ANNUAL REPORTSON THEPROGRESS OF CHEMISTRYGENERAL AND PHYSICAL CHEMISTRY1. INTRODUCTIONBy P. G. Ashmore(Department of Chemistry, Manchester University Faculty of Technology)ON the scores of scientific importance and research activity there are manytopics in General and Physical Chemistry that qualify for inclusion in eachissue of Annual Reports. The senior Reporter’s main task is to assess theseclaims against the available pages and the often fortuitous availability atthe time of Reporters for the chosen topics. It is extremely fortunate thatthere always seem to be enough able and willing authors to undertake thecollection of material throughout the year and to exert the very necessarycritical judgment at Christmas. There are some topics, however, whichhave not been dealt with in Annual Reports for such a long time thatrestarting them is a most formidable and daunting task.I would like tobe able to say that Dr. Haydon’s able article on liquid-liquid interfacesis the first of a revival group; there is no lack of work and interest insurface chemistry and there is a real need for frequent reports of a generalrather than a specialist kind. Unfortunately, the task of assessing theadvances and putting present work into focus on all fronts of surfacechemistry is too great for any one contributor, and provisional arrangementsfor further articles on rather more limited fronts suffered a severe setbackby the tragic death of Dr. Kipling. Nevertheless, I hope that surfacechemists will find some way of filling, before it is too late, a notable gapin the coverage of the past decade of Annual Reports.Three of the articles in this year’s Report cover a wide range of activityin reaction kinetics.The resurgence of work on direct investigations ofindividual elementary reactions of free atoms and radicals has receivedspecialist reports in various series of Advances and Progresses, but the moregeneral account of Drs. Thrush and Campbell will not only allow a widercircle of chemists to view the progress, but also provide a very valuableand convenient summary of rate data for kineticists.Two closely relevant types of investigation of elementary reaction (ion-molecule reactions and reactions in crossed molecular beams) are discussedby Dr. Henchman; as he points out, these extremely valuable and growingtechniques have been rather neglected in this country.Professor Bradleycontributes an extremely clear account of the coming-of-age of shock-waveinvestigations of chemical reactions, now that improvements in technique8 GENERAL AND PIIYSICSL CHEMISTRYallow us to take proper advantage of the high temperatures and homo-geneous conditions of the shocked gas to obtain accurate kinetic data.Dr. Collinson’s article on radiation chemistry reports a growth rate thatwould be enviable in many quarters of our national economy, but gavehim a particularly large field to cover. He has made a judicious selectionof work on some organic systems and on some aqueous solutions.The final article by Drs.Inman and White provides a very valuable assess-ment of the rapidly expanding work, both theoretically and industrially,on molten salts, a field where the interests and aims of physical and inorganicchemistry imperceptlibly merge2. LIQUID-LIQUID INTERFACESBy D. A. Haydon(Department of Colloid Science, University of Cambridge,Free School Lane, Cambridge)Introduction.-Liquid-liquid interfaces are usually discussed and re-viewed in association with air-water interfaces. This Report is thereforea departure from tradition which, whatever the disadvantages, at leastgreatly facilitates the presentation of the material. The work specificallycovered is that published during 1963-1965 although earlier publicationshave been mentioned where they are thought to be helpful. The occurrenceof liquid-liquid interfaces in emulsions and thin films, in many biologicalsystems, and in mass-transfer cells and columns makes the whole field verywide.Descriptions of various aspects of liquid-liquid interfaces at theintroductory level have been given by Davies and Rideall and, in the fieldof emulsions, Becher has produced a second edition of his book.2 Reviewscovering a number of aspects of liquid-liquid interfaces have appeared ina two-volume collection of article^.^ Although it is now t,hree years oldthe bibliography of work on liquid-vapour and liquid-liquid interfacescompiled by Stephens4 must also be mentioned.In the main part of this Report we shall discuss only systems in whichnew information has been obtained regarding the structure and propertiesof liquid-liquid interfaces.Of general interest, however, is the excellentwork of Princen and Mason on the shapes of fluid drops at fluid-liquidinterfaces in two- and three-phase systems 5 and also the recent work onthe formation and properties of thin hydrocarbon films under aqueoussolutions.6 In the stability of these films, which are approximately 60 Ain thicl~ness,~ we have a direct demonstration of the normal orientationand steric hindrance to interpenetration of adsorbed monolayers a t hydro-carbon-water interfaces, and a means to test in detail the hypothesis ofcolloid stability in hydrocarbons proposed by Mackor and Van der Waals.8Measurement of Interfacial Tension.-In most would-be accurate workJ.T. Davies and E. K. Rideal, “ Interfacial Phenomena, ” Academic, Press,Sew York and L:vdon, 1963.P. Becher, Emulsions, ” Rheinhold, New York, 1965.“ Recent Progress in Surface Science, ” Academic Press, New York and London:Vols. 1 and 2, eds. J. F. Danielli, K. 0. A. Pankhurst, and A. C. Riddiford, 1964. * D. W. Stephens, “ Gas/Liquid and Liquid/Liquid Interfaces-A Bibliography, ”Joseph Crosfield and Sons, London, 1962.H. M. Princen and S. G. Mason, J . Colloid Sci., 1965, 20, 156, 246.I. Langmuir and D. F. Waugh, J . Gen. Physiol., 1938, 21, 745; P. Mueller, D. 0.Rudin, H. Ti Tien, and W. C. Wescott, Nature, 1963, 194, 979; Circulation, 1962,Pt. 11, 26, 1107; J . Phys. Chem., 1963, 67, 534; P. Mueller and D. 0. Rudin, J .Theoret.Biol., 1963, 4, 268; T. Hanai, D. A. Haydon, and J. Taylor, J . Gen. Physiol., 1965,48, 59; C. Huang and T. E. Thompson, J . MoE. Biol., 1965, 13, 183; T. Hanai, D. A.Haydon, and J. Taylor, J . Theoret. Biol., 1965, 9, 278.C. Huang, L. Wheeldon, and T. E. Thompson, J . MoE. Biol., 1964, 8, 148;T. Hanai, D. A. Haydon, and J. Taylor, KoEloid-Z., 1964, 195, [l], 41; T. Hanai, D. A.Haydon, and J. Taylor, PTOC. Roy. Soc., 1964, A , 281, 377.E. L. Mackor and J. H. Van der Waals, J . Colloid Sci., 1952, 7, 53510 GENERAL AND PHYSICAL CHEMISTRYon liquid-liquid interfaces in recent years the drop-volume method ofinterfacial-tension measurement has been used. This entails the use of theempirical corrections proposed by Harkins and Brown for liquid-vapoursystern~.~ The measurement of the surface tension of pure liquids by thismethod is known to be reasonably accurate through comparison with resultsfrom the capillary rise and other techniques. Such confirmatory evidenceis, for obvious reasons, much more difficult to obtain for liquid-liquidinterfaces.It is therefore satisfying to note that quite close agreementbetween the drop-volume lo and the Wilhelmy plate l1 techniques has beeiiobtained for the n-heptane-water interface. A definitive check on theaccuracy of the drop-volume method for liquid-liquid interfaces is never-theless badly needed as it is evidently one of the most accurate of the moreconvenient methods. Another convenient method, that involving the duNouy ring, has recently been re-examined for liquid-liquid interfaces.l2 Asfor the Wilhelmy plate, zero contact angle is required between the measur-ing surface and one of the liquid phases.Rings wetted by aqueous solutionsare unsatisfactory owing to the influence of surface-active solutes on thecontact angle. It has been reported that if, however, the rings are madeoil wettable by coating with Teflon or polyethylene, the interfacial tensionsare considerably more reliable.l2 The du Nouy method nevertheless stilldoes not appear to be capable of such high accuracy as the drop-volumemethod. The development of the Wilhelmy plate technique for liquid-liquid interfaces has continued. Problems of buoyancy and contact anglefor oils on aqueous solutions of surface-active substances have now beensolved to the extent that interfacial tensions reproducible to h0.1 dynecm.-l have been reported for these systems.11, l3Interfaces Between Pure Liquids.-The investigation of the thermody-namic properties of liquid-liquid interfaces is usually complicated by thesignificant mutual solubility of the two phases.Thus, the interpretationof the temperature coefficients of the interfacial tensions to give interfacialexcess heats and entropies is by no means straightforward.l* The aliphatichydrocarbons against water, however, constitute one of the simpler inter-faces from this point of view, as the mutual solubility can, for most purposes,be neglected. Two investigations of the interfaces between normal aliphatichydrocarbons and water have been reported during the last five years.l5, loThe former is concerned only with hexane a t different temperatures butthe latter gives interfacial tensions at a range of temperatures for six hydro-carbons from n-hexane to n-hexadecane.From these data interfacial excessheats and entropies have been calculated, One of the more curious featuresof the results lies in the comparison of the entropies for hexane obtainedby the two sets of authors. Franks and Ives15 found a sharp maximumin the interfacial excess entropy at 34” while Aveyard and Haydon lo show* W. D. Harkins and F. E. Brown, J . Amer. Ohm. SOC., 1919, 41, 499.lo R. Aveyard and D. A. Haydon, Trans. Faraday SOC., 1965, 81, 2255.11 J. H. Brooks and B. A. Pethica, Trans. Faraduy Soc., 1964, 80, 208.J.A. Krynitsky and W. D. Garrett, J. C O W Sci., 1963, 18, 893.J. H. Brooks and Bc: A. Pethice, Tram. Paraday Soc., 1965, 61, 671.14 E. A. Guggenheim,15F. Franks end D. J. G. Ives, J . Chem. SOC.. 1960, 741.Thermodynamics,” North Holland Publ. Co., Amsterdam,1969HAYDON LIQUID-LIQUID INTERFACES 11entropies which are independent of temperature and similar for all thehydrocarbons. The maximum in the entropy was attributed to a changein the structure of water a t 34". However, it was found by Aveyard andHaydon that if 30% (v/v) n-octane was mixed with the n-hexane a pointof inflexion in the tension-temperature curve, giving a maximum in theentropy, appeared at about 34". On this evidence it seems likely that themaxima in the interfacial excess entropies originate in the hydrocarbonrather than in the water phase.Whether or not impurities in the hexaneused by Franks and Ives are the explanation of the discrepancy, the pheno-menon is of considerable interest and worthy of further investigation. Theinfluence of chain branching in the hydrocarbon on the properties of theinterface have been investigated by using 2,2,4-t1imethylpentane.~~ Whilethe tensions are significantly lower than for n-octane, the temperaturecoefficient of the tension was constant between 20" and 30" and equal tothat for the normal chain hydrocarbons.The poor prospects for the calculation of liquid-liquid interfacial tensionsfrom fundamental data is underlined by recent work on liquid-vapoursystems.ls Some investigators have, however, been concerned with theprediction of interfacial tensions from the values for the correspondingliquid-vapour interfaces. Fowkes l7 has argued that if two liquids interactonly through London dispersion forces then the interfacial tension, y12,should be given byY l 2 = 7 1 3- Y2 - (YldYzd)+ (1)where yf and y$ are the contributions of the dispersion forces to thetensions of the pure liquids.For saturated aliphatic hydrocarbons thewhole of the liquid-vapour tension is assumed to originate in dispersionforces and yd is equated to y. For interfaces between saturated aliphatichydrocarbons and other liquids, equation (1) therefore becomesY l 2 = Y1 + Y2 - (YldY2)* (2)From the surface tensions of various hydrocarbons and their interfacialtensions against wafer, the contribution of dispersion forces to the surfacetension of water has been estimated.A value of 214& 0.7 dynes cm.-lwas found for yla while y z changed from 18.4 to 29.9 dynes cm.-l. Inprinciple, therefore, the interfacial tension of any other non-polar liquidagainst water may be found. The equation works moderately well ifunsaturated hydrocarbons, aromatics in particular, are excluded. Themercury-water tension is predicted to be 425 & 2 dynes cm.-l as comparedwith the experimental value of 426 & 7 dynes cm.-l.Monomoleculaz Films At Liquid-Liquid IntePfaces.-These fihs may besubdivided into three main types which, in order of increasing complexity,are those of non-electrolytes, electrolytes, and polymers.At oil-water, andparticularly hydrocarbon-water, interfaces the behaviour of these films isrelatively uncomplicated by the large effects, found at air-water interfaces,which originate in the interactions between the non-polar parts of the filmmo1ecules.l To a fist approximation, the interfacial pressure of a film ofl6 T. S . Ree, T. Ree, and H. Eyring, J. Ohem. P h y ~ . , 1964, 41, 624,17F. M. Fowkes, J . Phys. Chem., 1962, 66, 382; 1963, 67, 263812 GENERAL AND PHYSICAL CHEMISTRYsmall moleciiles between dilute solutions can be predicted by means of asurface equation of state from a knowledge of the interfacial density ofthe film molecules. The form of these equations has remained a topic forlively discussion during the last few years.181 19Non-eZectroZytes.Little work has recently been reported on spread oradsorbed films of these substances at oil-water interfaces. Some data onwhat is probably the simplest of these systems, the normal aliphatic alcoholsadsorbed at aliphatic hydrocarbon-water interfaces, was reported in 1960.2OIn this work the interfacial tensions of dilute solutions of some lower alcoholsin water (equilibrated against the hydrocarbon phase) were measured asa function of alcohol concentration. The Gibbs adsorption equation wasused to calculate interfacial excess concentrations. In these systems theinterfacial excess concentrations are almost equal to the actual interfacialconcentrations, and thus it was possible to plot interfacial pressure againstinterfacial area per alcohol molecule.The results were closely representedby a two-dimensional Van der Waals equation with a zero intermolecularattraction term,I7(A - A,) = kTwhere 17 is the interfacial film pressure, ,4 is the interfacial area per inoleculeand A , is the value of A a t 17 = 00. The value of -4, was found to be18.5 A2 per molecule, in good agreement with comparable results froininsoluble films of alcohols at air-water interfaces. In this work, however,the experimental data were not of the highest precision, activity coefficientswere assumed to be equal to unity and the aliphatic hydrocarbon was amixture of straight- and branched-chain saturated molecules. With theinclusion of the activity coefficients and improvement of the accuracy theresults would nevertheless probably not change by more than a few percent. Subsequent work on dilute solutions of n-hexadecanol in n-heptaneagainst water has, subject to the neglect of activity coefficients, givensimilar good agreement with equation (3) with A , = 18.5 A2 per mole-cule.2l At the benzene-water interface equation (3) is also obeyed toa first approximation by n-propanol, n-decanol, and n-hexadecanol( A , = 18.5 &- 1 A2 per molecule) and n-heptyl acetate (A, = 43 & 4 A2 permolecule) .21 In this work, again, the activity coefficients were not availableand were assumed to be unity. More recently, alcohols from n-pentanol ton-tetradecanol were examined at the n-octane-water interface at tempera-tures ranging from 20" to 50°.22 The results are for rather high concentra-tions where the activity coefficients are not only unknown but are likely tohe appreciably less than unity.Until these activity coefficients are availablea test of equation (3) will not be feasible.While equation (3) is evidently applicable only to a two-dimensional18 G. M. Bell, S. Levine, and B. A. Pethica, Truns. Farday Xoc., 1962, 58, 904.19 F. M. Fowkes, J . Phgp. Chern., 1962, 66, 385; E. H. Lucmsen-Reynders andM. Van den Tempel, Proc. 4th Internat. Congr. Surface Activity, 1964 (in press);E. H. Lucassen-Reynders, Nature, 1966 (in press).20 D. A. Haydon and F. H. Taylor, Phil. Truns., 1960, 252, A , 225.21 G. T. Rich, Thesis, Cambridge, 1964.22 J.J. Jasper and B. L. Houseman, J . Phys. Chern.. 19G3, 67, 1548; 1965, 69,310; J, J. Jasper and R. D. Van Dell, ibid., 1965, 69, 481HAYDON : LIQUID-LIQUID INTERFACES 13gas it does, in fact, hold remarkably accurately for many films of non-electrolytes a t oil-water interfaces. Little attention has, however, beengiven to the theoretical basis of equation (3) for liquid-liquid interfaces.In principle the partition function for the interface from which the equationis derived must take some account of the nature of the solvent molecules,as, for instance, in the theory of the vapour interfaces of two-componentliquid mixtures.= No treatment of this type has, however, yet appeared.Some recent, but unpublished results of Aveyard, have shown that theadsorption of n-butanol a t water-n-aliphatic hydrocarbon interfaces isaffected to a small but significant extent, and in a systematic manner, byvariation of the chain length of the hydrocarbon.Electrolytes.Monomolecular films of ions a t liquid-liquid interfaces andthe problem of the structure of the associated electrical double layers werereviewed in 1964,24 although the publications covered were those up to1961 only. At this stage it was known that monolayers of strongly surface-active strong electrolytes such as alkyl trimethylammonium bromides andsodium alkyl sulphates obeyed, to a first approximation, the surface equationof state obtained by combining equation (3) with a term for the electro-static free energy of the interfa~e.~5 In its original form this latter termwas derived directly from the Gouy-Chapman theory for the potential ata planar impenetrable charged interface where all the ions were pointcharges.At high interfacial charge densities, however, experiments indi-cated that the interfacial pressures were seriously underestimated by thisequation of state 20, l3 and a theoretical analysis suggested that the principalreason for this was the neglect, in the expression of electrostatic free energy,of the finite size of the monolayer ions.26 More recently, attention has beencentred on the importance of the discreteness-of-charge effect in mono-molecular films a t liquid-air and liquid-liquid interfaces and an at tempthas been made to estimate the influence of this phenomenon on the inter-facial film pressure.18 Unfortunately, there are other corrections to theelectrostatic treatment which are of doubtful magnitude 24 and which tendto frustrate experimental checks of any specific correction. In short, thereare too many unknown, and too few measurable parameters.The ultimateneed for accurate experimental relationships between interfacial film pressureand interfacial area per ion has nevertheless stimulated some very carefulwork on both soluble and '' insoluble " ionic films at hydrocarbon-waterinterfaces-11, l3 The latter systems were studied by means of a liquid-liquid Langmuir trough technique, where the insoluble film molecules wereconfined with barriers and the interfacial pressures were measured by aWilhelmy plate meth0d.l' One particularly interesting conclusion has beenreached as a consequence of the repetition of earlier work on (insoluble)films at aliphatic hydrocarbon-water interfaces of equimolar mixtures of23 I.Prigogine and J. Marechal, J . Colloid Sci., 1957, 7 , 122; J. W. Belton andM. G. Evans, Trans. Puraduy SOC., 1945, 41, 1.z p D. A. Haydon, in " Recent Progress in Surface Science ", Academic Press, NewYork and London, 1964, Vol. 1, eds. J. F. Danielli, K. G. A. Pankhurst, and A. C.Riddiford, p. 94.25 J. T. Davies, Proc. Roy. SOC., 1951, A , 208, 224.26 D. A. Haydon and F. H. Taylor, Phil. Tram., 1960, 253, A , 25514 GENERAL AND PHYSICAL CHEMISTRYsodium octadecyl sulphate and octadecyl trimethylammonium bromide.11, 27At relatively high interfacial areas per ion equation (3) is closely obeyed,but at 57 A2 per ion a sudden condensation of the iilm occurs which con-tinues until it becomes almost incompressible at approximately 32 8 2 perion.The authors attribute this condensation to electrostatic interactionbetween the positive and negative film ions.An approach to electrical double-layer structure a t hydrocarbon-waterinterfaces which is complementary to the measurement of interfacial j i hpressures? is that of interfacial (contact) potential measurement. A crucialquestion in the setting up of an electrostatic model for ionized monolayersis the probability of the occurrence of counter-ions in the plane of, and inthe region behind the centres of the film ions.Some evidence has beenobtained from measurements of the interfacial potential 20, 28-30 which con-firms past suggestions that there are usually significant numbers of counter-ions in these regions.20, 25, 31 Interfacial-potential data have also been usedto give indications of the water dipole orientation in the presence of ionizedmonolayers a t n-decane-water interface^.^^ -30 Systems involving sodiumdodecyl sulphate and sodium chloride and dodecyl trimethylammoniumbromide and sodium chloride have been examined 28, 29 and, more recently,data on monolayers of cetyl trimethylammonium ions in presence of F-,C1-, Br-, I-, NO3-, and SCN- have been reported.30 Attempts to breakdown the net dipole term in the interfacial potential so as to give waterorientation a t the hydrocarbon surface and round the ionic groups haveyielded an apparently reasonable qualitative picture.A paper by Koenighas drawn attention to the need for care in the interpretation of surface-potential measurements and some indication of the complications that mayoccur in the interpretation of surface potentials for strongly dipolar mole-cules has recently appeared.33 For liquid-liquid interfaces the measurementof contact potentials has so far been achieved by means of placing it gold-plated vibrating electrode in the hydrocarbon phase. This techniqueobviously breaks down if there are ions or molecules present which aresoluble in the oil phase and hence are liable to adsorb on to the goldelectrode. A development which may help to overcome this dif6culty hasbeen reported by Zisman and his collaborators34 and consists of coatingthe electrode with Teflon.These authors used this electrode for studiesof films of volatile organic molecules on platinum where the electrode wasin the vapour phase, but it is possible that, in an oil phase, the adsorptionof many types of molecule on to Teflon may also be of negligible importance.The interpretation of electrokinetic potentials at liquid-liquid interfaceswas examined by several workers about five years ago. The systems studied27 J. N. Phillips and E. K. Rideal, Proc. Rog. Soc., 1955, A , 232, 149.2eD. A. Haydon, KoZloid-Z., 1962, 185, [2], 148.29 D. A. Haydon, KoZloid-Z., 1962, 187, [2], 146.a0 S. Minc and Z.Koczoromki, Roczniki Chem., 1966, 89, 469.31 A. J. Payens, Thesis, State University of Utrecht, 1955.32 F. 0. Koenig, Corniti! Ilbtern. Thermodyn., Cindt. E&rmhim., Cmpt. rend.33B. A. Pethica, M. M. Standish, 3. Minghs, and D. H. Ilea, Nature, 1966, 205,34K. W. Bewig and W. A. Zisman, J . Phys. Chem., 1963, 67, 130.Rdunion 3e, 1951, 299.348HAYDON : LIQUID-LIQUID INTERFACES 15consisted of dispersions of aliphatic hydrocarbons in aqueous solutions ofsurface-active anions and cations.35 The major diEculty in this field,namely the quantitative explanation of the large discrepancies bet weent'he electrokinetic potentials and those calculated from simple models ofthe electrical double layer, given the interfacial charge density, have stillnot been overcome.For small hydrocarbon droplets, as in oil-in-water emulsions, the presenceof the diffuse double layer in the hydrocarbon phase is usually inconse-quential as it is not able to develop fully. For small water droplets inhydrocarbon, on the other hand, the diffuse layer is more likely to be fullydeveloped in both phases, as was envisaged by Verwey and Niessen.%GElectrokinetic potentials of water droplets in benzene in presence of sub-stituted salicylates, oleates, sulphonates, or picrates have been shown tobe positive.37 Recent work, however, has shown that, on introducinginorganic electrolytes into the aqueous phase, it is the cations rather thanthe anions that influence the electrokinetic potential.38 The authors of thiswork have shown that an explanation of this phenomenon follows froman analysis of the equations for the double diffuse layer.Studies of protein films a t hydrocarbon-wa terinterfaces have continued to appear over the last few years.Althoughthere have been investigations of the rheology of protein films39 and ofthe partial displacement of proteins from interfaces, the basic problem withthese systems has remained their irreproducibility, stemming, presumably,from the tendency of proteins to adsorb and even spread irreversibly. Thelatter problem has been emphasized in an investigation of pepsin filmsusing a barrier technique and Wilhelmy plate.40 It was found that thepressure-area curve a t intermediate pressures moved to smaller areas thehigher the pressure of the film, as initially spread.In other words, thepepsin appeared to spread less completely the higher the pressure againstwhich it had to spread.The pendant -drop method of interfacial tension measurement has beenused to study the time dependence of the adsorption of bovine serumalbumin a t the n-octadecane-water interface.41 It was found, in agreementwith earlier work, that the interfacial area per molecule of the adsorbedprotein was considerably less than that found for protein spread at lowpressures. A discussion of the reasons for this situation has been givenby Ghosh and Bull in connexion with their investigation of chymotrypsinadsorption a t n-octadecane-water interfa~es.~2 A two stage process is en-visaged in which diffusion to the interface and spreading across the interfacePolymeric molecules.35 P.J. Anderson, Trans. Paraday Soc., 1959, 55, 1421 ; D. A. Haydon, Proc. Roy.36 E. J. W. Verwey and K. F. Niessen, Phil. Mag., 1939, 28, [7], 435.J. L. Van der Minne and P. H. J. Hermanie, J . Colloid Sci., 1952, 7 , 600; W.3a W. Rigole and P. Van der Wee, J . Colloid SCi., 1965, 20, 145.3D B. Biswas and D. A. Haydon, Proc. Roy. Soc., 1963, A , 271,296,317; Kolloid-Z.,40 L. Blight, C. W. N. Cumper, and V. Kyte, J, Colloid Sci., 1965, 20, 393.'IS. Ghosh and H. B. Bull, Biochim. Biophy8. Acta, 1963, 66, 150.42 S. Ghosh and H. B. Bull, Arch. Biochem. Biophys., 1962, 99, 121.Soc., 1960, A , W, 319.Albers and J. Th. G. Overbeek, ibid., 1959, 14, 501.1962, 186 [l], 6716 GENERAL AND PHYSICAL CHEMISTRYare t,he important factors.At low substrate concentrations the arrival ofprotein at the interface is suggested to be slow compared with the spread-ing, and in consequence, most of the interface should be covered withcompletely expanded molecules. At higher concentration the expansion ofthe protein molecules is more likely to be blocked by the presence of rapidlyaccumulating neighbours. The final result is thus likely to be a film ofonly partially expanded and some almost native protein which should showan area per molecule considerably less than tthe completely expandedmolecules3. REACTIONS IN DISCHARGE-FLOW SYSTEMlBy I. 116. Campbell and B. A. Thrush(Department of Physical Chemistry, Cambrdge)IN recent years the emphasis in chemical kinetics has been increasingl_vtowards the measurement of individual processes and an understanding ofthe rBle of the energy distribution in the reactants and products.So muchhas been published since Sugden’s Annual Report in 1959 on the kineticsof reactions of atoms and small radicals that this Report will be confinedto discharge-flow methods of studying such reactions. No attempt will bemade to include Molecular Beam or Shock Tube studies or any of the detailsof the work on chemiluminescence which can provide much information onenergy distributions.The use of electric discharges as a source of free atoms was pioneeredby Wood and by Bonhoeffer 2 some forty years ago, and a large numberof atomic reactions were studied semi-quantitatively by them and byH a r t e ~ k .~ These workers were limited, as the only quantitative detect,orsfor free atoms then available were the calorimeter and the Wrede-Hartecligauge, which are non-specific. The considerable increase in the use of dis-charge flow systems during the last ten years is almost entirely due to thedevelopment of specific methods for the measurement of the concentrationsof atoms and free radicals. These include chemiluminescence, gas-phasetitration, electronic absorption spectroscopy, mass spectrometry, and electronspin resonance.Chemiluminescence methods are highly specific and involve the measure-ment of the intensity of the yellow nitrogen afterglow, the air afterglowfrom 0 + NO, HNO emission from Htf NO in the red and near-infrared regions,or emission in these regions from recombining chlorine or bromine atoms.The intensities of these emissions are proportional to “I2, [O][NO], [H][NO]and [C1I2 and [BrI2 respectively, but the constant of proportionality maybe a function of total pressure or composition of the carrier gas.Suchsystems are normally calibrated by calorimetry or gas-phase titration ofthe active species. The behaviour of metastable electronically and vibra-tionally excited species can also be followed by their emission or absorptionspectra.Gaseous titration methods for free atoms are less specific and needindependent confirmation ; they are therefore discussed in the individualsections. Free atoms and radicals can readily be detected m a s spectro-metrically since their ionisation potentials are lower than the correspondingappearance potentials from the parent molecules.Calibration requires con-siderable care. Metastable species can also be detected from breaks in theionisation curves but their identification is not always certain.Unfortunately the absorption spectra of ground-state H, N, and 01 R. W. Wood, Proc. Roy. SOC., 1922, A , 102, 1.2 K. F. Bonhoeffer, 2. ph?/s. Chenz., 1924, 113, 199, 492; ibid., 1925, 116, 391.3 P. Herteck and I-. Kopsch, Natzcrwiss., 1929, 72718 GENERAL AND PHYSICAL CHEMISTRYatoms all lie in the vacuum-ultraviolet region, but, despite the difficultiesinvolved, this method has been used to study their reactions.4 The absorp-tion spectra of free radicals lie mainly in the more aceessible ultravioletregon, but so far only the hydroxyl radical5 has been studied in thisway.Although electron spin resonance provides a sensitive specific methodof detecting free atoms, the problems of calibration are considerable.Thespectrum is normally displayed in the differential form, whereas the area,under each peak is required and double integration is frequently needed.Errors due to high modulation amplitudes, power saturation, and thevariation of line width with relaxation time can readily occur. The sensi-tivity for the detection of free radicals is lower by a factor of the rotationalpa.rtition function due to the splitting of the spectrum into many lines.Such spectra generally involve electric dipole transitions, whereas the atomicresonances are magnetic dipole transitions.Useful discussions of calibra-tion methods have been given by Krongelb and Strandberg6 and West-enberg and de Haas.'Rate constants refer to room temperature (2Oo--25"c) unless otherwisestated in the following sections of the Report.=bogen Atoms.-Much of the recently published work on hydrogenactom reactions in the gas phase has used other sources of hydrogen atoms,such as radiolysis, mercury photosensitisation, or the hydrogen-oxygenreaction. A review of this topic has recently appeared.sThe study of hydrogen atoms in discharge flow systems appears simplesince there are no low-lying metastable or atomic states to complicate theinterpretation.There are, however, experimental difficulties, since the re-combination of hydrogen atoms on glass surfaces is frequently irreproducibleand affected by such species as water. Several workers have studied theheterogeneous recombination of hydrogen atoms by Smith's method inwhich the concentration gradient set up by hydrogen atom diffusion andsurface recombination in a side arm is measured. This yields a value of y,the fraction of collisions which lead to recombination on the walls or onan auxiliary probe. The theory of this technique has been improved byGreaves and Linnetti10 and extended by Wise and Ablow.ll There hashowever been controversy 1% l3 about the method of calculating the effectof the probe used by the latter authors.Glass and quartz surfaces give values of y of the order of 10-4 at roomtemperature 1 3 7 14 which is reduced to about 10-5 by coating with Teflon4F.A. Morse and F. Kaufman, J . Chem. Phy.9., 1965, 42, 1785.5 F. Kaufmm and F. P. Del Greco, Discuss. Farday SOC., 1962, 38, 128; idem.,* S . Krongelb and M. W. P. Strandberg, J . Chem. Phys., 1969, 81, 1196.7 A. A. Westenbe'? and N. de Haw, J . Chm. Phy8., 1964, 40, 3087.8B. A. Thrush, Progress in Reaction Kinetics," Pergamon, London, 1965,* W. V. Smith, J . Chm. Phys., 1943, 11, 110.9th Symposium on Cornbwtion, Academic Press, New York, 1963, p. 659.Vol. 3, a. 2.10 J. C. Greaves and J. W. Linnett, Trans. Furadq Soc., 1959, 55, 1338.11H. Wise and C. M. Ablow, J . Chem. Phys., 1958, 29, 634.1aK. Tsu and M.Boudart, Canad. J . Chem., 1961, 39, 1239.laB. J. Wood and H. Wise, J . Phys. Chem., 1961, 65, 1976.14B. J. Wood and H. Wise, J . Phy8. Chem., 1962, 66, 1049CAMPBELL AND THRUSH: DISCHARGE-FLOW SYSTEMS 19or a silicone.f5,16 Wood and Wise14 find that y for glass rises as thetemperature is lowered or raised, becoming a rapid second order processbelow -150"c or above 250"~. This is explained as a change from theRideal to the Hinshelwood mechanism of surface recombination. Valuesof y between 10-2 and 1 are found on Au, Al, Cu, Ni, Pd, Pt, Ti, and Wsurfaces,13 frequently with small positive temperature coefficients. Graphiteshows similar behaviour, but there is a parallel process with a larger activa-tion energy which yields small amounts of acetylene, ethane, and methane."Recent determinations of the rate of the homogeneous recombinationof hydrogen atoms gave rate constants (expressed as d[H,]/dt) of l0l6 (ref.18),8.9 x 1015 (ref. 19) and 3.4 x 1015 (ref. 20) in cm.6 mole-2 sec.-l for M = H2and 2.3 x 1015 for M = Ar (ref. 20). The values for hydrogen are generallylower than those obtained in older work.21-23 For argon the rate constantwas found to vary as T-* between 2 1 3 " ~ and 349'9, which if extrapolatedwould agree well with some shock-tube data on the reverse process 24although other data 2 5 s z6 from this source is consistent with an overall T-ldependence. A similar situation exists for M = H,. The importance of Has a third body in this reaction is still uncertain.An upper limit of 4.8 x 1014 cm.6 mole-, sec.-l has been reported forthe rate constant of the reaction(1)The two combination reactions of hydrogen atoms with diatomic mole-H + NO + M + HNO +- M + 48.6 kcal./mole (2)H + 0, + M + HO, + M + 46 kcal./mole (3)being almost equally exothermic and giving products with similar con-figurations with similar rate constants.Values of k, = 1.48 x 1016,8.7 x and 6.5 x 1015 cm.6 mole-2 sec.-l for M = H,, Ar, and He a t20"c have been determined28 by the HNO emission rnethod,a9 the rateconstant having a small negative temperature-dependence. A similarvalue of k, = 1.1 x l0l6 cm.6 mole-, sec.-l for M = H, was obtainedwith a calorimetric probe.30 The mechanism of the parallel formationH + N + M -+NH + Ma t room temperature .2cules studied are extremely similarl6 H.C. Berg and D. Kleppner, Rev. Sci. Imtr., 1962, 33, 248.l6 J. P. Wittke and R. IiE. Dieke, Phys. Rev., 1956, 103, 620.l7 A. B. King and H. Wise, J . Phys. C h . , 1963, 87, 1163.1* L. I. Avramenko and R. V. Kolesnikova, Izvest. Akact. Nauk S.S.S.R., Otdel.lo C. B. Kretschmer and H. L. Petersen, J . Chern. Phys., 1963, 39, 1772.ao F. S. Larkin and B. A. Thrush, Discws. Faraday SOC., 1964, 37, 112; idmyW. Steiner, Tram. Faraday SOC., 1936, 31, 623, 962.anH. M. Smdlwood, J . Amer. Chem. Soc., 1934, 56, 1642.asI. Amdur, J . Amr. C h . SOC., 1938, 80, 2347.24 J. P. Rink, J . Chm. Phys., 1962, 36, 262, 1398.a6 R. W. Patch, J . Chem. Phys., 1962, 38, 1919.E. A. Sutton, J. Chem. Phys., 1962, 38, 2923.C. Mavroyannis and C.A. Winkler, C a d . J . C h . , 1962, 40, 240.2% M. A. A. Clyne and B. A. Thrush, Discurs. Paraday SOC., 1962, B, 139.M. A. A.. Clyne and B. A. Thrush, Trans. Faraday SOC., 1961, 57, 1306.ao R. Simonaitis, J . Ph.ys. Chem., 1963, 87, 2227.khim. Nauk, 1961, 1971.10th Symposium on Combustion, Combustion Institute, 196420 GENERAL AXD PHPSICAL CHEMISTRYof electronically excited HNO in reaction (2) has also been discussed.Z8The subsequent reactionH + HNO -3 H, + NO (4)is very rapid, with a rate constant greater than 3 x 1O1O cm.3 mole-1 sec.-la t room ternperat~re.2~The reactions subsequent to (3) in discharge flow systems are rathermore complex.(5a)-+OH+OH (5b)+H,O + 0 (5c)(6)0 + O H + H + O , (7)also occur.Reaction ( 5 ) is very rapid; Clyne and Thrush31 have also shownthat 33 & 12% of it occurs by path (5a). They obtained a value ofk, = 8 x 1015 cm.6 mole-2 sec.-l for M = Ar a t 20"c using the HNOemission technique. A later calorimetric study under similar conditionsgave k, = 1.3 x 10l6 cm.6 mole-2 sec.-l in good agreenient.20 The muchlower value reported by Avrainenko and Kolesnikova 18 is based on theassumption that H02 is removed only by the reaction( 8 )which is unlikely in the presence of an excess of hydrogen atoms. Reaction(3) was found to have a negative temperature coefficient corresponding to aT-2 dependence or an activation energy of -2000 cal./mole. This extra-polates to give good agreement with measurements of the second explosionlimit of the hydrogen-oxygen reacti0n.~2~ 33 There is evidence that reac-tion (5b) yields vibrationally excited OH.34H + HOz+H, + 0,and in the absence of molecular hydrogenOH + OH+H,O + 0+ HO2 + H 2 0 2 + 0,The branching step in the hydrogen-oxygen reactionH + O , + O H + O (9)cannot be studied directly in discharge flow systems as the equilibrium istoo far to the left.Measurements of the reverse process, which are discussedlater, agree well with data from flames and explosion limits on the forwardreaction.31Schulz and Le have studied the isotope exchange reactionH + D2 +HD + D (10)between 100" and 2OO0c, finding a rate expression of 4.4 x 1012 exp(-7300/RT) ~ m . ~ mole-l sec.-l. The rate constants for this and otherisotopic combinations can be deduced from a recent study of the thermalexchange reaction between H2 and D2 36 in which it is suggested that higherrate constants in earlier work are due to oxygen diffusing through the31 M.A. A. Clyne and B. A. Thrush, Proc. F y . SOC., 1963, A , $3'75, 559.32 V. V. Voevodskii and V. N. Kondratiev, Progress in Reaction Kinetics ,"s3 R. R. Baldwin, 9th Symposium on Combustion, Academic Press, New Pork, 1963,34 P. E. Charters and J. C. Polanyi, Canad. J . Chem., 1960, 38, 1742.35 W. R. Schulz and D. J. Le Roy, Canad. J . Chem., 1964, 42, 4280.36 G. Boata, G. Careri, A. Cimino, E. Molinari, and G. G. Volpi, J . Chem. Php.,Pergamon, London, 1961, Vol. I, Ch. 2.p. 667.1956, 24, 783CAMPBELL AND THRUSH: DISCHARGE-FLOW SYSTEMS 21vessel walls. Shavitt 3' has examined this data and those from the ortho-para exchange in terms of transition-state theory.He shows that the morerecent data 36 give a better fit. His expressions deviate significantly fromthe Arrhenius form; readers are therefore referred to calculation I1 in hisfabulated rate constants.37Discharge-flow systems have been used more to study the energy dis-tribution in the products of hydrogen atom abstraction reactions than todetermine their rate constants. Garvin 38 and colleagues have shown thatthe reactionH + O,-+HO + 0,yields vibrationally excited OH in levels up to u = 9, which correspondsto the total exothermicity of the reaction, the effective "vibrational tem-perature " of the initial distribution being 9 0 0 0 " ~ . ~ ~H + N O , + O H + N O (12)appears to yield little vibrationally excited OH5 and its infrared emissionmust be partly due to H20 formed subsequently from OH.39 Mass spectro-metric studies 40 show that these reactions (11) and (12) have rate constantsof 1.6 x 1013 and 3 x 1013 (3111.3 mole-1 sec.-l respectively at 20"c.Polanyi and co-workers 34, 419 42 have also studied the infrared emissionby vibrationally excited species in the reaction of hydrogen atoms with Cl,,Br2, NOCI, etc.H + C1, -3 HCl + C1 (13)some 10% of the energy liberated appears as vibration of the HC1 formedwhich has up to five quanta of vibrational energy.41, *2 There is alsoevidence for excess rotational excitation, although the rapid rotationalrelaxation makes it difficult to determine the initial distribution.Thereaction D + C1, gives DC1 with a similar energy distribution to the HC1from H + Cl,.42 The other reactions have been examined in less detailbut their general features are similar. Bunker 44 and Polanyi 45 have madecalculations of the expected energy distributions in the products of thesereactions using Monte Carlo methods.1.1 kcal./mole and afrequency factor of about 10l1 ~ 1 1 1 . ~ mole-l sec.-l for the reactionH 3- CH, -3 H, -l- CH, (14)This frequency factor is much lower than that found by Kazmi, Diefendorf,and Le Roy4' for the reactionH + C&f,+H, + C,H, (15)The reactionIn the reactionJamieson 46 reports an activation energy of 7.437 I.Shavitt, J . Clzem. Phys., 1959, 31, 1359.38 J. D. McKinley, D. Garvin, and M. Boudart, J . Chem. Phys., 1955, 23, 784.3s D. Garvin, H. P. Broida and H. J. Kostkowski, J . ChenL. Phys., 1960, 32, 880.40 L. F. Phillips and H. I. Schiff, J . Chem. Phys., 1962, 37, 1233.41 J. K. Cashion and J. C. Polanyi, Proc. Roy. SOC., 1960, A , 258, 529.4 2 P. E. Charters and J. C. Polanyi, Discuss. Paraday SOC., 1962, 33, 107.43 E. W. R. Steacie, " Atomic and Free Radical Reactions," Reinhold, New York,44 N. C. Rlsis and D. L. Bunker, J . Chem. Phys., 1962, 37, 2713.4 5 J. C. Polanyi and S. D. Rosner, J . Chem. Phys., 1963, 38, 1028.4 6 J. W. S. Jamieson, Canad. J . Chem., 1964, 42, 1638.4 7 H. A. Kazmi, R. J. Diefendorf and D. J. Le Roy, Canad. J .Chern., 1963,41, 690.2nd ed., 195422 GBNERAL AND PHYSICAL OHlCIKISTRYtheir rate expression being 1.3 x lo1* exp (--8200/RT) ~ m . ~ mole-l sec.-l.Data from several sources support high frequency factors43 of about 1014 cm.3mole-1 sec.-l for hydrogen atom attack on paraffin hydrocarbons.8, 48The observed rate constant for hydrogen atom. removal in their reactionwith hydrogen peroxide is 6 x 1O1O ~111.~ mole-l sec.-l at room tempera-ture.49 This is an upper limit, since the mechanism and stoicheiometry arenot known. A value of 3.5 x loll exp (-2OOO/RT) ~111.~ mole-l sec.-lhas been found50 for the rate constant of the reactionH + NJ34 + NgH, + H, (16)Nitrogen Atom.-Although it has been accepted for some years that theyellow nitrogen afterglow arises from the recombination of ground state (48)nitrogen atoms, there has been considerable disagreement over methods formeasuring their concentration. The two methods are the titration ofnitrogen atoms with nitric oxideN + NO +N2 + 0 (17)i n which the end point is indicated by a change from the yellow nitrogen(N + N) and blue nitric oxide (N + 0) afterglows to the greenish airafterglow (0 + NO), and secondly the limiting HCN yield from reactionwith excess of ethylene or other olefin a t elevated temperatures.It wasgenerally agreed that the endpoint of the nitric oxide titration correspondedto oonsumption of ground-state nitrogen atoms, a point which has now beenestablished by e.s.r. ~tudies.5~~ Proponents of the HCN yield method52maintained that nitric oxide titration gave higher yields because an addi-tional energetic species in active nitrogen decomposed nitric oxide.Thisview is not consistent with the experiments cited 4, 7,51 or with the observa-tions using l5NO. In these, Kaufman and Kelso 53 have shown that theNO #I band emission in the products comes only from 14N + 0, and massspectrometric studies by Herron 54 and by Back and Mui S5 have shownthak the reaction with 15NO yields only 14N15N and no lW2, therebydemonstrating that 15N atoms are not liberated by decomposition of nitricoxide.Herron56 has recently shown that the reaction between nitrogen atomsand ethylene is not stoichiometric and occurs by a catalytic chain in-volving hydrogen atoms.The nitric oxide titration method can therefore be regarded as estab-lished; it involves a very rapid reaction; Clyne and Thrush57 obtained46 R.R. Baldwin, D. Jackson, R. W. Walker, and S. J. Webster, 10th Symposiumon Combustion, Combustion Institute, 1964.4 s S. N. Foner and R. F. Hudson, J. Chem. Phye., 1962, 36, 2676, 2681.so M. Schiavello and G. G. Volpi, J. Chem. Phys., 1962, 57, 1510.61 J. Kaplan, W. J. Schade, C. A. Barth, and A. F. Hildebrandt, C a d . J . Chem.,s*A. N. Wright and C. A. Winkler, Canad. J . Chem., 1962, 40, 5.6aF. Kaufman and J. R. Kelso, J . Chem. Phys., 1957, 27, 1209.84 J. T. Herron, J. Chem. Phys., 1961, 85, 1138.b6 R. A. Back end Y. P. Mui, J . Phys. Chem., 1962, 66, 1362.J. T. Herron, J. Phys. Cbm., 1965, 69, 2736.s7 N. A. A.Clyne snd B. A. Thrush, PTOC. Boy. Soc., 1961, A, 261, 269.1960, 88, 1688CAMPBELL AND THRUSH : DISCHABGE-PLOW SYSTEMS 23k17 == 3.0 x 101s exp (-200/RT) cm.3 mole-l sec.-l in a photometric study,in reasonable agreement with mass spectrometric values of (1.0 jc 0.5) x 1013and (5-07 -& 0.13) x 10l2 ~ m . ~ rnole-1 sec.-l by Herron 54 and by Phillipsand Schiff.68Other active species which might be present in the yellow afterglowinclude metastable N(2D) and N(2P) excited nitrogen atoms. Spectro-scopic 49 59 and mass spectrometric studies 60 have shown that these activespecies disappear rapidly, probably by wall deactivation.61 The wall decayof N(2.D) is inhibited by cooling to -195”c, when strong N, Second Positiveemission due to the combination of N(2D) + N(48) is produced.62 Theexcited nitrogen molecule in the A3&+ state is another metastable specieswhich exists in the afterglow.Its lifetime has been variously estimatedby a range of techniques; 0.07 seconds by e.s.r.,63 0.08 seconds by reactionwith ammonia,64 0.9 seconds from decay of its emission from a pulseddi~charge,~~ m. 1 second from its behaviour in active nitrogen at very highpressures,61 2.6 x 10-2 seconds from the intensity of the (6,O) band in theabsorption spectrum A3&+ +- XI&+ (ref. 66) and 2.0 -J= 0.9 seconds bycomparison of the intensities of its emission [ (0,6) Vegard-Kaplan] andabsorption spectra [ (1,O) First P0sitive].~7 Apart from the last two measure-ments, all these values are actual lives which are less than the radiativelife and depend on experimental conditions, since there is evidence thatN,(A3Eu+) is efficiently destroyed on the walls and by other species.61.68The discrepancy between the last two values which are rcldiative lifetimesis disturbing; if a more recent value of the life of N2(B3II,) is used,69Carleton and Oldenberg’s value 67 is increased by a factor of seven to15 seconds which widens the discrepancy; the v3 factor in the emissionprobability can only account for a factor of four between these values.Further work is clearly needed.Vibrationally excited ground-state nitrogen molecules have been detectedby heat release when nitrous oxide is added 70 and by vacuum-ultravioletspectroscopy, where populations of level v f f = 1 up to 10% of level wn = 0were observed.’l The antisymmetric stretch frequency of N,O is excitedby near resonance transfer from vibrationally excited nitrogen molecules.72The mechanism of the yellow afterglow has attracted considerableattention.It has been shown that emission in the near-infrared region58 L. F. Phillips and H. I. Schiff, J . Chem. Phys., 1962, 36, 1509.59 Y. Tanaka, A. S. Jursa, F. J. Leblanc, and E. C. Y . Inn, Planetary Space Science,6o S. N. Foner and R. F. Hudson, J . Chem. Phys., 1962, 37, 1662.61 J. F. Noxon, J . Chem. Phys., 1962, 36, 926.6 2 Y. Tanaka, F. J. Leblanc, and A. S. Jursa, J . Chern. Phys., 1959, 30, 1614.6s J. M. Anderson and J. N. Barry, Proc. Phya. Soc., 1961, 78, 1227.6 4 H. B. Dunford, E.R. V. Milton, and D. L. Whalen, C a d . J . Chem., 1964,6 5 E. C. Zipf, J . Chem. Phys., 1963, 38, 2034.6sP. G. Wilkinson and R. S. MuUiken, J . Chem. Phys., 1959, 31, 674.67 N. P. Carleton and 0. Oldenberg, J . Chern. Phys., 1962, 86, 3460.68 I. M. Campbell and B. A. Thrush, Proc. Chem. Soc., 1964, 410.69 M. Jeunehomme and A. B. F. Duncan, J . Chem. Phys., 1964, 41, 1694.‘OF. Kaufiaan and J. R. Kelso, J . Chern. Phys., 1958, 28, 510.71 K. Dressler, J . Chem. Phys., 1959, 30, 1621.72 E. L. Milne, M. Steinberg, and a. P. Broids, J . C h m . Phys., 1966, 42, 2615.1959, 1, 7.42, 250424 GENERAL AND PHYSICAL CHEMISTRYcomes from a newly discovered Y3&- state in addition to the familiarB3ng -+ A3&+ 74 Bayes and Kistiakowsky 75 have discussedthe mechanism in detail, dividing the bands into five groups; they proposedthat a steady state concentration of N,(5C,+) is the precursor of the emittinglevels assuming that the emission intensity was proportional to [N]2[M].Recent work shows that this relation only applies at low pressures,76 theintensity being independent of pressure above 1 mm. Hg.77 Campbell andThrush have recently summarised evidence on the pressure-dependenceof active nitrogen and have shown that the enhancement of the afterglowin argon or helium carriers is due to efficient quenching of N2(B3n,) byN,(X1&+). They conclude that N,(A3C,+) is the precursor of N2(B3n,)a,nd that about one third of the recombining molecules pass throughN,( B3rIg). Their value 78 for the luminescent rate constant in pure nitrogenis 7 x lo6 ~1111.~ mole-l sec.-l, in fair agreement with Young a,nd Sharpless'value 77 of 1.8 x lo7 ~111.~ mole-1 sec.-l.The rate constants for the three-body recombination of nitrogen atomshave been measured by several workers.The reaction was studied photo-metrically 19, 's and by mass spectrometry 79 using titration with nitricoxide to establish the absolute concentration of nitrogen atoms. The valuesgiven are for d[N,]/dt in cm.6 mole-2 sec.-l. Herron, Franklin, Bradt, andDibeler 79 obtained 5.7 x 1015 for M = N,, 2.8 x 1015 for M = Ar, and0.82 x 1015 for M = He, and these were independent of temperature in therange 195"-450"~, and agree with the relative efficiencies for bromine andiodine atom recombination.Mavroyannis and Winkler 8o found 1.06 x 10l6and Kretschmer and Petersen l9 obtained 8-0 x 1014, both for M = N2a t 300"~. Campbell and Thrush 78 have reported 1-38 x 1015 for N,,1.73 x 1015 for Ar, and 1.92 x lof5 for He at 298"~. The discrepanciesbetween these values may well be due to different methods of allowing forheterogeneous recombination.The short-lived pink afterglow of nitrogen was first reported by Bealeand Broida 81 in fast-flowing nitrogen at high pressures. The conditionsinvolved are somewhat unusual, since it can be preceded and followed bythe normal yellow nitrogen afterglow. It shows emission by species withenergies up to 22 ev above ground-state nitrogen molecules. These includeN(3s 2P), N2+(B2C,+), and N, First Positive and Second Positive emissionfrom levels above the dissociation limit into normal atoms.*, Absorptionstudies in the vacuum-ultraviolet region show ground-state nitrogen mole-cules with up to 20 quanta of vibrational energy 83 but no increase in theconcentration of atoms in the ground state (*X) or low-lying metastable73 K.D. Bayes and G. B. Kistiakowsky, J . Chem. Phys., 1958, 29, 949.7 4 F. J. Leblanc, Y. Tanaks, and A. S. Jursa, J . Chem. Phys., 1958, 28, 979.7 5 K. D. Bayes and G. B. Kistiakowsky, J. Chena. Phys., 1960, 32, 992.76 R. A. Young, R. L. Sharpless, and R. Stringham, J . Chem. Phys., 1964,41,1497.7 7 R. A. Young and R. L. Sharpless, J . Chem. Phys., 1963, 39, 1071.7'31. M. Campbell and B. A. Thrush, Chem. Comm., 1965, 250.7 9 J.T. Herron, J. L. Franklin, P. Bradt, and V. H. Dibeler, J . Chena. Php.,80 C . Mavroyannis and C. A. Winkler, Canad. J. Chem., 1961. 39, 1601.81 Q. F. Beale and H. P. Broida, J . Chem. Phys., 1959, 31, 1030.*2R. A. Young, J . Chem. Phys., 1962, 36, 3854.63 A. M. Bass, J . Chena. P h p . , 1964, 40, 696.1959, 30, 879CAMPBELL AND THRUSH : DISCHARGE-FLOW SYSTEMS 25states ( 2 0 and ") in the pink afterglo~.~4 The ion density is ca. 2 x 1O1Ocm.-3, which is about one hundred times greater than in the yellow after-gl0~;85 this explains efficient quenching by a 60 kc./sec. electric field.82Rag and Clark 85 suggest that N(6X) is the high-energy metastable speciesresponsible for the pink afterglow, but this was not detected in e.s.r.experiments.86 Young 82, 87 suggests a catalysed recombination of nitrogenatoms involving N,+ and N,+.have investigated the nitric oxide emission fromthe association of nitrogen and oxygen atoms.They show that the v' = 0level of NO(C2rl[) (6 bands) is populated by a two-body association andpostulate that NO(A2X+) ( y bands) is populated partly by radiation fromthe C state and partly by a pressure-dependent collisional mechanism.Formation of nitric oxide molecules in b4C- and B2II states also occursby a,n indirect collision-induced mechanism. Rate constants for the associa-tion of N and 0 with N, as third body have been measured as 1.85 x 1015mole-, sec.-l (ref. 80) and 3.3 x 1015 cm.6 mole-2 sec.-l (ref. 19) at300 OK.Young and SharplessThree recent studies of the reactionN +O,+NO + o (18)have yielded rate expressions of 1.7 x 1013 exp (-7500/RT) (ref.89)8.3 x 10l2 exp (-7200/RT) (ref. 57), and 2.3 x 10l2 exp (-5900/RT)cm.6 mole-, sec.-l (ref. go), all in good agreement with the value of2 x 10l2 exp (-6200/RT) found by Kistiakowsky and V01pi.~~ Agreementbetween the photometric value 9, (1.02 x 1010 cm.3 mole-1 sec.-l) and themass spectrometric 58 rate constant of (1.32 & 0.32) x loll ~111.~ mole-lsec.-l for the reactionN + O,-+NO + 0, (19)is less satisfactory. In this system catalysis by trace impurities of hydrogencan occur.The reaction of nitrogen atoms with nitrogen dioxide is more complex;the stoicheiometry shows that a t least three of the four possible initial stepsabN + NO,-+N,O + 0+NO +NO--+& + 0,+N2 + 2 0Cd84 C.E. Fairchild, A. B. Prag, and K. C. Clark, J. Chem. Phys., 1963, 39, 794.86 A. B. Pra.g and K. C. Clark, J . Chem. Phys., 1963, 39, 799.86 R. A. Young, R. L. Sharpless, and R. Stringham, J. Chem. Phys., 1964, 40,87 R. A. Young, C. R. Gatz, and R. L. Sharpless, J. Phys. Chem., 1965, 69, 1763.88 R. A. Young and R. L. Sharpless, Discuss. Faraday Soc., 1962, 33, 228.89 F. Kaufman and L. J. Decker, 7th Symposium on Combustion, Butterworths,90 C. Mavroyannis and C. A. Winkler, " Chemical Reactions in the Lower ands1 G. B. Kistiakowsky and G. G. Volpi, J . Chem. P h p . , 1957, 2'9, 1114.s 2 H. P. Broida, H. I. Schiff, and T. M. Sugden, Nature, 1960, 185, 759.251.London, 3959, p.57.Epper Atmosphere ," Interscience, New York, 1961, p. 28726 UENERAL AND PHYSICAL CHEMISTRYoccur, their relative proportions being independent of temperature. 93 Arecent mass spectrometric investigation 94 gives an overall rate constant of(1.11 & 0-12) x 1013 ~111.~ mole-1 sec.-l, and relative proportions ofa : b : c : d : : 0.43 : 0.33 : 0-10 : 0.13, in fair agreement with earlier work.93, 95The reaction of active nitrogen with bromine or hydrogen bromide givesorange NBr emission from near the walls of the vessel.96 It is probabletha-t NBr is formed heterogeneously and rapidly removed by reaction withnitrogen atoms.In contrast, the NS(B211 -+ X 2 n ) emission 97 from the reaction of activenitrogen with sulphur is confined to the centre of the flow tube, suggestingexcitation by a molecular species which is destroyed a t the walkgs Thesimilar beha,viour of emission by the CN impurity in active nitrogen hasbeen attributed to such a mechanism involving N2(A3Z,+) rather thanformation of excited CN in the catalytic recombination processN + CN + M-NCN + MN +NCN -+N2 + C NNS is apparently removed by nitrogen atoms; in their absence it can yieldN,S,, e k g 9The reaction between hydrogen atoms and nitrogen atoms gives NHemission i'rom levels v' = 0 and 1 of the A311 state as well as nitrogenFirst and Second Positive emission.100 Mannella lol has suggested that NHis excited by collision with N2(B3111,), v' = 12, but this species has a lifeof less than S ~ C .~ ~ and has no allowed vertical transition of sufficientenergy; his mechanism for the nitrogen Second Positive bands' excitation isN2(A3&+),p,6-9 f NH(ASII)~'=O + NH(X3C-) + Na(C 8n,)d = 0,lMany studies have been made of the reaction of nitrogen atoms withhydrocarbons. Linnett and Jennings1o2 have shown that, in addition tostrong CN red (A2n -+X2X+) and violet ( B 2 P -+X2X+) emission, thesereact'ions yield CH(2A), NH(311), C, Swan bands (from ethylene and acety-lene) and a spectrum now know to be due to NCN.lO3 The overall processesoccurring are clearly complex.Broida and co-workers 104-106 have studiedthe very strong CN emission from the reaction between active nitrogen andhalogenated hydrocarbons. They have shown that excited CN is producedpredominantly in high vibrational levels of the A211 state and that v' = 0level of the B22+ state is populated from A2II, v' = 10, by collision-93 M.A. A. Clpe and B. A. Thrush, Trans. Paraday SOC., 1961, 57, 69.Q 4 L. F. Phillips and H. I. Schiff, J . Chem. Phys., 1965, 42, 3171.95 F. Kaufman and J. R. Kelso, 7th Symposium on Combustion, Butterworths,9 6 E . R. V. Milton and H. B. Dunford, J . Chem. Phys., 1961, 34, 51.9 7 J. J. Smith and B. Meyer, J . Mol. Spectroscopy, 1964, 14, 160.g8 J. A. S. Bett and C. A. Winkler, J . Phys. Chm., 1964, 68, 2735.99 J. A. S, Bett and C. A. Winlrler, J. Phys. Chem., 1964, 68, 2501.London, 1959, p. 53.loo H. Guenebaut, G. Pametier, and P. Goudmand, Compt. rend., 1960, 251, 1480.Iol G. G. Dlannella, J .Chem. Phys., 1962, 30, 1079.lo2 J. W. Linnett and K. R. Jennings, Tram. Faraday SOC., 1960, 56, 1737.Io3 G . Herzberg and D. N. Travis, Canad. J . Phys., 1964, 42, 1658.I04 N. H. Kiess and H. P. Broida, 7th Symposium on Combustion, Butterworths,1O5 H. P. Broida and S. Golden, Canad. J. Chem., 1960, 38, 1666.I06H. E. Radford and H. P. Broida, J. Chem. Phys., 1963, 38, 644.London, 1959, p. 207CAMPBELL ABND THBUSH: DISCHARUE-FLOW 6YST1MS 27induced transitions between levels which are very near to resonance.Bayes,”o7 who studied CN red emission from the reaction of active nitrogenwith C,N,, HCN, ClCN, and CC1, showed that the vibrational distributionin this spectrum was the sum of two components, the P , arising from levels3 < v’ < 10 which corresponds to that described by Broida and thePI distribution where levels 0 < vr < 3 are populated.Setser and Thrush lo8showed that the P , distribution in the AZII state was accompanied bypopulation of higher vibrational levels (u’ > 5 ) of the B2Z+ state, whilethe P, distribution corresponded to population of level d = 0 of the B2Cfstate. These workers pointed out that the P, distribution is characteristicof reaction in which the CN linkage has to be formed and suggest that theexcited CN is produced in the reactionN + cx -+ x + CN(AQ,3 < v < 10)where X = H, C1, or Br. They also suggest that the PI distribution isdue to excitation of previously formed CN, either by metastable species orby CN acting as a third body. Brown and Broida 1°Q have suggested thatvery high vibrational levels of the A2II state (u‘ - 18) are precursors ofthe high vibrational levels of the B2X* state and suggest thatN + NCN --+ N, + CN”is the reaction responsible. Bayes 107 suggests that the P , distribution inthe CN red system is associated with the formation of excited CN in thedissociation of XCN.It seems unlikely that these processes would givethe observed excitations.09gen Atom.-Ground state (sP) oxygen atoms are readily producedby passing molecular oxygen through an electric discharge. They are easilydetected by the grey-green air afterglow emission caused by added nitricoxide. The intensity of this afterglow is proportional to [O][NO] and itprovides the indicator for Kaufman’s method of measuring oxygen atomconcentrations by titration with nitrogen dioxide.OfNO,+NO+O, (23)Krongelb and Strandberg 6 and Westenberg and de Haas 7 have determinedabsolute concentrations of O( 3P) by e.8.r.Vacuum ultraviolet absorptionspectroscopy is another method; Morse and Kadman have observedO(3P) but could not detect O(lD) or Q(W) in absorption in pure 0, or inthe presence of Ar, He, N,, or H,O. Emission by O(W) to O{lD), whichis the auroral green line at 5577 a, is observed when oxygen atoms areobtained by titrating active nitrogen with nitric oxide.61 This emissiondepends on the presence of nitrogen atoms 5l and the dominant excitationprocess is probablyN -I- N 4- 0 +N, 4- Ops)but some additional process giving an intensity proportional to the squareof the oxygen atom concentration must also participate.77 By comparisonof the intensity of the O(1S+3P) line at 2977 A with the auroral green linelo’ K.D. Bayes, Canad. J . Ohm., 1961, 39, 1074.lo8 D. W. Setser and B. A. Thrush, Proc. Roy. Soc., 1965, A, 288, 256.IoB R. L. Brown and H. P. Broida, J. Chem. Phys., 1964, 41, 2063.110 F. Kaufman, Proc. Roy. SOC., 1958, A , 247, 12328 GENERAL AND PHYSICAL CHEMISTRYNoxon has shown that collision-induced transitions must increase theprobability of the latter by a factor of about thirty.The general evidence is that excited oxygen atoms, unlike metastablemolecular species, are not an important constituent of discharged oxygen.Evidence for the presence of long-lived metastable oxygen molecules insuch systems came originally from mass spectrometric studies.1119112 Elias,Ogryzlo, and Schiff n3 showed that a detector coated with cobalt oxidecould detect a heat release after oxygen atoms had been removed by amercury mirror.These experiments suggested the presence of about 10%of O,( lA,). Such concentrations are confirmed by Noxon’s observation 114of its emission at 12,700 A and its detection with e.s.r. by Falick, Mahan,and Meyers.ll5Ogryzlo and his co-workers 116, 117 have shown that the broad emissionbands at 6340 A and 7030 A are due to concerted emission by two O,(lA,)molecules as an unbound complex. They deduce a radiative life of 2.5 x 10-3seconds for this process which is a factor of lo5 shorter than for a singleunperturbed 02(lA,) molecule.Their finding that the intensity of the‘‘ dimole ” emission is proportional to the square of 02(1Ag+3&-) emissionintensity at 12,700 was not confirmed by March, Fwnival, and Schiff>l*who obtained a first-power relation and suggested that the dimoles are alsoremoved by O,(lA,) possibly to yield 02(1Xff+). Young and Black 119 sug-gest that O,(lEg+) is formed from two 02(1Ag), although this process violatesspin conservation, since 02(3Cg-) must be the other product. O,(lA,) isnot significantly quenched by GO2, N20, N,, H,O, NO, Ar, He, CO, NH,,or HZ116 and only reacts slowly with ethylene.l15 It is not expected t obe highly reactive, as its excitation energy is only 22 kcal./mole.From the intensity of the atmospheric band emission at 7619 A Noxon 114concludes that the concentration of O2(lXC,+) in discharged oxygen is lessthan 1.07(,, in agreement with Clyne, Thrush, and Wayne,120* 12f who foundapproximately 0.1% of this species a t 1 mm.Hg pressure. Like 02(1Ag),this species is apparently mainly formed in the discharge, although it isless metastable, having a radiative life of 7 seconds. From experimentswhere oxygen atoms were prepared by titrating active nitrogen with nitricoxide, Young and Sharpless 77 showed that 02(lXg+) is a minor product ofoxygen atom recombination ; the absolute intensity of this emission foundwas 4 x lWO[0]2[MJ einsteins cm.4 sec.-l, collisional quenching beingnegligible up to 10 mm. Hg pressure of nitrogen.The concentrations ofO,(lZg+) produced were about 1% of those in discharged oxygen. As with02(1Ag), the emission spectrum of 02(1Zg+) shows that it is almost ex-ll1 S. N. Foner and R. L. Hudson, J . Chem. Phys., 1956, 25, 601.112 J. T. Herron and H. I. Schiff, Canad. J . Chem., 1958, 36, 1159.118 L. Elias, E. A. Ogryzlo, and H. I. Schiff? Canad. J . Chem., 1959, 37, 1680.ll* J. F. Noxon, Canad. J . Phys., 1961, 39, 1110.115 A. M. Falick, B. H. Mahan, and R. L. Myers, J . Chem. Phys., 1965, 42, 1837.11* L. W. Bader and E. A. Ogryzlo, Discuss. Furuduy Soc., 1964, 37, 46.S. J. Arnold, R. J. Browne, and E. A. Ogryzlo, Photochem. and Photob%oZ., 1965.118 R. E. March, S. G. Furnival, and H. I. Schiff, Photochem. and Plwtobwl., 1965.119 R. A. Young and G.Black, J . Chenz. Phys., 1965, 42, 3740.120 1%. A. A. Clyne, B. A. Thrush, and R. P. Wa-yne, Nature, 1963, 199, 1057.1z1R5. A. A. Clyne, B. A. Thrush, and R. P. Wayne, Photochern. and Photobiol.,1965, 4, 957CAMPBELL AND THRUSH : DISCHARGE-FLOW SYSTEMS 29clusively in the lowest vibrational level, which has an excitation energyof 37 kcal./mole. Thus it has sufficient energy to decompose ozone to O2and 0, a process which occurs quite rapidly.120* 118 The decomposition ofozone by 02(lAg) is 2 kcal./mole endothermic and probably has a rateconstant of -1 x 1010 cm.3 mole-1 sec.-l at room temperature.118 02(1X,+)is quenched by C02 and N20 77 but it is not strongly removed by NO,NO,, or O2.l2IOxygen molecules in the more highly excited metastable AS&+ statemust be formed by recombination rather than in the discharge, since thestate has a radiative lifetime of less than 0.1 second (Herzberg bands).The details of this process are not clear, but it does not appear to be animportant one .I22 9 7In assessing the following studies of oxygen atom reactions, it is im-portant to consider possible interference by the excited molecular speciesmentioned above.In the absence of molecular oxygen, oxygen atoms recombine by thereaction0 + 0 + M + 0 , + Mwhich has been studied a t room temperature by several groups. Harteckand Reeves 123 obtained a rate constant of 1-08 x 1015 cm.6 mole-2 sec.-lfor M = Ar and 0, in experiments where argon-osygen mixtures werepassed through a discharge.This is probably an upper limit, since theydid not allow for removal of oxygen atoms by other processes such assurface decay. Marshall 124 studied this recombination by e.s.r.and obtaineda rate constant of 1.35 x 1015 cm.6 mole-, sec.-l for M = Ar. For M = O,,Kaufman and Kelso 125 found a rate constant of (1.0 -J= 0.3) x 1015 cm.6mole-2 sec.-l. The most extensive study is by Morgan and Scl~ifT,l~~ whoproduced oxygen atoms in the absence of molecular oxygen by titratingactive nitrogen with nitric oxide to just past the end-point. They obtaineda rate constant of 1.01 x 1015 ~111.~ mole-2 sec.-l for M = N, a t roomtemperature (expressed as d[ 02]/dt) and gave the following relative efficienciesfor different third bodies:Ar : He : N, : N,O : GO, : SF, :: < 0-3 : 0.3 : 1.0 : 1.4 : 3.0 : 3.0The recombination of oxygen atoms in the presence of molecular oxygen0 + 0 2 + M + 0, + M (25)0 + 0, + 0, + 0, (26)is not only affected by excited oxygen molecules which can decomposeozone, but also by hydrogenous impurities which produce a hydrogen-atom-catalysed process with closely similar kinetics 110*20H + O2 + M -+ HO, + M (3)O + H O , + O H + O , (27)0 + OH+H + O2 (7)which occurs by the mechanismla* C.A. Barth and J. Kaplan, J. Mol. Spectroscopy, 1959, 3, 583.lasP. Harteck and R. R. Reeves, “Chemical Reactions in Lower and UpperAtmosphere ,” Interscience, New York, 1961, p. 219.lZ4 T. C. Marshall, Phys. Fluids, 1962, 5, 743.lZ5 F. Kaufxnan and J. R. Kelso, “ Chemical Reactions in Lower and Upper Atmo-sphere ,” Interscience, New York, 1961, p.255.lZ6 J. E. Morgan and H. I. Schiff, J. Chenz. Phys., 1963, 38, 149530 GENERAL AND PHYSICAL CHEMISTRYTo overcome these diiliculties Kauf'man and Kelso l a 7 produced oxygenatoms by flowing ozonised oxygen through a furnace at 100O"a Theyobtained a rate constant E25 of 2.7 x 1014 cm.6 mole-, sec.-1 for M = 0,in the subsequent recombination. Clyne, McKenney, and Thrush 128 passedargon containing minute amounts of oxygen through a discharge and addedmolecular oxygen downstream; they obtained E25 = 8 x 1012 exp (19OO/RT)cm.6 mole-2 sec.-l for M = Ar. In both studies the air afterglow was usedto measure oxygen atom concentrations and the studies agree very well ifthe relative efficiencies of molecular oxygen and argon are taken as 1.7 : I,the value found in the pyrolysis and photolysis of ozone,Studies of the reaction0 + 0, -3 0, + 0, (26)are also affected by the presence of excited oxygen molecules and byhydrogen atom catalysisH + O,+OH + 0,0 + OH+O, + HPhillips and Schif€68 report k26 = (1.5 &- 0.3) x 1O1O ~111.~ mole-1 sec.-lfrom a mass spectrometric study, whereas Mathias and ScW 129 obtainedk26 = 5.4 x lo9 ~111.~ mole-l sec.-l using mass spectrometry to determinethe steady-state ozone concentration in discharged oxygen, allowance beingmade for excited species.The latter value is in better agreement withstudies of the thermal decomposition of ozone.The reaction0 +NO + M - + N O , + Mand its associated air afterglow emission have been extensively studied.The rate constants obtained agree remarkably well 110,130-133 and onlycertain data are given. Clyne and Thrush 131 obtained k, = 9 x 1014 exp(+ 1800 & 400/RT) cm6 mole-2 sec.-l photometrically for M = 0, overthe range 212'-315'~ with relative third-body efficiencies ofHe : Ar : 0, : N, :: 0.58 : 0.87 : 0.87 : 1-0Klein and Herron 133 found it28 = (1.44 & 0.2) x 1016 exp (1930 &- 100/RT)cm.6 mole-a sec.-l mass spectrometrically for M = N,.Kaufman andKelso 132 give the following rate constants for various third bodies a t 3 0 0 " ~in units of cm.6 mole-2 sec.-l: He, 1.8 x 10l6; Ar,O,, 2.2 x 10ls;N2, 3.4 X lo1'; CO,,N,O, 4.7 X 1016; CH4,CF4, 5.0 X 10ls; SF6, 6.8 X 10l6;R20, 14.0 x lola.obeyed the relationKaufmanllo showed that the intensity of the air afterglow emissionI = I,[O]"O]~~ ~ ~la' F.Kaufman and J . R. Kelso, D.iscuss. Faraday SOC., 1964, 37, 26.lS8 M. A. A. Clyne, D. J. McKenney, and B. A. Thrush, Trans. Faradzy SOC., 1966,lSa A. Mathias and H. I. Schiff, Discuss. Paraday SOC., 1964, 37, 38.lSo E. A. Ogryzlo and H. I. SchZf, Canad. J. Chem,, 1959, 37, 1690.lS1 M. A. A. Clyne and B. A. Thrush, Proc. Roy. SOC., 1962, A, 269, 404.lSp F. Kaufman and J. R. Kelso, Symposium on Chemiluminescence, Duke Uni-lta F. S. K l e h and J. T. Herron, J . Chm. Phys., 1964, 41, 1286.61, 2701.versity, 1965CAMPBELL AND THRUSH: DISCHARGE-FLOW SYSTEMS 31and estimated the magnitude of I,. Although I , was independent of pres-sure over the range studied, Clyne and Thrush 131 showed that it dependedsomewhat on the nature of the carrier gas and varied with temperature asexp (+ 1500/RT). Fontijn, Meyer, and S ~ h i f f l ~ determined I, actino-metrically, obtaining 3.8 x lo7 ~ m .~ mole-l sec.-l at 300"~ for an oxygencarrier; they showed that the emission extends from about 3900 A to14,000 A with its maximum intensity at -6000 8.Broida, ScM, and Sugden 135 established that the air afterglow emissionhas banded structure near its short-wavelength Emit at 3975 A, and that itsapparently continuous nature at longer wavelengths arises from unresolvedoverlapping bands. They showed that I,, was independent of pressurebecause the rate-determining processes in the formation and removal ofeIectronically excited NO2 both involved the carrier gas; they proposed adetailed mechanism involving two excited states of NO,.By consideringdata on the quenching of the fluorescence of NO,, Clyne and Thrush 131showed that approximately half the recombination occurs via the emittingstate. New quenching measurements on the NO2 fluorescence and deter-mination of the I, for the corresponding carriers have led Kaufman andKelso132 t o the conclusion that between 50 and 90% of the recombinationoccurs via the emitting state. They also find that the quantity I, decreasesa t pressures around 100 p as predicted by the Clyne-Thrush mechanism.131Hartley and Thrush 136 have shown that these and other observations indi-cate that recombination occurs mainly into the ground state of NO, and thatrelatively free crossing occurs between the excited state and high vibrationallevels of the ground state.From experiments in the 1-2Op pressurerange, Harteck, Reeves, and Chace 137 conclude that the chemiluminescentcombination of 0 and NO is a two-body process. Doherty and Jonathan I38also found a second-order pressure dependence at low pressure where a third-order dependence would be expected, but they observed a dependence onthe nature of the carrier gas used. Experiments in which nitric oxide isreleased in the upper atmosphere suggest that there is a two-body radiativeprocess, but with its emission shifted to the red as compared with laboratoryexperiments at higher pressures.etc.etc.Ice and glasses. Investigations of water in the solid state have so farbeen relatively limited, but interest has revived in the past few years as it isrealized that the response of the solid, and the observation by electron-spinresonance and optical spectroscopy of stabilized intermediates, may beuseful pointers to equivalent occurrences in the liquid. The most interestingresults to date have been obtained with acid and alkaline glasses.Irradiation of glassy solutions of KOH or NaOH a t 7 7 " ~ gives rise totrapped electrons, 0- radicals, and small yields of H atoms.210-212 It hasbeen suggested that the production of the trapped transient species is largelydetermined by the physical state of the 212 and that the glassystate gives rise to higher yields than the polycrystalline state, in generalagreement with a new treatment of ion spurs in dipolar liquids a.nd solids.213On the other hand G(e-t) as measured in solid solutions of KOH or NaOHshows exactly the same smooth dependence on the concentration of OH-,even though the intermediate concentrations give rise to glasses with NaOHand crystalline solids with KOH.214 It is concluded that the only essentialrequirement for the observation of trapped species is a solution, and thatphysical state may not be very important.A limiting G(e-,) = 1.9 wasreached at high concentrations of OH-. Electron spin resonance measure-ments on solid solutions of different cations in H,O and D,O show that boththe cation and protons interact with the trapped electron, and are consistentwith delocalization of the electron over a radius of 3-4 from the centreof the trap.215Trapped electrons are not formed in sulphuric acid glasses because theconcentration of proton donors is too high.Nevertheless, by adding N,O205 T. J. Sworski, J . Arner. Chem. SOC., 1964, 86, 5034.206 T. J. Sworski, ref. 158, p. 263.207 C. Lifshitz and G. Stein, J . Chem. Phys., 1965, 49, 3330.208 A. Rafi and H. C. Sutton, Trans. Paraday SOC., 1965, 61, 877.E. Hayon, Tram. Paraday Soc., 1965, 61, 723.2 l o D. Schulte-Frohlinde and K. Eiben, 2. Nutzlrforsch., 1962, Ira, 445; 1963, 18a,99; B. G. Ershov, A. K. Pikaev, P. Y. Glazunov, and V. I. Spitsyn, Izuest. Akad. NaukS.S.S.R., Ser. Khim., 1964, 10, 1755.211 K.Eibon and D. Schulte-Frohlinde, 2. phys. Chenz. (Frankfurt), 1965, 45, 20.212 T. Henricksen, Radiation Res., 1964, 23, 63.21s R. Schiller, J . Chem. Phys., 1965, 43, 2760,214 L. Kevan, J . Chem. Phys., 1965, 69, 1081.215L. Kevan, J . Arner. Chem. SOC., 1965, 87, 1481COLLINSON : RADIATION CHEMISTRY 105or Fe2+ ions to the glass, evidence for the formation of mobile electrons(e-,) was obtained, giving C(e-J = 1.7 and G(H,) = 043.216 Again thedegree of crystallinity was found to be important, G(N,) from solutions ofN20 being larger in the glass, whereas G(H2) was constant. The effects ob-served by electron-spin resonance in acidic glasses containing a wide rangeof solutes have been interpreted in terms of reactions of mobile electronsand positive holes, which may initially exist in irradiated pure ice in anesciton-like bound216 F.S. Dainton and F. T. Jones, Trans. Paraday SOC., 1965, 61, 1681.217 P. N. Moorthy and J. J. Weiss, ref. 158, p. 1807. MOLTEN SALTSBy D. Inman and S. H. White(Clbemistry Department, IThe City University, London, E.C. 1)INTEREST in this field has rapidly expanded in recent years in parallelwith the increasing importance of high-temperature chemistry in moderntechnology. The practical applications have been reviewed.1 Although thistopic has not been covered before in this series, Reviews are availablewhich cover the literature up to 1964. Annual Reviews of molten-saltwork have appeared in the Russian literature since 1959. Two importantbooks on the subject have recently been p~blished.~ Bibliographies ofmolten-salt publications have been prepared.5The topic has also been featured at several international conferences[for example, 18th I.U.P.A.C.(Montreal, 1961) ; 7th I.C.C.C. (Stockholm,1962); 19th I.U.P.A.C. (London, 1963); 15th C.I.T.C.E. (London, 1964) ;1st Australian Conference on Electrochemistry (Sydney and Hobart, 1963)l.A discussion group has been organised in Great Britain. Gordon Conferenceson Molten Salts were held in 1959, 1961, 1963, and 1965 in the U.S.A.This report will mainly review the work carried out in 1965. The choiceof topics has necessarily been somewhat arbitrary. We shall survey thosetopics which, for other solvents, would be covered in reports on “ionassociation,” “ acid-base equilibria,” and ‘‘ electrode double-layer and elec-l1). Inman, New Scientist, 1965, 25, 220; R.Littlewood, British Chemical En-gineering, Oct., 1962: W. Sundermeyer, Angew. Chem. Infernat. Edn., 1965, 4, 222.a G. J. Hills, D. Inman, and L. Young, Proc. 8th C.I.T.C.E., 1956, 90; G. J. Jam,C. Solomons, and H. J. Gardner, Chem. Rev., 1958,58,461; “ Physico-Chemical Measure-ments at High Temperatures,” ed. J. O’M. Bockris, J. L. White, and J. D. Mackenzie,Butterworths, London, 1959; G. E. Blomgren and E. R. Van Artsdalen, Ann. Rev.Phys. Chem., 1960, 11, 273; H. Bloom and J. O’M. Bockris, in “Modern Aspects ofElectrochemistry No. 2,” ed. J. OW. Bockris, Butterworths, London, 1959;‘H. C. Gaurand B. B. Bhatia, J . Sci. I d . Res., India, 1962, 21, A , 16; R.W. Laity, in ReferenceElectrodes: Theory and Practice,” ed. D. J. G. Ives and G. J. Janz, Academic Press,New York, 1961 ; Yu. K. Delimarskii and B. F. Markov, “ Electrochemistry of FusedSalts,” Sigma Press, Washington, 1961; E. R. Van Artsdalen, in “ The Structure ofElectrolytic Solutions,” ed. W. J. Hamer, Wiley, New York, 1959; T. B. Reddy,Electrochem. Technol., 1963, 1, 325; H. C. Gaur and R. S. Sethi, J . Electroanalyt. Chem.,1964, 7 , 474; R. D. Reeves, Dim. Abs., 1965, 28, 118; J. D. Corbett, in “ Survey ofProgress in Chemistry,” ed. A. F. Scott, Academic Press, New York, 1964, Vol. 11;J. D. Corbett and F. R. Duke, in “ Techniques of Inorganic Chemistry,” ed, H. B.Jonassen, Interscience, New York, 1963, Vol. I ; G.J. Janz and S. C. Wait, Quart.Rev., 1963, 225; H. Bloom and J. W. Kastie, in “Non-aqueous Solvent Systems,”ed. T. C. Waddington, Academic Press, London, 1965; E. A. Ukshe, Uspekhi Khim.,1965,34,322; 0. J. Kleppa, Ann. Rev. Phys. Chem., 1965,16,187; G. Serravalle, Ricercasci. Rend., 1964,4, A, 549; A. F. Alabyshev, M. F. Lantratov, and A. G. Morachevskii,“ Reference Electrodes for Fused Salts,” Sigma Press, Washington, 1965; J. Lumsden,“ The Thermodynamics of Molten Salt Mixtures,” Academic Press, London, 1966.8 A. G. Morachevskii, Zhur. priklad. Khim., 1960, 33, 1434; 1961, 34, 1398; 1962,35, 1390; A. G. Morachevskii and B. V. Patrov, ibid., 1963, 38, 1374; 1964, 37, 1396;1965, 38, 2138.(a) “ Molten Salt Chemistry,” ed. M. Blander, Interscience, New York, 1964;( b ) “Fused Salts,” ed.B. R. Sundheim, McGraw-Hill, New York, 1964.6 G. J. Janz, Technical Bull. Series, Rensselaer Polytechnic Inst., Troy, NewYork., 1st edn., 1958; 2nd edn., 1961INMAN AND WHITE: MOLTEN SALTS 107trode reactions”; that is, we shall concern ourselves with the electro-chemistry of molten salts. Such important matters as phase diagrams,theories of pure molten salts, and electroanalysis may be incompletelycovered.Nitrates.--Becrtuse they are easily handled and have relatively low meltingpoints (cf. chlorides, sulphates, carbonates, silicates, etc.), inorganic alkaliand alkaline earth metal nitrates and their eutectic mixtures arc3 frequentlylstudied as examples of the molten state. The bulk properties such asdensity,6, 7 surface tension,* and conductance 6 , 7, 9 are well-established in many cases.Although the limit of electrochemical stability on the anodic side is theoxygen evolution process, the reduction of nitrate ions intervenes beforemetal deposition on the cathodic side (at least for the alkali metal andalkaline earth metal nitrates).This cathodic process may be at leastformally regarded as the reduction of the acid component NO2+ arisingfrom the acid-base dissociation of the anion :NO,- * NO,+ $- 0 2 -NO,+ i- 28 +NO,-Earlier work lo established the stoicheiometry of the overall nitrate reduc-tion process. Topol et aE.ll have disputed the reality of the species NO,+in nitrate melts on the basis of the results obtained by chronopotentiometryand linear sweep voltammetry in the NaN0,-KNO, solvent to which theyadded acidic and basic species. Their results suggest that the acid-baseequilibrium in the melt involves NO, rather than NOz+.On the otherhand, Brough and Kerridge l2 postulated the formation of NO,+ in account-ing for some metal reactions in fused nitrate media, Although NO,+ needonly have a transient existence in this context, its lifetime may tvell beenhanced in the presence of the acidic Li+ ions contained in their melts.Hladik and Norand l3 haye studied the reduction of NO,- and NO2- ionsin the LiC1-KC1 eutectic. The nitrate ions appear to be directly reduceda t the high temperatures ( > 350 “c) of their experiments, although the resultscould also be explained on the basis of a fast nitrate dissociation reactionprior to electron transfer.Small but significant quantities of NO,- ionshave been shown to be present in freshly fused NaN03-H;N0, melts byanodic voltammetry l4 and chronopotentiometry .15 The production of NO ,-6 G. J. Jam, A. T. Ward, and R. D. Reeves, Technical Bull. Series, RensselaerPolytechnic Inst., Troy, New York, 1964.J. P. Frame, E. Rhodes, and A. R. Ubbelohde, Trans. Farday SOC., 1959, 55,2039.C. C. Addison and J.@ N. P. Popovskaya, P. I. Protsenko, and A. F. Ehva, Zhur. rteorg. K h h ,1964, 9, 1211.lo G. J. Hills and K. E. Johnson, in “ Advances in Polarogmphy,” sd. I. Longmuir,Pergamon Press, 1960, vol. 111,982; J . Electrochem. SOC., 1961,108,1013; H. S. Swoffordand H.A, Laitinen, J . Electrochem. SOC., 1963, 110, 814.11L. E. Topol, R. A. Osteryoung, and J. H. Christie, J . EteCtrochem. Soc., 1965,San Francisco Programma, AbsGract No. 207.1% B. J. Brough and D. H. Kerridge, Inorg. Chem., 1965, 4, 1353.1s J. Hladik and G. Morand, Bull. XOC. chim. Frunce, 1965, 828.1 4 X. S . Swofford and 9. G. McComiek, Anulyt. Chm., 1965, 37, 970.1 5 D. Inman and J. Braunstein, Chem. Cmm., 1966, 148.Coldrey, J . Chem. Soe., 1961, 468108 GENERAL AND PHYSICAL CHEMISTRYis attributed to nitrate reduction by organic impurities and thermal decom-position. Duke and Kust l6 claimed to have genemted 02- ions in a nitratemelt by the reduction of 0, a t a platinum electrode, although Swoffordand McCormick l4 suggest that the 02- ions are formed by the reductionof the melt itself (see above).Bombi and Fiorani 17 have employed thelatter process to generate 02- ions in nitrate melts. Kust l8 used sodiumcarbonate as the source of oxide ions to confirm the E" values for theoxygen electrode reported earlier.16 Sodium carbonate was rejected 14 onthe basis of the incomplete dissociation of GO,,- ions in nitrate melts,although in Kust's work nitrogen purging of the melt would have driventhe reaction CO,,- 4 CO, + 0,- to completion. The E" values obtainedby this method agree to less than 0.3 mv with those obtained followingthe coulometric addition of 0,- ions. Geckle l9 has studied the effect onnitrate reduction of small quantities of water dissolved in the NaN0,-KNO,eutectic at 350"c using the rotating platinum disc indicator electrode anda controlled potential polarograph.The limiting current of the cathodicwave observed is proportional to the bulk concentration of H,O. Coulo-metry, visual observation and analysis of the melt, and mass spectrometryof the gases evolved from the melt were also employed to elucidate thereactions. He postulates that the reduction of NO3- to NO,- ions isinduced by the electroreduction of dissolved water which is subsequent.lyregenerated by a series of chemical reactions.Several workers have investigated the mechanisms of anodic reactionsin pure fused nitrates. Gupta and Sundheim 2o have examined the anodicreaction at a rotating platinum macro-electrode in molten silver nitrateusing both steady-state and fast-sweep voltammetry a t 238"~.Two steps,at +0.5 v and 3-0-9 v versus the Ag-Ag+ reference electrode, were observedin the current density-electrode potential plots. The process a t +0.5 v istransport-controlled. The reactant 02- ions arise from the dissociation ofNO 3- ions. The observed increase in concentration polarisation when acidicS,O,2- ions are added to the melt supports this view. The Tafel plot leadsto an na value of approximately 1.2 and thus to n = 2 and a = 0.6. Inaddition, they tentatively suggest that the ni5rate ion uizdergoes directelectron transfer at +0.9 v to give an NO, zadicml which then decomposes;na for this reaction was found to be 0.9. On interrupting the polarisingcurrent they found that a steady potential was reached within a few milli-seconds.Arvia and his co-workers have observed similar steady potentialsfor silver nitrate21 and sodium They interpret these results interms of reversible cells of the typeM 1 MNO, 1 NO,,NO,O, (Pi) M = Ag,&Cell e.m.f.s calculated from the thermodynamic data agree with the steadyF. R. Duke and R. N. Kust, J . Amer. Chern. SOC., 1963, 85, 3338.l7 G. G. Bombi and M. Fiorani, Talanta, 1965, 12, 1053.I*R. N. Kust, J . Phys. Chem., 1965, 69, 3662; Inorg. Chern., 1964, 3, 1035.l9T. A. Geckle, U.S. A.E.C. TID. 21511, 1965 (Chem. Abs., 1965, 63, 9455).2o N. Gupta and B. R. Sundheim, J . Electrochem. Soc., 1965, 112, 836.21 W. E. Triaca and A. J. Arvia, EZectrochim. Acta, 1964, 9, 919.A. J.Arvia, A. J. Calandra, and W. E. Triaca, Electrochim. Ada, 1964, 9, 1422INMAN AND WHITE: MOLTEN SALTS 109potentials obtained experimentally. Baraboshkin et u Z . , ~ have calculatedexchange currents for silver electrodes in molten silver nitrate in the tem-perature range 220-350" c from their earlier observations 24 of the formationof silver crystal nuclei on platinum during the electrolysis of silver nitrate.The exchange currents for silver electrodes in fused salts have also beenmeasured.25 Triaca and Arvia 26 have made a detailed investigation of theanodic reaction occurring on bright platinum during the electrolysis of moltennitrates containing silver ions, over a temperature range from 220 to 470"c.The processes are temperature-dependent.In the range 220-29O"c twoconsecutive Tafel slopes of 2 R T/F and R T/F were observed. Two potential-dependent rate controlling steps are proposed to explain these results.Above 350"c the Tafel slope is RTIF. At the higher temperatures therate of decomposition of the nitrate ions is presumably high enough tomaintain a finite concentration of 0 2 - ions at the electrode surface; thus,the direct discharge of this ion is maintained at all potentials. At the lowertemperatures the direct discharge of nitrate ions may also occur (see twoTafel slopes observed). These authors 27 have also published a preliminaryaccount of their studies of similar anodic reactions on graphite. Thesereactions are observed to be temperature-dependent.In a series of papers, Delimarskii and Shilina 28 report the cathodicreduction at a rotating platinum disc-electrode of the halogens formed insitu by reaction of the LiN0,-NaN0,-KNO, eutectic with small quantitiesof added halides.They determined the solubility, diffusion coefficient andits temperature dependence, and the thickness of the diffusion layer foreach of the halogens Cl,, Br,, and I,. The supporting electrolyte exhibiteda wave on the electrode potential-current density curve a t -2.6 v versusa massive platinum quasi-reference electrode which they attributed to thereactionNO,+ + e + NO,However, Hills and Johnson, and Swofford and Laitinen had earlier 10reported waves in nitrate melts a t much less negative potentials. Swoffordand Propp 29 have investigated the anodic oxidation of bromide and iodideions in the KN03-NaN03 eutectic a t a rotating platinum micro-electrodeat 250"~.The previously 2 s v 54 observed oxidation of halide ions by thenitrate melt is thus probably related to the acidic nature of the Li+ ionswhich increases the oxidising power of the nitrate ion. Reversible one-electron waves were obtained. The E" values for the reversible reactionsX, + 2e s 2X-are +0.126 & 0.005 v for iodine-iodide and f-0.644 & 0.010 v for bromine-23 A. N. Baraboshkin, L. T. Kosikhin, and N. A. Xaltykova, DoElady Akad. NaukS.S.S.R., 1965, 160, 145.24 A. N. Baraboshkin, L. T. Kosikhin, and N. A. Saltykova, Doklady Akad. NaubS.S.S.R., 1964, 155, 880.26 Yu. K. Delimarskii, E. V. Panov, and A.V. Gorodyskii, Ukrain. khim. Zhur.,1965, 31, 782.28 W. E. Triaca and A. J. Arvia, Electrochim. Acta, 1965, 10, 409.2 7 W. E. Triaca and A. J. Arvia, Electrochim. Acta, 1965, 10, 973.28 Yu. K. Delimarskii and G. V. Shilina, Electrochim. Acta, 1965, 10, 973; Ukrain.2B H. S. Swofford and J. H. Propp, Analyt. Chem., 1965, 37, 974.khim. Zhur., 1964, 30, 1045; G. V. Shilina, ibid., 1963, 31, 693110 GENERAL AND PHYSICAL CHEMISTRYbromide versus their Ag-Ag+ reference electrode. Novik and Lyalikov's 3*proposal that the reaction2KI + KNOs + KNO, + K,O + I,accounts for the increase in the limiting current of the nitrite oxidationwave upon addition of KI to this nitrate melt is thus shown to be incorrect.Swofford and Holifield31 have recently published a study of the anodicoxidation of halide ions in the molten sodium nitrate-potassium nitrateeutectic a t a dropping-mercury electrode.The oxidations were irreversibleand the processes rather complicated. The mercury itself is not inactive;the formation of insoluble mercury halides and their subsequent dispersalalong with adsorption phenomena certainly play a part in the overall processa t the anode. Delimarskii and Shilina 32 have examined the polarographicbehaviour of thallium ions in the LiN0,-NaN08-KN03 eutectic at 160"~.They calculate a diffusion coefficient of 7.55 x 10-6 cm.2 sec.-l from theirresults. Swofford and Holifield s3 have investigated the electroreduction ofTl+, Cd2+, and Pb2+ ions at a dropping-mercury electrode in the moltensodium nitrate-potassium nitrate eutectic.T1+ and Cd2+ ions gave reversiblepolarographic waves but the reduction of Pb2+ ions was apparently rathercomplicated. Two waves were observed a t the higher Pb2+ ion concentra-tions employed. The latter results are inconsistent with those of earlierworkers34 who employed nitrate melts containing Li+ ions a t lower tem-peratures. The possible significance of surface phenomena is discussed.Chronopotentiograms 34c for Pb2+ ions in NaN03-KN03 also exhibited twosteps a t high Pb2+ ion concentrations. The second step was tentativelyascribed to the electroreduction of lead oxide formed rapidly on the hangingmercury drop surface after the initial deposition of lead.The thermodynamic properties of solutions of AgNO, in mixtures ofalkali metal and alkaline earth metal nitrates have been determined frommeasurements of the e.m.f.s of cells with transferen~e.~~ Most of the systemsshow almost ideal behaviour. For AgNO, in RbNO, the activity coefficientis less than unity.An interpretation in terms of complex-formation issuggested. The plots of e.m.f. versus the logarithm of the reciprocal ofthe AgNO, mole fraction for the Rb-Ca, Na-Ba, and K-Ca nitrate solventsexhibit temperature-dependent inflections. (This could be interpreted interms of changes in the gross ionic associations of the solvent melt.)Murgdescu and Topor36 propose a form of ion-pairing to account forthe comparable mobilities observed for lithium ions and the proper alkalimetal cations during the electromigration of Lif ions through alkali metalnitrate solvents.The inversion of the relative mobilities of the alkali metalcations at a specific temperature and composition has been observed in30 R. M. Novik and Yu. S. Lyalikov, Zhur. analit. Khim., 1958, 13, 691.3lH. S. Swofford and C. L. Holifield, Analyt. Chem., 1965, 37, 1513.32 Yu. K. Delimarskii and G. V. Shilina, Elektrokhimiya, 1965, 1, 632.33 H. S . Swofford and C. L. Holifield, Analyt. Chern., 1965, 37, 1509.34 (a) N. Nachtrieb and M. Steinberg, J . Amy. Chem. Soc., 1950, '92, 3558; (b) J. H.Christie and R. A. Osteryoung, ibid., 1960,82, 1841; (c) D. Inman and J. O'M. Bockris,J . Electroanalyt. Chem., 1962, 3, 126.35 M. Bakhs, d. Guion, and J. P. Brenet, EZectTochim.Ada, 1965, 10, 1001.313 I. G. Murgulescu and D. Topor, Rev. Roumaine China., 1964, 9, 815INMAN AND WHITE: MOLTEN SALTS 111LiN03-KN03 mixtures by Chemla and Lantelme.37 However, the diffusioncoefficients 38 are in the order DLi+> DN&+ > D K + and are independent oftemperature and composition. The anion-cation interactions proposed toexplain these results have been discussed quantitatively using a form ofthe Nernst-Einstein equation. Some transition metal ions dissolved invarious molten nitrate solvents have been titrated conductometrically withhalide ions.39 Density and conductance data for the pure solvents havealso been reported. The conductance data do not compare well with data,reported for similar systems.4* Duke and King using a d.c.techniqueobtained results for KN03 which are in excellent agreement with those ofAngell41 who employed the a.c. method. It is suggested that these datacould be used for cell calibration, thus eliminating the use of aqueous stan-dards. Bredig,42 in criticising the work of Papaioannou and Harrington,39has pointed out errors in calculation and the incorrect choice of system fortheir proposed model to explain the large decrease in conductance on addingKBr. Bredig suggested that the startling effects observed may be artifactsof the system. However, as evidenced above, the presence of Li+ ions innitrate melts often leads to unusual effects, so further experimental investiga-tion of these systems under well-controlled conditions seems desirable.In recent years, the problem of whether complexes are present in moltensalt solutions has been the subject of many studies.These experimentalinvestigations may be divided into two groups, those which provide directevidence, e.g., spectroscopy and ionophoresis, and those which provideindirect evidence, e.g., potentiometry, polarography, chronopotentiometry,double-layer capacity measurements, etc. In the former group, the spectro-scopic techniques which have been used up to the present cannot alwaysdistinguish between the transient near-neighbour ion-ion interactions chaxac-teristic of these systems, and any discrete kinetic entities which may bepresent. These are best distinguished on a lifetime basis, although therelaxation times ( 10-11-10-12 sec.) of conventional u.v., visible, and i.r.spectroscopy are too short to effect this separation.The n.m.r. techniquemay provide a solution of this problem. The appearance of a new discreteband, which is not present in the solvent, in the Raman spectrum of themolten salt solution is strong evidence for comp1ex-formation.*~ The iono-phoretic technique may be used to deduce the presence of long-lived entities.Ni2+ ions move in the cathodic direction in the LiC1-KC1 eutectic.44 Thus,any discrete chloro-nickel complexes must be sufficiently short-lived for thegross movement of nickel to be towards the cathode. Co2+ ions wereobserved to move in the anodic direction,45 so species such as [COC~,](~-~)+37 31. Chemla and F. Lantelme, Electrochim.Acta, 1965, 10, 663.38 F. Lantelme and M. Chemla, Compt. rend., 1964, 258, 1484.39 P. C. Papaioannou and G. W. Harrington, J . Phys. Chem., 1964, 68, 2424, 2433.40 F. R. Duke and L. A. King, J . Electrochem. SOC., 1964, 111, 712.C. A. Angell, J . Electrochem. Soc., 1965, 112, 956; J . Phys. Chem., 1964, 68,4 2 M. A. Bredig, J . Phys. Chem., 1965, 69, 1753.43 For example, see J. K. Wilmshurst, J . CJwm. Phys., 1963, 39, 1779; D. E. Irisha*nd T. F. Young, ibid., 1965, 43, 1765; W. Bues, 2. anorg. Chem., 1955, 279, 104;H. Bloom and J. O'M. Bockris, ref. 4b.44 G. Alberti, S. Allulli, and G. Modugno, J . Chromatog., 1964, 15, 420.45 G. Alberti, G. Grassini, and R. Trucco, J . EZectroanalyt. Chem., 1962, 3, 283.1917112 GENERAL AND PHYSICAL CHEMISTRY(where n > 2) must be long-lived.Ion-exchange measurements 46 for Ni2fand Go2+ ions in melts also support these observations. A quantitative evalua-tion of this type of measurement has so far not been formulated. At present,the existence of a definite complex species is best decided on the basis ofevidence from several independent techniques. Gruen has reviewed thespect,roscopy of transition metal ions in nitrate^.^^ 47 The change in wave-length of the absorption band of Co2+ in the molten LiNO,-KNO, eutecticat 1 6 0 " ~ on addition of chloride ions hams been interpreted in terms of theformation of a tetrahedral chloro-complex. Tananaev and Dzhurinskii 48reported similar shifts. They interpret these in terms of the step-wiseformamtion of CoC1, and [ CoCI,]- before tetrahedral [ CoC1, ] 2 - .Padova andhis co-workers 49 have obtained density and spectroscopic data for neody-mium nitrate dissolved in the NaN0,-KNO, eutectic. The sharp changein the temperature coefficient of density at 131 together with large changesin molar volume in more dilute solutions suggest that the species presentin solution are concentration-dependent. This confirms their earlier spectro-scopic work.5o Baddiel, Tait, and Janz 51 have investigated the Ramanand i.r. spectra of pure molten silver nitrate a t 250"~. The splitting ofthe degenerate stretching vibration of NO,- observed indicates strong ion-ion interactions. Cigdn and Mannerstrand 52 have determined the solubilityproduct of AgBr and the stability constants of various bromo-complexes,by solubility and e.m.f.measurements. The latter measurements showfhat the Nernst equation is valid for the Ag-Agf electrode in their nitratesolvent over the concentration range 10-1-10-5 mole kg.-l. However, thevalidity of their results must be doubted as they obtained their stabilityconsDants from the e.m.f.s of cells where the Ag+ ion concentrations were102-103 times lower than mole kg.-l. The applicability of the Nernstequation in this concentration range was not established. Arnikar, Sharma,and Tripathi 53 have investigated Ag-Ag + concentration cells with trans-ference in molten KNO,, NaNO,, and their eutectic mixture, and show theliquid junction potential to be negligible in the concentration range examined.They determined the solubility product of silver chloride in the various nitratesolvents using NaCl and CdCl, as titrants.Ks was found to be independent'of the composition of the solvent media. They also calculate a degree ofdissociation for CdC1, (ctCdC12 = 0.65 if ctsaC1 = 1) and attribute this tocomplex-formation. Bombi, Fiorani, and Mazzocchin 54a have argento-metrically titrated halides and cyanides in the LiN03-KN0, eutectic a t150 Oc ; the titrations were monitored by potentiometry, bipotentiometry,and biamperometry. The solubility products and a stability constant for413 G. Alberti, A. Conte, and S. Allulli, J . Chromatog., 1965, 18, 564.4 7 D. M. Gruen, Quart. Rev., 1965, 349.48 I. V. Tananaev and B.F. Dzhurinskii, Doklady Akad. NauE S.S.S.R., 1960,49 J. Padova and J. Soriano, J . Chem. and Eng. Data, 1964, 9, 510.50 J. Padova, 19th I.U.P.A.C., 1963, Abstract No. B3-6.61 C. B. Baddiel, M. J. Tait, and G. J. Jam, J . Phys. Chem., 1965, 69, 3634.5 2 R. Cigen and N. Mannerstrand, Acta Chem. Scand., 1964, 18, 2203; 1755.53 H. J. Arnikar, D. K. Sharma, and R. Tripathi, Indian J . Chem., 1965, 3, 7.64 (a) G. G. Bombi, M. Fiorani, and 0. A. Mazzocchin, J . Electroanalyt. Chem.,134, 1374; 1960, 135, 94; 1961, 139, 120; 1961, 140, 374.1965, 9, 457; ( b ) H. S . Swofford, Analyt. Chem., 1965, 37, 610INMAN AND WHITE: MOLTEN SALTS 113[AgCNJ- were calculated. CN- and I- ions are both unstable in thissolvent, which contains Lif ions. As discussed above, this can be relatedto the acidic nature of the Lif ion.Swofford54b has titrated halide ionsin the NaN0,-KNO, eutectic a t 250"~. Bornbi and Fioranil7 have in-vestigated the acidic nature of various cations in the NaN0,-KNO, eutecticusing a Pt-0, indicator electrode to monitor 0 2 - ion concentrations duringthe titration of the cations by 0 2 - ions generated in situ by the electro-reduction of NO,- ions at a Pt cathode. Cd2+, Co2+, Hg2+, Mg2+, &In2+,Ni2+, Pb2+, and Zn2+ ions gave stable nitrate solutions, but of these onlyCd2f and Zn2f ions gave sharp inflexions, evidencing their relatively strongacid nature. Tien 55 has determined the stability constants B1 and B2 forsilver cyanide complexes in the NaN0,-KNO, eutectic at 248" c using a,gallium-in-glass reference electrode.The Agf and CN- ion concentrationswere such that no precipitation occurred. The p-values do not agree withthose of earlier ~ o r k e r s . ~ ~ - ~ * Binuclear complexes were shown to be absent.Chamberlin 59 has calculated stability constants for silver chloride complexesin molten potassium nitrate from polarographic data, using a modificationof the DeFord and Hume 6o method. Inman 61 has reported stability con-stants for the chloro- and bromo-complexes of cadmium in the moltenNaN0,-KNO, eutectic at 250"c. An electrode of the third kind, reversibleto Cd2f ions, to monitor the changing metal-ion concentration duringtitration with KCI and KBr was established. The e.m.f. data were usedto calculate successive stability constants for these systems using the methodpublished earlier.62 The stability constants are compared with those deter-mined by other methods and discussed in terms of the co-ordination ofCd2f ions in nitrate melts.Narayan and Inman 63 have described theapplication of current-reversal chronopotentiometry to the investigation ofmetal ions dissolved in molten nitrates where the metal produced by cathodicreduction is oxidised by the solvent. Schlegel 64 has continued the studyof acid-base reactions in molten nitrates with an investigation of thedichromate-chlorate system in the presence of excess chloride. TheequilibriumCr,0,2- + C10,- + C10,+ + 2Cr0,2-is followed by the slow stepThe equilibrium constant ( K ) at 250"c is 1-6 x 10-lo mole kg.-l and therate constant (k) 0.208 kg.mole-l min.-l These values are compared withthose established earlier for nitrate-dichromate 65 and bromate-dichromate.66xkc10,+ + c1- + c1, + 0,65 H. Ti Tien, J . Phys. Chem., 1965, 69, 3763.66 S. N. Flengas and E. Rideal, Proc. Roy. SOC., 1956, A , 233, 443.57D. L. Manning and M. Blander, Inorg. Chem., 1962, 1, 594,68 J. Jordan and J. Prendergrast, Proc. 7th I.C.C.C., 1962, 102.59 J. Chamberlin, Diss. Abs., 1964, 25, 2749.6o D. D. DeFord and D. N. Hume, J . Amer. Chem. Soc., 1951, 73, 5321.61 D. Inman, Electrochim. Acta, 1965, 10, 11.szD. Inman, I. Regan, and B. Girling, J . Chem. SOC., 1964, 348.s3 R. Narayan and D. Inman, J. Polarog. SOC., 1965, 11, 27.134 J. Schlegel, J . Phys.Chem., 1965, 69, 3638.6 5 F. R. Duke and S. Yamamoto, J . Amer. Chem. SOC., 1959, 81, 6378.66 F. R. Duke and E. Shute, J . Phys. Chem., 1962, 66, 2114114 GENERAL AND PHYSICAL CHEMISTRYNokhosoev and Aleikina 67 have shown that chlorides and dichromatesreact a t temperatures greater than 400"c. Shams El Din and his CO-workers 68 continue their potentiometric studies of acid-base reactions inmolten salts. They have shown that the reaction between chromic oxideand molten potassium nitrate is2Cr0, + 2N0,- -+ Cr,O,*- + N,O, --+ 2N0, + &02with an activation energy of 25.3 kcal. mole-l. Paul and Dev 69 haveexamined the nature of Lewis acids in molten acetamide. A review ofacid-base and redox reactions in ionic melts has appeared.70 Courmert,Porthault, and Merlin 71 have studied the acid properties of some condensedphosphates in alkali metal nitrates at 350"c using the technique developedby Shams El Din.72 Sodium and trisodium orthophosphate were used asreactive bases.The results are in accord with the relative acid strengthsfound for condensed phosphates in the solid state. Courgnaud and Tr&millon 73 have investigated acid-base a,nd electrode reactions in moltencalcium nitrate te trahydrate .Halides.-Fluorides. They are highly corrosive and consequently difEcultto handle, but nevertheless a considerable amount of information on themis available. They are very important industrially in metal-winning, -plating,and -forming. The system NaF-AIF', has been extensively studied becauseof its importance in the aluminium industry.The nuclear reactor at OakR'idge which employs a molten fluoride reactant 7* has also necessitated alarge number of fundamental studies of these systems.75Their use as electrolytes for the electrodeposition of the transition metalsof Groups IVA, VA, and VIA has recently been reported. 76 Previous attemptst o deposit these metals from aqueous, non-aqueous, and other molten saltsystems in general either failed or gave rise to non-coherent or dendriticdeposits. The deposition of niobium from a solution of KF (26.2 wt. yo),LiF (10-5 wt. %), NaF (47.0 wt. %), and K2NbF7 (16.2 wt. yo) at 775"cand a current density of 50 ma ~ r n . - ~ has been reported.76 The importantvariables are temperature, valency state, and impurity level. Substantiallypure fluorides are required, and an explanation is advanced in terms ofcomplex ions of the appropriate stability for reduction at the cathode.The effect of the specific adsorption of impurity anions (e.g., chloride) ORthe electrode double layer may also be an important factor.77 Chrono-67 M.V. Mokhosoev and S. M. Aleikina, Zhur. neorg. Khim., 1964, 9, 1684.6BA. M. Shams El Din and A. A. El Hosary, J. Electroanalyt. Chem., 1965, 9,349; A. M. Shams El Din, A. A. El Hosary, and A. A. A. Gerges, ;bid., 1964, 8, 312.6 Q R. C. Paul and R. Dev, Indian J. Chem., 1965, 3, 315.7 O I. Slama, Chern. listy, 1965, 59, 792.7 1 N. Courmert, M. Porthault, and J. C. Merlin, Bull. SOC. chim. France, 1965, 910.78 A.M. Shams El Din, Electrochim. Acta, 1962, '7, 285; A. M. Shams El Din, A. A.7 3 R. P. Courgnaud and B. T. Trhillon, Bull. Soc. chim. France, 1965, 752; 758.74 W. R. Grimes, Nuclear News-A.N.S., May, 1964, 3.75 E.g., M. Blander, W. R. Grimes, N. V. Smith, and G. M. Watson, J . Phys. Chem.,1959, 63, 1164; 5. P. Young, Analyt. Chem. Div. Ann. Progr. Report, O.R.N.L. 3243,Dec., 1961; S. Cantor, R. F. Newton, W. R. Grimes, and F. F. Blankenship, J. Phys.Chem., 1958, 62, 96: K. Grjotheim, 2. phys. Chem. (Frankfurt), 1957, 11, 150.76 G. W. Mellors and S. Senderoff, J. Electrochem. SOC., 1965, 112, 266.77 A. D. Graves and D. Inman, Ekctroplating and Metal Finishing, in the press.El Hosary, and A. A. A. Gerges, J. Electroaaatyt. Chem., 1963, 6, 131INMAN AND WHITE: MOLTEN SALTS 115potentiometric measurements 7s have been made on dilute solutions10-4 mole kg.-1) of potassium fluorotantalate in the LiF-NaF-KF eutectiobetween 650 and 850"c to obtain a better understanding of the mechanismsinvolved in metal deposition in these systems.The reduction of quh-quevalent tantalum occurs by a two-step process which may be writM[TaF,I2- + 38 + TaF,(s) + 5F-andTaF,(s) + 28 --+ Ta, + 2F-The first step is diEusion-controlled, as charge transfer is rapid. Thediffusion coefficient for the [TaF7J2- ion is 1.5 x cm.2 sec.-l a t 750°c,and the activation energy between 650 and 800"~ is 8.5 kcal. mole-l.The low value of the diffusion coefficient and the high activation energyare consistent with a highly co-ordinated species such as [TaF,]".Thesecond step is not diffusion-controlled and poorly understood. Mellors andSenderoff 7 9 have indicated that the sexivalent fluorides of tantalum andniobium can be prepared by the electrolysis of molten alkali metal fluoridescontaining the quinquevalent ions using an inert anode. The oxidationprocess has been studied by chronopotentionietry. The latter authorshave also measured the densities and surface tensions of the systems LiF-Nap-ZrF, and LiF-KF-ZrF4. Plots of surface tension uemus log CzrpFIgive straight lines with inflections at concentrations corresponding to thestoicheiometries of complex compounds. The strength of the complex ionsis less in LiF-NaF-ZrF, than in LiF-KF-ZrF8, The presence of [ZrF7]3-is postulated, but [ZrF,J2-, the most stable configuration in the solid, isabsent in the melts containing sodium ions.The decrease in strength ofthe complexes in La-NaF-ZrF, may be due to the presence of the morehighly polarising Naf ion. Polishchulr 81 has measured the electrical con-ductivities of the ICF-KBF,-3K2ZrF6/2B 203 system. The temperaturedependence of electrical conductivity indicates an ionic structure and alsoa change in melt structure which is interpreted in terms of the formationof [ZrF,I3- ions. Sheiko et aLS2 have examined the anodic dissolution ofzirconium in fluoride and mixed fluoride-chloride melts at 800"c over arange of current densities. ZrF2 is found t o be insoluble in 60 mole yoKF40 mole yo NaF, but in mixed melts an exchange reaction occurs givingZrC1, (see also Smirnov and Kudyakov 83), The products of anodic dis-solution and corrosion are found in the anodic sludge where zirconium isfound in the bivalent form.Investigation of the fundamental processesinvolved in the winning of aluminium continue t o receive attention. Pion-telli and his co-workers s4 have investigated anodic phenomena at carbon7a S . Senderoff, G. W. Mellors, and W. J. Reinhart, J . Electrochem. SOC., 1965,112, 840.7 9 G. W. Mellors and S. Senderoff, J . Ekctrochem. Soc., 1965, 112, 642.G. W. Mellors and S. Senderoff, Proc, 1st Australian Conference on Electro-chemistry, Pergamon Press, New York, 1964, 578: J . Electrochem. Soc., 1964, 111, 1366.P. A. Polishchuk, Z h u ~ . neorg.Khim., 1964, 9, 921.8p I. N. Sheiko, T. N. Grechina, and V. T. Barchuk, Ukruin. khim. Zhur., 1964,80, 1055.88 M. V. Smirnov and V. Ya Kudyakov, Zhur. neorg. Khim., 1965, 10, 1211.€2. Piontelli, B. Mazza, and P. Pederferri, Electrochim. Acta, 1965, 10, 1117;Atti. Accad. w z . Lincei, Rend. Classe Sci. $8. mat. nut., 1964, 37, 3116 GENERAL AND PHYSICAL CHEMISTRYelectrodes of varying shapes and sizes in cryolite melts containing variousconcentrations of aluminium oxide. Mergault and Jacoud, 85 continuingearlier studies, have determined the quantity of electricity necessary toproduce the anode effect in cryolite solutions containing B203, SiO,, Tho,,La&,, Fe2O3, Cr203, Nb205, Ta205, CeO,, TiO,, and ZrO, at 1025"~ a t acurrent of 10 A for various concentrations of oxygen in the bath.The oxideswere classified into four groups according to their behaviour. Rey 86has investigated two cells of the typeA1 I Al,O,,cryolite I 0, and 0, I Al,O,,cryolite I C.The back-to-back combination of these cells represents the aluminiumelectrolysis cell. The experimental e.m.f.s are compared with those cal-culated from thermodynamic data for these systems. The formation ofoxide complexes of carbon adsorbed on the carbon electrode followed bytheir slow evolution is proposed to explain the time dependence of thee.m.f. of the cell 0, I Al,O,,cryolite I C. Pizzini and Agace 87 have measuredanodic and cathodic polarisations on lead electrodes in molten PbF,-NaFbetween 600 and 800"~ using the galvanostatic and potential sweep tech-niques.Passivation is important a t lead anodes in these melts if tracesof oxide impurities are present. Pizzini and Morlotti 88 have studied thecathodic evolution of hydrogen from LiF-NaF-KF melts contaminatedwith water by polarisation as a function of current-density measurements.They assume that hydrolysis leads to dissolved HF from which hydrogenis deposited according toHF +e--+*H, +F-They also assume that the HF is present in the melt as the complex [HE',]-,the slow dissociation of which to give HF determines the measured rate ofevolution. Polarisafion measurements were also employed to study theanodic evolution of oxygen following the addition of oxides, hydroxides,and water t o the melts. It is difficult to account for the experimentalTafel slope (2RT/F) but the formation of the peroxide ion 0 2 2 - is thoughtt o be important.The preparation of pure germanium by the electrolysisof GeO, in mixed fluoride melts a t 900"c has been reported.89 The authorssuggest that the dissociationis followed by the direct discharge of these ions.Chlorides. Double layer capacitance and electrode kinetic measurementsin molten salts and aqueous solution have recently been reviewed.90 Earlierwork,gl largely by the Russian schools, had shown that the minimum capa-GeO, + Ge4+ + 202-8s P. Mergault and R. Jacoud, Comnpt. rend., 1965, 260, 529.86 M. Rey, Compt. rend., 1965, 260, 5528.87 S. Pizzini and L. Agace, Corrosion Sci., 1965, 5, 193.e 8 S. Pizzini and R.Morlotti, Electrochim. A d a , 1965, 10, 1033.89 R. Monnier and P. Tissot, Helv. Chim. Acta, A964, 47, 2203.So A. D. Graves, G. J, Hills, and D. Inman, in Advances in Electrochemistryand Electrochemical Engineering," ed. P. Delahay and C. W. Tobias, Interscience,New York, 1965; A. N. Frumkin, Svensk kern. Tidskr., 1965, 77, 300; E. A. Ukshe,N. G. Bukun, D. I. Leikis, and A. N. Frumkin, Electrochim. Acta, 1964, 9, 431.91 E. A. Ukshe, N. 0. Bukun, and D. I. Leikis, Izvest. Akad. Nauk S.S.S.R., Otdel.khim. Nauk, 1963, 1, 31; Zhur. $2. Khim., 1962, 36, 322INMAN AND WHITE: MOLTEN SALTS 117citances of metal-melt interfaces were very dependent on the melt com-position but largely independent of the metal. Ukshe and Bukun 92 havemeasured the electrical double layer capacitance of a lead electrode in avariety of binary alkali metal a.nd alkaline earth metal chlorides of thetypes MgCl,-MCl (M = Li, Na, K, Rb, Cs) and XC12-YC1 (X = Ca, Ba, Sr;Y = Na or K).The measurements were made with an impedance bridge 93operating at 20 kc./sec. The reference electrode was Pb-PbC1, (2.5 mole yoin the appropriate solvent melt). The double layer capacitance-electrodepotential curves were symmetrical parabolas about the potential of zerocharge, although in some systems, e.g., MgC1,-LiCl, CaC1,-KCl, CaCl,-NaCl,BaC12-KC1, unexplained steps in the cathodic and/or anodic branches wereobserved. The minimum double layer capacitance (Cmin) was plottedagainst the mole fraction of the components. There do not appear to beany systematic trends in these data except in those cases where complexformation is suspected. In these, a deep minimum is observed at the com-position corresponding to the stoicheiometry of the complex.The data foreach system were interpreted in terms of the structure of the melt com-ponents and in terms of electrostriction. The temperature coefficient ofthe minimum double layer capacitance is large for MgCl, (0.2 pF cm.-2deg.-l) compared with LiCl or NaCl (0.11 pF crn.--, deg.-l) and KC1 orCsCl (0.03 pF (3111.12 deg.-1). There appears to be a simple relationshipbetween Cmin and the cation radius for the pure chlorides of both Group IAand Group IIA metals. This is probably related to both the close packingof the ions in the double layer and electrostrictive effects. The influenceof charge might be resolved if data were available for the system CaC1,-BaCl,to compare with KC1-NaC1 (the latter system appears to be free fromcomplications such as complex-formation) in which a steady decrease inCmin from NaCl through KC1 is observed.The structure of the doublelayer is not clearly understood in molten salt systems, and more work mustbe done before a clear picture can emerge. Delimarskii and Kikhno 94report the values for the points of zero charge on the following solid metalelectrodes: Ni, Ag, Cr, M i , Fe, Zr, Ti, Ta, Be, in BaCl,-KCl a t 700"c verswtheir Ag-Ag+ reference electrode. Ukshe et aLg5 have measured the capaci-tive and resistive components of the impedance of Ni, Fe, and Ti electrodesin molten KCI.The results were discussed in terms of oxide film formation.Oscillographic studies 96 of interelectrode capacitance in molten salts forcells with a small phase shift have been made for porous electrodes. Opti-mum conditions and a rapid method for determining the ratio of the surfaceareas of two pairs of electrodes are reported. Graves and Inman 97 havepresented some differential capacitance measurements of the electrical9 2 E. A. Ukshe and N. G. Bukun, Zhur. neorg. Khim., 1964, 9, 1766, 2494; 1965,10, 551; 729; 731; 1008; Elektrokhinziya, 1965, 1, 113.O 3 D. I. Leikis and B. N. Kabanov, Trudy. Inst. $2. Khim. Akad. Nauk S.S.S.R.,1957, Vp. 7, 5.94 Su. K. Delimarskii and V. S. Kikhno, Ukruin. khim. Zhur., 1964,30,1156; 1965,31, 116; 872.95 E.A. Ukshe, S. I. Stepanov, and N. G. Bukun, Izvest. Akad. Nauk S.S.S.R.,Metal, 1965, 1148.96A. V. Goroclyskii and E. V. Panov, Ukruin. khim. Zhur., 1964, 30, 1060.97 A. D. Graves and D. Inman, NaEure, 1965, 208, 481118 GENERAL AND PEYSICAL CHEMISTRYdouble layer a t platinum-halide-melt interfaces. They identify the ano-malously high capacitances of the anodic branch reported earlier by Laitinenand Roe 98 with the formation of a layer containing oxygen on the electrodesurface. The adsorption of platinum ions on to the charged metal surfaceis evidenced by the double-layer capacitance-potential shifts. These changesare explained by assuming Pt2+ to be present as the [PtC1,12- species.Grislain 99 has reviewed and assessed methods for determining thevapour pressure of inorganic compounds over a wide range of temperature.Kuz’menko and N o v i k o ~ ~ ~ ~ have examined equilibria of the formMRbC1, + MC1+ RbCl where M = K or Na, using vapour compositiondata and a treatment developed by the authors.The nature of the mixeddimer is discussed. Changes in the composition of initially equimolarKCl-KBr solid solutions due to differing degrees of vaporisation have beeninvestigated.lol The effect of alkali metal halide diluents on the solubilityof metallic cadmium in CdC1,-alkali metal chloride melts has been ex-amined.lo2 The solubility and decomposition overvoltage for titaniumtetrachloride in molten KC1 have been measured.103 Threadgill 104 hascontinued his studies of the preparation of metallic calcium by the electro-lysis of calcium oxide dissolved in molten calcium chloride.He found aeutectic composition for CaO-CaC1, a t 17 wt. yo CaO. The solubility ofcalcium in the electrolyte is very low -0.02-0.05~0. The sodium-sodiumchloride system has also been investigated.lo5h g e l l 1 0 6 has published a free volume model of transport for both con-ductance and diffusion in molten salts. This treatment has been used lo’to rationalise departures from the Nernst-Einstein equation in a varietyof pure molten salts. They ara correlated by means of “glass transitionbased ” corresponding temperatures suggested by the free volume model.Coupled with molar volume data these correlations are consistent with theview that mutual ionic interactions cause the Nernst-Einstein equation tofail in these liquids.Angell and his collaborators108 have presented pre-liminary data showing the effect of pressure on transport rates in a repre-sentative melt, Ca(N03)2-KN0, (38 : 62 mole yo). The data are analysedin terms of the free volume model. Grantham lo9 has measured the con-ductivity of molten Bi-BiBr, solutions as a function of temperature overthe whole range of compositions. The conductivity of Bi-BiC1, solutionswas also measured in the composition range 0-30 mole yo of metal. Above400 O c these systems resemble the Bi-BiI, system investigated earlier,llO9s H. A. Laitinen and D. Roe, Coll. Czech. Chem. Comm., 1960, 25, 3065.99 B. Grislain, Bull. Xoc.chim. France, 1965, 879.100 A. L. Kuz’menko and G. I. Novikov, Vestnik Leningrad. Univ., 1964, No. 22,101P. Luova and P. Tuominen, Suomen Kem., 1964, 37, B, 207.l o a Yu. N. Rodionov and V. R. Klokman, Radiokhirniya, 1965, 7, 159.108 M. V. Smirnov and V. S. Maximov, Elektrokhimiya, 1965, 1, 727.lo4 W. D. Threadgill, J . Electrochem. Soc., 1964, 111, 1408; 1965, 112, 632.106 E. I. Adaev and A. G. Morachevskii, Zhur. priklad. Khim., 1965, 38, 2105.106C. A. Angell, J . Phys. Chem., 1964, 68, 218; 1917.10’ 0. A. Angell, J . Phys. Chem., 1965, 69, 399.108 C. A. Angell, L. J. Pollard, and W. Straus, J . Chem. Phys., 1965, 43, 2899.1oSL. F. Grantham, J. Chem. Phys., 1965, 43, 1415.IlOL. F. Grantham and 8. J. Yosim, J . Chem. Phys., 1963, 38, 1671.102INMAN AND WHITE: MOLTEN SALTS 119in which the conductivity increases monotonously with metal content.Above 500 O c the three predominating transport mechanisms are believedto be ionic below 10% metal, semiconducting between 20-60% metal,and metallic above 80% metal.Bronsteinlll has described an all-metalcell for e.m.f. measurements in corrosive metal-metal halide systems (forfurther discussion of solutions of metals in molten salts see ref. 4). Laityand Moynihan 112 have carried out transference experiments to determinethe relative mobilities of Lif and Kf ions with reference to the C1- ionover a wide range of composition in the KC1-LiC1 system a t 640"~. Thecombination of these results with conductivity data yield equivalent con-ductivity isotherms for each cation.These decrease monotonously withincreasing K+ ion concentration. The Lif ion mobility decreases rapidlyand crosses that for Kf ion a t 20 mole yo of KCl. It would eventuallyreach 20% of its initial value, in pure KC1. Similar behaviour has beenreported for other systems.113 Chelma et aE.113 have discussed their resultsin terms of complex ion formation. Laity and Moynihan take a less extremeview and make use of the model proposed earlier by Lumsden114 to explainheat of mixing data for alkali metal halides. In this model the effect ofthe highly polarising Li+ ion on the polarisable anion (Cl- ion in this case)in a group such as Li+, C1-, K+ is important. Several Li+ ions can becomeimmobilised on the addition of small quantities of KC1 through inducedpolarisation, and this results in the rapid fall in Li+ ion mobility.Onlya slow rise in K+ ion mobility is observed on addition of LiCl to pure KClowing to the consequent weakening of the K+-Cl- attractive forces. Thisqualitative model also explains the viscosity behaviour of LiCl-KC1 melts.Some of the results of Papaioannou and Harrington 39 discussed previouslymay be understood in the light of this model. Using the " diEusion intoa capillary " technique, Bockris, Richards, and Nanis 115 have obtainedself-diffusion coeficients for the Group I1 metal chlorides over a 200"temperature range a t constant pressure. The data have been used to testseveral models for liquids. A version of the hole model enables heats ofactivation to be predicted.Some structural information was deduced byconsidering the deviations from the Stokes-Einstein and Nernst-Einsteinrelationships. Ichikawa, Shimoji, and Niwa 116 have determined the self-diffusion coefficient of calcium ions in molten calcium chloride and trace-ion dif!fusion coefficients of calcium ions in 99.8 mole yo NaCl + 0.2 mole yoCaC1, and 99.8 mole % KC1 + 0-2 mole yo CaC1, using the " diffusion outof a capillary " technique. The results for the self-diffusion coefficient ofCa2+ ions in CaCl, are lower than those obtained by Bockris et al. ThisWerence may arise from errors inherent in the (different) techniques em-ployed. In a survey of self-diffusion techniques, Angell and Tomlinson117111 H.R. Bronstoin, J . Electrochem. SOC., 1965, 112, 1032.lla R. W. Laity and C. T. Moynihan, J . Phys. Chem., 1964, 68, 3312.113 J. A. A. Ketelaar and E. P. Honig, J . Phys. Chem., 1964, 68, 1596. F. Lantelme114 J. Lumsden, Discuss. Furuduy SOC., 1961, 32, 138.lX5 J. O'M. Bockris, S. R. Richards, and L. Nanis, J. Php. Chern., 1965, 69, 1627.116 K. Ichikawa, M. Shimoji, and K. Niwa, Ber. BunsengeseUschaft phys. Chem.,11' C. A. Angell and J. W. Tomlinson, Trans. Faraday SOC., 1965, 61, 2312.and M. Chemla, Bull. Xoc. chim. France, 1960, 2200.1965, 69, 248120 GENERAL AND PHYSICAL CHEMISTRYsuggested that the " diffusion out of a capillary " technique is the mostreliable. They also suggest that the method of diffusion in a porousrefractory strip should be further exploited.These authors employed theformer method for determining the self-diffusion coefficient in molten leadand thallous chlorides. The results are discussed in the light of varioustheories of transport in ionic liquids. Laity and McIntyre 118 have developeda rigorous treatment which relates the chronopotentiometric diffusion co-efficient, Dch, t o other transport properties in fused salt systems. It isshown that D,, becomes equal to when the concentration is sufficientlysmall (<0.04~). A new method is proposed to take account of the doublelayer charging current in determining the value of z to be used in thecalculation of Dch. Smirnov et have measured the diffusion coefficientof molybdenum ions in several molten alkali metal chlorides and theirmixtures over a range of temperatures by chronopotentiometry.The rateof diffusion of Mo3+ ions decreases as the solvent cation radius increases.The '' jump over " mechanism proposed is similar to that proposed byLaity and Moynihan (see above). Diffusion coefficients l20 of Pb2+, Cd2+,and Zn2+ ions in fused NaC1-KCl have been measured by chronopotentio-metry.High temperature modifications of the Unicam S .P.500,121 the Cary14 H,122, 123 and the Unicam S.P.700 124 spectrometers have been described.A system for preparing pure melts in situ is described.124 The spectrumfor neodymium chloride in the LiC1-KC1 eutectic illustrates the performanceof the S.P.700 spectrometer. The spectra of fourteen lanthanide chloridesin LiC1-NaC1-KC1 in the range 0.8-245 p between 400 and 800"c arereported 122 and compared with similar data in water and molten nitraternedia.47 The effect of the 4f electrons is manifested by the absence ofbands for certain elements.Mamiya l22 has also investigated the absorptionspectra of Ni2+ and Co2+ ions in the LiC1-KC1 eutectic over the temperaturerange 360-7OO"c. Below 390"~ the splitting of the near-i.r. band is in-terpreted in terms of complex species. The spectra of Co2f and Ni2+ ionsare temperature-independent above 500 Oc and ascribed to a single species.Kukk 125 has examined the visible and near-i.r. absorption spectra of &In2+,Co2+, and Ni2+ ions in alkali and alkaline earth halide solvents. Comparisonof these spectra with those obtained in other solvents and in the solid stateindicates that these metal ions are tetrahedrally co-ordinated in LiC1-KCI,LiCl, LiBr-KBr, and CaCl2-MgCl2.E.s.r. spectra are also reported forMn2+ ions in KC1, LiCl, and LiC1-KC1, over the temperature range 25-900"~. N.m.r. studies in molten salts have been reported by Hafnerand Nachtrieb l 2 6 using 205Tl in various pure thallium salts and their118R. W. Laity and J. D. E. McIntyre, J . Amer. Chem. Soc., 1965, 87, 3806.M. V. Smirnov, 0. A. Ryzhik, and G. N. Kazantsev, Elektrokhimiya, 1965,1, 59.120 T. Yanagi, T. Ikeda, and M. Shinagawa, J . Chem. Soc. Japan, 1965, 86, 898.1 2 1 D. N. Henty and D. H. Kerridge, J . Sci. I~zstr., 1965, 10, 756.1z2 M. Mamiya, Bull. Chem. SOC. Japan, 1965, 38,178; Japan Analyst, 1965,14,519.123 C .R. Boston and G. P. Smith, Rev. Sci. Instr., 1965, 36, 206.134 R. A. Bailey and J. A. McIntyre, Rev. Sci. Imtr., 1965, 36, 968.125 M. Kukk, Diss. Abs., 1964, 25, 1602.126 S . Hafner and N. H. Nachtrieb, J . Chem. Phys., 1964, 40, 2891; 1965, 42, 631;Rev. Sci. Instr., 1964, 35, 680INMAN AND WHITE: MOLTEN SALTS 121mixtures with alkali metal halides. The results yield information on cation-anion interactions in molten salts. A detailed spectroscopic study oftransition metal ions in molten AlCl, has been reported.127 The authorsshowed that bivalent Ti, V, Cr, Mn, Fe, Co, Ni, and Cu are octahedrallyco-ordinated in this solvent. This may be compared with the tetrahedralco-ordination exhibited by all except V2f in alkali metal chloride media.l**They1,' have also examined the effect of varying the KCl concentrationin the solvent KC1-AlC1, on the spectrum of Co2+.Below 42 mole % ofKCl, the spectrum is interpreted in terms of Co(A&Cl,), species with theCo2f ion in an octahedral co-ordination. A mixed complex Co(A12C17)(AlC14)is in equilibrium with C0(A12C17), and [AlCl,]- in the region 42-49 mole yoof KCl. The equilibrium constant a t 300"c was found to be 3.5 x ~ O - , M .The octahedral co-ordination of the CoZf ion in the mixed complex isseverely distorted. Viscosity data 129 and Raman spectra 13* obtained forthe KC1-AlCl, (51 : 49 mole %) and NaCl-AlC1, (51 : 49 mole yo) confirmthe presence of [A1C14]- species a t these compositions. X-Ray data 131 forthe compounds formed from the bivalent metal chloride-aluminium chloridefused salt systems have been tabulated.Ibers 132 reports the formation ofa compound CoC1,,2A1C13 from the molten mixture of CoC1, and AlCl,.Spectra for Tlf, Pb2+, and Bi3f ions in LiC1-KC1 have been r e ~ 0 r t e d . l ~ ~Greenberg and Warshawsky 134 have continued their study of alkali metalsdissolved in fused salts. The state of the metal in solution depends onthe concentration (see also ref. 109). In dilute solution (less than l W 3mole yo), F-centres are formed. As the concentration increases, atomic andmolecular species form. At high metal concentrations and above the con-solute temperature, the metal exists in the metallic state. Charge transferand ligand field spectra of (tetrahedral) tetrahalogenonickel( 11) ions inmolten dimethyl sulphone and molten organic halide salts have beenre~0rded.l~~ It is suggested that molten dimethyl sulphone is a particularlyuseful solvent for the study of transition metal complexes.The clearestexamples of complex formation in molten salts are found when the organicquaternary ammonium salts are used as solvent. The Raman spectra 136of BaC12--NaN03 and BaCl,-AgNO, melts have been examined at 430"c.The Raman lines obtained indicate complete dissociation, in agreement withcryoscopic data. On the other hand, BaC1,-KC1 melts give lines whichmay correspond to the species [BaCl,]-. No such definite indication wasreported by Ukshe and Bukun (see above) although there is some indica-tion of a minimum in the minimum double layer capacitance versus mole yoplot for this system.Hamer, Malmberg, and Rubin 137 have completedIz7 H. A. 0ye and D. M. Gruen, Inorg. Chem., 1964, 3, 836; 1965, 4, 1173.12* D. M. Gruen and R. L. Mcbeth, Pure Appl. Chem., 1963, 6, 23.12s I. A. Kryagova, Zhur. fiz. Khim., 1939, 9, 1759.13* K. Balasubrahmanyan and L. Nanis, J . Chem. Phys., 1965, 42, 676.131 R. F. Belt and H. Scott, Inorg. Chem., 1964, 3, 1785.132 J. A. Ibers, Acta Cryst., 1962, 15, 967.133 G. P. Smith, D. W. James, and C. R. Boston, J . Chem. Plzys., 1965, 42, 2249.13* J. Greenberg and I. Warshawsky, J . Amer. Chem. Soc., 1964, 86, 3572; 5351.135 G. P. Smith, C. H. Liu, and T. R. Griffiths, J . Amer. Chem. SOC., 1964, 86, 4796.136 J.Valiier and R. Lira, Compt. rend., 1964, 259, 4579.13' W. J. Hamer, M. S. Malmberg, and B. Rubin, J . Electrochem. Soc., 1965, 112.7 50.122 GENERAL AND PHYSICAL CHEMISTRYtheir calculations of e.m.f.s from thermodynamic data for cells containingsingle solid or liquid fluorides, chlorides,138 bromides, and iodides for thetemperature range 25-1500"~. Hamer 139 has also published data forsingle metal oxide systems over the temperature range 25-3000"~. Thisseries is not as complete as that for the halides because of the thermalinstability of many of the metal oxides. Sethi and Gaur 140 have publishedsimilar data for formation cells containing oxides of lead and bismuth aselectrolytes. Their results agree fairly well with available experimentaldata.Leonardi and Brenet 141 describe a new design of chlorine referenceelectrode. Their E" for the cell Ag i AgCl 1 C1, is in reasonable agreementwith that of Hamer et ~ 1 . l ~ ~ Kisza 142 has established an e.m.f. series basedon the hydrogen electrode in molten dimethylamine hydrogen chloride inthe temperature range 170-200"c for some metal chlorides. The couplesstudied were Cu2+-Cu+, Cu2+-Cu, Ag+-Ag, Bi3+-Bi, Ni2+-Ni, Cuf-Cu,Co2+-Co, Pb2+-Pb, Sn2+-Sn. Activity coefficients calculated for thesesystems indicate strong metal-solvent interactions. Baboian, Hill, andBailey143 report E" values for Ti"-Ti and Ti3+-Ti2+ versus the Pt-Pt2+reference electrode in the LiC1-KCl eutectic at 450" and 550"c. Baboian 14phas also studied zirconium and hafnium ions in this eutectic by potent.io-metry and polarography.The equilibrium potentials of zirconium ions inmolten czsium chloride have also been rep0rted.1~~ Munday146 has in-vestigated the lower oxidation sta,tes of Cd, Pb, Sn, and Zn in moltensodium tetrachloroaluminate. He showed that Cd22f, Pbf, Pb23+, Pbg5+,etc., and Sn,+, Sn,3+, Sndf, etc., are the reduced species present. In thezinc system the electrode potentials were unstable. Okada, Yoshida, andHisamatsu 147 have shown that Cd,2f is the predominant species in solutionsof cadmium metal in molten cadmium chloride and LiC1-KC1 containing30 mole yo CdCl,. Measurements of the e.m.f.s of a series of concentrationcells in the system SnC1,-T1C1 indicate the presence of [SnClJ- and [SnC1J3-species.l48 The SnC12-KC1 system has also been ~tudied.l4~ In the systemTaCl,-MCI, where M = Nay K, Rb, Cs, thermographic analysis 150 suggeststhe presence of [TaCl5I2-.Saeki and Sakulei151 have studied the equi-librium between niobium and niobium subchloride in the LiC1-KC1 eutecticby an e.m.f. method. The presence of NbCl, andNb ,Cl, is suggested. Therefractive indices of NaC1, KCl, CdCl,, PbCl,, and their binary mixtures103, 8.138 W. J. Hamer, M. S. Malmberg, and B. Rubin, J. Electrochem. SOC., 1956,139 W. J. Hamer, J. Electroanalyt. Chern., 1965, 10, 140.140 R. S. Sethi and H. C. Gaur, Indian J . Chem., 1965, 3, 177.1 4 1 J. Leonardi and J. Brenet, Compt. rend., 1965, 261, 113.142 A. Kisza, Bull. Acad. polon. Sci., 1965, 13, 409; 415.143R.Baboian, D. L. Hill, and R. A. Bailey, Canad. J . Chem., 1965, 43, 197.144 R. Baboian, Diss. Abs., 1965, 26, 6426.145 V. Ya. Kudyakov and M. V. Smirnov, Elektrokhimiya, 1965, 1, 143.146 T. C. F. Munday, Diss. A h . . 1965, 25, 6216.14' M. Okada, H. Yoshida, and Y. Hisamatsu, J. Electrochem. SOC. Japan, 1964,148 J. Josiak and J. Terplowski, Roczraiki Chem., 1965, 39, 805.Peng Hsui-Wu and Wan Sew-chew, Sci. Sinica, 1965, 14, 1379.1 5 O V. V. Safonov, B. G. Korshunov, Z. N. Shevtsova, and S. T. Bakum, Z h w .151 Y. Saeki and T. Sakulei, J. Less-Conamon Metals, 1965, 9, 362.32, 99.neorg. Khim., 1964, 9, 1687INMAN AND WHITE: MOLTEN SALTS 123have been measured over a range of temperature^.^^^ Deviations fromadditivity for the CdCl,-KCl and PbC1,-KCl systems provide further evi-dence for complex-formation in these melts.Bayanov and Serebrennikov 15shave investigated the properties of ceriuni and erbium dissolved in Zn, Pb,Cd, and Bi by the e.m.f. method using the LiC1-KCl eutectic as the moltenelectrolyte. This method has also been employed t o examine the reductionof molten cerium trichloride by liquid cerium.154 Roms and Delimarskii 155have studied the thermodynamics of dilute solutions of AgCl dissolved inPbCl,-KCl-NaCl eutectic by the e,m.f. method. The excess functionscalculated for AgCl indicate the presence of complex species. The inter-pretation of excess functions in this way is always, of necessity, ambiguous.The electrodeposition of nickel from the molten NaC1-PbCl eutectic con-taining Ni2+ ions has also been investigated.lS6Work using dropping-metal electrodes has been reviewed.157 Delimarskiiand Kuz’movich 158 report the successful use of a dropping-bismuth electrodein NaC1-KC1 a t 700”~.The apparatus described enabled the drop timeand drop size to be varied within wide limits. The Heyrovsliii-Ilkovicequation described the Cd2+ ion reduction wave, but it was necessary toemploy the Kolthoff-Lingane equation for Pb2+ and TI+ ions although thisonly applied to a first approximation. The limiting current (in spite oflarge oscillations) was shown to be proportional to concentration. Diffusioncoefficients were calculated but the values appear to be high. This maybe due to the extreme mobility of the bismuth drop which leads to largemaxima.Panchenko has discussed some limitations of dropping moltenmetal electrodes and the use of va.rions other electrode materials.159 Heproposes the use of an automatic dropping-bismuth electrode. The flowof bismuth is controlled by a valve which is operated by a time relay anda magnetic chopper. He has shown that the limiting current is proportionalto concentration for Ag+ ions in molten LiC1-KC1 at 420”~. The Kolthoff-Lingane equation describes the observed reduction wave. A value of4.9 x loA5 cm.2 sec.-l for the diffusion coefficient is reported to comparewell with the early data of Lorenz l 6 * for this system. However, thisvalue is about twice that reported in more recent work.l61,162 Schmidtet ~ ~ 1 .1 ~ 3 have obtained polarograms for Pt2f, P d z f , Bi3+, and Sb3+ ionsdissolved in LiCl-KCl eutectic a t 450 O c with an intermittently polarisedplat’inum indicator electrode. The waves for Pd2+ ions a t palladium-H. Bloom and B. M. Peryer, Austral. J . CJzem., 1965, 18, 777.153 A. P. Bayanov and V. V. Serebrennikov, Zhur. $2. Khim., 1965, 39, 717.ls4 M. V. Smirnov and V. S. Lvov, Elektrokhirniya, 1965, 1, 833.156 Yu. G. Roms and Yu. K. Delimarskii, Ukrain. khim. Zhur., 1964, 30, 1151,156 S. Ziolkiewicz and G. Morand, J . Chim. phys., 1965, 62, 312.157 See ref. 4a, 681; ref. 4b, 255.15* Yu. K. Delimarskii and V. V. Kuz’rnovich, J . Appl. Chem. (U.X.S.R.), 1964,15* I. D. Panchenko, Zhur. $2. Khim., 1965, 39, 514; I. D. Panchenko and K. M.I6O R.Lorenz, “ Raumerfullung w. Jonen beweglichkeit,” 1922, Leipzig.161 H. A. Laitinen and W. S . Ferguson, Analyt. Chem., 1957, 28, 4.182 C. E. Thalmeyer, S. Bruckenstein, and D. M. Gruen, J . Inorg. Nuclear Chem.,163 E. Schmidt, H, Pfander, and H. Xiegenthaler, Electrochim. Acta, 1965, 10, 429.3, 1484.Boiko, Ukrain. khim. Zhzlr., 1965, 31, 190.1964, 26, 347124 GENERAL AND PHYSICAL CHEMISTRYcovered platinum electrodes and Pt2f ions at pure platinum surfaces weredescribed by the Kolthoff-Lingane equation. Bismuth and antimony formalloys with the platinum substrate. Diffusion coefficients for these metalions were calculated. Tanaka et aL16* have summarised recent advancesin polarography in Japan, including molten-salt work. The electrodepositionof bismuth from BiCl,, Bi203, and BiOCl dissolved in chloride and boraxmelts has been r e ~ 0 r t e d .l ~ ~ The cathodic deposition from KC1-LiC1-BiCl,melts at low concentrations of BiC1, at temperatures between 400 and 600"cis diffusion-controlled. At higher concentrations the mass-transfer polarisa-tion is negligible. The limiting current densities observed at molybdenumelectrodes for Bi,O, dissolved in CaCl,-NaCl between 600 and 9OO"c in-crease with temperature. The cathodic deposition of Bi3+ ions from BiOCl-BaC1,-CaCl,-NaCl at 900 Oc is also mass-transfer polarised. Anodic chlorineevolution 166 on graphite electrodes has been investigated in A1C13-NaC1 a t19O"c and PbC1,-NaC1 at 430"~. Tafel equation coefficients and electricaldouble layer capacitances have been obtained from decay and chargingcurves at both temperatures. Roughness factors are calculated and usedt o correct the exchange current densities calculated from the analysis ofthe charging curves.The overall electrode process was found to be reaction-polarised at the lower temperature and charge-transfer polarised at 430 "c.Fondanaiche and Kilundai 167 have shown that chlorine evolution on graphitein molten NaCl (820-9OO"c) and NaC1-KC1 (700-900 "c) occurs withoutappreciable polarisation. The thermodynamics of the hydrogen electrodeand the rate and mechanism of hydrogen deposition in solutions of hydrogenchloride in fused chlorides have been studied.lasb Mass-transfer polarisationcontributes to the overall polarisation observed. The proton may be presentin the melt as the [HCl,]- ion (see ref.88).The variation of surface tension lag with composition andtemperature in the system Cd2+, K+, C1-, Br- has been used to investigatethe presence of complex species in the melt. The results indicate thepresence of [CdX,]-, [CdX,]2-, and mixed halogen complexes. Stern 17Ohas continued investigations of electrode potentials in fused salts with astudy of the liquid junction potentials in the AgC1-AgBr system. Theassumption that liquid- junction potentials are negligible in molten salts isonly strictly valid in dilute solutions where the solvent ions carry the bulkof the current through the melt. A parallel assumption that transportnumbers in molten salts are -0.5 has been shown to be unreali~tic.l7~, 172Anion transport numbers are largely uninvestigated. Although liquid-junction potentials in binary melts containing two cations have been ex-164 N.Titnaka, E. Itabashi, and T. Ito, J. Electrochem. SOC. Japan, 1964, 32, 119.1135 F. Colom and L. Alonso, Electrochim. Acta, 1965, 10, 835.166 H. J. Vandenbroele, Rev. Fac. Cienc. quim., Univ. m c . La Plata, 1962/3, 34,16' J. C. Fondanaiche and T. Kikindai, Cornpt. rend., 1965, 260, 2801.16* (a) H. A. Laitinen and J. A. Plambeck, J. Amer. Chem. SOC., 1965, 87, 2202;lG9 R. B. Ellis and A. C. Freeman, J. Phys. Chem., 1965, 69, 1443.1'1 E. P. Honig, Ph.D. Thesis, Amsterdam, 1964.1~ A. Berlin, F. Mkncs, S. Forcheri, and C. Monfrini, J. Phys. C'hem., 1963, 67,Bromides.215.(ZI) E. A.Ukshe and V. N. Devyatkin, EEektrokhinaiya, 1965, 1, 627.K. Stern, J. Electrochem. Soc., 1965, 112, 1049.2505INMAN AND WHITE: MOLTEN SALTS 125tensively investigated, similar systems containing two anions have receivedlittle attention.The system becomes less ideal with increasing AgCl concentration. tBr-decreases and eventually becomes negative as the mole fraction of AgBrin the mixture is reduced. It was thus necessary to take account of complexspecies such as Ag,Rr+. Van Norman173 has measured the solubility ofsilver in molten silver chloride and silver bromide at 520"c, using in situchronopotentiometry. The solubility of silver in AgCl is 0.0123 mole yoand in AgBr 0.0118 mole yo. The diffusion coefficient of the soluble specieswere found to be 2.0 x cm.2 sec.-1 in AgCl at 620°c, and 4.2 xcm.2 sec.- 1 in AgBr at 500"~. Analysis of the anodic chronopotentiogramsindicated that the oxidation was a reversible one-electron process.Markvvand Podafa174 have measured the redos potential of the couple Ti2+-'l'i3+in molten KBr-NaBr at 700"~.Karl and Klemm 175 have determined the electrical conduc-tivities of lithium iodide, rubidium bromide, and lead(@ iodide over a widerange of temperatures. Sternberg et a,Z.176 have established a reversibleI,-I- electrode in molten AgI.MiScellaneous.--Thiocyunutes. The ionic nature of some alkali metalthiocyanates has been deduced by Raman spectroscopy 1 7 7 a t temperaturesabout 20"c above the melting point.The SCN- group is a discrete kineticentity in the melts. The symmetric stretching frequencies v1 are comparedwith those obtained for aqueous solutions of the thiocyanates. The v1frequencies are found to depend on the cation field strength, as the nearestneighbours of anions in melts are on average cations. In aqueous solutionthe frequency v1 does not vary with the cation. This may be attributedto the shielding effect of hydration.Polarographic, chronopotentiometric, and conductometric studies ofthiocyanates have been employed t o evaluate these melts for high-tempera-ture batteries.178 Their electrochemical decomposition is limited on theanodic side by the reactionsStern employed the cellAg I &GI( Xi) ,AgBr( X,)iiAgCl( Xi1) A@r( XZ1) 1 Ag.Iodides.BCNS- -+ (CNS), + 28at +0*25v versus their Ag-Ag+ reference electrode, and on the cathodicside by the reactionCNS-+2e --+ CX- + Ss-at -1-75 V.Alkali metal deposition occurs at more negative potentials.(The reaction K+ + e + K occurs at - 2 . 9 3 ~ and the reactionNa* + e --+ Na at -2-35 v.) The conductance plot log K versw 1/T w-asshown to be essentially linear. The addition of AgBr, NaI, NaBr, NaC1,KC1, LiCl, and LiSCN did not markedly affect the conductance of the melt.x(CNS), + (CNS)*Z173 J. D. Van Norman, J . Electrochem. Soc., 1965, 112, 1126.17* B. F. Markov and B. P. Podafa, Ukruirz. khim. Zhur., 1965, 31, 873.175 W. Karl and A. Klemm, 2. Naturforsch., 1964, Ma, 1619.176 S. Sternberg, I. Adorian, and I. Galasiu, J .Chim. phys., 1965, 62, 63.177 C. B. Baddiel and G. J. Janz, Trans. Furuday SOC., 1964, 60, 2009.17* R. E. Panzer and 31. J. Schaer, J. Electrochem. SOC., 1965, 112, 1136126 GENERAL AND PHYSICAL CHEMISTRYA series of factorially designed tests were carried out t o investigatesuch variables as anode and cathode materials, electrolyte, atmosphere,resistive load, temperature, and electrolyte matrix in order to designoptimum working conditions for the voltaic cell. At 200"c and withcurrent densities of 100 m~ cm.-2, closed-circuit voltages ranged from 1-5to 2.5 for at least 5 minutes. This performance is nearly equivalent tothat of the best previous thermal cells in LiCI-KC1 operating a t 450"c.The conductivities and viscosities of inorganic melts and organic quaternaqammonium salt melts have been compared.179 In the quaternary ammoniummelts E,/EA is approximately unity, whereas in inorganic salt melts theratio is closer to five.In the former systems, EA and Ev seem to be almostindependent of the cation a t a given temperature. The results for theorganic melts can be readily explained in terms of steric effects; densitymeasurements, and studies with models of tetraisopentylammonium thio-cyanate were employed. The results for inorganic melts such as nitrateshave been explained in terms of ion association.ls0 The SCN- ion can fitinto the small interstices formed between the close-packed quaternaryammonium ions. Thus, the free rotation of the cation and movement ofthe SCN- ions from one interstice to another is hindered.The large experi-mental values for the activation energy of conductance (-8-5-1 1 kcal.mole-1 ; cf. 2 kcal. mole-1 for inorganic melts) support this view. Moltenpotas-sium thiocyanate can act as a reducing agent for transition-metal oxyanions.lslSuZphutes. The electrical conductance l S 2 of lithium sulphate has beenmeasured over the temperature range 575-970"~. It was found that anempirical equation of the form K = a + bt + ct2 (t = "c) fits the data,better than an Arrhenius type of equation. The ratio between the equi-valent conductance and the self-diffusion coefficient is less than that requiredby the Nernst-Einstein equation. The results are analysed in terms of thefrictional coefficient formalism la3 using earlier self-diffusion data.18* Thenegative value of r++ indicates some ion-association in the melt.ls5 Ramanspectra 166 in alkali metal sulphate melts also provide evidence for strongcation-sulphate interactions.The plot of Av(v,,) versus Zi/q (for thecations) is linear. The spectra for Cr3+ and Cu2f ions in molten sulphateshave now been analysed in terms of ligand-field theory.187 The resultsshow that the Cr3+ ion is octahedrally co-ordinated in sulphate melts (withA = 14.3 kK, t9 = 0.72). It is suggested that three bidentate sulphate ionsprovide the six oxygen atoms at the corners of the octahedron. The octa-hedral co-ordination of the Cu2+ ion on the other hand is severely distortedand involves two bidentate sulphates in the xy-plane and possibly twounidentate groups a t a greater distance from the metal ion along the z-axis.17O G.J. Janz, R. ID. Reeves, and A. T. Ward, Nature, 1964, 204, 1188.l80 E. Rhodes, W. E. Smith, and A. R. Ubbelohde, Proc. Roy. SOC., 1965, A, 285,181D. I€. Kerridge and M. Mosley, Chenz. Comm., 1965, 505.laa A. LundBn, 2. Naturforsch., 1965, 20a, 235.18a R. W. Laity, J . Chem. Phys., 1959, 30, 682; Discuss. FaTaday Soc., 1961, 32, 172.184 A. LundBn, 2. Naturforsch., 1962, Ira, 142.186 R. W. Laity, Ann. New York Acad. Sci., 1960, 79, 997.186 G. E. Walrafen, J . Chem. Phys., 1965, 43, 479.1 8 7 K. E. Johnson, R. Palmer, and T. S . Piper, Spectrochim. Acta, 1965, 21, 1697.263INMAN AND WHITE: MOLTEN SALTS 127Miller and Seward 188 have investigated galvanic cells in molten bisulphatesolvents (mainly ammonium bisulphate).The solubilities of HgI, HgII, andAgI sulphates were found to be 0*00065, 0.0085, and 0.0215 moles per moleof ammonium bisulphate respectively a t 160"c. The cellsPt I Hg,S0,(sst.),HgS0,(Cl),NH4HS04 ii ref. (A)have been investigated using the reference electrode Pt I Hg,SO,(sat.),HgSO,(sat.), NH,HSO, ii . The e.m.f. of cell A was found to be adequatelyrepresented by the expressionand Ag I Ag,S0,,NR4HS0,,ii ref. (B)E = Eo + Eref. - (0.0861/2) log (CH~SO,~K~S~/CSO,,-~~S)a t high sulphate concentrations. The silver electrode was found to behaveideally with respect to the Ag+ ion concentration in cell B. The e.m.f. ofthe cell varied linearly with the acidity or basicity of the system.How-ever, it decreased anomalously as the SOk2- ion concentration was increased.The viscosity of ammonium bisulphate with various additions of ammoniumsulphate was measured over the temperature range 160-180"~. No struc-tural changes to account for the anomalous behaviour mentioned abovewere found. The kinetics of hydrogen evolution on bright and blackplatinum electrodes during the electrolysis of molten potassium hydrogensulphate have been studied.189 The results have been discussed in relationto the earlier work of Shams El Din in ref. 90. Laitinen190 has reviewedthe high-temperature polarography of oxyanions (including Sod2-). Thereactions are complicated, and poorly understood in many cases, and muchwork remains to be done.The anodic discharge of the SOg2- ion a t acarbon electrode in molten Na,SO, a t 700"c has been studied by producta,nalysis, and the results compared with thermodynamic predictions.lgl Themost likely reaction was found to beSO,2- - 2e + C + SO, + CO,Carbonates. The electrical conductivities, densities, and surface tensionsof some binary and ternary alkali metal carbonate systems have beenreported.lg2 Tracer-diffusion coefficients lg3 have been measured for Naf,K+, and C032- ions in the Li,CO3-Na,CO3-K2CO3 eutectic using the" diffusion out of a capillary " technique. Errors caused by end-effects inthis technique are extensively discussed. The diffusion coefficient of theC032- ion in sodium carbonate is approximately one half that of the Na+ion, whereas in the ternary eutectic the difksion coefficient is some 6-10times smaller than those of Naf or K+ ions.The molar conductance of188 J. P. Miller and R. P. Seward, J . Phys. Chem., 1965, 69, 3156; J. P. Miller,lS9 A. J. Arvia, A. J. Calandra, and H. A. Videla, E1ectrochi.m. Acta, 1965, 10, 33;lgo H. A. Laitinen, Talanta, 1965, 12, 1237.lD1 I. T. Guldin and A. V. Buzhinshya, Elektrokhirniya, 1965, 1, 716.lS2 A. T. Ward and G. J. Janz, Electrochim. Acta, 1965, 10, 849.lD3 P. L. Spedding and R. Mills, J . Electrochem. SOC., 1965, 112, 594.Diss. Abs., 1965, 25, 4428.H. A. Videla and A. J. Arvia, ibid., 1965, 10, 21128 GENERAL AND PHYSICAL CHEMISTRYsodium carbonate calculated from the tracer diffusion coefficients is coni-pared with the experimental data of Janz and Lorenz.194 The calculatedvalues exceed the measured values above 9OO"c.These deviations fromthe Nernst-Einstein equation may be explained in terms of a Grotthus-type mechanism involving rotation of ion-pairs such as Mf C032- assuggested earlier.194 Rolin and Recapet l g 5 have continued their study of thethermodynamic properties of alkali metal carbonates. They report enthalpyand heat of fusion data for Li2COB, Na2C03, and K2C03, and their eutecticmixtures. They have also investigated the dissociation of NaC1, NaOH,and Na20 in molten Na2C03 cryoscopically. Ingram and Janz 196 havepresented a thermodynamic analysis of corrosion in molten carbonatessimilar to that of Pourbaix 197 in aqueous solutions.This had previouslybeen extended to molten salt systems by Littlewood.198 The predictionsof this analysis require experimental verification. An electrochemical seriesin molten carbonates is formulated for several metals, and acid-base pro-perties in these systems discussed.196 Outhier 199 has discussed the propertieswhich metallic oxides should possess if they are to be used as porous oxygenelectrodes in molten carbonates. Busson et uZ.200 have investigated theinteraction of the acid-ba,se systems based on the carbonate ion and waterin molten alkali metal carbonates. K = for the reactionCO, + 20H- + H20 f -k GO,,-They have also carried out a thermodynamic analysis of the oxygen electrodein the Li2C03-Na2C03-K2C0, melt. Delimarskii et uE.201 have studied thereactions occurring during the electrolysis of fused carbonates. Carbon isa reduction product a t the cathode. A limiting current for oxide ionoxidation was observed a t the anode prior to the direct discharge of thecarbonate ions. The cathode reaction 202in fuel cells employing carbonate electrolytes has been investigated onplatinum and palladium electrodes.Shvedov and Ivanov 203 have measured the transportnumbers of Na+ and K+ ions in their respective molten hydroxides overa temperature range 380-5OO"c using an all-nickel cell assembly and aHydroxides.lg4 G. J. Janz and M. R. Lorenz, J . Electrochem. Soc., 1961, 108, 1052; J . Chenz.195 M. Rolin and J-M. Recapet, BulZ. SOC. chim. France, 1964, 2504, 2511.lQ6 M. D. Ingram and G. J. Janz, Electrochim. Acta, 1965, 10, 783.197 M. Pourbaix, " Thermodynamics of Dilute Aqueous Solutions," Arnold, London,198 For example, R. Littlewood, J . Electrochem. Soc., 1962,109,525; Trans. A.I.M.E.,lSQ G. Outhier, Compt. Tend., 1964, 259, 3249; 1965, 261, 986.2oo N. Busson, S. Palous, R. Buvet, and J. Millet, Compt. rend., 1965, 260, 6097;2O1 Yu. K. Delimarskii, V. F. Grishchenko, and A. V. Gorodyskii, Ukrain. khim.2O2 A. V. Silakov, G. S. Tyurikov, and N. P. Vasilistov, Elektrokhimiya, 1965, 1,2O3V. P. Shvedov and I. A. Ivanov, Zhur. Jiz. Khim., 1965, 39, 756.and Eng. Data, 1961, 6, 321.1949.1965, 233, 772.1965, 261, 720.Zhur., 1965, 31, 32.613INMAN AND WHITE: MOLTEN SALTS 129porous corundum diaphragm. The transport number found for the Naf ionwas 0.1 & 0-03 and for the K+ ion 0.03 -& 0-03. Afanas'ev and Gamazov 204have investigated the platinum electrode in molten sodium hydroxide meltsat 500"c. The electrode potential depends on both the melt composition,e.g., the oxide and hydroxide ion concentrations and the gas phase compositionabove the melt. The E versus log a0-2, and log aOH-, plots were straightlines whose slopes were almost the theoretical values. Delimarskii and hisco-workers 205 have reported investigations on some intermetallic compoundsin alkaline melts. Delimarskii and Zarubitskii 206 have also studied the im-pedance of nickel electrodes in fused NaOH at 340"c. The capacitanceof the nickel electrodes rises markedly on passivation. The changes ofthe impedance components with cathodic polarisation were studied anddiscussed.The solubility of TiO, in a variety of molten salts(mainly borates and fluorides) has been in~estigafed.~~' Delimarskii andNazarenko 208 have measured the solubilities of metal oxides in fused borax.Kamyshov et aZ.209 have studied the effect of electric current on the rateof nitrogen dissolution in molten oxides. Kimura and Hayakawa 210 haveinvestigated some metal oxides in molten borax using a voltammetrictechnique with a micro-platinum cathode. Both E, values and the limitingcurrents are shown to be independent of concentration. This latter, perhapssurprising observation, is discussed in terms of variations in the gross meltstructure. The limiting current is stroiigly dependent upon temperaturediL/dt - 2-60/, deg.-l compared with 1% deg.-l in other fused salt systems).Colom and Alonso have also carried out a voltammetric study of Bi20,in molten borax using both molybdenum and platinum micro-electrodes.With the platinum electrode, polarograms could only be obtained between800 and 900"~ and when the mole fraction of Bi,O, was less than 0.01.Mitchel1211 has studied the structure of boric oxide melts containing F-ions by vibrational spectroscopy. The melts are diluted with KBr andquenched prior to measurement. The results show close agreement withexisting frequency and structural data 212 obtained on the quenched un-diluted melts. No evidence was found for the incorporation of the F- ionin the boron-oxygen network. The effect of F- and 02- ions on the moltenboric oxide are further examined213 using as probe the electronic spec-trum of C02+. At high F-Polymeric systems.The results confirm the earlier conclusions.204A. S. Afanas'ev and V. I?. Gamazov, Zhur. $z. Khim., 1964, 38, 2823.205 Yu. K. Delimarskii, 0. G. Zarubitskii, and I. G. Pavlenko, Ukrain. khim. Zhur.,1964, 30, 1289; 1965, 31, 573; Yu. K. Delimarskii and 0. G. Zarubitskii, DopovidiAkad. Nauk Ukrain. R.S.R., 1965, 619.206 Yu. K. Delimarskii and 0. G. Zarubitskii, Dopovidi Akad. Nauk Ukrain. R.S.R.,a07 I. N. Anikin, I. I. Naumova, and G. V. Rumyantseva, Kristallograjya, 1965,20* Yu. K. Delimarskii and G. D. Xazarenko, 'Cikrain. khint. Zhur., 1965, 31, 813.200 V. M. Kamyshov, 0. A. Esin, S. K. Chuchmarev, and A. A. Dobryden, Elektro-210 Y . Kimura and Y . Hayakawa, J . Electrochem. SOC. Japan, 1964, 32, 37.211 A. Mitchell, Trans. Faraday SOC., 1965, 61, 5,212 I. C. Hisatsune and N. H. Suarez, Inorg. Chem, 1964, 3, 168.213 A. Mitchell, Trans. Faraday SOC., 1965, 61, 2295.1965, 485.10, 230.khimiya, 1965, 1, 227130 GENERAL AND PHYSICAL CHEMISTRYconcentrations, however, the reaction1\O F I4- BOF B + B B B B \ 1 \ / \ / \ /0' '0' \ / \ 0 0 0 0 0 0-Ioccurs.have studied the reactions between metal oxidesand molten sodium metaphosphate up to 900"~. The products werenot simple pyro- or ortho-phosphates but polyphosphates of the typeMnNa15-,P,025, where M = Ti, Fe, Co, Cu, Pb, Bi. Kingsley et aZ.215 havemade a spectroscopic study of the Mn043- ion in calcium halogenophos-phates. Riebling and Gabelnick 216 report measurements of electrical con-ductivity in alkali metal germanate melts at 1300"~ using the cell whosedesign was reported earlier.217 The results are discussed in terms of cationvolumes and cation-oxide ion interactions. Some of their results are ex-plained in terms of the GeO, octahedra which may be present. The resultsare also compared with those for other glass-like melts and molten salts.Viscosity and density measurements 218 of sodium aluminogermanate meltshave provided structural information.214 Yu. K. Delimarskii, V. N. Andreeva, and T. N. Kaptsova, Izvest. Akad. NaukS.S.S.R., Neorg. Materialy, 1965, 1, 150.216 J. D. Kingsley, J. S. Prener, and B. Segall, Phys. Rev., 1965, 137, 189.216 E. F. Riebling and S. D. Gabelnick, J . Electrochem. Soc., 1965, 112, 822.217 E. F. Riebling and P. C. Logel, Rev. Sci. Instr., 1965, 36, 425.218 E. F. Riebling, J . Chem Phys., 1965, 43, 1772.Delimarskii e