ANNUAL REPORTSON THEPROGRESS O F CHEMISTRY.GENERAL AND PHYSICAL CHEMISTRY.The Velocity of Chemical Reactions.THE difficult fundamental problems of the kinetics of chemicalreactions have continued to attract much discussion and experi-ment. The accepted view that reaction occurs through the agencyof activated molecules which have acquired a critical incrementof energy over the average tends to become more definite by theidentification of activated states with higher quantum states of themolecule, and a solution of the problem may be expected to followthe further development of our knowledge of quantum mechanicsas it affects the energy exchange between molecules.R. C. Tolman 1 points out that the well-known expressions foruni- and bi-molecular reactions : -dC/dt = kC = k’e--E/RTC, and- dC/dt = kCC‘ = k’dT .e-(E+E’)IRTCC’, have previously beendeduced only by assuming some possible but not inevitable mechan-ism, one fast enough to maintain the full Maxwell-Boltzmann quotaof active molecules in the reacting system or one involving radi-a t i ~ n . ~ By a mathematical analysis based on the “principle ofmicroscopic reversibility ” * he derives these equations withoutassuming any specific mechanism, and concludes that they givethe energy of activation under a wider variety of conditions thanhave hitherto been considered.The main problem is to find a mechanism of energy transfer1 J . Amer. Chem. SOC., 1925, 47, 2652.2 €4. Marcelin, Ann. Pihysique, 1915, [ix], 3, 120; A., 1915, ii, 328; J.Rice,Brit. Assoc. Reports, 1915, 397; W. H. Rodebush, J . Amer. Chem. SOC.,1923, 45, 606; A., 1923, ii, 303; J. A. Christiansen and H. A. Kramers,2. physikal. Chem., 1923, 104, 451 ; A., 1924, ii, 28.J. Perrin, Ann. Physique, 1919, [ix], 11, 5 ; A., 1919, ii, 177; W. C. McC.Lewis, J., 1918, 113, 471 ; Phil. Mag., 1920, [vi], 39, 26; A , , 1920, ii, 100.4 “ If we have a system in statistical equilibrium, the principle requiresnot only that the number of molecules in any given state shall remain con-stant, but that the number leaving that state in unit time by any particularpath shall be made up by the entrance of an equal number of molecules by thereverse of that partidar path ” (R. C. Tolman, Zoc. cit.).A* 12 ANNUAL REPORTS ON THE PROGRESS OF CHEMISTRY.adequate to supply activated molecules a t least as fast as they areused up by reaction.The position has been reviewed by M. Boden-stein,5 R. C. Tolman,6 and others.' In a purely thermal homo-geneous gas reaction activation may be effected (a) by collisions,( b ) by absorption of radiation. The simplest view is that activationresults from the inelastic collision of two molecules of sufficientlyhigh energy of translational motion relative to each other, thelatter being converted into internal energy of activation. In abimolecular change A + B = AB, reaction may depend on collisionbetween * (a) A* + B, or ( b ) A* + B*, or ( c ) A + B, in the last caseboth being activated by the collision itself. Tolman finds that thedata for well-defined second-order reactions are compatible with( b ) and (c), not with (a), but regards this as no proof of activationby collision.C. N. Hinshelwood lo regards the experimentalevidence as definitely in favour of ( c ) , the heat of activation, E,being a real measure of the relative kinetic energy that the moleculesmust possess in order that reaction may occur on collision. Onthe other hand, activation by simple collision has been proved tobe very far from adequate to maintain unimolecular reactions.The chance of two molecules of nitrogen pentoxide colliding withthe high energy of 24700/N cal. is very small. The view recentlyput forward by Sir J. J. Thomson l1 that certain molecules accumu-late by collision sufficient energy to cause dissociation implies anaccelerating effect by an inert diluent, which is contrary to recentobservation. l2The failure of simple collisions to account for the high rates offkst-order reactions accompanied by such large energy of activationled J.A. Christiansen and H. A. Kramers l3 to propose the " hotmolecule " theory whereby molecules are activated mainly bycollisional transfer of the heat of reaction ; thus for a unidirectionaldecomposition (regarded as exothermal) : AB (AB)* + [A* +B*]+A + B. The complex of resultants immediately after spon-taneous decomposition of (AB)* contains the heat of reaction as wellZ. Elektrochem., 1925, 31, 343.J . Arner. Chern. SOC., 1925, 47, 1524; A., ii, 799.A* signifies an activated molecule of A.' E.g., G.N. Lewis and D. F. Smith, ibid., p. 1508; A., ii, 799.9 LOG. cit.lo C. N. Hinshelwood and J. Hughes, J., 1924, 125, 1841; C. N. Hin-shelwood and R. E. Burk, Proc. Roy. SOC., 1924, A, 108,284; A . , 1924, ii, 751 ;C. N. Hinshelwood and C. W. Thornton, Phil. Mag., 1925, [vi], 50, 1135;Ann. Report, 1924, 11.l1 Phil. Mag., 1924, [vi], 47, 337; A., 1924, ii, 222.l2 See also criticisms by J. Rice, Trans, Puruduy BOG., Oct., 1925;l3 LOG. cit., Ref. 2.A,,ii, 1076 (part of a general discussion on photochemical reactions)GENERAL AND PHYSICAL CHEMISTRY. 13as the original energy of activation. This high-energy product(" hot molecule ") can then revert to normal by transfer of itsexcess energy to the reactant AB on collision, thus activating it,and so on.These reaction chains may be started by ordinaryhigh-energy collisions. There is strong experimental evidencethat such high-energy molecules can exist, and transfer theirenergy in this way.14 The theory leads to velocity equations ofthe right form, e.g., for a unimolecular reaction :-dC/dt = kC =AC* = A(p*/p)e-ElzTC, where C and C* are the concentrationsof the normal and activated reactants, respectively, p and p* thecorresponding a priori probabilities l5 of these states (p*/p is notvery different from unity), while A is the probability per secondthat an activated molecule will decompose ( l / A , the mean life-period of the latter, is about sec.). Although there is goodevidence that molecular chain mechanisms actually occur in somereactions, difliculties arise which appear to rule this out as a generalexplanation.lG Thus to keep a first-order course each decornpos-ition must be immediately followed by activation of reactant inorder to maintain the same statistical equilibrium A B e (AB)* aswould obtain in the absence of reaction, which involves the highlyimprobable assumption that the " hot molecule," in spite of itsnumerous collisions with indifferent molecules, preserves its highenergy content until it meets a reactant.The unimolecular rate ofdecomposition of nitrogen pentoxide is not influenced by di1uents.l'There is also the objection that decompositions are endothermic, sothat supplementary activation would appear to be necessary ;J. Rice l8 has, however, given reasons for doubting the validity ofthis criticism.No form of collision theory alone appears to beadequate in the case of unimolecular reactions.The only alternative is activation by absorption of radiationpresent in temperature equilibrium with the system. The well-known Lewis-Perrin theory l9 of activation by selective absorptionof radiation of frequency v ( A + hv +A*, and E = Nhv per mol.)l4 See, e.g., 0. Klein and S. Rosseland, 2. Physik, 1921, 4, 46; A., 1921,ii, 291; J. Franck, ibid., 1921, 4, 89; 1922, 9, 259; A., 1922, ii, 464; G.Cario and J. Franck, ibid., 1922, 11, 161 ; A,, 1922, ii, 809.15 " The a priori probabilities take account of the fact that each quantumstate of dehite energy content of a molecule may be realised in differentways, the p's representing the number of possible modes of realisation of eachstate" (A.C. McKeown, Phil. Mag., 1923, [vi], 46, 323).la R. C. Tolman, ref. 6; M. Bodenstein, ref. 5; G. N. Lewis and D. F.Smith, ref. 7; J. Rice, ref. 12.18 J. Rice, loc. cit., ref. 12.19 Trans. Paraday SOC., 1922, 1'7:l7 See refs. 33.Discussion on the Radiation TheorySee also a review of radiation theories by of Chemical Action, p. 545 et seq.H. S. Harned, J. B'ranklin Inst., 1923, 196, 1811.1 ANNUAL REPORTS ON THE PROGRESS OF CHEMISTRY.has been abandoned because the frequency predicted from thetemperature coefficient of reaction velocity does not in generalagree with absorption bandsY2O nor does it accelerate the reaction,and because activation by this means cannot occur with the speedrequired to account for known first-order rates.21 R.C. Tolman’scalculations 22 indicate, however, that this mechanism is a possibleone for bimolecular reactions. Attempts to improve the radiationtheory 23 found less favour than theories of collision, but on accountof the difficulties of the latter, especially as regards unimolecularreactions, attention has again been focussed, by G. N. Lewis,R. C. Tolman, E. K. Rideal, J. Rice, and others, on radiation as themost probable activating agency. T ~ l m a n , ~ ~ reviewing the varioussuggestions for a less restricted radiation theory, finds most promis-ing the view that, on the assumption of a nearly continuous seriesof high energy quantum states corresponding to the idea of theweakening of a chemical bond, energy may be absorbed over acontinuous range of frequencies, thus allowing an ample inflow.Somewhat similar ideas had been expressed by E.K. Rideal andW. C. McC. Lewis. G. N. Lewis and D. F. Smith,25 who support a“ general ” radiation theory, employ the concept of discrete lightquanta of relatively large size, thus facilitating collisional exchangeof energy between molecules and quanta. J. Rice 26 has subjectedthese latter developments to a searching criticism, without, how-ever, questioning the necessity of some form of radiation theory.These more recent views have not yet been formulated withsufficient precision for experimental test.W. E. Garner 27 maintains that the close agreement between thecritical increment from temperature coefficient with that calculatedfrom the kinetic theory ( E = 2 x total number of collisions xe-ElRT. P), assuming P == 1, i.e., every collision between activatedmolecules is “ fruitful,” cannot, as Hinshelwood 28 supposes, beaccepted as evidence that P is unity, or that the critical incrementfrom temperature coeEcient is the true energy of activation.20 The interesting observation has been made by W.T. David, Proc. Roy.Soc., 1925, A , 108, 617; A . , ii, 980, that the rate of explosion of gas mixturesis increased when the infra-red radiation superimposed on that emitted bythe burning gases is of the type that can be absorbed by the gases.21 J. A. Christiansen and H. A. Kramers, ref.2; R. C. Tolman, ref. 6.22 Ref. 6.23 See ref. 19 and R. C . Tolman, J . Amer. Cham. SOC., 1920, 42, 2506;1921, 43, 269; A., 1921, ii, 99, 248; J. Perrin, ref. 3.24 Refs. 6 and 23; J. Rice, ref. 12.25 Ref. 7.26 Ref. 12.27 Phil. Mag., 1925, [vi], 49, 4G3; 50, 1031; A., ii, 552, 1167.28 Ibid., 1925, [viJ, 50, 360; A., ii, 874.See also S. C. Roy, Z. Plqsili, 1925, 34, 400; A., ii, llG7GENERAL AND PHYSICAL CHEMISTRY. 15M. Born and J. Franck29 stress the importance of attackingproblems of chemical kinetics from the point of view of the Bohratom and the quantum laws of energy exchange. They considerthat in the collision reaction A + B+ C* + C + Q (heat ofreaction), the fact that the primary complex, C*, formed at themoment of collision contains Q plus the relative kinetic energy, X ,of A and B before collision does not of itself, as has often beenassumed, prevent the formation of C, for molecules can existtemporarily having a quantised energy content much greater than &.What does, however, prevent the formation of C is the infinitelysmall probability that Q + X will agree with any quantum state( X being continuous).The energy of this incompletely quantisedprimary molecule (or " quasi-molecule ") can adjust itself to adiscrete value neither by radiation nor conversion into translationalenergy, consequently C can be formed only if, during the period ofcollision, the complex is struck by a third atom or molecule, whichcarries off the excess energy (heat of reaction) as kinetic energy oftranslation, thereby converting C* into C.Such ternary collisionsare evidently unnecessary in the reaction type AB + C -+ AC + B,since the continuous energy of translation is available for adjust-ment to a definite quantum state, and it is suggested that thecatalytic effect of water vapour may be associated with a transitionfrom the addition to the relatively faster exchange type of reaction.They are also unnecessary in the addition type when one reactantis large enough for the quantum states to be practically continuous.Surface catalysis provides an extreme example of the latter ; whenB strikes adsorbed A, energy adjustment to the appropriate quantumstate can be established by the continuous energy reservoir of thesolid catalyst.M. Polhyi and E. WignerY3O while admitting themechanism of ternary collision, differ from Born and Franck inregarding the probability of combination resulting from binarycollision as finite, though small. M. Bodenstein31 points out thatthe ternary collision theory suggests an acceleration of reaction inthe presence of inert gases, which is not in accord with experiment.Whilst the existence of homogeneous bimolecular gas reactionsfree from " wall effect,s " has been established, the fact that reputedexamples of such unimolecular reactions have proved to be notindependent of surface/volume ratio or of pressure has led to seriouadoubts as to their actual existence. Thus the best defined example,that of the thermal decomposition of nitrogen pentoxide, was29 2.Physik, 1925, 31, 411; A., ii, 266; Ann. Physik, 1925, [iv], 76, 225;A., ii, 365; J. Franck, 2. Elektrochem., 1925, 31, 350; Natumuiss., 1924, 47,1063; A., 1925, ii, 836.30 2. Physik, 1925, 33, 429.31 Ref. 516 ANNUAL REPORTS ON THE PROGRESS OF CHEMISTRY.apparently catalysed by nitrogen peroxide.32 In view of its greattheoretical importance, this reaction has been carefully re-examinedby four independent investigator^,^^ who agree in finding no evidenceof catalysis or wall-effect and no variation of rate ( a ) over an enor-mous range of pressure down to less than 1 mm., (b) in the presenceof excess of argon, air, and nitrogen peroxide. Further, the rate ismuch the same in inert solvents over wide concentration ranges.=The absence of any retarding influence of diluents or of very lowpressures renders improbable any chain mechanism. According toH.S. Hirst and E. K. Ridea1,35 however, who worked a t pressuresdown to 0.01 mm., the rate, constant above 0.25 mm., begins toincrease a t this critical pressure, approaching a limiting valuefive times the normal as the pressure diminishes. This limitingvalue is in remarkably close agreement with the Dushman-Rideal 36equation :--dC/dt = ve-shv/RT. C, which suggests some form ofradiation mechanism of activation. The reaction is consideredto take the following course : one-fifth of the activated moleculesalways decompose independently of pressure, whilst four-fifthsdecompose only if, after activation, they fail to collide during aperiod of about 10-6 second (collision within this period causessimple de-activation).There seems to be no reason for doubtingthe unimolecular homogeneous character of this reaction.D. F. Smith 37 finds that the thermal decomposition of sulphurylchloride, which has been stated3* to be a wall reaction, is a first-order reaction independent of surface/volume ratio except to aslight extent at lower temperatures. According to D. L. Watson,39the thermal decomposition of four derivatives of oxalacetic esterin the pure liquid phase follows a unimolecular course, and inertsolvents are without influence on the velocity coefficients. Thephenyl derivative decomposes autocatalytically. It is shown that,by molecular chain activation, first-order reactions should be auto -32 F.Daniels and E. H. Johnston, J . Amer. Chem. SOC., 1921, 43, 53; A.,1921, ii, 249; R. H. Lueck, ibid., 1922, 44, 757; A., 1922, ii, 433; F. Daniels,0. R. Wulf, and S. Karrer, ibid., 1922, 44, 2402; A . , 1923, ii, 24.33 J. K. Hunt and F. Daniels, ibid., 1925, 47, 1602; A., ii, 801 ; E. C.White and R. C . Tolman, ibid., 1925, 47, 1240; A., ii, 682; H. S. Hirst, J.,1925, 127, 657; H. S. Hirst and E. K. Ridoal, Proc. Roy. SOC., 1925, A , 109,526.34 R. H. Lueck, ref. 32.36 S. Dushman, J . Amer. Chem. SOC., 1921, 43, 397; A . , 1921, ii, 315;E. K. Rideal, Phil. Mug., 1920, [vi], 40, 461; A., 1920, ii, 676; J. Rice,ibid., 1923, [vi], 46, 312; A., 1923, ii, 622; A. McKeown, ibid., 1923, [vi],46, 321 ; A., 1923, ii, 623.35 Ref.33.37 J . Amer. Chem. Xoc., 1025, 47, 1862; A . , ii, 876.38 C. N. Hinshelwood and C. R. Prichard, J., 1923, 123, 2725.39 Proc. Roy. SOC., 1925, A , 108, 132; A., ii, 556.see C. N. Hinshelwood, J., 1920, 117, 156; 1921, 119, 721.For similar reactionsGENERAL AND PHYSICAL CHEMISTRY. 17catalytic, and this is considered to be the essential character of allfour reactions. The thermal decomposition of ozone 40 is mainlyhomogeneous (second order), but not free from wall effect. Oxygenretards and indifferent gases accelerate the reaction. The mechan-ism proposed is : 0,e03* ; 03* + 0,e-Complex ; Complex +30,' it being assumed that the complex may break up into 0, or0, either spontaneously or on collision ; collision with oxygenfavours Complex + O,, and collision with an indifferent gas favoursComplex + 0,.This mechanism includes the possibility of mole-cular chain activation and Born-Franck ternary collisions.Theories which provide for energy transfer adequate to maintainknown reaction rates touch but one aspect of the question. Theenergy of collision in gases dried to inertness is presumably notaltered in the presence of minute traces of water, and in this sensemost reactions are catalytic. This point has been emphasised byR. G. W. NorrishY4l who considers that the inert " dry molecules "are partly activated by close association with water, which, in virtueof its high polarity, weakens the structure of the " resting '' mole-cule, so that the supplementary activation necessary for reactionmay be attained by collision.The catalytic activity of surfaces isdue to similar causes, non-polar surfaces such as paraffin wax beingnon-catalytic. The rate of photochemical union of hydrogen andchlorine is independent of the pressure of water vapour down toloe4 mm., when it begins to fall, reaching zero a t lo-' mm. water-vapour pressure.42 This corresponds to a gradual removal of thewater film. It is therefore suggested that the removal of thiscatalytically active film rather than " ultra-dryness " of the gasesis responsible for the suspension of reactions of dry gases in general.Quantitative studies of moisture effect on reaction rate are muchneeded.Heterogeneous Catalysis and Adsorption.A very considerable amount of work on this subject has beendone since it was last noticed in these Reports.43 I.Langmuir'swell-known theory43 has on the whole been substantiated and is40 R. 0. Griffiths and A. McKeown, J., 1925, 127, 2086.*l J., 1923, 123, 3006; H. S. Taylor, J. Physical Chem., 1924, 28, 897.42 M. Bodenstein and W. Dux, 2. physikal. Chem., 1913, 85, 297; A., 1913,ii, 1039; A. Coehn and G. Jung, ibid., 1924, 110, 705; A., 1926, ii, 142;R. G. W. Norrish, J., 1925, 127, 2316; A., ii, 1179; Trans. Faraday XOC.,Oct., 1925; A., ii, 1080.43 See the Reports of the Committee on Contact Catalysis (NationalResearch Council), especially the second report : W. D. Bancroft, J. PhysicalChew&., 1923, 2'7, 801; and third report : H. 5. Taylor, ibid., 1924, 28, 898.An excellent account is given by H.S. Taylor in his " Text-book of PhysicalChemistry " (Macmillan, 1924). See also the discussion on catalysis, J .Paraday SOC., 1922, 17, 60718 ANNUAL REPORTS ON THE PROGRESS Ol? CHEMISTRY.very generally accepted. A consideration of kinetic studies, of theparallel investigation of adsorptive capacity and reaction rate fort’he same catalyst, and of work on catalyst surfaces has, however,led H. S. Taylor to a “ concept of the catalytic surface which is,perhaps, more comprehensive than earlier efforts and which leadsto interesting general conclusions with reference to matter in thesolid state.” It has been fully established that the Langmuiractive areas or centres occupy only a small fraction of the surface,that they vary in their capacity both to adsorb and promote reaction,and that metallic catalysts are extraordinarily sensitive to heattreatment and poisoning, but the consequent reduction of catalyticactivity greatly exceeds that of adsorptive capacity.To accountfor these facts, a modification of the Langmuir concept is proposed,and illustrated by reference to a metallic catalyst such as nickel.Whilst giving no information about the surface, X-ray examin-ation 45 shows that active catalysts prepared by low-temperaturereduction of oxides consist of fine granules having the definitelattice structure of the crystals. The mode of production andsensitivity to moderate heat treatment (incipient sintering) suggestincomplete surface crystallisation, i.e., occasional groups of atomsfixed in metastable positions associated with high energy andchemical unsaturation relative to the atoms in the regular latticebelow (Fig.l).46NiINi Gas Phase NiI 1Ni- Ni Ni-I I INi- Ni- Ni- Ni- Ni-I l l 1 1I. -Ni- Ni- Ni- Ni- Ni- Ni- Ni- Ni-Ni- Ni-Ni- Ni- Ni-11. -Ni- Ni- Ni- Ni- Ni- Ni- Ni- Ni-Ni- Ni-Ni- Ni- Ni-l l l l 1 1 . 1 I I I I l lGranule ProperFIG. 1.From layer I1 (where each Ni is surrounded by six others) out-wards, the degree of constraint or saturation decreases to a varyingextent, the “ peak ” atoms differing from gaseous atoms only bythe single valence which holds them to the solid. While a freeatom can bind four CO groups, a singly anchored atom may adsorbthree such groups (or their equivalent), and a doubly anchored,two.Corner and edge atoms of a crystal are also unsaturated to44 Proc. Roy. Xoc., 1925, A, 108, 105; A., ii, 562.4 j G. L. Clark, W. C. Asbury, and R. 31. Wick, J . Arner. Chem. Soc., 1925,46 Bigure taken from H. S. Taylor; ref. 44.47, 2661 ; R. W. G. Wyckoff and E. D. Crittenden, ibid., p. 2866GENERAL AND PHYSICAL CHEMISTRY. 19different extents. Reactants such as hydrogen and ethylene maybe held by the same Ni atom (Langmuir postulated adsorption ofreactants on adjacent centres). The rise of activity of platinumand silver in oxidation catalysis is due to surface disintegrationleading to an increased number of unsaturated atoms. Moderateheat treatment causes atomic displacement to a more regularsurface with loss of energy, the more mobile singly-bound andhighly active atoms being the more readily displaced.That thesaturation capacity of a surface varies for different adsorbed gasesis thus to be expected ; there are, for example,47 more copper atomscapable of holding carbon monoxide than hydrogen. At highertemperatures, adsorption is increasingly confined to the moreunsaturated surface atoms and it is t o these that poisons firstbecome atta~hed.~8 Correlating the adsorptive and catalytic powerof copper for the reaction C2H4 + H, = C2H,, R. N. Pease 49found that poisoning by mercury, whilst not appreciably affectingthe weaker (high pressure) adsorption, destroyed both the stronger(low pressure) adsorption and the catalytic activity, and a coppercatalyst which strongly adsorbed 5 C.C.of carbon monoxide suffereda 90% loss of catalytic activity by adsorbing only 0-05 C.C. of thispoison, which means that the surface owed 90% of its activity toless than 1% of its strongly adsorbing centres.5* Thus even low-pressure adsorption measurements may give no true index ofcatalytic activity. That the varying degree of saturation ofsurface atoms involves varying catalytic activity is well illustratedby the progressive poisoning experiments of G. Vavoii and A.H u s s o ~ , ~ ~ who found that when platinum had been poisoned bycarbon disulphide just sufficiently to suppress the hydrogenation ofdipropyl ketone, it could still hydrogenate piperonal and nitro-benzene, whilst a further dose of poison stopped the former, butnot the latter reaction, which again could be poisoned by morecarbon disulphide.The amount of active surface is thus deter-mined by the reaction catalysed.Support for the theory outlined above is found in the high valuesof heats of adsorption, e.g., of hydrogen on nickel 52 : 13,500--20,5004 7 R. N. Pease, J. Amer. CI:em. SOC., 1923, 45, 1106, 2235; A., 1923, ii, 472,4 8 See, e.g., data for hydrogen on nickel: A. W. Gauger and H. S.49 R. N. Pease, ibid., p. 2296; A . , 1923, ii, 862.50 R. N. Peaso and L. Stewart, &id., 1925, 47, 1235; A., ii, 691; see also13. B. Maxted, Trans. Puraday SOC., 1917, 13, 36; E. B. Armstrong andT. P. Hilditch, ibitl., 1922, 17, 669.6 1 C'ompt.rend., 1922, 175, 277; A . , 1922, ii, 631.52 R. A. Beebe and H. S. Taylor, J. Amer. Chern. Xoc., 1924, 46, 43; A.,1024, ii, 159; B. Foresti, Guzzetta, 1923, 53, 487; 1924, 54, 132; 1925, 55,185; A., 1923, ii, 747; 1924, ii, 320; 1925, ii, 692.842.Taylor, ibid., p. 920; A., 1923, ii, 39820 ANNUAL REPORTS ON THE PROGRESS OF CHEMISTRY.cal. per mol., depending on its history. E. A. Blench and W. E.Garner 53 find values for oxygen on charcoal varying from 60,000cal., a t low temperatures and high adsorptions, to the surprisinglyhigh value of 220,000 cal. a t high temperatures and low adsorptions.As expected, there is a greater heat effect for initial than for subse-quent adsorption. The heat of formation of carbon dioxide fromsolid charcoal, which includes the endothermal breaking of carbonlinkings, is 97,000 cal., whilst from gaseous carbon it is 380,000 ~ a l .~ ~The high heat of adsorption indicates that some carbon links onthe surface are already broken, yielding atoms in a highly un-saturated and active state, so that their combustion resembles thatof gaseous rather than of crystalline carbon. The fact that theheat of combustion of the incompletely crystallised charcoal exceedsthat of graphite is also in agreement with the theory. E. F. Arm-strong and T. P. H i l d i t ~ h , ~ ~ who have made a thorough study ofcatalytic hydrogenation at nickel surf aces, express general agree-ment with this theory, but consider that the acting nickel atommay at the moment of catalytic change be actually deta?ched fromthe metal as a complex of nickel, oil, and hydrogen, which breaksup with deposition of the nickel.According to A. W. Gauger,5spure nickel and platinum distilled in a vacuum on to the surfaceof glass wool are catalytically inactive, so that activity depends oncondition rather than on extent of surface. He regards the moleculesor atoms a t active centres as having electrons in energy levelshigher than normal. 31. Bodenstein 57 refers to deformation of suchadsorbed molecules.Special interest attaches to the investigations of W. G. Palmer 58and F. H. Constable 59 dealing mainly with the catalytic dehydro-genation of alcohols (vapour) in the presence of copper. Thereaction rate is independent of pressure over a 12-fold range,showing that reaction occurs only in the layer in immediate contactwith the copper.The primary alcohols, ethyl, propyl, butyl,isobutyl, and isoamyl, decompose a t the same rate (isopropyl53 J , , 1924, 125, 1288; Nature, 1924, 114, 932; A., 1925, ii, 140.54 K. Fajans, Ver. Deut. Physikal. Ges., 1913, 14, 324.5 5 Proc. Roy. SOC., ,1925, A, 108, 111 ; A., ii, 562.5 6 J. Anter. Chem. SOC., 1925, 47, 2278; A., ii, 1072.5 7 Annalen, 1924, 440, 177; A., 1925, ii, 216.5 8 W. G. Palmer, Proc. Roy. SOC., 1920, A, 98, 13; 1921, A , 99, 412; A . ,1920, ii, 609; 1921, ii, 542; D. M. Palmer and W. G. Palmer, ibid., p. 402;A., 1921, ii, 541; W. G. Palmer, ibid., 1922, A , 101, 175; A., 1922, ii, 437;W. G. Palmer and F. H. Constable, ibid., 1924, A, 106, 250; 1925, A , 107,255; A., 1924, ii, 843; 1925, ii, 311.59 F.H. Constable,ibid., pp. 270, 279; A., ii, 311; Nature, 1925, 116, 275;A., ii, 983; Proc. Roy. SOC., 1925, A , 108, 355; A., ii, 804; Proc. Cuinb.Phil. SOC., 1925, 22, 738; A., ii, 881GENERAL AND PHYSICAL CHEMISTRY. 21alcohol five times faster) and with the same temperature coefficient,indicating identical mechanism and energy relations. The adsorbedmolecules are oriented with the hydroxyl group only in closeassociation with the surface. At the active centres the OH groupis distorted so that the hydrogen atom readily oscillates to thecopper, a second atom of hydrogen of the -CH2*OH group breakingaway automatically with formation of aldehyde. The energy ofactivation is supposed to be concentrated in the hydrogen atom ofthe OH group.The catalyst, prepared by successive oxidationand reduction at low temperature, is perfectly reproducible and adetailed study leads to views concerning the nature of the surfacewhich differ in no essential respect from those of H. S. Taylor.The active centres form only a very small fraction of the totalsurface. They are associated with varying critical increments,most of the reaction occurring at the centres of low increment.A mathematical analysis of the process is given.The varying catalytic and adsorptive activity of different portionsof a charcoal surface has been studied by E. K. Rideal and W. M.Wright,GO who express views in close agreement with those ofTaylor and Constable.By measuring the rate of absorption ofoxygen and evolution of carbon dioxide by charcoal suspended inwater (a zero-order reaction), they find the active fraction of thesurface to be 0*38%, this being the ratio of the number of moleculesof poison just sufficient to stop the reaction to the number necessaryto saturate the surface. In the catalytic oxidation of organicacids, 40% of the surface is active.Much discussion has centred round the question as to whetherthe active substance in a metal catalyst is the metal or an oxide.Thus M. C . Boswell and his associates maintain that the incom-pletely removed oxygen in nickel granules is vital for their activity.Recent work 62 appears, however, to have proved conclusively thatthe active substance is nickel.It is unlikely that the complex phenomena of promoter actionwill be explicable by any one theory.W. W. Hurst and E. K.Rideal 63 ascribe the promoting action of palladium on copper tointerface effects, the molecules a t the boundary between two solidphases being in a specially active condition. The promoting actionof irreducible oxides on nickel in the reaction GO2 + 4H2 = CH, +61 See H. S . Taylor, Report on Contact Catalysis, ref. 43; M. C. Boswelland C. H. Bayley, J . Physical Chem., 1925, 29, 11; A., ii, 215.62 A. W. Gauger, J . Amer. Chem. SOC., 1925, 47, 2278; A., ii, 1072; H.Adkins and W. A. Lazier, ibid., 1924, 46, 2291 ; A., 1924, i, 1278; C. Kelber,Ber., 1924, 57, [B], 136, 142; A., 1924, ii, 243, 244.J., 1925, 129, 1317.J ., 1924, 125, 685, 69422 ANNUAL REPORTS ON THE PROGRESS OF CHEMISTRY.2H,O is attributed by S. Medsforth 64 to dehydration or decom-position of intermediate compounds, and to selective adsorpt)ion.E. F. Armstrong and T. P. Hilditch 65 consider the effect mainlydue to an increase of surface. W. W. Russell and H. S. Taylor 66find no proportionate increase of adsorption with catalytic activityof nickel when promoted by thoria, so that the effect is to renderthe surface more active rather than more extensive. In conformitywith Taylor’s theory of the catalytic surface, an irreducible oxideprevents coalescence (sintering) of the nickel atoms, its chief functionbeing thus to produce a larger number of unsaturated atoms ofhigh activity.R. W. G. Wy~koff,~’ by an X-ray examination ofammonia-iron catalysts promoted by potash and alumina, findsthat the latter prevent the growth of the iron crystals.W. A. Bone and G. W. Andrew G8 have investigated the catalyticoxidation of carbon monoxide a t a gold surface a t 300”. Thetheoretical mixture 2CO + 0, reacts a t a rate proportional to itspressure when the catalyst has reached its normal activity bycontinued reaction. This normal activity is strongly stimulatedby previous exposure to either reactant or by reaction in a mixturecontaining excess of either reactant, and greatly reduced by keep-ing a t room temperature or a t 300” in a vacuum. After such alter-ations it reverts to its normal activity on continued reaction in thetheoretical mixture.They consider, therefore, that whilst bothgases are “activated” by association with the surface, suchactivation is by no means strictly confined to the surface layer,but extends to the more deeply “occluded” gases. This is ofconsiderable interest in view of the general conviction that surfacecatalysis is determined by a single adsorbed layer.C. N. Hinshelwood69 and his associates have investigated theinfluence of catalytic surfaces (heatcd wires) on gas reactions whichare bimolecular in the homogeneous phase. If the active surfaceof the catalyst remains saturated with adsorbed molecules, thereaction rate is independent of pressure (zero order), and no con-clusion can be drawn as to the number of molecules involved inthe catalytic reaction.With small adsorption, however, thisnumber is given by the order of reaction measured in the usual way.Intermediate states give illusory results. Zero orders are given by64 J., 1923, 123, 1452.6 5 Proc. Roy. SOC., 1923, A , 103, 586; A., 1923, ii, 551.66 J. Physical Chem., 1925, 29, 1325.6 7 LOC. cit., ref. 45.68 PTOC. Roy. Soc., 1925, A , 109, 459.69 C. N. Hinshelwood and C. R. Prichard, Proc. Roy. SOC., 1925, A , 108,211; A., ii, 567; J., 1925, 127, 327, 1552; C. N. Hinshelwood and R. E.Burk, ibid., pp. 1105, 2896GENERAL AND PHYSICAL CHEMISTRY. 23NH31W and HIlAu ; first orders by N,OIAu, N201Pt, HIIPt, NH31Pt,NH3(Si02. The catalyst promotes rapid first-order reaction, e.g.,of N,O = N, + 0, by “accepting ” oxygen atoms which thenevaporate as molecules.It also accelerates reaction by loweringthe energy of a c t i ~ a t i o n , ~ ~ which (calculated in the usual way) isfrequently about half the bimolecular value. There is no justi-fication, however, for attaching any important significance to theratio 2 : 1. A similar ratio is obtained by H. A. Taylor 71 for thedecomposition of hydrogen iodide a t glass walls and in the homo-geneous phase. Kinetic studies of the reduction of carbon dioxideby hydrogen at the surface of hot platinum and tungsten wires byC. N. Hinshelwood and C. R. Prichard 72 lend strong support tothe theory of H. S. Taylor.The activating effect of a polar surface has been strikinglydemonstrated by R. G. W. Norrish.73 Both dry chlorine andbromine react with ethylene rapidly on glass, still more rapidly onstearic acid, and scarcely at all on paraffin wax.This is regardedas strong evidence that molecular activation depends on inducedpolarity by association with a polar molecule. This reaction has beenemployed as a measure of the polarity of various surfaces.74The Surfaces of Liquids.According to the Langmuir-Harkins theory of the orientedunimolecular layer, which has been so firmly established as regardslong-chain substances with a polar end-group on water, wheresurface concentration and pressure can be directly measured, theexcess surface concentration of a solution is similarly restricted toa unimolecular layer, and the surface tension of the solution is dueonly to the stray fields of force of the molecules in this layer.Thelatter state is obviously less simple than the and recentwork 76 suggests that the phenomena at the surfaces of solutionscannot always be interpreted in quite such a simple manner. I nthe absence of trustworthy direct methods of measuring surfaceconcentration, r, it has generally been calculated from the surfacetension, U, of a solution of concentration C by the Gibbs equation :I’ = - C/BT . d ~ / d C , a form which is valid only for ideal solutions.The area, A , per molecule in the layer, obtained by assuming it tobe unimolecular, is generally in sufficiently good agreement with70 See also C. N. Hinshelwood and B. Topley, J., 1923, 123, 1014.72 J., 1925, 127, 806, 1546.73 J., 1923,123, 3006; 1925, 127, 2318; 1926, 55.74 N.K. Adam, R. S. Morrell, and R. G. W. Norrish, J . , 1925, 127, 2793.75 F. G. Donnan, Brit. Assoc. Rep., 1923, 59.‘13 See, e.g., 5. Sugden, J., 1924, 125, 1167; Ann. Report, 1924, p. 8.J. Physical Chem., 1924, 28, 984; A., 1924, ii, 74524 ANNUAL REPORTS ON THE PROGRESS OF CHEMISTRY.expected values to confirm the theory, but conclusions as to thestate of the surface molecules based on these values of A are notalways so certain as in the case of insoluble films. Recent measure-ments indicate that the acetic acid molecule occupies 28% morearea at an air-water than a t a hydrocarbon-water interface,77 andthat whilst pyrogallol (air-water) occupies only 16% more area thanphenol, which stands vertical, the o- and m-dihydroxy-derivativesmust be considerably inclined to, and the p-derivative flat on thesurface, giving incompressible although not close-packed films.78Ih recent work the importance of employing the Gibbs equationin its exact form has been emphasised, and activities have beenused instead of concentrations. Except for very dilute solutions ofnon-electrolytes, the concentration formula gives quite erroneousresults.A. K. Goard and E. K. Rideal 79 find that with increasingactivity of phenol in aqueous salt solution (due to increase of saltas well as phenol concentration) the adsorption increases to a maxi-mum, leading to values of A and of surface thickness in excellentagreement with accepted values of the dimensions of the benzenenucleus.The adsorbed phenol is thus contained in a single layerof close-packed, vertical molecules. Although this film is probablythe chief factor in determining the surface tension, the latter isdefinitely influenced through the film by molecules below it, indi-cating (contrary to the Langmuir-Harkins theory) the operationof forces exceeding molecular dimensions.and others points in the samepdirection, while E. Edser 81 considersthat attractive forces may be appreciable over a range of manymolecular diameters from the surface. The foundation of thefilm as well as the film itself cannot be neglected in a completetheory of surface tension.Salt solutions having a higher surface tension than water arecovered, according to Langmuir, with a single layer of orientedwater molecules.According to recent measurements 82 on suchsolutions, based on the Gibbs equation (activities), the apparentthickness of this layer decreases considerably with increasing con-centration, the effect depending on the nature of the salt. This isdifficult to reconcile with the single layer theory. Various sug-gestions are made, e.g., that the increasing diffusion pressure of1610; A., ii, 771.The work of T. Iredale7 7 W. D. Harkins and H. M. McLaughlin, J. Amer. Chem. SOC., 1925, 47,7 8 W. D. Harkins and E. H. Grafton, &id., p. 1329; A., ii, 658.7* J., 1925, 12'7, 1668.8o Phil. Mag., 1923, [vi], 45, 108s; 1924, 48, 177; 1925, 49, 603; A . , 1923,81 Brit. Assoc. Fourth Report on Colloid Chemistry, 1922, p.40.82 A. K. Goard and E. K. Rideal, J., 1925,127, 1668; IV. D. Harkins andH. M. Me Laughlin, J. Amer. Chem. SOC., 1925, 47, 2083; A , , ii, 959.ii, 379; 1924, ii, 663; 1925, ii, 508GENERAL AND PHYSICAL CHEM1S”RY. 25the ions forces them nearer the surface and thus perhaps modifiesthe orientation of the water molecules, and that the water shellof the ions becomes smaller or more tightly packed with increasingconcentration.The equation of state proposed by Langmuir g3 (following J.Traube) g4 for the “Gibbs layer” is FA = RT, where F , thesurface pressure, is the excess surface tension of the solvent overthat of the solution, and A is the area occupied by a gram-moleculeas surface excess. This two-dimensional analogue of the ideal gaslaw, whilst approximately valid for low values, fails at high valuesof F , suggesting a behaviour similar to that of compressed gases.R.K. Schofield and E. K. Rideal,s5 using the Gibbs formula(activities), find that for aqueous ethyl alcohol, with increasingconcentration, A first diminishes to a minimum, which is onlyslightly larger than the value corresponding to a close-packed film,then rises to a steady value three times the minimum for somereason not fully explained. Pyridine at a water-mercury inter-face shows similar behaviour. When, for aqueous solutions offatty acids, C, to C,,, a t benzene-water as well as air-water inter-faces, FAIRT is plotted against F , curves strongly resemblingthe corresponding ones for highly compressed gases are obtained,leading to P(A - B) = xRT as the equation of state of the surfacelayer for high values of F.B (compare b of the gas equation) isthe minimum area of an adsorbed gram-molecule under high surfacecompression, and 1 /x represents the lateral molecular cohesion,which increases with length of the carbon chain. This cohesion isgreatly reduced when the chains are in a hydrocarbon phase. Forsucrose at a water-mercury interface, x = 1 (no cohesion), givingthe exact analogue of the Sackur osmotic pressure equationP(V - b ) = RT, and the value of B, which is in striking agree-ment with the dimensions of the molecule from other sources,indicates that its long axis lies in the plane of the interface. Thegeneral conclusion is reached that the molecules adsorbed in aunimolecular layer from a weak solution affect the surface tensiononly by their thermal agitation. “At a given temperature theireffectiveness depends solely on their surface concentration, inter-facial areas, and lateral cohesion.” M.Volmer and P. Mahn&rt,86from direct measurements of the amount of benzophenone adsorbedfrom a crystal on the surface of mercury, obtain the same equationof state. It has also been deduced theoretically by S. C. Kar.8’83 J. Amer. Chem. Soc., 1917, 39, 1848; A., 1917, ii, 525.84 Annalen, 1891, 265, 27; A . , 1891, 1468.8 5 Proc. Roy. Soc., 1925, A , 109, 57; A., ii, 960; Nnlure, 1925, 116, 8%.a6 2. physikal. Chem., 1925, 115, 239; A., ii, 508.Physikal. Z., 1925, 26, 615; A., ii, 104526 ANNUAL REPORTS ON THE PROGRESS OF CHEMISTRY.An equation of this kind had previously been shown by N.K.Adam *8 to represent the behaviour of expanded films of insolublefatty acids.Organic vapours are reversibly adsorbed on mercury surfaces,giving, according to T. Iredale,8g unimolecular films. The con-siderable drop of surface tension caused by condensation from anearly saturated vapour indicates that the interfacial tensionbetween two liquids is not due exclusively t o the arrangement of asingle layer of molecules, but results from attractions extendingthrough a layer many molecules thick. Iredale's mode of usingthe drop-weight method has been adopted by othersg0 as beingmore rational than that of J. L. R. Morgan and W.D. Harkins.The continued study of long-chain surface films by N. K. Adam 91has shown that methyl esters of the alcohols do not form stablefilms, the methyl radical destroying the anchorage of the polargroup on water, and that with sufficient complexity of the mole-cular heads (e.g., the substituted ureas), a " two-dimensionalallotropy " occurs, depending on different types of packing in thefilm, and associated with a definite transition temperature. Penta-erythritol tetrapalmitate exists as a stable film with its four longchains parallel and perpendicular to the surface.A. P. Cary and E. K. Rideal,92 by placing crystals or lenses oflong-chain fatty acids and esters (instead of the usual solutionin a volatile solvent) on the surface of water, find that surfacesolution occurs, not by bulk spreading, but from the edge of contactof the lens or crystal,93 and a t a measurable rate until kineticequilibrium is reached between the substance and the film.Theprocess is reversible and occurs in two stages, first the productionof an expanded film under zero compression, F =crwater-um ;secondly, its condensation to an equilibrium pressure, F,, character-istic of the substance. The system is then the two-dimensionalanalogue of a saturated solution in equilibrium with the solidsolute. Contrary to the conclusions of A. Mar~elin,~~ there is a8 8 Proc. Roy. Soc., 1922, A , 101, 516; A., 1922, ii, 687; Ann. Report, 1923,I>. 22.Ref. 80.00 See A. K. Goard and E. K. Rideal, J., 1925, 12'9, 780.91 Summarising paper : J .Physical Chem., 1925, 29, 87; A., ii, 195;N. K. Adam and J. W. W. Dyer, Proc. Roy. SOC., 1924, A , 106,694; A., 1925,ii, 32.92 PTOC. Roy. SOC., 1925, A , 109, 301, 318, 331; A., ii, 1046-1048; Nature,1925, 115, 457; A., ii, 388.93 M. Volmer and P. Mahnert, Zoc. cit., find that benzophenone spreadsfrom a crystal placed on mercury directly, and not by vaporisation or bulksolution.94 Compt. rend., 1921, 173, 38; A., 1921, ii, 488GENERAL AND PHYSICAL CHEMISTRY. 27limit to the expansibility of the film due to the mutual attractionof the carbon chains. With increasing temperature dFe/dT has aconstant positive value through a considerable range, including thatover which the film changes from the condensed to the expandedstate. At the melting point of the crystal it changes sharply,becoming in some cases zero, in others negative, and finally aftera transition point less negative.The curves for different sub-stances are straight lines, generally parallel. The breaks areattributed to expansion of the lens-water interfacial film. Filmexpansion is regarded as due to hydration of the polar heads,this effect being opposed by attraction of the hydrocarbonchains for each other. An application of the Clapeyron equationto the two-dimensional system gives the latent heats of the changes,including approximately correct values of the latent heats of fusionof the crystals.The question of the rate of evaporation of water through a filmof insoluble fatty acid has been discussed. The conclusion ofG.Hedestrandg5 that such a film has no retarding effect has beenrefuted by N. K. Adamg6 and by E. K. Rideal,97 who has shownexperimentally that considerable retardation occurs, the effectbeing increased by an increase in surface concentration.Sir William Hardy,98 in a lecture to the Chemical Society, hasdiscussed the problems of interfaces. His well-known work onlubrication leads to the view that the surface forces responsiblefor the primary unirnolecular layer of oriented molecules on asolid, although acting directly over ranges comparable with moleculardimensions, are nevertheless by some means transmitted over muchgreater distances through secondary films, which, possibly onaccount of induced polarity, themselves are structured to a graduallydecreasing extent owing to the operation of thermal agitation.The essential differences between primary and secondary layers,and the factors governing their formation have been summarisedby Cary and Rideal.ggStrong Electrolgtes.Previous Reports have indicated that in recent yews the attemptto interpret the properties of solutions of strong electrolytes interms of a degree of dissociation measured by the conductivityratio has been abandoned.Instead much effort is being devotedto the exact determination of the thermodynamical properties of5 1 ~ J . Physz'cccl Chenz., 1924, 28, 1245; A., 1925, ii, 102.9 G Ibid., 1925, 29, 610; A., ii, 658.9g J., 1925, 127, 1207; Brit. Assoc. Fourth Report on Colloid Chemistry,9 7 Ibid., 1925, 29, 1585.1922, 18.5.LOC.cit., ref. 92, p. 30128 ANNUAL REPORTS ON THE PROGRESS OF CHEMISTRY.solutions, and for their expression the activity concept of G. N.LewisIt may perhaps be emphasised here that activities are not, asthey have been often regarded, empirical quantities which mustbe inserted into the theoretical equations in place of concentrations,in order to produce the observed results. They are in fact exactexpressions of the partial free energies of substances in solution. If AFis the free energy change (per mol.) in the transfer of a substance froma solution in which its concentration is c to a second solution in whichits concentration is co, then the ratio of the corresponding activitiesis given by AF = RT log, a/ao.If the solution were “ ideal,” thefree energy change in the transfer would be AF = RTlog, c/co.Now if the second solution be that which has been chosen as standard(usually an infinitely dilute solution), we may put a. = co, sinceonly the ratio of the activities has been defined. It is now evidentthat a/c, which is known as the activity coefficient f , is a measureof the deviation from the ideal relation.A number of important measurements of activities have appearedduring the year, among which the following may be mentioned :sodium hydroxide in aqueous solution,2 sodium hydroxide insodium chloride sol~tions,~ potassium hydroxide in potassiumchloride solution^,^ water in sodium chloride and potassium chloridesolution^,^ sulphuric acid in aqueous sulphate solutions,6 hydrogenchloride in ethyl-alcoholic s~lution,~ hydrogen chloride in methyl-alcoholic solution,s calcium, strontium, and barium chlorides inaqueous sol~tion,~ barium chloride in aqueous solution, lo hydrogenfluoride in aqueous solution.ll G.Scatchard has redetermined theactivities of hydrogen chloride in aqueous solution l2 and hasrecalculated the activities of potassium, sodium and lithium chlor-1 G. N. Lewis and M. Randall, “Thermodynamics and the Free Energyof Chemical Substances,” McGraw, Hill Book Co., 1923. Compare also H. S.Harned in “ A Treatise on Physical Chemistry,” edited by H. S. Taylor(Macmillan, 1924), p. 701.2 H. S. Harned, J . Amer. Chem. SOC., 1925, 47, 676; A., ii, 397.3 Idem, ibid., p.684; A., ii, 398.4 Idem, ibid., p. 689; A., ii, 398.6 Idem, ibid., p. 930; A., ii, 538.6 H. S. Harned and R. D. Sturgis, ibid., p. 945; A., ii, 538.7 H. S. Harned and M. H. Fleysher, ibid., p. 82; A., ii, 538.has been almost universally adopted.Many of theresults in refs. 2 to 7 are collected in a summarising paper by H. S. Harned,2. physikal. Chem., 1925, 11’7, 1 ; A., ii, 977.8 G. Nonhebel and H. Hartley, Phil. Mag., 1925, [vi], 50, 729; A., ii, 1061.9 W. W. Lucasse, J . Amer. Chem. SOC., 1925, 47, 743; A., ii, 399.10 J. N. Pearce and R. W. Gelbach, J . Physical Chem., 1925, 29, 1023;11 J. D. C. Anthony and L. J. Hudleston, J., 1925, 127, 1122.12 J, Amer. Chem. SOC., 1925, 47, 641; A., ii, 397.A., ii, 867GENERAL AND PHYSICAL CHEMISTRY.29ides and potassium hydroxide by the use of a form of the Debyeequation for extrapolation to zero concentration. l3 The activitiesand activity coefficients of a number of salts in aqueous solution(and in some salt solutions) are thus known with considerableaccuracy. The activity coefficients of a number of salts are plottedin Fig. 2 against their concentrations.It will be observed that with increasing concentration the activitycoefficient first diminishes, reaches a minimum, and rises again,FIG. 2.-J1 2Square root of concentration.becoming greater than unity at high concentrations. This behav-iour appears to be general.The first attempt to obtain a general formula for the change ofactivity coefficient with concentration was made by G.N. Lewisand G. A. Linhart,14 who showed that in very dilute solution theobserved values could be represented by an equation of the typelogf = - Pea', where p and a’ are constants. G. N. Lewis andM. Randall l5 suggested the empirical rule that for uni-univalentelectrolytes a’ = 1/2 and J. N. Bronsted 16 employed a similarl3 G. Scatcherd, J. Arner. Chem. Soc., 1925, 47, 648; A., ii, 397.l4 Ibid., 1919, 41, 1951; A., 1920, ii, 97.l6 “ Thermodynamics,” p. 345.l6 J . Amer. Chem. Soc., 1922, 44, 938; A., 1922, ii, 48230 ANNUAL REPORTS ON THE PROGRESS OF CHEMISTRY.relation, while B. A. M. Cavanagh l7 showed that an equation ofthis form followed from S. R. Milner’s theory of electrolytes. Whilstsuch an equation covers the diminution of the activity coefficienta t small concentrations, it is necessary to introduce another termto include the subsequent rise.Thus H. S. Harned18 has giventhe empirical equation logf = - pea' + ccc, where a is a thirdconstant, and has shown that values of the constants can be foundwhich satisfactorily reproduce the observed values over a widerange of concentration. J. N. Bronsted employed the similarequation logf = - 0 . 4 2 ~ ~ ’ ~ + ac for uni-univalent electrolytes.19The activity coefficient of an electrolyte at fixed concentrationin the presence of increasing concentrations of other salts followsa similar course, but the magnitude of the effect varies considerablyaccording to the valence type of the salts. In order to eliminatedifferences of the electric forces in equal concentrations of ions ofdifferent valencies, Lewis and Randall 2o introduced a quantitycalled the ionic strength (p), obtained by multiplying the con-centration of each ion by the square of its valency and dividingthe sum of the products by two.They were then able to statethe general principle that in dilute solutions “ the activity coefficientof a given strong electrolyte is the same in all solutions of the sameionic strength.” In stronger solutions, deviations appear owingto the individual behaviour of ions.Turning now to the attempts to explain the behaviour of strongelectrolytes in terms of interionic electric forces, the Debye-Huckeltheory 2 l has been greatly extended.22a P. Debye has given a moredirect derivation of his fundamental equation of the relationbetween activity coefficient and concentration.22b The first step1 7 Phil.Mag., 1922, [vi], 44, 226, 610.18 J . Amer. Chem. SOC., 1920, 42, 1805; 1922, 44, 252; A., 1920, ii, 664;l9 LOC. cit., ref. 16.20 J . Amer. Chem. Soc., 1921, 43, 1112; A., 1921, ii, 427.2 1 Ann. Report, 1924, p. 25.22 Bibliography below :-1922, ii, 255.(a) I?. Debye and E. Huckel, “ On the Theory of Electrolytes.”I. The freezing point lowering and reIated phenomena, Physikal. Z.,1923, 24, 185; A., 1923, ii, 459. 11. The limiting law of electrical con-ductivity, ibid., 1923, 24, 305; A., 1923, ii, 724. (b) P. Debye, “ TheOsmotic Equation of State and Activity of Dilute Strong Electrolytes,”ibid., 1924, 25, 97; A., 1924, ii, 386.(c) 0. Scharer, “The Theory ofSolubility Influences in Strong Electrolytes,” ibid., 1924, 25, 145; A.,1924, ii, 455. (d) I?. Gross and 0. Halpern, “ On the Dilution Law andDistribution of Strong Electrolytes,” ibid., 1924, 25, 393; A., 1925,ii, 117. (e) A. Frivold, “Contribution to Knowledge of So-calledAnomalous Properties of Strong Electrolytes,” ibid., 1924, 25, 465 ; A.,1925, ii, 396. df) P. Debye and J. McAulny, “The Electric Field oGENERAL AND PHYSICAL CHEMISTRY. 31is the calculation of the electrical potential of an ion in the solutiondue to all the other ions about it, which is effected by the combineduse of Boltzmann’s equation and Coulomb’s law. To a firstapproximation (neglecting the dimensions of the ion itself) it hasthe value I) = - eF,/D, where ei is the charge on the ion, D thedielectric constant of the medium, and K a quantity characteristic4x 8xe2 Nof the solution which is given by K~ = = B ~ P , DETin which ni is the number of ions of the ith kind per C.C.in the solu-tion, E the charge on a univalent ion, N the Avogadro number,E the gas constant per molecule, and p the ionic strength. Insteadof calculating the total electrical energy as before,22a Debye nowdeduces the work, W , which must be done in giving the ions theircharge, assuming them to be initially uncharged; W = X~niei$i/3for all the ions in one C.C.The free energy of the system is greater by this amount than ifthere were no electric forces between the ions, i.e., CAP =ENiET log ci/ciO + W ; * actually ZAF = ZN&T logfici/c,O, whenceit is only a matter of mathematics to deduce the result,for a salt giving vi ions of valency 2%.For an electrolyte giving ionsof valencies xlzz this reduces to loglOf = zlz,BY/Land at 0” B = 0.5.It must be clearly understood that these equations represent* For brevity, summation is taken to include the solvent and all the kindsof ions present.~~ ~~~~~~~Ions and Neutral-salt Action,” ibid., 1925, 26, 22; A., ii, 171. ( 9 ) E.Huckel, “ On the Theory of Concentrated Aqueous Solutions of StrongElectrolytes,” ibid., 1925, 26, 93; A., ii, 513. (h) 0. Redlich, “ On theTheory of Electrolytic Conductivity,” ibid., 1925, 26, 199; A., ii, 541.( i ) P. Gross and 0.Halpern, “ On the Temperature-dependent Para-meter in Statistics and the Debye Theory of Electrolytes, ibid., 1925,26, 403; A., ii, 566. ( j ) A. A. Noyes, “ Interionic Attraction Theoryof Ionised Solutes.” I. “ Critical Presentation of the Theory,” J . Amer.Chem. SOC., 1924, 46, 1080; A., 1924, ii, 658. 11. “ Testing the Theorywith Experimental Data,” ibid., p. 1098; A., 1924, ii, 659. (k) A. A.Noyes and W. P. Baxter, 111. “ Testing the Theory in Alcohol Solvents,”ibid., 1925, 47,2122; A., ii, 970. ( I ) P. Debye and L. Pauling, IV. “ TheInfluence of Variation of Dielectric Constant on the Limiting Law,” ibid.,p. 2129; A., ii, 970. (m) P. Gross and 0. Halpern, “Electrolytes inSolutions of Low Dielectric Constant,” Physikal. Z . , 1925, 26, 636; A.,ii, 1152.(n.) Compare also J. J. van Laar, ‘‘ On the Theory of StrongElectrolytes and its History,” 2. anorg. Chem., 1924,139, 10832 ANNUAL REPORTS ON THE PROGRESS OF CHEMISTRY.the “ limiting law ” a t very small concentrations; in most casesconsiderable deviations occur below 0.W. A. A. Noyes z z j dis-cussed a large body of data and came to the conclusion that althoughdeviations begin a t quite small concentrations, the evidence sup-ported the truth of the equation in limiting cases. J. N. Bronstedand V. K. LaMer 23 have since determined the activity coefficientsof a number of complex cobaltammines of various valence typesin dilute salt solutions up to 0*02N, and find that they are com-pletely in accordance with the Debye equations.F. Hovorka andW. H. Rodebush 24 have determined the freezing points of solutionsof seven electrolytes between 0.01 and 0*001M, ‘‘ which show ratherremarkable agreement with the values derived from the formulaof Debye and Huckel.”On the other hand, M. Randall and A. P. Vanselow 25 find thatthe freezing points of solutions of hydrogen chloride, thallouschloride, and lead nitrate are not in good agreement. G. Scatchard,26however, claims that they are in agreement with the modifiedequation discussed below.The first step in the extension of the theory to more concentratedsolutions is to take account of the size of the ions. If the ion hasfinite size, the expression for the potential becomesz.$ K+i = - -- - D 1 +- Kas’where ai is the mean least distance of approach of the ccntre ofany ion to that of the ion in question.The activity coefficientequations becomeandassuming that the mean ionic radius has a mean value, a, for allions.It has been shown by Scatchard26 that this is equivalent to theempirical equation logf = - pc1I2 + CLC, and p = 0.5 for uni-univalentsalts. 0. Scharer has applied equation (4) to the solubilities of saltsin salt solutions.22c The ionic radius term a is unknown and is ailadjustable variable in the equation. He finds the values whichreproduce the experimental data best by a graphical method, andthus obtains equations which completely represent the solubilities23 J . Amer. Chem. SOC., 1924, 46, 555; A., 1924, ii, 30624 Ibid., 1925, 47, 1614; A., ii, 772.z 5 Ibid., 1924, 443, 2418; A , , 1925, ii, 33. 26 LOC.cit., ref. 12GENERAL AND PHYSICAL CHEMISTRY. 33of calcium sulphate, silver sulphate, thallous chloride, and variouscobaltammine salts in salt solutions, with and without a common ion.A. A. Noyes 27 pointed out that there is an irreconcilable differencebetween the results of the Debye and the Milner 28 calculations ofthe inter-ionic energy. The two theories have been further criticallycompared by S. R. Pike and G. N ~ n h e b e l , ~ ~ who come to the con-clusion that neither can do more than indicate the form of theequation. G. Nonhebel and H. Hartleym have examined theactivity coefficients of hydrogen chloride in methyl alcohol, ethylalcohol, and water with the same object.Whereas the Debyecalculation leads to the expression - logf = pdc- CYC, theMilner equation becomes - logf = I.OlSpf(c)d< where f(c) is atabulated function of the concentration. These authors concludethat the Milner equation without any adjustable constant cc repre-sents the data better than the Debye formula with one, and findthat the activity coefficients are well represented by empiricalequations of the form - logf = p’dc and that the values of P’/aare in good agreement with the Milner function f(c) at 000005n.Further work on this discrepancy will be welcome.R. H. Fowler 31 has examined the range of validity of the com-bined use of Boltzmann’s and Poisson’s equations with particularreference to the Debye-Hiickel theory and concludes that thevalidity of this theory cannot be regarded as established except forsmall values of the ionic concentrations.The Debye equation has been tested on another point (in whichit agrees with the Milner equation).The fundamental equationindicates that at the same concentration in different solventslog f K 1 /D3I2. This requirement has been confirmed by A. A. Noyesand W. P. Baxter,32 using the data for lithium chloride, hydrogencKloride, and sodium ethoxide in ethyl alcohol, and by G. Nonhebeland H. H a r t l e ~ . ~ ~It will be observed that equations (1)-(4), in which logf is neces-sarily negative, cannot account for the rise of the activity coefficientabove unity in concentrated solutions. I n order to explain this,Hucke122g has considered the change of dielectric constant of thesolution with concentration.It is assumed that the addition of2 7 LOC. cit., ref. 22, j , I .28 Phil. Mag., 1912, [vi], 23, 551; 1913, 25, 742; 1918, 35, 214, 362; A . ,1913,ii, 481; 1918, ii, 54, 148; Trans. Faraday Soc., 1919,15, 148; A . , 1920,ii, 152.2s Phil. Mag., 1925, [vi], 50, 723 ; A , , ii, 1061.31 Proc. Camb. Phil. SOC., 1925, 22, 861.32 LOC. cit., ref. 22, k.33 LOG. cit., ref. 8.47, 2098 ; A., ii, 971.30 LOC. cit., ref. 8.Compare also G. Scatchard, J . Amer. Chein. Soc., 1925,REP.-VOL. XXII. 34 ANNUAL REPORTS ON THE PROGRESS OF CHEMISTRY.salts causes a lowering of the D.C. by an amount proportional tothe concentration, as expressed by the formula D = Do - XSc,where 6 is the lowering produced by unit concentration of oneparticular kind of ion.A change of D.C. involves, in addition toits effect on the Coulomb law expression, a change in the electricalenergy of the ions themselves in the medium. If it be supposedthat the charge ed is located on a sphere of radius bi, the potentialat the surface is ei/Dbi and the corresponding electrical energyei2/2Dbi. I n bringing the ions from an infinitely dilute solution(D.C. = Do) to a solution of D.C. = D, the electrical energy willincrease byand this term must be added to the Coulomb expression in orderto obtain the complete electrical work term. In order to simplifythe calculation, Hiickel puts ai = bi and uses a mean value, a, forall the ions present, thus obtaining the result :wheref(K) is a complicated function of K which is found on cal-culation to be nearly proportional to the concentration.Hencefor a uni-univalent saltwhere A and B have the same values as before and C is a thirdconstant. Values of C corresponding to various values of 6 arecalculated and that value which fits the observed data best isselected. The activity coefficients of lithium, sodium, and potass-ium chlorides over the whole range of concentration are givenby the equation, assuming the values of the D.C. lowering coefficient,6, to be 20, 9, and 6, respectively. Equally good agreement withthe observed results for hydrogen chloride was obtained both onthe assumption that the hydrogen ion is free (H+) and that it existsas the hydrate (H30+), different values of the constants beingrequired for the two cases.It would appear that a considerable step forward in our knowledgeof the factors determining the behaviour of salt solutions has beenmade.It must be remembered, however, that in the final equationthere are two variable factors which can be selected to fit theobserved results. The outcome is not, therefore, a calculation ofactivities from fundamental data, and until the constants havebeen obtained independently the theory must be regarded as GENERAL AND PHYSICAL CHEMISTRY. 35derivation of the form of the empirical equation, taking accountof inter-ionic forces and the dielectric constant.The somewhat artificial character of the calculation is realisedby Debye and Hiickel, and the latter has given an exhaustive dis-cussion of the factors concerned in concentrated aqueous solutions.34The high dielectric constant of water means high polarisability inan electric field and the presence of molecules which act as electric“ dipoles.” I n the intense electric fields in the vicinity of ions thedipoles must be oriented and owing to the intensity of the forces 35it is probable that ‘‘ electric saturation,” i.e., total possible orien-tation, occurs. Each ion is therefore surrounded by an atmosphereof water “ bound ” by electrostatic forces.An applied electricfield must have a smaller effect on solvent molecules in the vicinityof ions than on the unoriented solvent, hence the dielectric constantis lowered by the presence of ions.This view was originally putforward by K. Fajans,36 and M. Born 37 has shown that the hydrationenergy of ions can be accounted for on the same basis.The salting-out effect of non-electrolytes by salts has beendiscussed by P. Debye and J. McAulay 225 from the same point ofview. The effect of introducing a strong electrolyte into a solutionof a non-electrolyte, according to these authors, is to cause themore polarisable components to amass themselves round the ions,where the electric field is great. If the solvent has the greaterpolarisation capacity, the effect is a displacement of the non-electrolyte from the vicinity of the ion (ie., an apparent diminutionof the solution volume) and a rise in its activity. If the non-electrolyte is more polarisable than the solvent and raises its D.C.,the effect is the reverse.Calculations of the increase of the activityof non-electrolytes by salts on these lines agree with the magnitudeof the effect.38 The importance of these theories on the questionof hydration in solution is obvious, but it is felt that the time isscarcely ripe for a general discussion.P. Gross and 0. Halpern have further considered22m the behaviourof electrolytes in solvents of low dielectric constant. They findthat the effect of the ionic charges on the osmotic properties maybe not only small, but also in the reverse direction of the Coulombeffect described above. The solute will thus appear to be associ-34 LOG. cit., ref. 22, g.35 At 3 x 10-8 cm., Hiickel estimates the potential gradient at 2 X lo636 Ber.Deut. Physikal. Ges., 1919, 21, 549, 709; A., 1920, ii, 12, 154.volts per cm.. 3 7 2. Physilc, 1920, 1, 45; A., 1921, ii, 166. Compare also 0. Bliih,ibid., 1924, 25, 220; A., 1924, ii, 824; Naturwiss., 1921, p. 732.38 Compare also E. A. Hafner and L. von Kiirthy, Arch. exp. Path. €’harm.,1924, 104, 148; A., 1925, ii, 283.B 36 ANNUAL REPORTS ON THE PROGRESS OF CHEMISTRY.ated. They are able to account qualitatively for the various typesof anomalous behaviour observed in non-aqueous solutions.The determination of the dielectric constants of solutions ofelectrolytes has thus become a matter of considerable importance.Until recently no direct means of measurement was available, butthe difficulty has been overcome by the use of electric oscillationsof short ~ a v e - l e n g t h .~ ~ The lowering of the D.C. of water by saltsis confirmed. I n the case of electrolytes such as tetrapropyl-ammonium iodide in alcoholic solvents, the initial decrease isfollowed by a rise.Conductivity of Strong Electrolytes.On the theory of complete dissociation the change of the equivalentconductivity of a strong electrolyte with dilution is ascribed tochanges in the inter-ionic forces. The theory of J. C. Ghosh did notsurvive P. Debye and E. Hiickel41 have treated theconductivity problem on the same basis as their theory of osmoticproperties. It was there postulated that every ion is immediatelysurrounded by an excess of ions of opposite charge.When theion is in motion through the solution, a finite time is required forthe redistribution of ions in this fashion (period of relaxation)and there will always be an excess of ions of opposite sign in itsrear, hence it will be subject to a retarding force. Further, sinceions of opposite sign are moving in different directions and dragwith them a certain amount of solvent, the viscous resistance t othe motion of an ion will be greater than if the other ions were atrest. Assuming Stokes’s law, Debye and Hiickel have computedthese effects and deduced the relation for dilute solutions :A0 - + K,b)d%,where A, and A, are the equivalent conductivities at infinite dilutionand at concentration c, wl = 1/2(1a/lc + &/la), la and Zc being themobilities of anion and kation, b the mean ionic radius, Kl and Kzconstants depending on the temperature and D.C.of the medium.The first term in the bracket represents the period of relaxationeffect, the second the viscosity effect. This equation agrees withthe empirical relation found by Kohlrausch 42 for the conductivityof dilute salt solutions, A. - A, = x l / c : Using a modified form of39 R. T. Lattey, Phil. Mag., 1921, [vi], 41, 829; A., 1921, ii, 426; K.Theodortschick, Physikal. Z., 1922, 23, 344; P. Walden, H. Ulich, and 0.Werner, 2. physikal. Chew., 1924, 110, 43; 1925, 115, 117; 1925, 116, 261;A., ii, 512, 773.40 Ann. Report, 1918, 11; 1919, 14.4l LOG. cit., ref. 22, a, 11.42 2. Elektrochem., 1907, 13, 333; A,, 1907, ii, 600GENERAL AND PHYSICAL CHEMISTRY.37Stokes’s law, 0. Redlich has obtained a modified equation whichgives exact agreement with the experimental results for aqueoussolutions. J. E. Frazer and H. Hartleya have determined theconductivities of fifteen univalent salts in methyl alcohol and findthat although they agree with the form of the equation, they arenot consistent with the value of w given above, indicating thatsome revision of the theory is necessary in this respect. C. W.Davies 45 also has given an empirical equation which is in goodagreement with the data of aqueous solutions, viz. : A, - & =Kdz(dc + dz). According to Frazer and Hartley, this is notapplicable to methyl-alcoholic solutions. A. Ferguson and I. Voge146have used an empirical equation of the form : A, - A, = Ben.Activities in Binary Liquid Mixtures and Deviations fromRaoult’s Law.Activities in binary liquid mixtures of non-electrolytes have alsoreceived much attention in recent years.The most general expres-sion of Raoult’s law of perfect solutions is that the activity of acomponent (taken as unity for the pure liquid) is equal to the mo1.-fraction. The deviations from Raoult’s law which occur in manysystems have been widely discussed in the past and have beenattributed exclusively to chemical and to physical effects. Theattempt to explain all deviations chemically (e.g., by solvation andassociation) failed, although they certainly occur in some systems.Attention has therefore been focussed on purely physical effects.A comprehensive theory of liquid mixtures has been developed byJ. H.Hildebrand and collab0rators.~7 In a perfect solution amolecule of any,component behaves exactly as if in its own pureliquid, and any deviation from ideality must be due to the differencebetween the forces acting on a given molecule in the solution andwhen it is surrounded solely by molecules of its own kind (Le., inthe pure liquid component). Deviations will obviously be leastwith liquids of very similar nature in which the intermolecular forcesare nearly identical. As a measure of the forces between moleculesin different liquids Hildebrand has taken the internal pressure.He has obtained concordant values of this quantity from a varietyof properties, the constants of van der Waals’s equation, surface43 LOG.cit., ref. 22, h.44 Proc. Roy. SOC., 1925, A , 109, 351; A., ii, 1163.45 J . Physical Chem., 1925, 29, 473, 973, 977; A., ii, 541, 871.46 Phil. Mag., 1925, [vi], 50, 971; A., ii, 1163.4 7 J . Amer. Chem. SOC., 1916, 38, 1452; 1017, 39, 2297; 1919, 41, 1067;1920, 42, 2180, 2213; 1921, 43, 500, 2172; 1923, 45, 682, 2828; A., 1916,ii, 518; 1918, ii, 36, 65; 1919, ii, 392; 1921, ii, 23, 24, 307; 1922, ii, 141;1923, ii, 315; 1924, ii, 94; J. H. Hildebrand, “Solubility” (Chem. Cat. Coy).,192438 ANNUAL REPORTS ON THE PROGRESS OF CHEMISTRY.tension, heat of vaporisation, and from solubility data, and hasshown that deviations from Raoult’s law are qualitatively pro-portional to the internal pressure difference of the components.This theory has been extended to solubility relations by theconsideration of the solubility curve up to the melting point of thesolute, at which the solubility expressed as mo1.-fraction in theliquid phase is unity. The “ ideal ” solubility curve for solutionswhich obey Raoult’s law is given by the relation : d log N/d( 1 /T) =- Lf/4-58, where N is the mo1.-fraction of solute, and Lf the latentheat of fusion of solute.Hence, if log N is plotted against 1/T, astraight line is obtained for an ideal solution having the slope- Lf14.68. When the solubility curves of a given solute in anumber of solvents are plotted in this way a family of curves isobtained deviating more or less from the ideal line.Such sets ofcurves have been obtained for na~hthalene,~8 iodine,4* s~lphur,~gand a number of organic compounds.50 The ratio of the actualslope of the solubility curve t o the ideal slope (a factor f ) has beenshown by F. S. Mortimer 50 to be related to the internal pressuresof the solvent and solute by the equation f = Kl - Kz + 1, whereKl and Kz are the relative internal pressures, referred to naphtha-lene as unity. He has described a chart by which the properfactor can be read off for any pair of substances the relative internalpressures of which are known. It is thus possible to predict themutual solubilities of many non-polar or slightly polar substanceswith some precision.Whilst Hildebrand’s theory gives a qualitative indication ofdeviations from Raoult’s law over a wide field, it has not beendeveloped quantitatively.J. A. V. Butler 51 considers that thepartial heat of solution is a better measure of the difference betweenthe forces acting on the molecules of a component in solution andin their pure liquids, and shows that in the case of thallium amal-gams,52 the logarithm of the activity coefficient, which representsthe deviation from Raoult’s law, is almost exactly proportional tothe partial heat of solution. Close proportionality is also shownby the less accurate data of a number of binary alloys.534 s J. H. Hildebrand and C. A. Jenks, J . Amer. Chenz. XOC., 1920, 42, 2180;49 Idem, ibid., 1921, 43, 2172; A., 1922, ii, 141.60 F. S . Mortimer, ibid., 1922, 44, 1416; 1923, 45, 633; A ., 1922, ii, 621;61 Ibid., 1925, 47, 117; A., ii, 539.62 T. W. Richards and F. Daniels, ibid., 1919, 41, 1732 ; G. N. Lewis and53 N. W. Taylor, ibid., 1923, 45, 2865; A., 1924, ii, 80.A . , 1921, ii, 23.1923, ii, 299.M. Randall, ibid., 1921, 43, 233; A., 1920, ii, 34; 1921, ii, 241GENERAL AND PHYSICAL CHEMISTRY. 39The Galvanic Cell and Potential DiSferences.The seat of the electromotive force in the galvanic cell has beendiscussed afresh by J. A. V. Butler.54 The establishment of theGibbs-Helmholtz relation between the electrical energy producedand the total energy of the reaction going on in the cell, and thewide use of the Nernst relation between the electrode potential andconcentration of metal ions caused attention to be focussed almostexclusively on the chemical effects a t the electrodes as the principalsources of the electromotive force.Recent physical investigations 55on the thermionic and photoelectric properties of metals have,however, demonstrated conclusively the existence of large metalcontact P.D.'s. The question arises, How are these to be reconciledwith the " chemical '' theory and in particular with the correspond-ence between the electromotive force and the energy of the chemicalreaction ? Butler meets this by showing that the energy of transferof electrons from one metal to another, on which the metal contactP.D. depends, is an integral part of the energy of the chemicalreaction. Thus the reaction Zn + Cu" = Cu + Zn" which occursin the Daniel1 cell consists of three parts : (1) the passage of zincions from the metal into the solution, (2) the deposition of copperions on the copper, and (3) the transfer of electrons from themetallic zinc to copper. It is shown that the energy change of thethird part is large, in fact of the same order of magnitude as thetotal energy of the reaction.In the galvanic cell the three stagesoccur at the two electrodes and a t the metal junction, and eachgives rise to its appropriate P.D. It thus appears that the twoconflicting theories of the galvanic cell, the " chemical '' and the" physical," are finally reconciled.It is further pointed out that the existence of large metal contactP.D.'s has an important bearing on the determination of theabsolute electrode potentials.The possibility of a metal junctionP.D. between the metal investigated and the reference electrodehas been overlooked, and it is likely that the discordant resultsobtained in methods employing mercury and those using othermetals is due to this cause.J. A. V. Butler has further given a kinetic theory of the P.D. atmetal electrodes (Nernst P.D.),56 a t oxidation electrode^,^' and at64 Phil. Mag., 1924, [vi], 48, 927; A., 1925, ii, 42.6 5 0. W. Richardson and K. T. Compton, ibid., 1912, [vi], 24, 592; A.,1912, ii, 1039; R. A. Millikan, Physical Rev., 1916, 7, 18; 1921, 18, 236;A. E. Hennings and W. H. Kadesch, ibid., 1916, 8, 209.66 Trans. Paraday SOC., 1924, 19, 720; A., 1924, ii, 598; compare Ann.Report, 1924, p.17.Trans. Paraday Xoc., 1924, 19, 734; A . , 1924, ii, 59840 ANNUAL REPORTS ON THE PROGRESS OF CHEMISTRY.metal junctions.58 The expression obtained for the electrodepotential of a metal is :U RT RTnF nP nF E = + __ logK + -- log C,in which U is the total energy change of the transfer of metal ionsfrom the metal to the solution and K is a small constant charac-teristic of the metal. The first two terms are equivalent to theNernst “ solution pressure.” By means of a cyclic process,SB it hasbeen shown thatU = X+ J - H - +F,where X is the latent heat of vaporisation of the metal, J is theenergy required to ionise it in the vapour state, H is the hydrationenergy of metal ions, and - +F the heat evolved in returning theelectrons to the metal (4 = thermionic work function).J. Heyrovskf 6o also has discussed the factors determining theNernst potential and by means of a thermodynamical argumenthas oht,nined a somewhat similar expression :H RT log P - + __ log c,RT= - P H Fin which P includes quantities which depend on the metal aloneand is regarded as the real “ solution pressure.’’ However, it maybe noted that the electrode potentials calculated by this expressiondiffer from the observed much more widely than those obtained byassuming that the electrical energy is simply equivalent to the totalenergy of the cell, and it is concluded from a similar argument thatthe metal contact P.D. is small (not greater than 0.2 volt), whichis contrary to a large body of experimental evidence.J. Heyrovsky 61 and collaborators have published an extensiveseries of researches on electrolysis with the dropping mercurycathode. An instrument called the polarograph is described 62which automatically records decomposition curves. It is claimedthat the dropping mercury cathode has a number of advantagesover electrodes hitherto employed. A fresh surface is continuallyexposed, reversible polarisation curves are obtained with veryminute currents and the hydrogen overvoltage is the highest whichhas been observed. A variety of problems has been investigated5 8 Phil. Mag., 1924, [vi], 48, 746.59 Loc. cit., ref. 54.60 J . Physical Chem., 1925, 29, 344, 406; A., ii, 404, 544; Rec. trm. chim.,61 Trans. Paraday Soc., 1924, 19, 692; A., 1924, ii, 598; RPC. truv. chim.,62 J . Heyrovsk$ and M. Shilrata, ibid., p. 496,1925, 46, 447; A., ii, 672; Compt. rend., 1925, 180, 1653; A . , ii, 793.1935, 44, 488-606; A., ii, 673-678GENERAL AND PHYSICAL CHEMISTRY. 41by its use. Its value has been strikingly shown by its independentindication of dvi-manganese in manganese salts. 63Two important papers on the Donnan membrane equilibriumhave appeared. The data of F. G . Donnan and A. J. Allmand 64on the distribution of potassium chloride between a pure aqueoussolution and a solution containing potassium ferrocyanide separatedby a membrane impermeable to the ferrocyanide ion, have beenrecalculated in terms of activities instead of concentrations byN. Kameyama.65 The distribution equilibrium found is in accord-ance with the ionic strength principle. R. Azuma and N. Kame-yama 66 have investigated the potential difference and equilibriumof sodium chloride and Congo-red across a semipermeable collodionmembrane. Assuming that Congo-red ionises as a uni-univalentelectrolyte and the ionic strength principle holds, the equilibriumresults are in agreement with the theory. It was not possible toprove that the P.D. was in agreement with the theoretical value,owing to the difficulty of eliminating liquid-liquid potentials.The theory of membrane equilibria has been restated by E. HiickelY6'in relation to the Debye-Hiickel theory.In conclusion, the Reporter desires to thank Mr. J. A. V. Butlerfor the great help he has given in the preparation of this Report.J. E. COATES.63 V. DolejEiek and J. Heyrovskfr, Nature, 1925, 116, 782; A. N. Campbell,64 J . , 1914, 105, 1941.66 Phil. Mag., 1925, [vi], 50, 849; A., ii, 1062.66 Ibid., p. 1264.6 7 Kolloid-2. (Zsigmondy Festschrift), 1925, 36, 204 ; A., ii, 528.ibid., p. 866; J. Heyrovskf, ibid., 1926, 117, 16