57, 1669; H. S. Harned andG. L. Kazanjian, ibid., 1936,68, 1912GLASSTONE : ELECTROCHEMISTRY. 103p i o n i ~ , ~ ~ chlor~acetic,~~ sulphuric (second formic,Q7 boric,48b u t y r i ~ , ~ ~ phosphoric (first and second stages),m carbonic (bothstages),5f lactic,52 and glycollic.53 A corresponding procedure forobtaining the dissociation constant of bases by the study of cellswithout liquid junction has been proposed,5* but apparently not yettested.55If it is required to study the actual extent of ionisation of a weakacid in the presence of an added salt, e.g., a chloride MCl, a t variousconcentrations, cells of the typehave been used; from the measurements the quantity KA =m,.rnAtl* in the salt solution may be determined. The values ofKA at 25", in solutions of chlorides a t the same ionic strength, havebeen found to decrease in the order BaCl,>LiCl>NaCl>KCl.These results may prove of importance in connexion with the'L secondary kinetic salt effect '' 56 in acid-base catalysis.Dissociation of Water.-A potentiometric method for the deter-mination of the ionic activity product of water, i.e., K , = %.aoH,,which does not involve liquid junctions, has also been devised : 57it depends on measurements with the cellAgIAgCl (s) NaCl (m,) NaOH (m,)lH,the E.M.F.of which, like that of the cell described on p. 101, isgiven byE = E, - (PtT/P)log,a,.a,,, . . . . . (12)where K is ~.aOH,IaEzO, that is Kw/agpo. Substituting the activities44 H. S. Harned and R. W. Ehlers, J. Arner. Chem. Soc., 1933, 55, 2377.45 D. D. Wright, ibid., 1934, 50, 314.48 MT. J. Hamer, ibid., p. 860.4 7 H. S. Harned and N. D. Embree, ibid., p. 1042.48 B. B. Owen, ibid., p. 1695.49 H. S. Harned and R. 0. Sutherland, ibid., p. 2039.50 L. F. Nims, ibid., 1933, 55, 1946; 1934, 58, 1110.6 1 D. A. MacInnes and D. Belcher, ibid., 1933, 55, 2630; 1935, 57, 1683.52 L. F. Nims and P. K. Smith, J.Riol. Chem., 1936,IIa, 145.63 L. I?. Nims, J . Arner. Chem. SOC., 1936, 58, 987.54 E. J. Roberts, ibid., 1934, 56, 878.5 5 See, however, H. S. Harned and B. B. Owen, ibid., 1930,52, 5079, 5091 ;66 H. S. Harned and F. C. Hickey, ibid., 1937, 59, 1284, 2303.57 E. J. Roberts, ibid., 1930, 52, 3877.€3. B. Owen, ibid., 1934, 56, 24104 GENERAL AND PHYSICAL CHEMISTRY.of the ions by the products of activity coefficient and molality,equation (13) becomesAs before, if the values of the left-hand side for various concen-trations of electrolyte are plotted against the ionic strength, theintercept obtained by extrapolation to infinite dilution gives bothK and K,, since h20 and the activity coefficients are then equal tounity. An alternative method for determining K, that does notrequire a knowledge of E, has also been used : it involves the em-ployment of the cells just described together with similar cellscontaining hydrochloric acid instea>d of sodium hydroxide.58 Thismethod has been adapted to the measurement of the activity ionicproduct of heavy water.59By re-arrangement, equation (14) may be written in the formRTP E - E, +-loge % =mOHnwhenceIf the value of ymyc1,, which is equal to the square of the meanactivity coefficient of hydrochloric acid at the total ionic strengthexisting in the cell, is known from measurements on cells containingthis acid, it is possible to determine ~ H ' ~ O H ~ / ~ 2 0 , i e ., the " ionicactivity coefficient product " of water in the given cell, sinceall the other quantities are available.Further, since K , i.e.,~H.yoH.m,.m,H'/aa20, is known, for it is numerically equal to K , atinfinite dilution, it is possible to evaluate mH.moH, : this gives theactual extent of the dissociation of water in the solution present intheDissociation Constants and Temperature.-The dissociation con-stants of several acids have been observed to pass through a maximum58 H. S. Harned and W. J. Hamer, J . Amer. Chem. SOC., 1933, 55, 2194.5D E. Abel, E. Bratu, and 0. Redlich, 2. physilcal. Chem., 1935,173, 353.60 For applications, see H. S . Harned et al., J. Amer. Chem. Soc., 1932, 54,3112; 1933, 55, 2194, 2206, 4496; 1935, 57, 1873; 1937, 59, 1280, 3033,2304GLASSTONE : ELECTROCHEMISTRY. 105as the temperature is increased : this fact led to the developmentof the equationlogK - log K, = - a(t - 0)2 .. . * (17)relating the dissociation constant K at the temperature t to themaximum value K , at the temperature 0, the quantity a being auniversal constant. This equation appears to be applicable toglycine, alanine, and formic, phosphoric (second stage), propionic,chloroacetic, sulphuric (second stage),61 b u t y r i ~ , ~ ~ lactic,52and glycollic 53 acids.It has been suggested that the agreement of equation (17) with theexperimental results is to some extent fortuitous; by means of asemi-theoretical treatment, involving the empirical facts that theentropy change and the change in heat capacity for first-stageionisations are constant, the equationlogK = A + B / T - C b g T .. . (18)has been deduced.62 This can be shown to be virtually identicalwith equation (17), provided t and 0 be not too far from 25"; withwater, however, for which the maximum dissociation constantprobably occurs near 300°, the two equations give different results,equation (18) being preferable.If dissociation constants are determined over a range of tem-pcratures, then application of the van 't Hoff isochore gives the heatof ionisation ; the appropriate calculations have been made forvarious acids 43-53 and for water ; 6o and in several cases the resultshave been found to be of the same order as those determined calori-metrically.62 The heat of ionisation of water is of particular interestas it is numerically equal to the heat of neutralisation of a strongacid and strong base in dilute solution; values have been estimatedfor a series of temperatures from 0" to 60°, the results being probablymore accurate than the best calorimetric values.Activity Coeficients obtained by using Celts with Transport.-The E.M.F.method for the determination of the activity coefficientof an electrolyte in dilute solution is one of the most valuable, but ithas been limited in its application by the necessity of setting up acell " without transport " having its two electrodes reversible withrespect to the two ions of the electrolyte. Although this can bedone in many cases, it is not always possible, e.g., for nitrates, orconvenient, e.g., for salts of alkaline earth and other metals.A newmethod has, however, been developed by D. A. MacInnes 63 whereby81 H. S. Harned and N. D. Embree, J. Amer. Chem. SOC., 1934, 56, 1050,2797 (correction).63 A. S. Brown and D. A. MacInnes, ibid., 1935, 57, 1356; T. Shedlovskyand D. A. MacInnes, ibid., 1936,58,1970; 1937,59,603; D. A. MacInnes andA. S. Brown, Chsm. Rewkws, 1936,18, 335.62 K. 8. Pitzer, ibid., 1937, 59, 2365106 GENEXAL AND PHYSICAL CHEMISTRY.the activity of an electrolyte can be determined by means of cells" with transport,'' provided the transport numbers of the ions overthe concentration range to be studied are known. If the transportnumbers are constant, the E.M.F. of a cell of the typeAglAgNO,(c) i Ag~O,(c,)lAgis given by E = (2n&T/P) loge a/al, where n#a is transport numberof the anion, and a and al are the mean ionic activities in the twosolutions of silver nitrate.In actual practice the transport numbersare not constant, however, so the equation for the E.M.P. must bewrittenE =or dE =where f is the mean activity coefficient in terms of volume concen-tration, i.e., a = fc. The transport number at any concentrationmay be expressed aswhere nl is the value at some reference concentration cl. If thisvalue for n, is substituted in equation (20) and the resulting expres-sion integrated between the limits cl, i.e., the reference concentration,and any concentration c, it follows thatna=nl+An . . . . . (21)./A%. d log, c - '1 An.d(Alog,f) . . (22)n1 c1 n1 ClThe term A log,f is equal to bgef - Iog,fi, where fi is the meanactivity coefficient in the reference solution and hence may be takenas constant. The first two terms on the right-hand side of equation(22) may be evaluated directly from the experimental data, the thirdis obtained by graphical integration of An against log,c, from theknown variation of transport number with the concentration, andthe fourth term, which is small, is computed by similar integrationof An against A log, f, preliminary values of the latter quantity beingobtained from the first three terms of the equation.To convert theA logef values into actual activity coefficients based on the usualstandard, i.e., the coefficient approaches unity as the dilution isincreased, use is made of the Debye-Hiickel equation applicable todilute solutions, vix., - log,f = a f i / ( l + p&), where a is a uni-versal constant and p an adjustable one depending on the size of thQLASSTONE : ELECTROCHEMISTRY.107ions.constant, this equation may be re-written asFor dilute solutions, therefore, the plot of A log,f + C& against( A - Alogef)dF should be a straight line with intercept A andslope p; the value of A , which is required for the plot, is obtainedby a short series of approximations. The method has been used todetermine the mean activity coeficients of the ions in sodium,potassium, calcium, and hydrogen chloride, and silver nitrate solu-tions. When the E.M.F. of the concentration cell can be expressedin terms of a simple function of the concentration, as is sometimesthe case, direct substitution may be made in equation (20), which isthen integrated between the concentration limits of 0 and c, thecorresponding values of the activity coefficient being 1 andf.Thisprocedure has been used for silver nitrate solutions up to O - ~ N . ; 64the results, which do not involve the Debye-Hiickel equation, are inexcellent agreement with those obtained by the previous method.Cathodic and Anodic Phenomena.-Hydrogen overvoltage. Therehas been in recent years a revival of interest in the problems ofhydrogen overvoltage, partly because of the intrinsic importance ofthe subject and partly on account of its possible relationship to theseparation of hydrogen and deuterium by electrolysis.66 It isgenerally agreed that for a given metal the overvoltage (q), i.e.,the difference between the potential a t which hydrogen gas isliberated from a cathode and the theoretical, or reversible, value inthe same electrolyte, is related to the current density (C.D.), repre-sented by I, by the equationwhere a and b are constants.According to the older theories, whichattributed the overvoltage to the slowness of the reaction 2H --+ H,at the electrode, b should have a value of 2.302 x HeT/2P, Le.,0-029 at 17", whereas the modern viewpoint, which considers theslow stage to be H' + E + H,* i.e., the union between the ion andan electron is slow, requires b to be equal to 2.302 x 2RT/P, i.e.,0.116 at 17". At silver, gold, copper, mercury, and nickel electrodesin dilute acid, values of b approximately 0.12 have been found atordinary temperatures,66 but there is no general agreement on thisSince A IOgef = logef - log,fl = log,f + A , where A is aA log,$ + adC= A + P(A - Alog,f)dc .. . (23)r = n + b l o g I . . . . . . (24)64 D. A. MacInnes and A. S. Brown, Zoc. cit., ref. (63).G 5 See, e.g., J. A. V. Butler, 2. Elektrochern., 1938, 44, 55.66 E. Baars, Sitzungsber. Cfes. Bef6rd. Naturw. Marburg, 1928, 63, 213;F. P. Bowden and E. K. Rideal, Proc. Roy. SOC., 1928, A , 120, 59; S. Levinaand V. Sarinski, Acta Physicochim. U.R.S.S., 1937, 6, 491.* Although written H', it is probable that in aqueous solution the hydrogenion is H,O'108 GENERAL AND PHYSICAL CHEMISTRY.matter for b values from 0.072 to 0.126 and between 0.055 and 0-075have been reported for a mercury cathode in 2~-hydrochloric acidand 2~-sulphuric acid re~pectively,~~ whereas for copper b has beenfound to vary from 0.070 t o 0.116, and a figure as high as 0-3 has beenrecorded;6* high values have also been found for lead in acetic acidand for tantalum in sulphuric acid sol~tions.~g On the other hand,low values, vix., 0-025 and less, have been obtained for metals of lowovervoltage, e.g., platinum and palladium,66* 689 70 but there is atendency for them to increase with continued use of the cathodewhich becomes partly poisoned. The potential of a platinumcathode in potassium hydroxide solution increases with time, butwhen a steady condition is reached the b value is stated to be 0.11-0.12.71 This result should, however, be accepted with reserve, forthe experimental arrangement is open to criticism : not only didoxygen from the anode appear to have access to the cathode, butthe “ direct method ” of measurement was used with currents whichmust have exceeded 0.5 amp., thus introducing appreciable resistanceerrors. Mention may be made in this connexion of the developmentof a novel technique for the measurement of the potentials of polarisedelectrodes whereby this source of error is avoided.72It was indicated in a previous Report 73 that more than onemechanism appears to be necessary to account for all the facts ofovervoltage, and the results recorded above support this view. Forhigh overvoltage metals it seems probable that, at least at relativelyhigh C.D.’s, the discharge of hydrogen ions is the process whichdetermines the rate of hydrogen evolution, but at low C.D.’s, andespecially for metals with low overvoltages, e.g., platinum andpalladium, another process is the determining factor.When theenergy of adsorption of hydrogen atoms on a metal is appreciable,then the atoms can be deposited at a lower potential than thatrequired for the discharge of “ free ” hydrogen.65’ 74 If the rate ofdesorption of hydrogen from the surface in the form of molecules isG 7 St. von Nhray-Szabo, Naturwiss., 1937, 25, 12.6 8 K. Wirtz, 2. physikal. Chem., 1937, B, 36, 435.70 L. P. Hammett, J . Amer. Chem. SOC., 1924, 46, 7 ; C. A. Knorr and E.Schwartz, 2.Elektrochem., 1934, 40, 38; 2. physikal. Chem., 1936, 176, 161 ;M. G. Raeder and K. W. Nilsen, quoted by C. A. Knorr and E. Schwartz,Zoc. cit., 1936; see A., 1936, 1207; see also L. Iiandler and C. A. Knorr, 2.Elektrochem., 1936, 42, 669.T. Erdey-Grtiz and H. Wick, ibid., 1932, A, 162, 53.71 G. Masing and G. Laue, 2. physikal. Chem., 1936, 1’48, 1.7 2 A. Hickling, Trans. Faraday SOC., 1937, 33, 1540.73 Ann. Reports, 1933, 30, 37.74 J. A. V. Butler, Proc. Roy. SOC., 1936, A, 157, 423; cf. J. Horiuti and M.Polanyi, Acta Physicochim. U.R.S.S., 1935, 2: 505; J. Horiuti et al., Sci.Papers Inst. Phys. Chem. Res. Tokyo, 1936, 29, 223GLASSTONE : ELECTROCHEMISTRY. 109rapid, then the overvoltage will remain low ; in these circumstancesit can be shown that the overvoltage is a linear, instead of a logarith-mic, function of the C.D.75 At higher C.D.’s, and particularly withmetals for which desorption is slow, the discharge of ions to formadsorbed atoms can no longer keep pace with the requirements of thecurrent, the surface becoming saturated, and an alternative process,viz., the formation of “ free ” atoms, at a higher potential mustensue.The overvoltage will now vary with C.D. according toequation (24), b being approximately 0.12 at ordinary temperaturesas is frequently found.The slow change in cathode potential with time, which is a familiarphenomenon of overvoltage, has been attributed to the slow rate ofattainment of adsorption saturation ; 76 some writers have ascribedthe observation to changes in the electrode surface, e.g., inactivationof active centres,71 but this is regarded as improbable, for the effecthas been observed with a liquid mercury cathode.It should beemphasised that modern theories of overvoltage do not take suffi-ciently into account the influence of interfacial forces, especially forhigh-overvoltage metals, resulting from changes in the nature of theelectrolyte; it has been known for several years that certain batho-tonic substances lower overvoltage 77 and this ha’s been confirmedrecently for mercury with a solution of hydrochloric acid in ethylalcohol as electrolyte.78A theory of overvoltage which differs completely from thosedescribed above has been proposed to account for observalions madewith a dropping mercury cathode in electrolytes containing light andheavy water in various proportions : 79 this theory supposes theformation of molecules of hydrogen to take place through inter-action of the deposited atoms and hydrogen ions in solution.Since the latter are formed by the ionisation of water molecules, therate of this process, as indicated by the ionic product, is of import-ance.On the basis of these postulates an equation is deduced relatingthe overvoltage to the isotopic composition of the water. It isconcluded that the hydrogen-deuterium separation coefficientshould have a mean value of 5.4, the ratio of the ionic productsof the two isotopic forms of water, which should decrease to 2.7 inordinary water and increase to 50 in concentrated heavy water;76 J.A. V. Butler and G. Armstrong, J., 1934, 743; M. Volmer and H.Wick, 2. physikal. Chem., 1935,172, 429.7g St. von Naray-Szabo, ibid., 1937, 178, 356.7 7 See S . Glasstone, “ The Electrochemistry of Solutions,” 1937, pp. 427,7 8 S . Levina and M. Silberfarb, Acta Physicochim. U.R.S.S., 1936, 4, 275.7Q J. Nov&k, Coll. Czech. Chern. COM., 1937, 9, 207; J. Heyrovskf, ibid.,431.pp. 273, 346110 GENERAL AND PHYSICAL CHEMISTRY.this tvpe of variation with composition appears to be contrary toexperiment. 8oAnodic Oxidation. It has been generally accepted that in pro-cesses of electrolytic oxidation, as well as in reduction, each definiteelectrode potential stage corresponds to a different process : it is,therefore, somewhat unexpected to record that in the oxidation ofthiosulphate ions, in a buffered neutral solution, two stages ofpotential can be observed a t a smooth platinum anode, but thenature and efficiency of the oxidation process, vix., SOYo of tetra-thionate and 20 yo of sulphate, approximately, is independent of thepotential.81 The electrolysis commences at a potential of about0.8 volt, but after an interval of time, which is greater the smallerthe current, there is a rapid rise to a higher stage, about 1-5-1.6volt.The change of potential is not connected with impoverish-ment of the electrolyte, as has been but to changes in thselectrode, for previous anodic polarisation will cause the potentialto rise to the higher stage immediately after the commencement ofelectrolysis of the thiosulphate solution. Addition of small amountsof mercuric cyanide, e.g., 0.001~., also results in a rapid rise ofpotential, but the efficiency of oxidation is only affected to a, smallextent.Similar results have been obtained in the electrolysis ofsulphite solutions, the products being dithionate (40-60%) andsulphate (50-60y0) .83 Two different potential stages, withoutany apparent change in the nature of the products, have also beenobserved at a platinum anode in the oxidation of methyl alcohol,formaldehyde, formic acid,84 and ethyl alcohol.85In order to account for the independence of anode potential andthe oxidation efficiency, it has been suggested that hydroxyl ionsare primarily discharged at the anode and that pairs of the resultingradicals combine irreversibzy to form hydrogen peroxide : 20H' +2~ + 20H --+ H,O,.The latter can either oxidise it depolariserin the electrolyte, e.g., S,03" or SO3", or it can decompose intooxygen and water. It is supposed that this oxygen becomes asso-ciated with the platinum electrode and is responsible for its potential,the two stages corresponding to two modes of attachment. Theoxidation reactions occur independently, and the potentials are notindicative of these processes but of a simultaneous side reaction.Mercuric cyanide is strongly adsorbed, so that the platinum-oxygen80 Cf. 35. P. Applebey and G. Ogden, J., 1936, 163.81 S. Glasstone and A. Hickling, J., 1932, 2345.82 J.A. V. Butler and W. M. Leslie, Trans. Paraday Soc., 1936,32,435 (444).8s S, Glasstone and A. Hickling, J., 1933, 829.84 E. Miiller et al., 2. Elektrochem., 1923, 29, 264; 1927, 33, 561 ; 1928, 34,85 C. Marie and G. Lejeune, J. Chirn. physique, 1929, 26, 237.266, 704; S. Tanah, &id., 1929, 35, 38GLASSTONE : ELECTROCHEMISTRY. 111association of the first type does not occur; the stability of thehydrogen peroxide is, however, not appreciably affected and theanodic oxidation process is unchanged. The presence in the solutionof substances able to catalyse the decomposition of hydrogen per-oxide suppresses the oxidation efficiency and favours the rapid rise ofpotential which soon reaches that necessary for oxygen evolution.The change of efficiency resulting from the use of different electrodematerials can be accounted for in the same way.Under suitableconditions, ammonium molybdate can bring about a marked increasein the formation of sulphate when thiosulphate is oxidised : 86this is exactly analogous to the behaviour in the purely chemicaloxidation with hydrogen peroxide. Other evidence for the theoryof anodic formation of the peroxide is found in the increased propor-tion of oxygen obtained in the electrolysis of acidified solutions ofpermanganate and dichr~rnate.~’Some authors 82# 88 have preferred to consider the primary anodicprocess, in the electrolysis of solutions of S203”, SO3” and other ionsundergoing analogous oxidation, as being the discharge of theseions, the radicals subsequently reacting, e.g., S203” --+ 2~ + S20:,followed by S203 + S203” + S40s” ; or S203” __p E + S203 ,followed by 2S203’ --+ SaOs“.The influence of various addedsubstances has then been attributed to the deposition of oxides,e.g., of lead, manganese, and silver, on the anode which thus alterits nature. It is not easy to understand, however, why any changein the anode material should affect ionic discharge and the subsequentprocesses. This daculty is accentuated by the fact that the effectsare virtually the same for a number of oxidisable ions of differenttypes, e.g., sulphite, chloride, and acetate (see below). Further,some added substances, e.g., powdered silver, carbon, or iron andcopper salts, are able to influence the oxidation efficiency withoutforming any deposit on the anode. In the electrolysis of halidesolutions 89 the presence of catalysts for hydrogen peroxide decom-position produces a marked change in anode potential as well as inthe yield of halogen : it is not easy to account for these results on thebasis of the preferential discharge of halogen ions, since this is areversible process and should be independent of electrode material.The observations are, however, in harmony with the view thathydroxyl-ion discharge also occurs, leading to the formation ofhydrogen peroxide which acts as the oxidising agent. The sug-gestion has been made 88 that the behaviour of different electrode8 6 S. Glasstone and A. Hickling, J., 1932, 2800.8 7 A. Hickling, J., 1936, 1453.88 0. J. Walker and J. Weiss, Tram. Paraday Soc., 1935, 31, 1011; W. D.88 S. Glasstone and A. Hickling, J., 1934, 10.Bancroft, Trans. EEectrochem. SOG., 1937, 71, Preprint 7,53112 GENERAL AND PHYSICAL CHEMISTRY,materials in the oxidation of ions may be explained by the differencein anodic overvoltages, but this does not seem to be in harmony withthe experimental facts.90The anodic oxidation of chromic salts to chromic acid has longpresented a diflicult problem : the process occurs more readily a tlead dioxide and platinised platinum anodes than at one of smoothplatinum, in spite of the higher potential of the latter. A recentinvestigation 91 has shown that for electrolysis in acid solutionfactors which would tend to decompose hydrogen peroxide favourthe oxidation reaction : this is to be expected, since the peroxide isable to reduce chromic acid to chromic ions. The actual oxidationin these circumstances is apparently brought about by active oxygenthrough the intermediate formation of metallic peroxides. Inneutral and alkaline solutions, however, the results are quite dif-ferent for, in addition to this type of process, hydrogen peroxide canalso oxidise the chromic ions. The interesting results obtained inthe electrolysis of solutions of chromic salts can be explained in asatisfactory manner on the assumption that hydrogen peroxide isformed a t the anode, although it is not necessarily the effectiveoxidising agent, but do not seem capable of any other simple inter-pretation. It may be appropriate to emphasise here that otheranodic reactions are known in which hydrogen peroxide does notappear to be the essential oxidant.92The formation of hydrogen peroxide by the combination of hydr-oxyl radicals in the gas phase is generally accepted,93 but the produc-tion of the peroxide from hydroxyl ions at the anode in solution isnot commonly observed; this is probably the result of catalyticdecomposition by the anode material. At low temperatures,Mhowever, or in the presence of fluoride ionsY95 which act as a catalyticpoison, appreciable amounts of hydrogen peroxide have been ob-tained at platinum anodes in the electrolysis of solutions of alkalihydroxides. The formation of the peroxide, in amounts corre-sponding to the requirements of Faraday’s laws, has been observedwith a glow-discharge anode; 96 the electrode is then not immersedin the solution and catalytic decomposition is prevented.S. Glasstone and A. Hickling, Trans. Faraday SOC., 1935, 31, 1656.91 R. F. J. Gross and A. Kickling, J., 1937, 325.92 S. Glasstone and A. Hickling, “ Electrolytic Oxidation and Reduction,”1936, pp. 336-338, 350.08 W. H. Rodebush and M. H. Wahl, J . Chem. Physics, 1933, 1, 696; 0.Oldenberg, ibid., 1935, 3, 266; R. W. Campbell and W. H. Rodebush, ibid.,1936, 4, 293.84 E. H. Riesenfeld and B. Reinhold, Ber., 1909, 42, 2977.96 Riw 9 Miro, Helv. Chim. Acta, 1920, 3, 355.O 6 5. Glasstone and A. Hickling, J . , 1934, 1772; see also A. Klemenc andT. Kantor, 2. physikal. Chem., 1934, B , 2’9, 359GLAqSTONE : ELECTROCEEMISTRY. 113The Kolbe Reaction. Since the discovery in 1849 that ethane andcarbon dioxide are obtained by the electrolysis of aqueous solutionsof acetates, the electrolytic Kolbe reaction has attracted muchattention and two main theories of its mechanism have been held.These are (1) that acetate ions are directly discharged at the anode,the radicals then interacting, thus : 2CH3-CO*O* + C2H, + 2C0, ;and (2) that active oxygen is first produced and this oxidises theacetic acid or acetate ions to acetyl peroxide which then decomposesto give ethane and carbon dioxide. The supporters of the respectivemechanisms have put forward a large amount of experimental evid-ence in favour of their apparently opposing views : 97 it is, therefore,of interest to mention that a comprehensive theory, involving boththe above, together with the concept of the anodic formation ofhydrogen peroxide, has been proposed.g8 To explain a, variety ofobservations, e.g., influence of anode material and of added sub-stances, it is suggested that in aqueous solution the hydrogen per-oxide first produced reacts with the acetate ions to form acetateradicals ; these combine in pairs yielding acetyl peroxide, whichsubsequently decompose^,^^ thus :H,O, + BCH,*CO*O’ --+ 20H’ + 2CH3*CO*O* +In the presence of catalysts for hydrogen peroxide decomposition, orif the acetate-ion concentration at the anode is kept low, e.g., whenacetic acid solutions are electrolysed in the presence of small amountsof neutral salts, an alternative process occurs. The hydrogen peroxide,or active oxygen, oxidises the acetic acid, or acetate ions, to peraceticacid, which on decomposition gives methyl alcohol :The formation of this substance, known as the Hofer-Moest reaction,has been confirmed under the conditions mentioned. In non-aqueous solutions the formation of hydrogen peroxide is not possible ;direct discharge of the acetate ion must then occur, with the conse-quent formation of acetyl peroxide and finally the decompositionproducts of the latter, thus :ZCH,-CO*O‘ + 2~ + 2CH3*CO*O* --+A side reaction which sometimes occurs is oxidation of the non-aqueous solvent by the acetyl peroxide, thus decreasing the yield of(CJ&j*CO*O*),+ C2HG + 2C02.CH3*CO*OH + [O] .+ CH,*CO*O*OH + CH,*OH + CO,.(CH3*CO*O*), --+ C.3, + 2C0,.Q7 For summary, see S. Glasstone and A. Hickling, op. cit., Chapter VIII.98 Idem, J., 1934, 1878; 1936, 820.SQ For evidence collected by F. Fichter and collaborators, see S. Glasstoneand A. Hickling, op. cit., pp. 293-296 ; also, H. Wieland et al., Annalen, 1934,513, 93; 0. J. Walker and G. L. E. Wild, J . , 1937, 1132114 GENERAL AND PHYSICAL CHEMISTRY.ethane. Amongst the many achievements of the theory may bementioned its interpretation of the hitherto perplexing fact thatalthough the Kolbe reaction at a platinum anode takes place equallywell in aqueous and non-aqueous solutions , with a gold electrodethere is no formation of ethane in aqueous solution although it isproduced with a high efficiency in non-aqueous media. In aqueoussolution a gold anode becomes covered with a layer of oxide which is avery effective catalyst for the decomposition of hydrogen peroxide,so that the Kolbe reaction is inhibited ; the oxide, however, appearsto have little influence on the stability of acetyl peroxide and hencethe formation of ethane should occur in non-aqueous solutions.Some observations have been made recently of the electrolysis ofacetic acid and potassium acetate in deuterium oxide, and ofdeuterium trideuteroacetate and sodium trideuteroacetate in water ;1only when the deuterium is present in the acetic acid, or acetate, doesthe ethane evolved contain any appreciable amount of this isotope.The results appear to throw little light on the mechanism of theKolbe reaction; they merely show that if methyl radicals are anintermediate product in the formation of ethane they do not reactwith the solvent. Electrolysis of different deuteropropionic acids 2may, however, provide a clue as to the origin of the ethylene, whichis one of the chief products in aqueous solution : this is, however, adeviation from the usual type of Kolbe reaction. S. G.J. A. V. BUTLER.S. GLASSTONE.M. RITCHIE.W. F. K. WYNNE-JONES.1 H. Erlenmeyer and Vi. Schoenauer, Helv. ChE.irn. Acta, 1937, 20, 222;3 Idem, Ber., 1937, 70, 819.P. Holemann and K. Clusius, 2. physikal. Chern., 1937, B, 35, 261