ANNUAL REPORTSON THEPROGRESS OF CHEMISTRY.GENERAL AND PHYSICAL CHEMISTRY..MUCH of the work published during the past year has been devotedto extending or consolidating advances to which reference has alreadybeen made in these Reports. This circumstance has not renderedany easier than usual the selection of matters to be discussed. Thefollowing pages, therefore, can in no sense claim to be complete : a tbest it may be hoped that they will provide a more or less representa-tive sample of the kind of investigation which has been going on. Inthe newer quantum theory no fundamental advances have been made,but there have been several important applications to problems ofchemistry. The interest of these applications is so great that thepossibility of simplifying the mathematical technique of quantummechanics, even with some sacrifice of formal accuracy, is one whichshould be very well worth exploring.Definite advances are being made in solving the problem of thestructure of the atomic nucleus, but so far there is not even theglimmering of a theory about the forces which hold the nucleustogether.There has been much valuable systematic work in connexion withthe theory of solutions, crystal structure, and especially the elucid-ation of molecular structure with the aid of measurements of dipolemoments and of the Raman effect.In the field of chemicalkinetics there has also been considerable activity.Atoms, Electrons, and Protons.Much work has been published dealing with the theoreticalimplications of the quantum mechanics.The success of this ingiving a quantitative account of a very large variety of physicalphenomena is more and more demonstrated: but the feeling isalso growing that something more fundamental than the formalismof this calculus is desirable, and that there may, for example, besomething to be said for re-introducing phase relationships in dealin14 HINSHELWOODwith atomic interactions,l instead of treating the matter entirely interms of the rather difficultly understandable exchange energy. Butthe time is hardly ripe for the discussion of this in these Reports.There have been several more detailed discussions of the theory ofmolecular forces,2 including interesting attempts to draw con-clusions about molecular structure (see following section) .3 Anotherdevelopment of interest to chemists is the discussion from thequantum mechanical point of view of collision processes and the con-ditions governing the interchange of energy which may occur duringthem. C.Zener,* for example, considers the collision between anatom and a diatomic molecule : it appears that in simple systems theprobability of an interchange of vibrational energy is quite a smallone. Ionic and electronic collisions have also been considered by anumber of authors. The variation of effective atomic cross sectionfor various types of electronic collision is discussed by H. S. W.Massey and C. B. 0. Mohr.5 I n a general way it appears that thecomplex experimental relationships can be accounted for.Experimental work on the diffraction of electrons has continued.6P.M. S. Blackett and F. C. Champion have studied the scatteringof slow a-particles by helium and confirmed Mott’s theory of theprocess, which yields results differing from classical theory.The most interesting development in the whole field is probablythe experimental one due to Rutherford and his school, which carriesus a step further in the understanding of the nature of the atomicnucleus. For the last year or two interest has been rather centredon the quantum mechanical theory of nuclear disintegration, whichgives in general terms a roughly quantitative description of thephenomena of a-particle emission, but, from its nature, does not tellmuch about the structure of the nucleus or the “mechanism” ofdisintegration.The success of the new investigation depends upona method for counting a-particles accurately : * the small ionisationcurrent caused by the particle traversing a few millimetres of its1 Compare R. M. Langer, BuffaIo Meeting of Amer. Chem. SOC., Sept. 1931.2 W. Heitler and G. Rumer, 2. Physik, 1931, 68, 12; A., 547; R. de L.Kronig and W. G. Penney, Proc. Roy. SOC., 1931, [ A ] , 130, 499; A., 407; M.Delbruck, ibid., 1930, [ A ] , 129, 686; A., 17; J. E. Lennard-Jones, ibid., p .538; A., 17; F. Hund, 2. Physik, 1931, 73, 1.3 L. Pauling, J . Amer. Chem. SOC., 1931, 53, 1367; A., 670.Physical Rev., 1931, 37, 556; A,, 543.Proc. Roy. SOC., 1931, [A 1,132, 605.F. L. Arnot, ibid., 1931, [ A ] , 130, 655; A., 542; R.Wierl, Ann. Physik,Proc. Roy. SOC., 1931, [ A ] , 130, 380; A., 280.1931, 8, 521 ; A., 665.* (Sir) E. Rutherford, F. A. B. Ward, and C. E. Wynn-Williams, ibid.,1930, [A], 129,211 ; A., 1930,1338; C. E. Wynn-Williams and F. A. B. Ward,ibid., 1931, [A], 131, 391 ; A., 666GENERAL AND PHYSICAL CHEMISTRY. 15path is amplified by means of a valve. The corresponding ionisationproduced by a p-particle is too small to be recorded. By suitablearrangement of the amplifying and recording system it is possible tocount a-particles even in presence of a strong y-radiation. Adifferential method was then developed which allowed the countingof those particles having ranges between say x and x plus a fewmillimetres, instead of as usual the total number with rangesexceeding a given amount.This was rendered possible by the useof a double ionisation chamber of special construction : the numberstopping within a region about 2 mm. broad could be detected andcounted. With the new counter it was possible to detect the missingshort-range or-particles from the dual disintegration of radium-C-those corresponding to the C" branch.Since 1919 it has been known that radium-C, in addition to thenormal 7-cm. range particles, gave rise to a group of long-rangeparticles. Analysis of these by the new method reveals the factthat they can be resolved into a t least 9 homogeneous groups between7 cm. and 12 cm. range, the ranges and relative numbers in eachgroup being determined. Prom the ranges the velocities and energiescan be estimated by the usual methods.The existence of these long-range groups is now correlated with the emission of y-rays in aremarkable manner.The general theoretical picture is as follows. Radium-C emits a@-particle giving rise to radium0 : some of the a-particles in thenew nucleus are in excited levels, from which they either escape aslong-range particles, or fall to a lower level with emission of a y-ray.It is from the ground level of the radium-C' nucleus that the normal7-cm. range particles are supposed to come. (In certain cases theremay be emitted instead of the y-ray itself a @-particle liberated fromthe outer atom by some process of internal conversion.)l* The firstpoint which emerges is that the maximum energy which could bereleased in a transition from a level corresponding to the particles ofhighest energy to the ground level is about equal to the maximumenergy of any @-ray line in the magnetic spectrum of radium-C (theselines come largely from the conversion of y-rays).From the energies of the fast a-particles a series of energy levelsabove the normal level can be calculated.Assumed transitionsbetween these or from them to the ground level can then be comparedwith the energies of y-rays or the corresponding p-rays of the magneticspectrum. The intensity of the y-rays is of the order of one quantum(Lord) Rutherford, F. A. B. Ward,and W. B. Lewis,Proc. Roy. SOC., 1931,lo Compare R. H. Fowler, ibid., 1930, [A], 129, 1; A., 1930, 1338; C.D.[A], 131,684; A., 890.Ellis and G. H. Aston, ibid., p. 180; d., 1930, 133916 HTNSHELWOOD :for each two or three disintegrations, but the number of nuclei whichdisintegrate to give the fast particles is of the order of 1-10 in amillion, so that the probability of a change in energy level withemission of radiation is very much greater than that of the escape ofone of the “excited” a-particles. The energy differences of thea-particle states are not only of the same order of magnitude as they-ray energies, but in several examples numerical agreement wasfound, lending still stronger support to the idea that the quantum ofy-radiation is emitted during an a-particle transition in an excitednucleus.Working on this basis, Lord Rutherford and C.D. Ellis l1 analysethe numerous y-rays from radium-C‘ into an orderly scheme. Twokinds of energy-level system are theoretically possible. In the first,a given particle could occupy a series of energy levels of differentquantum numbers, in a manner analogous to the successive electronicstates of the hydrogen atom. In the second, the different levels aredue to the excitation of varying numbers of particles to the samehigher state. The Pauli principle rules out this second class ofsystem for electronic states, but permits it in the case of a-particles.Rutherford and Ellis take as their working hypothesis the simpleassumption that transitions occur whereby the number of particlesin a single excited state changes in one act from r to r’ withemission of a quantum of radiation.If E~ and c2 are the energies ofeach of the particles in the ground state and the excited state beforethe transition, and E,‘ and E ~ ‘ the corresponding quantities after thetransition, then, if n is the total number of particles, the totalenergies before and after the transition are given respectively by(n - r ) ~ ~ + m2 and by (n - r ’ ) ~ ~ ’ + TIEZ’ ( E ~ ’ and E ~ ’ are changedbecause there is necessarily a mutual action of the a-particles). Theseconsiderations indicate that the energies of the y-rays should berepresentable in the form hv = p E , - qEz, where p and q areintegers. The experimental data are found to be compatible withsuch a scheme, though it is not excluded that slightly differentschemes might fit the facts equally well.R.H. Fowler,12 considering the quantum mechanical aspect ofthe problem in the light of these new results, finds by the analysis ofa very rough and ready model, vk., a one-dimensional model in whicha-particles act on one another with potential energies of a quadraticform, that the energy levels are such as to give rise to y-rays offrequencies represented by an expression of the form hv = p ( E , -qE,). The necessary modification of the formula of Rutherford andEllis would not do violence to the experimental facts.11 Compare R. H. Fowler, Proc. Roy. Soc., 1031, [A], 132, 667.12 Nature, 1931, 128, 453GENERAL AND PHYSICAL CHEMISTRY. 17Physical Theory and Benzene Structure.In the last Report reference was made to the work of E.Huckel onthe quantum mechanical interpretation of the " rigidity " of chemicaldouble bonds. The simpler structural problems of organic chemistryare so beautifully and completely treated with the aid of the tradi-tional ideas that the translation of these ideas into the language ofwave mechanics must be regarded rather as a satisfaction to thephysicist than as a real advance in chemistry. But there arecertain problems which the ordinary conceptions of chemistry havenot been able to solve without the aid of auxiliary ideas sometimes ofa rather indefinite character. Prominent among these problems arethe questions of the structure of aromatic compounds and theorientation of substituents in the benzene nucleus. Here especiallythe introduction of new guiding principles is particularly valuable,and we may well look with interest to see whether a quantummechanical treatment does indeed go a step beyond re-interpretingwell-tried old ones.It must be remarked at the outset that any-thing like a complete quantum mechanical analysis, starting fromfirst principles, of a problem so complex as the structure of benzeneis impossible. Even the treatment of the hydrogen molecule has tobe based upon approximations, which already take for grantedcertain empirically established facts. For example, as F. Hundpoints when Heitler and London neglected the Coulomb forcesbetween two hydrogen atoms in comparison with the " exchangeforce" they were virtually taking advantage of the empiricalinformation that the inert gases have in fact no chemical affinity.Indealing with benzene, therefore, Huckel l4 has to begin by makingvarious assumptions the validity of which cannot be estimatedexcept by reference to the results. Nevertheless, certain discoveriesemerge which appear to be of great importance.The method of analysis is based upon the fact that in benzene thereare six bonds over and above those engaged in the single linkagesof carbon to carbon or carbon to hydrogen. Thus there are sixelectrons not accommodated in ordinary non-polar single bonds(each carbon brings four outer electrons, each hydrogen one : total30. Huckel takes themore general case of n CH groups in a ring and n electrons, and triestwo methods of approximation.In the first, each electron is thoughtof as assigned to one ring atom in a definite quantum state, and theinteraction between them is investigated by methods analogous tothose which have been successfully applied to the determination ofinteraction in many-electron systems in connexion with ferro-magnetic phenomena. It transpires that rings with an odd numberlS 2. Phyeik, 1931, 73, 1. l4 Ibid., 1931, 70, 204.The 12 single bonds require 24, leaving 6)18 HINSHELWOOD :of carbon atoms should possess a greater energy content thanthose with an even number, but there appears to be no other resultof obvious chemical interest. The second method of approximationneglects at first the “ exchange energies ” of the electrons, andconsiders the n electrons in a field of force which varies periodicallyround the ring.The possible quantum states of these electrons areworked out, and the important result appears that, when the Pauliprinciple is introduced (Le., that only two electrons of opposite spinscan occupy the same quantum state), systems of electrons of 2, 2 +4 = 6, 2 + 4 + 4 = 10 are found to constitute “ closed groups,”somewhat analogous to the closed groups in the atoms of the periodicsystem. Since C2H2 is not a ring and C,,H,, is not known at all, thesix-electron system occupies a unique place, and the stability of theclosed group is evidently responsible for some of the characteristicproperties of benzene and the aromatic compounds. The 8-group isnot a closed one, which may well explain the well-known lack ofaromatic character shown by cyclooctatetraene.Heterocyclic six-membered rings such as pyridine, or five-membered rings such asthiophen, in which the heterocyclic atom can contribute more thanone electron can also show aromatic properties. Although the effectof the disturbed cyclic symmetry on the behaviour of the electronsin these examples cannot be calculated, it is not ah all unreasonablet o suppose that the importance of the closed groups of six persists.To sum up, we may say that while quantum mechanics cannot solvethe benzene problem in anything like a complete way, it shows thatthere is good ground for believing that ring systems in which a groupof six electrons unaccommodated in single bonds occurs will possessspecial properties and probably unusual stability.This appears tobe a result of great importance.Huckel has also dealt with the problem of benzene substitution,lsmaking approximate estimates of the disturbance of charge distribu-tion of the electrons caused by the presence of a substituent. Accord-ing t o his arguments, the negative charge in, for example, nitro-benzene is increased relative to benzene in the ortho- and para-positions and reduced in the meta-position, Le., in this example theelectrons are “ repelled from the carbon atom where the substitutionhas taken place and from themeta-carbon atom towards theortho- andpara-atoms.” This is an effect of the opposite sign from that postu-lated by the simple theory of “ alternating polarities,” but accordingto Huckel a carbon atom where there is a defect of negative chargewill repel a hydrogen nucleus, thereby lessening the work which mustbe done in removing it.Substitution thus becomes easier. Meta-substitution occurs therefore in the nitrobenzene example.lo 2. Phyeik, 1931, 72, 310GENERAL AND PHYSICAL CHEMISTRY. 19It should be mentioned in this connexion that L. E. Sutton 16 hasshown by direct comparison of the dipole moments of correspondingaromatic and aliphatic compounds that " electron drifts " away fromor towards the substituent group can in fact be detected. This isimportant, since for the first time it establishes the physical reality ofthe electrical changes which various current theories of organicreactions postulate.The conception of the " electronic cloud " which Schrodinger'stheory introduces has hitherto proved useful in forming a picture ofseveral phenomena : e.g., the interaction of the clouds was used byHuckel in interpreting the properties of the double bond, and histheory of induced polarities deals with disturbances of the electricdensity associated with the 6 '' ring " electrons of benzene.In a stillsimpler'example repelling or attracting hydrogen atoms with the aidof an unequal density distribution can be pictured thus :-It seems possible that an interpretation of some of the mostimportant quantum mechanical results relating to molecular forcescould be worked out, which would enable chemists to solve theirparticular problems with the aid of rules determining the behaviourof electron clouds.These rules would be simple and approximate andutilisable in their proper sphere in much the same waty as that inwhich Faraday used the lines of force, or van 't Hoff the rigid directedbond. It is to be hoped that the possibility of constructing somesuch theory will be investigated.Potential-energy Curves and their Applications.For the interpretation of the physical and chemical behaviour ofdiatomic molecules consideration of the potential-energy curve isoften of great help. This curve (Fig. 1) represents the potentialenergy of the molecule as a function of the distance between theatoms, and has the form shown in the figure. At the point wherer = r,, the energy is a minimum and the molecule is in its normalequilibrium state.When it is vibrating the atoms may be atl6 Proc. Roy. SOC., 1931, [A], 133, 66820 HINSHELWOOD :distances greater or smaller than ro. For the smaller separations theenergy increases very rapidly towards indefinitely great values, sincethe compressibility of a molecule is very limited. As the separationof the atoms exceeds the equilibrium value the energy increases atfirst fairly rapidly and in approximate accordance with the law ofa, simple harmonic oscillator, and then more slowly as the bindingforces between the atoms get weaker and weaker. Finally theenergy reaches a constant limiting value corresponding to completedissociation of the molecule. The distance D in the figure representsFIQ.1.the energy of dissociation. An exact theoretical equation for theform of this curve is hardly possible to obtain, since it will dependupon the variation of binding force with distance, but various semi-empirical equations are used, the constants in which can be evaluatedby the aid of data obtained from band spectroscopy. One of themost convenient of these equations is that of P. M. Morse 1' whichwill be briefly explained, since it is likely to become of increasingimportance in chemical calculations.The energy of the molecule is represented by the expression2De - - t o ) - - 2a(r - r,,) E(r) = Del7 Ph.ysical Rev., 1929, 34, 57; A., 1929, 975GENERAL AND PHYSICAL CHEMISTRY. 21E(r) is the energy for a displacement r of the two atoms, D is thedissociation energy, and a is a constant.This expression possesses the following properties : as r approaches00, E comes asymptotically to the value zero, i.e., the energy of thecompletely separated atoms is taken as the standard level; E has asingle minimum of - D a t r = r o ; when r = 0 the value of E ,although not infinite as it should be, is very great, which is a goodenough approximation.Thus the general shape of the curve isprovided for. The equation has a further important property : ifwe write down Schrodinger’s wave equation for an oscillator andsubstitute the above value for the potential energy and then deter-mine the allowed values for the various energy levels of the vibratingsystem, we obtain a series of the same form as one of the best ofthe empirical spectroscopic equations.The constants of the twoexpressions can be equated, and thus a can be expressed in terms ofspectroscopic data. For the nth level Morse’s equation gives W(n) =hao(% + 4) - (h2aO2/4D)(n + +)2. The spectroscopic vibrationallevels can be well expressed by the formula W(n) = hwo[(n + 8)- x(n + +)2] : wo is the frequency of oscillations small enough tobe simple harmonic ; from the original equation it is easily found tobe equal to - - , where p is the ‘c reduced mass.” Afterequation of constants and simplification, the value finally found fora is 0-2454(&fo0x)t, where M = MIM,/(M1 + M,), M , and M ,being the atomic weights of the two nuclei on the oxygen scale andwo being expressed in wave numbers.Morse also gives an empiricalrule for finding ro, wiz., riao = 3000 A.3/cm. : the latter, however,is not regarded as of universal validity.Among the applications of the potential-energy curve, we mayfirst consider the approximate estimation of the energies of activationof chemical reactions. Reference was made in the last Report l8 tothe calculations of Villars. The idea underlying these has beenextended and modified by Eyring and Polanyi, in a series of veryinteresting calculations based on the London theory of valency andof intermolecular forces.lg The force between two atoms is made upof two parts : an electrostatic part called the Coulomb force, and aforce depending on the quantum mechanical phenomenon ofresonance,20 and going by the name of the “exchange force.” Invery simple examples the magnitude of these forces can be estimated.In molecule formation the exchange force is the more important.In2x TD>f PAnn. Reports, 1930, 27, 19.l9 H. Eyring and M. Polanyi, 2. physikal. Chem., 1931, [B], 12, 279; A.,Zo Ann,. Reports, 1930, 27, 24.688; H. Egring, J . Amer. Chsrn. SOC., 1931, 53, 2637; A., 114022 HINSHELWOOD :calculating an energy of activation we must be able to estimate theforces acting between a number of atoms present together: e.g.,between three atoms when we are considering a reaction of the typeY + XZ = YX + Z ; or between four atoms in a reaction of thetype WX + YZ = WZ + XY. As we have just seen, informationabout the energy of diatomic combinations can be derived from theband spectra of the diatomic molecule ; the next step is renderedpossible by some formulze of London’s which give the total energy ofa system of three or four atoms in terms of the energies which theseatoms would possesss if they existed as isolated pairs (diatomicmolecules) with nuclear separations equal to their actual distancesapart in the polyatomic codiguration. For three atoms the totalenergy is given bywhere Q = A + B + C, the sum of the three Coulomb energies ofthe three atoms taken in pairs, and a, 8, and y are the exchangeenergies of the three possible isolated molecules. For four atomsthe expression becomesE = Q + [${(a - P)’ + (a - yI2 + (P - y)’P= Q + “a1 + a2 - 81 - PJ2 + (a1 + a2 - y1 - yJ2 +(P1 + P2 - y1 - y2)2)lfwhere Q is the sum of six Coulomb energies, and al, a2, .. . , etc.,are the exchange energies of the six diatomic combinations. Toproceed further some rather bold approximations are necessary.The total energies of the diatomic combinations are first obtainedfrom the band spectra of the corresponding molecules. A roughestimate, based upon the analogy of very simple cases such as H2where actual calculation is possible, is then made of the proportionof the total energy which is Coulomb energy and the proportion whichis exchange energy : thus the separate terms of the above formulze arefound. To assume that the proportion is constant for differentdistances of the atoms is obviously a rough and ready procedure, butis more or less justified by the fact that the Coulomb energy is ingeneral only a small fraction of the total. We now can calculatein principle the total energy of any configuration of three or fouratoms and can thus study its variation as one molecule approachesanother from a great distance and comes into chemical reaction withit. If we startwith XZ at the normal molecular distance, and Y far removed, andthen bring up Y, the energy increases : it passes through a maximumand can then fall again if, for example, Z is removed to infinityleaving the molecule YX behind.The “ reaction ” which hasoccurred has involved passage over an “energy pass.” The minimumheight which can be found for this pass for any separation andThe results of such a calculation are as followsGENERAL AND PHYSICAL CHEMISTRY.23direction of approach is the energy of activation of the reaction.Corrections for change in the zero point energy of the molecules whichaccompanies the reaction must also be applied in a rough manner.The calculation of the course of the energy as the molecules approachfor “reaction” is simplified by the fact that London showed in ageneral way that reactions of the type Y + XZ require least activ-ation when the three atoms remain in a straight line. Eyring andPolanyi have estimated the heats of activation of the reactions H,(para) + H = H2 (ortho) + H; H + HBr = H, + Br; and H $Br, = HBr + Br. For the first they obtain 13 kg.-cals., for thesecond 10, and for the third zero.For the first the experimentalvalue is uncertain but lies between 4 and 11 kg.-cals. ; the second andthird are merely known to belong to the class of atomic reactionswith very small inertia.Eyring has made similar calculations for reactions of the typewhere X is a halogen. The numerical results are very uncertainbecause it is necessary to assume ad hoc that the Coulomb energy is36% of the total energy and the figure assumed makes a good dealof difference. The most interesting result, however, is a comparativeone independent of this assumption; namely, that the reactionbetween molecular iodine and hydrogen should occur more easilythan the reaction by way of iodine atoms, whereas with the otherhalogens the atomic mechanism is energetically preferred.This ofcourse is in striking accord with experiment.This is a convenient place to refer to the theory of adsorptioncatalysis of M. Born and J. Franck and M. Born and V. Weisskopf,21which, although not making use of actual potential-energy curves,depends upon general theoretical considerations concerning thepotential energy of molecules in relation to the atomic separations.The idea underlying the theory can be illustrated by an analogy. Ina homogeneous collision reactiqn two molecules must meet possessingthe energy of activation; quantum mechanically there is a finiteprobability of transformation by leakage 22 whether they possess thisenergy or not. But during the short time of a collision the probabilityis vanishingly small unless the energy is practically equal to orgreater than the activation energy : when adsorbed on a surface thesojourn of molecules in proximity to one another is long enough forthe probability of transformation to attain a finite value.Thetheory is thus a combination of the old chemical idea that thesurface keeps in proximity molecules which otherwise would havelittle opportunity to react and the principle underlying the GamowH2 + X = HX + H ; H + X, = HX + X ; X2 + X‘= XX’+ X21 2. phyeikal. Chem., 1931, [BJ, 12, 206; A., 576.22 Ann. Repom, 1930, 27, 2624 HINSHELWOOD :theory of radioactive change with its extensions to the spontaneouschemical changes in molecules.23 Born and Weisskopf take anidealised picture of a molecular rearrangement, namely two masspoints acting on each other with forces and capable of existing in twopositions of stable equilibrium at different separations.Thus therewill be two sets of states corresponding to various energy levels of theinitial and final products of the quasi-chemical rearrangement. I fthere were no interaction between the two sets of states the Schro-dinger equations for them would be independent, but a perturbationterm is introduced on account of the mutual potential energy of theatoms and the crystal surface on which they are assumed to beadsorbed. When there are two energy levels, one in the initial setand one in the final set, which correspond, transition by quantummechanical resonance occurs. The exact correspondence of energywould be extremely unlikelywithout the crystalsurface, which may actin one of two ways.It either takes up itself the difference between thetwo energies, or it makes the transition possible in virtue of the strongand variable mutual potential energy between atoms and surface. Arather elaborate calculation leads to the result that the probabilityof transition depends very markedly on the displacement which theatoms have to suffer in the process. If this is of the order of 0-5 A.it appears that the transformation can occur in a few seconds. Toarrive at this result special assumptions have to be made about theoscillation quantum numbers. The authors remark that this pictureof adsorption catalysis is not the only possible one : “ lowering of theenergy threshold, resolution of chemical linkages under the influenceof the adsorption forces, intermediate reactions with the atoms of thesurface, etc., may be essential factors.” It must be remarked herethat the experimental evidence has already shown that all thesefactors and especially the first are in fact of great importance.Howmuch room this leaves for the operation of the quantum mechanicalleakage phenomenon is a difficult question.A very different but equally interesting type of potential-energycurve is that representing the potential energy of an ct-particle inthe neighbourhood of a nucleus. When the distance is greater thanabout 10-12 cm., the particle is repelled in accordance with Coulomb’slaw, and the potential energy is represented by the part ab of thecurve sketched in Fig.2. For smaller distances the curve mustfollow a course such as bcd, eventually passing into the region of“ negative energy,’’ since a-particles inside nuclei are in fact in astable condition, and under the influence of attractive rather thanrepulsive forces.24 The shape of the curve nearer to the centre of23 Compare ,4nn. Reportcr, 1930, 27, 26, 317.24 G. Gamow, “ Constitution of Atomic Nuclei,” Oxford, 1931GENERAL AND PHYSICAL CHEMISTRY. 25the nucleus than b is a fascinating question connected with the mostfundamental problems of nuclear structure and stability. Atpresent only rough and tentative guesses can be made, but eventhese are proving helpful. For example, the course abet may bepostulated as an approximation, the potential energy being sup-posed to drop suddenly to the value in the stable position.Thetwo characteristic magnitudes are then x and y in the figure.H. M. Taylor,25 working on this basis, has shown that Rutherfordfand Chadwick’s results on the deviations of a-particle scattering inhelium from that required by the Coulomb law can be accountedfor if suitable values are chosen for the two constants z and y.Knowing these, the energy of binding of an a-particle inside anucleus can be estimated : here, however, the result is not quitesatisfactory, showing that the schematic simplilication is too greator is of an inappropriate form. It turns out that the energy lostwhen two a-particles unite to form a nucleus of mass 8 wouldcorrespond to a mass defect of 0.28%, which is very much too26 Pmc.Roy. SOC., 1931, [A], 134, 10326 HINSHELWOOD :large, compared with the known mass defects of elements such ascarbon and oxygen.Liquids and Solutions.Work continues on the properties of " intensively dried " liquids.The fundamental experimental fact, stated without prejudging anytheoretical issue, is that a liquid which has been sealed up withphosphorus pentoxide for a considerable time frequently will notgive a rapid continuous stream of vapour unless it is heated to aconsiderably higher temperature than would have been necessarybefore the drying process. This means either that some innerequilibrium in the liquid has been displaced, or that the rate ofevaporation from the liquid surface has been considerably lowered,rendering the liquid very liable to what is virtually superheating.26On general theoretical grounds the latter alternative appears muchmore probable, since it is easy to understand that during thestorage with phosphorus pentoxide either something (such as wateror colloidal dust particles) is removed which would have facilitatedevaporation, or that something (such as phosphorus pentoxideitself) is introduced which by concentrating itself in the surfacelayer impedes the free passage of molecules into the vapour phase.The former alternative would be very surprising from the thermo-dynamic standpoint, since it would mean that a small quantityof a foreign substance produced a change in free energy of thesystem out of all proportion to that produced by further additions.This would be extremely remarkable since thermodynamic functionsin general are linear in the concentration for small additions.Froma kinetic point of view it would also be remarkable, since if a verysmall addition of water produced a finite shift in equilibriumthrough the bulk of a liquid phase it would mean that each watermolecule exerted an influence far beyond the range of ordinarymolecular forces. This could only happen by some kind of trans-mitted polarising effect, the existence of which would imply thatliquids possess a " structure." J. W. Smith 27 has recently studiedthe distillation of ethyl bromide from one evacuated bulb to anotheracross a, constant-temperature gradient, and finds that the ratedecreases considerably as the '' drying " of the liquid proceeds.28The vapour pressure, however, remains unaltered, and moreover,there is no evidence of any fractionation of the dried liquid during26 Compare inter alia E.Cohen and W. A. T. Cohen-de Meester, Proc. K .Akad. Wetsnsch. Amsterdam, 1930, 33, 1003; A., 294; 8. Lenher, J . PhysicalChem., 1929,33,1579; A , , 1929,1372.2' J., 1931, 2573.28 Compare the work of R. Stumper, Kolloid-Z., 1931, 55, 310; A., 906,on the acceleration of evaporation by colloidal matterGENERAL m D PHYSICAL CHEMISTRY. 27distillation, such as would be expected if a shift in an innerequilibrium had occurred. This observation is parallel to that ofRodebush and Michalek, who some time ago observed that the rateof evaporation and of condensation of ammonium chloride werediminished by a process of drying.The earlier observation ofSmith and Menzies that dry calomel had zero vapour pressure alsomeans that the rate of evaporation was zero under the conditionsof the experiment rather than that the equilibrium value had beenreduced to nought. There are two analogies, differing in principle,in terms of which these various results may be explained. One isthe apparent zero value of the vapour pressure which would befound for slightly contaminated mercury a t ordinary temperatures,the other is the complete non-efflorescence of a perfect salt hydratecrystal in the absence of nuclei.The literature contains a number of confiicting results about theinfluence of drying on static properties such as surface tension andvapour pressure.(Miss) E. J. Greer 29 found that the addition ofvery minute traces of water to well-dried benzene gave rise to apartial pressure of several mm. a t 20" : this would indicate that ameasurable change in the total vapour pressure may occur withoutthe shifting of any equilibrium in the " dried " liquid itself.A. W. C. Menzies,30 on the other hand, finds that it is quite easyby simple distillation to free benzene from water to such an extentthat further changes in vapour pressure on further drying arenegligible. S. L. Wright, junr., and Menzies 31 have also failed tocodinn the existence of delays in the establishment of vapour-pressure equilibrium, or to find an influence of pre-treatment onthe boiling points of such liquids as benzene, carbon tetrachloride,and bromine. W.A. West and Menzies38 explain Baker's resultswith acetic acid, where changes of, vapour pressure were observedduring several days, as due to fractionation of a liquid containingsome water. Smits and others33 have recently shown that thecomplete removal of dissolved gases from liquids is a much morem c u l t matter than had been previously supposed. This raisesthe question whether a number of earlier observations on thechange of vapour pressure on drying are not redly to be explainedby the presence of permanent gases. Repetition of some of theearlier experiments has failed to show any change in the vapour20 J .Amer. Chem. SOC., 1930, 52, 4191; A., 34.30 J . Physical Chem., 1931, 35, 1655; A., 901.31 J . Arner. Chem. SOC., 1930,52, 4699; A., 1931, 294.32 J . Physical Chem., 1929, 33, 1893; A., 1930, 145.33 A. Smits, E. L. Swart, P. Bruin, and W. M. Mazee, 2. phyaikd. Chem.,1931,153,255; A., 43028 HINSHELWOOD :pressure when adequate precautions are taken to free the liquidsfrom gases.While no doubt a number of unexplained observations remain,the impression grows in strength that changes in static propertiesof liquids on “drying” are usually spurious effects, while thedynamic effects are genuine but not more mysterious than otherinfluences of nuclei, surface films, or catalysts.If some of the statements in the literature about the influence ofexcessively minute traces of impurity on the physical properties ofliquids are true, then it is difficult to escape the conclusion thatsomething of the nature of a structure exists in certain liquidsunder suitable conditions, and that infiuences from a single moleculeare transmitted by some kind of polarising influence, affecting stringsor clusters of molecules. Experimentally the situation is not unlikethat prevailing in psychical research where, we are told, most of theevidence can be ruled out, but a small obstinate residuum has tobe contended with.A matter of considerable interest in connexion with this generalquestion of the structure of liquids is the possible existence of twoliquid forms of certain substances.Two liquid forms of helium 34with a transition temperature depending on the pressure have beendescribed. From discontinuities in the curves showing the vari-ation with temperature of such properties as density, specific heat,viscosity, and dielectric constant, the existence of two liquid formsof nitr~benzene,~~ carbon di~ulphide,~~ and ether 37 has been inferred.But with nitr~benzene,~~ at least, careful re-investigation of theproblem indicates that the effects claimed are not real. It is t Qbe hoped that the phenomenon will be quite definitely establishedas real in certain examples or explained away in the near future.Among the most clarifying advances of the last ten years inphysical chemistry has been the recognition of the part played bypurely electrostatic forces in determining the properties of dilutesolutions of electrolytes. Conductivity, osmotic pressure, and thethermodynamic properties deducible from it, mutual solubilityinfluences of electrolytes, and the so-called salt effect on reactionvelocity all depend to a greater extent on the interionic forcesthan on any other single factor : indeed, the Debye-Hiickel theory34 W.H. Keesom and K. Clusius, Naturwiss., 1931, 19, 462; A , , 1004;Proc. K . Akad. Wetensch. Amtlterdam, 1931, 34, 605; A . , 1004.35 J. Mazur, Nature, 1930, 126, 993; 1931, 127, 741, 893; A . , 1931, 148,792, 899.36 M. Wolfke and J. Mazur, ibid., p. 926; A . , 896.37 J. Mazur, ibid., 1930,126, 649; 1931,127, 270; M. Wolfke and J. Mazur,38 N.B. Massey, F. L. Warren, and J. H. Wolfenden, J., 1932, 91.ibid., 1930,126, 684GENERAL AND PHYSICAL CHEMISTRY. 29of electrostatic interaction is so closely followed by many strongelectrolytes at high dilution that deviations from it can now beregarded, not so much in the light of a failure of a theory as ofuseful positive information about the intrusion of other specificfactors. These specific factors do in fact continue to reveal them-selves in a number of instances. For example, among the morerecent results are the following : in nitromethane solution tetra-ethylammonium salts behave in agreement with the Debye-Hiickel-Onsager theory of conductivity, but other salts, which behave asstrong electrolytes in alcoholic solution, are weak electrolytes innitromethane39 in spite of its greater dielectric constant.Thusmolecule formation is by no means controlled by electrostaticforces, even though the behaviour of ions can be predicted in termsof them if molecule formation does not actually occur.Two further important consequences of the Debye-Huckel theoryof interionic forces in dilute solutions have been investigated. Thefirst, relating to the viscosity of dilute solutions of electrolytes, isexplained in another section of the report. The second is that indilute solution the partial molar volume of an electrolyte, or therelated quantity, the apparent molecular volume, is proportional tothe square root of the concentration. In 1929 Masson 40 showed thatsuch a square root relation held for a number of electrolytes, andlater W.Geffcken4l elaborated. the matter. 0. Redlich and P.Rosenfeld42 have now shown that such a relation is a necessaryconsequence of the Debye-Eiickel theory. They derive theequationV: is the limiting value of the partial molar volume T2 at idbitedilution; w = QCv,zS2, where v, = number of ions of kind S permolecule, and z is the valency of the ion; q is a constant dependingupon the temperature, the dielectric constant of the solvent, thevariation of the dielectric constant with pressure, and the com-pressibility of the solvent. The " apparent molecular volume," 4, is related to the " partial molar volume " by the relation m G =d - ) , where m = the number of g.-mols. of solute per 1000 g.ofsolvent. For 4 the following equation holds : 9 = $0 + t q . w3I2. c*/~.From the examination of the experimental data made by Masson- - v, = v; + qUF= . c1'2-dm8s C. P. Wright, D. M. Murray-Rust, and (Sir) H. Hartley, J., 1931, 199;compare also the results for nitrobenzene, W. F. K. Wynne-Jones, ibid., p.795.40 (Sir) D. 0. Masson, Phil. Mag., 1929, 8, 218.41 2. phy8ikal Chem., 1931, [A], 155, 1.4a Ibid., p. 65; 2. EEektrochem., 1931, 37, 70530 HINSHELWOOD :and by Geffcken, the square-root relation appeared to hold for agiven electrolyte up to fairly high concentrations, but the slope ofthe curves exhibiting the relation varied from example to examplein an apparently individual manner. Redlich and Rosenfeld,examining the best experimental data, come to the conclusion,however, that at higher concentrations the lines are really curvedand that at great dilution they all converge to a single straight line,with the anticipated slope.While the best test can be made withthe data for uni-univalent electrolytes, the influence of the valencyfactor with multivalent electrolytes is, as far as can be seen,approximately in accordance with the theory. Examination ofdata for methyl-alcoholic solutions indicates that the influence ofthe solvent on the constant q can also be accounted for.There is no doubt that electrostatic forces may play a considerablepart in determining solubility, but it is also clear that their influenceis only one among many factors. By comparing theoretical pre-dictions of solubility, based upon a purely electrostatic theory,with the experimentally found relations, an idea can be formedof the importance of the electrical factor.The general principleunderlying such calculations is as follows. Some simple model ofan ion or molecule is adopted and its electrical energy in a mediumof given dielectric constant is worked out. The difference betweenthe electrical energies it possesses in two different media is thenequated to the work of transfer from one medium to the other,i.e., to ET In K , where K is the partition coefficient. Such calcul-ations are successful only in a very genera1 and qualitative way forionic partition coefficients. R. P. Bell43 has recently extendedthem to the solubility of dipole molecules, and thus obtained aninteresting picture of the extent to which the electric moment ofsuch molecules determines their behaviour.As a model he assumesa spherical molecule of dielectric constant unity, with a dipole a tits centre, the distance between the two charges of the dipole beingsmall compared with the radius of the sphere, He points out thatonly in rather extreme examples can the treatment give quantitativeresults, since it would predict, for example, equal solubilities fornon-polar gases such as hydrogen, oxygen, and nitrogen. Withstrongly polar molecules, such as ammonia and water, there is amuch better chance of success, and Bell finds for the relativesolubilities of these two substances in a number of solvents adependence upon the dielectric constant agreeing fairly well withthat anticipated by the theory, especially for the solvents of higherdielectric constant.For the solubility of the mercuric halides, regular relations areJ., 1931, 1371; Trane. Paraday Soc., 1931, 27, 797GENERAL AND PHYSICAL CHEWISTRY.31also found. The conditions under which the method of calculationis more or less adequate appear to be, therefore, when the dipolemoment of the solute is large, the dielectric constant of the solventis large, and also when the dipole molecule itself has a large diameter,since then the approximation involved in regarding the solvent asa continuous medium is better justified.As exemplified by examples referred to in a later section, theelectric moment of molecules does not appear to determine theiradsorption on surfaces to any important extent.However, asmight be expected, the degree of ionisation of a dissociable group inan interface influences the interfacial tension to a marked degree.The interfacial tension between aqueous and benzene solutions ofthe higher fatty acids varies with the hydrogen-ion concentrationof the aqueous phase over just the range which would be expectedif it were determined by the degree of ionisation of the unimolecularinterfacial layer, although in certain examples specific effects aresuperimposed. R. A. Petersu has recently shown that the inter-facial tension of the system benzene-water-hexadecylamine varieswith hydrogen-ion concentration in just the opposite sense to thatfor the long-chain acids, but over approximately the same range,and indeed that the interfacial-tension changes follow rather closelythe dissociation curve for a weak base.Here, therefore, we haveanother example of a phenomenon where electrostatic forces are ofpredominant importance.The Viscosity of Solutions of Electrolytes (by H. W. THOMPSON).Measurements of the viscosity of dilute solutions of electrolyteshave recently acquired considerable theoretical interest. TheDebye-Huckel interpretation of the conditions existing in suchsolutions, and the theoretical derivation of the Kohlrausch square-root relation between equivalent conductivity and concentrationwere based upon the idea that in the neighbourhood of any ionthere will be more ions of unlike than of like sign.The “ionicatmosphere” will normally be symmetrical, but will become dis-torted when the central ion moves. It possesses a c‘ thickness ”and requires a definite time for its establishment-these magnitudesbeing determined by the ionic concentrations, the valencies, thetemperature, and the dielectric constant of the solvent. Thevelocity of the central ion, moving under the influence of an electricfield, will depend on the properties of the atmosphere.It is not unnatural to suppose that the ionic atmosphere plays apart in determining the viscosity and in particular its variationwith salt concentration.44 Proc. Roy. SOC., 1931, [A], 133, 14032 KINSHELWOOD :The viscosity of liquids and solutions has been studied experi-mentally for a long time.Earlier workers confined themselvesprimarily to the construction of reliable viscometers : later workdealt primarily with the connexion between viscosity and molecularstructure, and with questions such as that of molecular association.Hubner first investigated the viscosities of a series of salt solutionsand found that addition of salt sometimes decreased the viscosityof the solvent. Sprung concluded that salts could be divided intotwo groups : those of the first depress the viscosity of water at lowtemperatures but increase it somewhat a t higher temperatures,while those of the second group always increase the viscosity.The work of Slotte, Arrhenius, Wagner, Ranken, and Taylor andGetman led to the suggestion of various empirical relationshipsexpressing the coefficient of viscosity, q, as a function of the con-centration.Griinei~en?~ working a t greater dilutions, discoveredthat the curves of viscosity plotted against concentration exhibita negative curvature a t the dilute end. He attempted, but not verysatisfactorily, to explain the results in terms of the electrolyticdissociation theory. M. P. Applebey 46 confirmed the existence of‘‘ negative curvature ” and also the fact that electrolytes maycause either an increase or a decrease in the viscosity of the solvent.He suggested that two opposing factors operate in determining theviscosity of a solution : on the one hand, there is a depolymerisationof triple water molecules which tends to decrease viscosity, and onthe other, there is ionic friction which increases it.With feeblyhydrated ions the second factor will be small compared with thefirst, and accordingly a net decrease in viscosity will be observed.I n general, however, ions will be hydrated to such an extent that,at any rate at the higher dilutions, a net increase of viscosity willbe found.G. Jones and M. Dole 47 have recently redetermined the viscositiesof solutions of barium chloride at various concentrations. Theirresults are not in agreement with any of the earlier empirical rela-tionships. While at the higher concentrations there is a certaindegree of proportionality between fluidity (reciprocal viscosity) andconcentration, yet a t the lower ones, i.e., in the region of the so-callednegative curvature, a systematic discrepancy exists.Jones andDole propose to account for this qualitatively as follows. Accord-ing to the Debye-Huckel theory of the ionic atmosphere, the effectof the interionic forces in opposing the motion of an ion in anelectric field is proportional to the square root of the concentration4ti Wise. Abhandl. Techn. Reichsanstalt, 1905, 4, 151, 237.46 J., 1910, 97, 2000.47 J . Amer. Chem. SOC., 1929, 51, 1073,2950; A., 1929, 767, 1385GENERAL AND PHYSICAL CHEMISTRY. 33in very dilute solutions. Thus for the fluidity the following rela-tionship is indicated : + = 1 + A d z + Bc. For all strong electro-lytes, A will have a negative value (the fluidity decreasing withconcentration), and for non-electrolytes it will be zero; B can onlybe regarded in the first instance as an empirical coefficient.Therelation can be re-expressed as (+ - l)[<c = A + Bdc, or if thehigher concentration term be neglected, q/qo = 1 + K&, where Kis a fresh coefficient.Jones and Dole re-examined the older data, which appeared toobey the relation, although there is a lack of data for high dilutions.They accordingly predicted that at high dilutions the viscosity ofsolutions of all strong electrolytes will be greater than that of purewater even in the case of salts which a t ordinary concentrationsshow a diminution of viscosity. This requirement was alreadysatisfied by the results of K. Schneider 48 working with solutions ofpotassium chlorate : an increase in viscosity occurred between 0and 0-05N, but at higher concentrations a falling off was observed.The same is true of nitric acid solutions, where W.Bousfield49found that at 11" and below N/32 the viscosity exceeds that ofwater. At higher temperatures the viscosity exceeded that ofwater at all concentrations examined.A mathematical investigation of the problem from the stand-point of the Debye-Hiickel theory enabled H. Falkenhagen andM. Dole 50 to derive the relation q/qo = 1 + K 6 . The simplecase of a uni-univalent electrolyte with ions of equal mobility wasfirst studied. The calculation leads to the resultK = (e/60q0udDx} x 0.491 x lolo,where e is the electronic charge, q the viscosity of the solvent, u thereciprocal of the frictional force acting on the ions, D the dielectricconstant of the solvent, k: the Boltzmann constant, T the absolutetemperature, and x the valency of the ions.H. Falkenhagen hasrecently extended the calculations to include any single ~ a l t . ~ 1The only salt with ions of approximately equal mobility ispotassium chloride, for which the calculated value of K is 0.0046.A series of theoretical values for various salts are given by Falken-hagen and Dole. W. E. Joy and J. H. Wolfenden52 have nowconfirmed the value of H predicted for solutions of potassiumThe latter form is now most common.4 8 Di88., Rostock, 1910.so Phyaikal. Z., 1929, 30, 611; A., 1929, 1389; 2. physikal. Chern., 1929[B], 6,159 ; A., 1930,155 ; also H. Felkenhegen, " Reviews of Modern Physics,"1931, 3, 412; Nature, 1931,127, 439; A., 560.5 1 2.physikal. Chem., 1931, [B], 13, 93; A., 905; Physikal. Z., 1931, 32,365 ; A., 686.63 Proc. Roy. SOC., 1931, [A], 134, 413.J., 1915, 107, 1781.REP.-VOL. XXVIU. 34 HINSHELWOOD :chloride, nitric acid, rubidium nitrate, and potassium chlorate.Some earlier determinations of the viscosity of rubidiumnitrate solutions by H. G. Smith, J. H. Wolfenden, and (Sir) H.Hartley 53 also exhibit the Falkenhagen-Dole effect. This effectprovides a further confirmation of the idea of the ionic atmosphere,and the influence of electrostatic forces in determining the propertiesof dilute solutions.Surface Chemistry.So many papers have been devoted to various aspects of surfacechemistry that it is necessary to devote a section again this yearto a few at least of the divisions of this rapidly expanding subject.The explanation of the nature of adsorptive forces in terms of thenewer theories of molecular forces has been discussed by F.Londonand M. P01anyi.~~ It appears, as one might have expected, thatthe dipole moment of the adsorbed molecule is of secondary import-ance only in adsorption phenomena, a conclusion which is supportedby the experiments of H. Cassel and F. Salditt 55 on the adsorptionof various vapours by mercury. It is also of interest to note inthis connexion that C. A. Sloat and A. W. C. Menzies 56 find theadsorption of various alkali bromides by lead sulphide not to berelated in any way to the lattice dimensions of the deposited salt.Three of the six salts examined give orientated deposits on leadsulphide, but there appeared to be no preferential adsorption ofthese.Among many other papers on the more general aspects ofadsorption phenomena may be mentioned a detailed discussion ofthe nature of the boundary layer in " lubrication " phenomena bySir W. HardyY5' in which chains of highly polarised moleculesstretching from one of the solid surfaces to the other are pictured.A general theory of the quantum mechanics of adsorptioncatalysis is dealt with in another section. has alsoderived a theoretical expression for the rate of a heterogeneous gasreaction, using assumptions analogous to those employed in theconsideration of homogeneous reactions : the results generally arein good agreement with experiment.Even if the details of Topley'scalculation are not accepted in their entirety, his results seem toshow that the introduction into the theory of adsorption catalysisof quantum mechanical " leakage " effects is far from necessary.On Dhe whole, much more work has been done on adsorptionconsidered as a statical (equilibrium) phenomenon than on theB. Topley53 J., 1931, 403.5 5 Ibid., 1931, 19, 110; A . , 421.5 6 J . Physical Chem., 1931, 35, 2022; A , , 904.6 7 Phil. TTans., 1931, [A], 230, 1 ; A., 559.6 8 Nature, 1931, 128, 115; A., 918.64 Naturwiss., 1930, 18, 1099; A., 1931, 161GENERAL AND PHYSICAL CHEMISTRY. 35dynamical aspect of the process.We may, however, now considersome papers dealing with this important side of the subject. Ex-periments on the exchange of energy between helium atoms andsurfaces of tungsten and of nickel are described by J. K. Roberts,59the method of investigation being that of studying the heat lossfrom a wire of the metal surrounded by the gas. The accommodationcoefficient is found to vary with the state of adsorbed gas films onthe wires. T. Alty,SO by measuring the rate of evaporation from awater surface and extrapolating to zero pressure, obtains the idealrate of evaporation into a vacuum : by comparing this with therate a t which molecules strike the surface from the saturatedvapour it is found that, in order that there may be equilibrium,only about 1% of the incident molecules actually enter the liquid.G.Veszi 61 estimates that when the atoms in a metal vapour arereflected from the surface of a stream of oil, they sojourn in somecases for a period of the order of 104 to 10-5 see. on the oil surface.An interesting relation between catalysis and accommodation comesto light in the work of K. F. Bonhoeffer and A. Parkas 62 on thetransformation of para-hydrogen into the equilibrium mixtureunder the influence of a heated platinum wire. The suggestedmechanism of the catalysis is that the para-hydrogen is adsorbedby the metal in the atomic form, and that re-evaporation in themolecular form subsequently takes place, ortho- and para-hydrogennaturally coming off in the equilibrium proportion.The usualplatinum catalytic poisons inhibit the transformation, while a traceof oxygen accelerates it. Now at lower temperatures, where nocatalysis occurs, heat exchange between the wire and the gas takesplace only by the process of reflexion of molecules from the surface.But a t higher temperatures adsorption of the striking moleculesand their subsequent evaporation play a considerable part in theenergy exchange, and one finds that the accommodation coefficient,as inferred from the amount of electrical energy required to main-tain the wire at a given temperature, increases abnormally, and thatthis increase sets in just as the catalytic activity appears. Poisonedplatinum a t the same temperatures continues to show a normalcoefficient.Under normal conditions both direct experiment and indirectevidence from the kinetics of heterogeneous reactions show that therate at which an adsorption equilibrium is established is very great.Slow processes are usually interpreted as solution phenomena or59 Proc.Roy. SOC., 1930, [ A ] , 129, 146; A., 1930, 1340.60 Ibid., 1931, [A), 131, 554; A,, 904.62 Ibid., 1931, [B], 12, 231; A., 691.2. physikab. Chem., 1930, [B], 11, 211; A., 1931, 16236 HINSHELWOOD :chemical changes. H. S. Taylor 63 has, however, recently suggestedthat these slow changes are of more fundamental importance thanhas hitherto been assumed. He says “there are now numerousdata showing abnormal variations in the extent of adsorptionwith both temperature and pressure, large variations in the velocityof attainment of equilibrium in different adsorption systems andin the velocity of evaporation of adsorbed gases inconsistent withthe present adsorption theory.” Taylor suggests that true adsorp-tion per se is not necessarily a rapid process, but one which takesplace with a characteristic velocity determined by the same kindof factors as those which govern the rate of chemical changes, andin particular that an activation energy may in some cases benecessary for a molecule to be adsorbed by a surface.He pointsout that a diatomic molecule can in principle be adsorbed in twoways, either molecularly or with accompanying resolution intoatoms. The first kind of process would be expected to have a verysmall heat of activation, but the second may well require theabsorption of very considerable amounts of energy; it would thusbe improbable a t low temperatures, but would have a high temper-ature coefficient.Some of the conclusions which may be drawnfrom these postulates are as follows. At low temperatures molecularadsorption will predominate, the total amount adsorbed in this waydecreasing as the temperature rises. In a certain higher range oftemperature the total adsorption increases again, not on account ofchanged equilibrium conditions, but because the veZocity of theactivated atomic adsorption, hitherto negligible, now becomes appre-ciable. In this region equilibrium will not be attained instan-taneously, and it may be found that the large amounts of gastaken up at a higher temperature are more or less tenaciouslyretained on rapid cooling to a temperature where they would notoriginally have been taken up.Furthermore, the heat of adsorp-tion will vary with temperature according to whether the molecularor the atomic type of adsorption is predominating. Some examplesof the evidence that effects of this kind are indeed observed maynow be quoted. The adsorption of hydrogen by nickel above- 100” is rapid : at lower temperatures, A. F. Benton and T. A.WhiteG4 found the adsorption a t a given temperature to varyaccording as the saturation was effected at the actual experimentaltemperature or at a higher one with subsequent cooling. Between- 183” and - 191” adsorption was rapid but a t higher temperaturesit was slower, suggesting according to Taylor’s point of view that aprocess with an activation energy is coming into play.Taylor63 J . Arner. Chern. SOC., 1931, 53, 578; A., 421.64 Ibid., 1930,52,2326; A., 1930, 990GENERAL AND PHYSICAL CHEMISTRY. 37quotes other evidence of a similar kind, such as the diminishingadsorption by platinum as the temperature falls, the “ plurality ofprocesses ” involved in the fixation of oxygen by charcoal, and thehysteresis phenomena observed in the adsorption of oxygen bysilver and gold.It will be readily admitted that the processes postulated byTaylor may in principle play an important part in adsorptionphenomena, and that the conclusions he draws from his postulatesare correct. But it is open to considerable doubt at the momentwhether the experimental observations which he quotes in his firstpaper are really to be identified with the operation of such pro-cesses. It is necessary first to show that, the abnormal temperatureinfluences and the hysteresis effects are not really due to solution.Taylor argues that the amounts of gas involved are higher than theknown solubilities, and that in these adsorption effects the surfacefactor is all important, which it would not be in a solubility pheno-menon.But to this must be said that it is not a question of theequilibrium solubility, but rather of rate of solution. At lowtemperatures the penetration of a gas into the lattice of a, metalwill in general be excessively slow: at higher temperatures theporosity increases and gas can penetrate a few layers of atoms notfar removed from the surface, especially when the surface structureof the metal has been loosened and to some extent opened up byvarious chemical or physical treatments. Where gas has oncepenetrated, gas can penetrate again more easily than before, andin the case of a metal which has had the outer layers of its massopened up, we may have a fairly easy absorption or removal of gaswhich appears something like the rapid establishment of anequilibrium.But this is very far from being the true solubility :persevering treatment at higher and higher temperatures will makethe metal more and more porous. It is improbable that with amassive metal the true solubility would be attained except underquite exceptional conditions, so that Taylor’s argument about themagnitude of solubility effects seems inconclusive.Increasinglygreat ease of penetration of metals by gases as the temperaturerises (the penetration being in any case slow and difficult, and,moreover, retarded by the gas already taken up), is a, factor whichundoubtedly must be taken into account, and which will inevitablymask the effects discussed by Tayl0r.6~ These important effectsought, however, to exist, and it may be possible with experimentalwork of some delicacy to isolate them from the other complications,as we shall see below in dealing with some of the most recent work.(It may be remarked here that failure of the so-called Polanyi-65 Compare E.W. R. Steacie, J . Physical Chem., 1931,35, 2112; A., 90438 MNSHELWOOD :Hinshelwood equation G6 when a measured heat of adsorption isused does not necessarily indicate that another energy term, Taylor'sactivation energy of adsorption, is to be introduced. Accordingto Taylor's well-established theory of active points, the heat ofadsorption on the active points responsible for the catalysis willnot bear any definite relation to the average heat of adsorptionrevealed by a calorimetric measurement .)In pursuing the idea of activated adsorption, Taylor and William-son 67 have found some interesting results on the adsorption ofhydrogen by manganous oxide and manganous oxide-chromiumoxide mixtures. The process is slow with manganous oxide at 0"and loo", and between 184" and 305" increases in rate ten-fold,corresponding to a " heat of activation " of about 10,000 calories :with the promoted oxide both kinds of adsorption can be studied,one with a negligible "heat of activation" and the other withthe high and variable value for this quantity.The adsorption isreversible. W. E. Garner and F. E. T. Kingman 68 find that acatalyst of zinc and chromium oxides adsorbs hydrogen and carbonmonoxide a t low temperatures " without dissociation of the mole-cules," whereas above 100" the adsorption becomes irreversible andthe gases can only be removed as water or carbon dioxide.F. E. T. Kingman has also made observations on the criticalincrement of adsorption of hydrogen by charcoal. 69In connexion with the discussion of the part played by actualpenetration, rather than surface adsorption, in the processes classedunder the general heading of sorption, the work of M.G. Evans 70on the sorption of ammonia by chabazite is of great interest. Hefinds that when ammonia is taken up the X-ray diagrams of thechabazite show distortion of the space lattice. This distortion,caused by actual penetration of the ammonia molecules, mayultimately shatter the crystal. While it is not probable that allslow sorption depends upon such effects, this work shows thatspecific allowance for the possibility of such penetration must bemade in any given example.In certain examples, notably with oxygen and charcoal, the heatof adsorption for the first portions of gas taken up is greater thanfor subsequent portions.A. F. H. WardY7l however, has foundthat the heat of adsorption of hydrogen by activated copper isindependent of the concentration of gas on the surface. In these6 6 J. K. Dixon, J. Amer. Chem. Soc., 1931, 53, 17G3; A., 803; H. Dohse67 H. S. Taylor and A. T. Williamson, J . amer. Chem. SOC., 1931, 53, 813,6 8 Trans. Faraday Soc., 1931, 27, 322.70 Proc. Roy. SOC., 1931, [ A ] , 134, 97.and W. Kalberer, 8. physikal. Chem., 1929, [B], 5, 131 ; A., 1929, 1231.2168 ; A., 421,902.69 Nature, 1931, 128, 272.7 1 Ibid., 1931, [ A ] , 133, 506, 522GENERAL AND PHYSICAL CHEMISTRY. 39experiments there was a clear separation observable between theinstantaneous adsorption and a slow subsequent process of solution.Ward adduces evidence to show that the dissolved hydrogen is notsplit up into atoms as has often been supposed, and also givesreasons for supposing that penetration of the copper by the hydrogenoccurs by diffusion along the boundaries of metal grains ratherthan by diffusion into the lattice.These considerations are relevantalso in connexion with the above remarks on Taylor’s views.An interesting new method of investigating surface films hasbeen developed by J. H. Schulman and E. K. Ridea1,72 followingup a technique introduced by Guyot and by Frumkin. Thepotential difference at the surface of an aqueous solution bearinga thin film of a substance such as a fatty acid is measured electro-metrically, one electrode being a calomel electrode making contactwith the liquid in the trough, the other being an air electrode,consisting of a metal plate above the solution with the interveningair gap ionised by a radioactive preparation.The acidity of theaqueous layer can be varied, the state of compression of the filmcan be altered as in the Langmuir-Adam trough, and the variationsin surface potential studied, Theoretically the difference ofpotential at the surface is given by AV = 4mp, where n is thenumber of molecules per sq. cm. of surface and p is the effectivevertical component of the average electric moment. By thismethod transitions, for example, from vapour to liquid expandedfilm can be observed. The values of p in different states 6f thefilms can be found and conclusions can be drawn about the structureand configuration of the films.The authors give evidence for theexistence of a new phase appearing in the transition from theliquid to the vapour state of films of fatty acids. There are alsoindications that in the vapour films the molecules are horizontallyorientated.A. J. Allmand and others 73 have continued a detailed andsystematic study of the adsorption isotherms of gases and vapours.One of the most important results 74 of the more recent of theseinvestigations is the strong indication found that the adsorptionof vapours by porous solids such as charcoal is definitely discon-tinuous in nature : the curves representing amount adsorbedplotted against pressure of the vapour in the gas phase showingbreaks, or even horizontal steps where the amount adsorbed72 PTOC. Roy.SOC., 1931, [A], 130, 259, 270, 284; A., 299.73 A. J. Allmand and R. B. King, ibid., p. 210; A., 160; A. J. Allmand andA. Puttick, ibid., p. 197; A., 160.7 4 A. J. Allmand and L. J. Burrage, ibid., 1931, [A], 130, 610; A., 558;compare also A. F. Benton and T. A. White, J . Amer. Chem. SOC., 1931, 53,2807; A., 100540 HINSHELWOOD :increases suddenly for a small increase of pressure. The inter-pretation of these observations depends upon the fact that thesurfaces possess regions of different adsorptive power, and uponthe fact that the adsorbed vapour may change from a two-dimen-sional gas to a two-dimensional liquid at certain critical pressures.Mention should also be made of the work of N.I. Kobosew andW. L. A n ~ c h i n , ~ ~ who study adsorbed gas films by the method ofelectron bombardment, measuring, not the ionisation potential ofthe adsorbed gas as in the experiments of J. H. Wolfenden 76 andof G. B. Kistiak~wsky,~~ but the actual rate at which the gasbecomes desorbed under the influence of the electron stream.While the interpretation of some of their observations seems alittle speculative, it is interesting to note that Kobosew and Anochindistinguish three " desorption potentials " of hydrogen fromplatinum, corresponding to the release of the gas from plane sur-face, edges, and corners of the solid respectively. Complex relationsare found when oxygen and hydrogen are simultaneously presenton the surface.Xtructure of Natural Products.As an example of the way in which complex natural structuresare being investigated, reference may be made to the work ofJ.B. Speakman 7* on the micelles of wool fibre. The methodsused are interesting : the elastic properties of wool fibres in dryair and in water are quite different; fibres in methyl or ethylalcohol behave as in water, while fibres in butyl or amyl alcoholbehave as though dry. Thus the capillary space in the wool fibreis probably of about the same size as the molecule of propyl alcohol,in which liquid an intermediate behaviour is found. When ittakes up a liquid, a wool fibre may swell and subsequently beable to take up another liquid which could not have penetratedinto the dry wool.By studying the elastic properties of the fibresin mixed liquids, information about the structure of swollen fibrescan be obtained. This is supplemented by observations on changeof length on swelling, comparison with X-ray data, and deductionsfrom various other facts to give a detailed picture. The woolfibres have a " plate-like " structure, and are arranged with theirlong axes parallel to the fibre length: they are probably aboutten times as long as thick. The micellar thickness is about 200 8.,and the intermicellar distance about 6 A., increasing on swelling in7 5 2. physikal. Chern., 1931, [B], 13, 63; A., 903.7 6 Proc. Roy. SOC., 1926, [ A ] , 110, 464; A., 1926, 217.7 7 J.PhysicalChem., 1926,30, 1356; A., 1926, 1188.7 8 Proc.Roy. SOC., 1931, [ A ] , 132, 167; A., 1003GENERAL AND PHYSICAL CHEMISTRY. 41water to about 41 8. This makes the internal surface of woolsomething of the order of a million sq. cm. per gram.An extremely interesting account of X-ray studies of hair andother animal fibres has been published by W. T. Astbury andA. Street,79 who find that all animal hairs appear to give sub-stantially the same X-ray diagram. When the hair is stretched,one form of X-ray photograph (the " a ") gives place to another(the " p ") corresponding to a unit cell of quite changed dimensions.This transformation seems to depend upon the reversible change ofa certain group of length 5.15 8. into another of length 6.64 a.,and plays a very important part in determining the elastic propertiesof hairs, which are in many ways remarkable.The " p " form ismore vulnerable to the attack of sodium sulphide, and in thisconnexion it is a suggestive fact that the most intense X-rayreflexion given by stretched hair has the same spacing as the mostimportant reflexion which the substance cystine gives. In thiswork we see, however incompletely, that interrelation of micellarstructure, chemical structure and behaviour, and mechanicalproperties upon which the function of living matter must depend.Chemical Kinetics.In the Annual Report for 1927 it was remarked that if the theoryof collisional activation in unimolecular reactions proved to begenerally applicable, then the distinction between unimolecularand bimolecular gas reactions would become one of degree only.This anticipation i s apparently being fulfilled, and the classicaldivision of reactions into orders is losing a good deal of its signifi-cance.Naturally in a change of the type X + Y = XY we stillhave a bimolecular reaction in the older sense, but complicationsarise in reactions of the types XY = X + Y, and 2XY = X, + Yz.In all known thermal reactions of the first of these two latter types,activation is by collision (though in principle we can hardly yet ruleout the possibility that real non-collisional unimolecular changesmay be discovered), and thus these changes do not differ in principlekinetically from those of the second type. The essential differenceis merely one in the time which elapses between activation andreaction, and therefore in the relative importance of deactivationof the active molecules by subsequent collision before the chemicaltransformation has occurred.In general we may say: rate ofactivation = k,c2; rate of deactivation = k,c . a, where u is theconcentration of active molecules ; rate of actual chemical trans-formation = k,a. For a stationary concentration of active mole-713 Phil. Tram., 1931, [ A ] , 230, 75; A., 897.B 42 HINSHELWOOD :cules k1c2 = E,c . a + E3a; whence the rate of reaction is Ic3a =klc2/(1 + k 2 / k 3 . c ) . According as the ratio E2lI1.3 is great or small,the reaction approximates to the classical unimolecular or bimole-cular type. Only when the chance of deactivation is absolutelyzero can the reaction remain bimolecular up to indefinitely greatpressures.I n reactions of the type 2XY = X, + Y, this con-dition might be fulfilled if both molecules had to be transformed atthe moment of impact or never, as, for example, when the processXY = X + Y is energetically impossible. Even here, however,deactivation at really high pressures could occur by ternary collision,so that in general we must not expect simple definite orders ofreaction, but only approximations to a given order in a givenregion of pressure. The principle that simple molecules givebimolecular reactions while complex molecules give unimolecularreactions should be stated in the form that complex moleculesundergo reactions which become kinetically unimolecular at rela-tively low pressures while simple molecules react bimolecularlyup to relatively very high pressures.Turning to some of the recent experimental evidence illustratingthese ideas, we may first take the decomposition of nitrous oxide.When this was first investigated, a reaction depending upon collisionof two molecules was called bimolecular, and unimolecular reactionswere supposed to be changes of isolated molecules produced by anagency such as radiation: in this sense the decomposition ofnitrous oxide proved to be definitely bimolecular.However, overits whole course the reaction did not give good bimolecular con-stants, nor did the straight line obtained by plotting the reciprocalof the time of half change against the pressure pass through theorigin, as it should in a classically perfect bimolecular reaction :complicating factors were known to exist.N. Nagasako andM. Volmer,go making measurements up to pressures of 10 atmo-spheres, state that the reaction tends to become kinetically uni-molecular. H. C. Ramsperger and G. Waddington,gl treating thereaction as a quasi-unimolecular one, show that only two squaredterms are involved in the activation process, so that the reactionstill belongs to the category of those with a simple activationmechanism. Whether we should regard the actual process of thechemical transformation a t higher pressures as N,O = N, + 0, orwhether the chemical process is still to be regarded as essentially2N,O = 2N, + 0, with a marked deactivation at high pressuresby ternary collisionsthe binary collisions being assumed to havea small but finite duration-is a t the moment an open question.2.physikal. Chem., 1930, [B], 10,414; A . , 1931, 174.Proc. Nut. Acad. Sci., 1931, 17, 103; A . , 671GENERAL AND PHYSICAL CHEMISTRY. 43If the former is the real mechanism it is very interesting, since a tquite low pressures another homogeneous unimolecular decomposi-tion of nitrous oxide has recently been detected.sa Thus therewould be two independent modes of activation of the N,O molecule.The decomposition of ozone,83 bimolecular as ordinarily measuredin the gas phase a t comparatively low pressures, becomes uni-molecular in carbon tetrachloride solution. While it is possiblethat the mechanism of the solution reaction may be quite different(involving a chain process in which the solvent takes part) there isat least an interesting possibility that the solvent merely acts asan enormous concentration of inert gas, and so aids the quasi-unimolecular reaction of ozone to attain its limiting unimolecularrate, the action being analogous to that of hydrogen and otherinert gases on certain gas reactions definitely known to be of thequasi-unimolecular variety.Further interesting examples of quasi-unimolecular reactions havebeen discovered.In the unimolecular decomposition of germaniumtetraethyl,M the rate of reaction begins to fall off at about 70 mm.The limiting velocity is given by the expression In k = 32.8 -51,00O/RT; 8-10 degrees of freedom appear to be involved in theactivation process.The thermal decomposition of dimethyltri-azene 85 is a homogeneous reaction of the same type; the ratefalls off a t initial pressures of about 10 mm., the constant beinggiven by Ink = 26.74 - 33,80O/RT.H. J. Schumacher and G. Sprenger 86 have studied the thermaldecomposition of nitryl chloride, 2N0,CI = 2 N 0 , + Cl,, which theydescribe as a reaction of the first order. As far as the influence ofpressure on the rate of change goes, it approximates much moreclosely to a reaction of the second order, since the influence ofpressure on the unimolecular velocity constants even at 10 atmo-spheres shows no sign of diminishing (as it should in a quasi-unimolecular reaction). Moreover, the relation between the heatof activation and the absolute rate of change is more nearlycharacteristic of reactions of the bimolecular type.The heat ofactivation is 20,500 calories. Taking, for example, the experi-mental data for 140" and 506 mm., the number of molecules whichwould react in a simple bimolecular reaction with activation in82 F. F. Musgrave and C. N. Hinshelwood, Proc. Roy. SOC., 1932, in the8s E. J. Bowen, E. A. Moelwyn-Hughes, and C. N. Hinshelwood, ibid., 1931,84 R . L. Geddes and E. Mack, J . Amer. Chem. SOC., 1930, 52, 4372; A.,8 5 H. C. Ramsperger and J. A. Leermekers, ibid., 1931,53,2061; A . , 916.86 2. physikal. Chem., 1931, [B], 12, 115; A., 1931, 915; ibid., 13, 267.press.[A], 134,211.1931, 7844 HINSHELWOOD :two squared terms only is about 5 x 1017, while the actual numberreacting is about 2.3 x Thus we are within about one powerof ten of the theoretical number ; this is more characteristic of the" bimolecular " region than of the " unimolecular " region.The authors apparently prefer t o call the reaction unimolecular,principally because the complete course of the change for a giveninitial concentration is adequately described by an equation of thefirst order.But it appears also that the products of reaction havea specific accelerating influence, so that this constancy of theunimolecular k must be largely accidental. Although this matteris really one of nomenclature, it is mentioned here because a gooddeal of confusion sometimes arises now about the use of the term" order of reaction " and some convention should perhaps be agreedon.As a result of complications due to catalysis by products andso on, the order deduced from the influence of the initial concen-tration is not always the same as that which gives the best equationto express the course of the reaction with time for a given initialpressure. This comes to light particularly clearly in the nitrylchloride reaction. It seems to the reviewer that the more significantorder is that which describes the influence of changing pressure.The slow decomposition of diazomethane 87 between 140" and220" is stated to be bimolecular ; probably over an extended rangeof pressures the order would also prove variable here. The thermaldecomposition of methane, according to G.C. Holliday and W. J.Gooderham,88 takes place in two stages in the first of which twomolecules of methane react together to give C,H,. Interestingpreliminary observations 89 on the reaction C1, + Br, = 2BrC1indicate that it may be a bimolecular reaction proceeding accordingto the simplest possible law, with a heat of activation of about14,000 cals.The observation 90 that the decomposition of nitrogen pentoxideceases entirely below a certain pressure does not appear t o becorrect, though a definite falling off in the specific rate in the neigh-bourhood of 0-06 mm. now seems to be e~tablished,~~ in accordancewith theoretical expectations. This reaction therefore is also of thequasi - unimolecular class.E. W.R. Steacie, J. Physical Chern., 1931, 35, 1493; A., 916.** J., 1931, 1594; A., 915.89 W. Jost, 2. physikal. Chem., 1931, [BJ, 14, 413.O0 C. Sprenger, ibid., 1928, 136, 149; A., 1928, 1099.01 H. C. Ramsperger and R. C. Tolman, Proc. Nat. Acad. ScG, 1930, 16,6 ; A., 1930,547; H. J . Schumacher and G. Sprenger, ibid., p. 129; A . ,1930, 708; J. H. Hibben, J. Physical Chem., 1930, 34, 1387; A., 1930, 1127;J. H. Hodges and E. F. Linhorst, Proc. Nat. Acad. Sci., 1931, 17, 28; A,,436GENERAL AND PHYSICAL CHEMISTRY. 45If it turns out to be true that quite simple molecules such as N,Ocan survive the activating collision and decompose spontaneouslygiving a free atom after a certain finite lapse of time, then thereverse process, namely, the combination of two simple structuressuch as atoms to give one molecule, becomes a thermodynamicnecessity.In these circumstances the generally accepted principlethat ternary collisions are necessary for- combination between twoatoms to occur loses its general validity. According to M. Volmer 92the newly formed molecule may be in a state where the quantisationof vibrational energy is no longer sharply defined. If this is so thenecessity of the ternary collision to allow the correct adjustment ofthe energy of the molecule formed t o the quantum requirements nolonger exists.An interesting study has been made by G. B. Kistiakowsky andM. Nelles 93 of the slow isomerisation in the gas phase of dimethylmaleate into dimethyl fumarate, which occurs in the region of300".The reaction is of the quasi-unimolecular type, but unlikemost reactions of this kind it shows an abnormally small rate, beingonly about 103 of that calculat'ed on the assumption that twosquare terms are involved in the activation process. Thus, not onlyare the degrees of freedom of the molecules mostly inactive, but inaddition a great many of the collisions are from the chemical pointof view inelastic.In the following will be briefly summarised a few results ofoutstanding importance in connexion with the mechanism ofchemical reactions.D. L. Chapman and F. B. Gibbs 94 have shown that if chlorine issufficiently purified from oxygen it combines with hydrogen at arate proportional to the square root of the light intensity. This isbecause the chlorine atoms propagating the reaction chains areremoved by recombination rather than by the chemical action offoreign substances (as occurs in the incompletely purified gasesj.Thus the hydrogen-chlorine combinations falls into line with thecorresponding bromine reaction, the rate of which is usually pro-portional to the square root of the intensity, but becomes propor-tional to the fist power of the intensity in the presence of certaindeliberately introduced impurities.In connexion with the well-known induction period in the hydrogen-chlorine reaction, shownlong ago by Chapman to be due to the presence of nitrogen tri-chloride, it should be mentioned that the photosensitised decompos-ition of this latter substance has been studied by J.G. A. Griffithssa 2. phyaikal. Chem., 1931, [BJ, 13, 299.O3 Ibid., 1931, Bodenstein Festband, p. 369; A., 1230.s4 Nature, 1931,127, 854; A,, 80646 HINSHELWOOD :and R. G. W. NorrishYs5 and correlated with the inhibition of thehydrogen-chlorine combination. M. Bodenstein and W. Unger 96have discovered another mechanism by which the chains in thephotochemical hydrogen chloride formation are broken, namely, adirect chemical reaction in t'he gas phase between some of thechain carriers and a gaseous silicon compound formed by theaction of activated chlorine on the glass or silica walls of thevessel.Franck and Bodenstein 97 have suggested the following modific-ation of the Nernst chain in the combination of hydrogen andchlorine :(1) c1, + hv = 2c1;(2) C1+ H20 + H, = HCl + H,O + H ;(3) H + C1, + H, = 2HC1+ H ;(4) H + 0, + H, = H,O + OH;(5) H + 0, + C1, = HCl + (210,.For the rate of reaction this scheme gives the equationwhich expresses adequately most of the experimental results ( I isthe intensity).The chief experimental reason for introducing thenew hypothesis is the necessity for the presence of a small con-centration of water vapour reported by Coehn and Jung. Thusstep (2) is one of the simplest reactions which can follow (1) if ahydrogen atom is to be produced as in the original Nernst chain.The concentrations of steam molecules and chlorine atoms are,however, very minute, and if it be supposed that at each step inthe cycle of changes a chlorine atom and a steam molecule mustmeet, there is the greatest difficulty in accounting for the observedrate of reaction. To avoid this difficulty, chlorine atoms areassigned no r6le except in the first link of the chain, the subsequentsteps all depending on the reaction and re-formation of hydrogenatoms in the series of termolecular processes represented above.It is also interesting to note that the photochemical formation ofhydrogen chloride takes place with an effective quantum yieldwhen the light is absorbed in the band region of the chlorinespe~trum,~8 i.e., where activated molecules and not free atoms are9 5 Nature, 1931,127, 14; Proc. Roy.Soc., 1931, [A], 130, 591; A., 179, 578.B6 2. physikal. Chem., 1930, [B], 11,253 ; A . , 1931, 319.9 7 M. Bodenstein, Faraday Society, Liverpool Meeting, 1931 ; Trans.98 E. Hertel, 2. physikal. Chem., 1931, [BJ, 14, 443.FaraduySoc., 1931, 27, 413; A., 1136(3ENERA.L AHD PHYSICAL CHEMISTRY. 47primarily formed. These molecules are supposed to give atoms insubsequent collisions.The following investigations illustrate interesting points ofprinciple in connexion with the theory of chain reactions. H. W.Melville and E. B. Ludlam 99 have studied the influence of a longseries of foreign gases on the lower critical limit encountered in theoxidation of phosphorus vapour, and have established a definitecorrelation between the influence of the gas and the diffusioncoefficient through it of the chain-propagating molecules, thus con-firming the view that at the critical limit the branching of thechains just ceases to be kept in check by deactivation at the wall ofthe vessel. In this connexion also reference may be made to anelaborate mathematical treatment of chain reactions by theapplication of the classical diffusion equations.A curious phenomenon shown by mixtures of hydrogen sulphideand oxygen has been observed by H. W. Thompson : underappropriate conditions a series of successive explosions takes place,each being preceded by an increasing time lag. The effect seems tobe connected with the formation of a catalytic substance during aninduction period and the occurrence of an explosion when a criticallimit is reached, the explosion being incomplete, however. Duringan ensuing induction period a critical concentration is againreached.Norrish3 has observed a sharp limiting pressure of chlorine inmixtures with hydrogen and oxygen above which an explosionoccurs when the gases are exposed to the light of a mercury lampat 300".In the last report reference4 was made to the fact that atomicreactions of the type Na + C1, = NaCl+ C1 may not alwaysoccur without activation, as the work of Polanyi and others ondilute alkali metal-halogen flames had suggested. The wholequestion has been dealt with in a paper by H. von Hartel andM. P ~ l a n y i , ~ who find that in the reactions between sodium vapourand the methyl halides the activation energy is nil for the iodideand rises regularly through the bromide and chloride to the fluoride.There is also an appreciable " inertia " associated with the reactionbetween sodium vapour and cyanogen, but since this is independentV. Bursian and V . Sorokin, 2. physikal. Chem., 1931, [BJ, 12, 247; A.,2 Nature, 1931,127,629; A., 689; also H. W. Thompson and N. 8. Kelland,99 Proc. Roy. SOC., 1931, [A], 132, 108; A., 1014.688.J., 1931, 1809; A., 1014.Nature, 1931,127, 853; A., 805.Ann. Reports, 1930, 27, 20.5 2. physikal. Chem., 1930, B, 11, 97 ; A., 17448 GENERAL AND PHY SIC& CHEMISTRY.of temperature it must depend upon some kind of " steric ) ) factorrather than upon the necessity for ,activation. Although theprinciple that atomic reactions do not require activation ceases t obe absolute, it remains true to say that many such reactions doproceed without activation, and also that the heat of activation ofothers is generally quite small.C. N. HINSHELWOOD