HEYROVSEP :~ELECTROAFFINITY OF ALUMINIUM. PART I. 11 III[.-y'he h'leclroafinity of A lwminiufir. par$ 1. Y'he Ionisation and H yily*olysis of A lurniniurn Chloride. By JAROSLAV HEYROVSK~. ALUMINIUM chloride is a quaternary strong electrolyte exerting a fourfold osmotic pressure in dilute solutions. I n a solution there may exist the ions AlCl,' AlCl" Al"' and C1'. There is no evidence for the existence of complexes such as (AlC13)2; the single molecules seem to possess more affinity for water than for each other. The cations combine to a certain degree with the hydroxyl ions to form AlCk-OH AlCl(OH), and Al(OH),; as however this hydrolysis does not exceed 3.8 per cent. even in the most dilute solutions concerned it will first be neglected in considering the dissociation of aluminiuni chloride.Let the concentration of aluminium chloride in gram-equivalents per litre be denoted by c and that of the chloridions resulting from the dissociation be cx (where x is less than 1). Let the concentration of the Al"' cations expressed in gram-cations per litre be further let [Al'"] = * Y3 (Y,<Q)> [*1C1,'1 = c * Yl (Y1<3) and [*1C137 = c Yo (YO<*). [AlCl'.] = c . y2 (pa<+) Then We shall first consider how the dissociation would change with dilution if the law of mass action were valid 12 HEYROVSK+ ELECTROAFFINITY OF ALUWNIUM. PART 1. which gives yz = K . cx . y3 y1 = K R 2 . (cx)' . ~3 yo = K,K,K? * * y3' Substituting in equation (l) we get ~ J ~ { K . K ~ K ( C X ) ~ + K,K,(cx)~+ K,CX + l } = 33 Similarly Since x==y1+2y,+3y3 .. . . . . . ( a ) when c becomes great x approaches zero and therefore yl approaches zero YZ 7 9 ) ?/3 7 7 7 7 And from (l) yo approaches 4 (a maximum). As c approaches zero x approaches unity; it therefore follows from the above equations that y3 approaches 4 (a maximum), yl y2 and yo approach zero. We can find the maximum values of y1 and y2 as follows: Therefore By equating t o zero we find that yz is a ma.ximum when x=-: Similarly we may show that' and therefore y1 is a maximum when x = - ) i . Summing up, yo is a maximum when c is great and x is small and a minimum when c is very small and x = 1 HEYROVSK* ELECTROAETINITY OF ALUMINIUM. PART I. 13 y1 is a maximum when x = Q and a minimum when c is very y2 is a maximum when x=$ and a minimum when c is very y3 is a maximum when c is very small that is when x = 1 and The changes of yl ye y3 and x corresponding with the gradual This was deduced for an ideal electrolyte obeying the law of great or very small that is when x=O or x=1.great or very small that is when x=O or x=1. a minimum when c is very great that is when x = 0. splitting into simpler ions on dilution are plotted in Fig. 1. Fx. 1. 7 3 0Y i 0 mibas action. It can however be shown that the ,same mode of dissociation takes place a.s indicated in Fig. 1 whatever the law may be so long as it is of the general form = K . Cation x C L o n C:,Olt!Cllle ~~ (For van't Hoff's constant a= b = $ p ; for Ostwald's a= b =I).) progression we obtain Thus assuming that the indices a b c .. . are in arithmeti 14 HEYROVSd ELECTROAFFINITY OF ALUMINIUM. PART I. which means that y2 is a maximum when x = 3 just as in the previous case and similarly for y1 and ys. The coefficient x representing the ionised fraction of chlorine, may be determined from the potential of a calomel electrode filled with the respective aluminium chloride solution. These results however would give only the maximal and minimal value of y3 that is of the dissociation into Al"' ions but no evaluation whatever as to y and y2. Further conclusions respecting the [Al' ''1 concentration can be drawn from the hydrolysis of aluminium chloride solutions when solid aluminium hydroxide is present. Then [Al"'] . [OH/]3= K , the concentration of [OH/] being determinable by the potential of a hydrogen electrode.To evaluate the remaining ionic concentrations specific con-ductivit]ies of aluminium chloride solutions must be considered. Measurements of Electromotive Force. Measurements of cells of the type Hg J calomel solution of AlCl I Hz(Pt) were made with all precautions in the manner described by Tol-man and Ferguson ( J . Amer. Chem. SOC. 1912 34 232) and Acree (Amer. Chem. J. 1911 46 585 621 638) using platinised glass electrodes. Readings on the potentiometer could be made t o a tenth of a millivolt. In order t o examine the reproducibility of the electrodes used, the cell was filled with 0.100N-hydrochloric acid. The E.M.F. a t 18O and 760 mm. was found to be constant for three days a t 0.3958 f 0*0003 volt.The mercury used was purified and twice redistilled. The hydrogen was passed through alkaline permanganate and lead nitrate solutions and was finally bubbled through the same solu-tion as was being measured in the cell. The aluminium chloride was purified by precipitation three times from solution by means of hydrogen chloride and the solutions were kept a t 2 5 O in bottles the corks of which were covered with paraffin. The cell was kept in a thermostat at 2 5 O and the readings were repeated after three or four weeks. The effect of barometric change was much less than the variations due t o experimental errors which became appreciable in dilute solutions. Two hydrogen electrodes were used simultaneously dipping into the same solution; they did not differ by more than 0.5 millivolt HEYROVSKP ELECTROAFFINITY OF ALUMINIUM.PART I. 15 In this way the following results were obtained for the cell: +Hg I (Hg,Cl solid) AlC1 solution 1 H,(Pt) -Concentration in gram-equiv. per litre. of AICI, 2.29 0-368 0.0184 0.0092 0.006 13 0-0046 0.00306 E.M.F. observed at hfferent times. L c I 0.3739 0-3735 0.3725 0.3725 0.4810 0.4810 0.4808 0.4807 0.6000 0-6005 0-5930 0.5924 0.6190 0.6186 0.6170 -0.632 - 0.634 -0.639 0.646 0.643 0.641 0-659 0.658 0.661 0.655 Mean E.M.F. in volt. 0.373 0.481 0.595 0-618 0.633 0.643 0.658 I n other experiments the aluminium chloride solutions were shaken with precipitated aluminium hydroxide for several weeks previous to being introduced into the cell.With these solutions, the following results were obtained : Concentration of AlCl, solutions .in gram - equiv. per litre. 2-88 0.934 0- 1845 '0.0675 0.0337 0.02134 0.01067 0-00675 0.002 13 E.M.F. observed at different times. r 0,4221 04225 0.4940 0-4934 0.5351 0.5623 0.5804 0.5965 0.5980 0.6147 0.6158 0.6376 -0.6685 -, 0.4218 -0:4939 0.4934 0.5359 0.5614 0.5818 0.5807 0.5981 0-5996 0-6360 0.632 0.6667 - -Mean E.M.F. in volt. 0.4222 0.4937 0.5355 0-5620 0.5813 0.5984 0-6150 0.635 0.668 I n order t o find the potentials of the single electrodes the aluminium chloride-calomel electrodes were compared with normal and tenth-normal calomel electrodes at 2 5 O .Saturated half-saturated and quarter-saturated potassium chloride solutions were used successively as intermediate solutions and the values extra-polated. This method of eliminating the liquid potential was found to be much more trustworthy than the use of potassium nitrate or ammonium nitrate solutions. All glass tubes were a t least 0.5 cm. wide and no glass taps were used. The following values were found for the potential 7rN of the electrode. Hg I (Hg2Cl,)A1Cl solution taking the normal calomel electrode as zero 16 HEYROVSKP ELECTROAFFINITY OF ALUMINIUM. PART I. Concen-tration of AICl,. 2-29 0.368 0.0409 0-01 84 0.0092 0.006 13 0.00460 0.00306 Quarter half and fully saturated potassium chloride solution as intermediate solution. - P.D.against P.D. against TK N-calomel electrode. N/10-calomel electrode. extra- - polated. -0.0090 -0.0114 -04137 -0.0615 -0.0625 -0.0664 -0.0140 +0.0294 0.0282 0.0278 - 0.0249 - 0.0252 - 0.0267 + 0.0273 0.0766 0.0768 0.0772 .+ 0.0238 0.0238 0.0234 +0*0772 0-0970 0.0976 0.0978 0.0444 0.0444 0.0444 + 0.0986 0-1174 0-1168 0.1155 0.0636 0.0634 0-0633 fO.1150 0.1257 0.1255 0.1253 0.0721 - -(-0*1250 - + 0.1320 0.1352 0.1328 0-1320 - - - 0.1420 0.1420 0.0868 0.0870 0.0880 +0*1420 Similarly the values of 7rN of calomel electrodes filled with aluminium chloride solutions saturated with aluminium hydroxide were obtained : P.D. against N-calomel electrode using qmrter hdf and fully saturated Concentration potassium chloride solution. RK h of AlCI,. r- \ extrapolated.2.88 -0.0105 -0.0132 -0.015 -0.017 0.934 + 0.0106 0.0092 0.0076 + 0.007 0.0675 +0*0721 0.0720 0.0710 + 0.070 0.00675 +0*1185 0.1180 0-1172 +0*117 For the calculations of conductivities Jones' data (Carnegie Inst. Publ. No. 170) which agree with those of Ley (Zeitsch. physikal. Chem. 1899 30 206) were used. Molecular Dilution u Molecular Dilution 97 conductivity 1 mol. in conductivity 1 mol. in nt 25". u litres. I at 25". u litres. 220.86 360.56 1024 265.12 3; 381.44 2048 308.80 128 393.79 4096 193.51 4 ' 341 -24 512 Calculation of Dissociation.. The single potentials rN were plotted against logc and the values of 7rN for use in the following calculations were taken from the smooth curve. The concentrations of chloridions were found in the following way.I f 7rN were the potential of a calomel electrode in a given aluminium chloride solution then the concentration of a potassium chloride solution in which the calomel electrode had this same potential 7rN was found from the curve showing the relation between the potential of a calomel electrode and the logarithm of the concentration of potassium chloride. It was assumed that the concentration of chloridions in these two solutions was the same HEYROVSK$ ELECTROAFFINITY OF ALUMINIUM. PART r. 17 and the absolute value of the chloridion concentration was then calculated from measurements of the conductivity of potassium chloride solution. The following values of rN for calomel elec-trodes in potassium chloride solutions were used (Abegg Auerbach, and Luther Abhandl.Bunsen Ges. No. 5 ) : xx of AT- calomel electrode at 25’= 0.000 volt 7 ) N/10- 7 9 7 , 0.0541 ? 7 N/100- ! 9 7 7 0.108’7 7 9 N/1000- ? ,? 7 0.164 Column 3 (table I) was calculated in this way. From this the ratio [ g = x column 4 was obtained. c For example in an 0*0092N-solution of aluminium chloride the potential of a calomel electrode is equal to the potential of a calomel electrode in potassium chloride solution for which logc=3*9114 that is 0.008155N. This being ionised to the extent of 94.6 per cent. has [C1’]=0.007715 which must be identical with the concentration of chlorine ions in an 0.0092iV-solution of aluminium chloride. Hence x = [‘K1=O*838l. C The [Cl’] was not calculated directly from the formula 7r = - 0.0591 10g,,[Cl’], hecause second-class (anionic) electrode potentials do not agree exactly with the values calculated from condudivity data possibly due to the formation of complex mercury ions.Hydrogen electrode potentials however were found t o vary strictly according to the formula r = - 0.0591 log,,C, when the concentration of hydrogen ions C, is determined from conductivity (Bjerrum Zeitsch. physikal. Chem. 1905 53 428; 1907 59 341). Thus column 7 (table I) was calculated from the potential of hydrogen electrode 7~~ (referred to the normal potassium chloride-calomel electrode as zero) the difference of the two normal electrodes having been taken as 0-2837 volt. The ratio cH* (column S) shows the degree of hydrolysis h . For example the potential of a hydrogen electrode in 0*0092X-aluminium chloride solution is - 0.5037 volt 18 HEYROVSKP ELECTROAF-FINITY OF ALUMINIUM.PART I. TABLE I. 1 2 3 4 5 6 7 s 9 c = con-con tration o f AlCI in gram - Maxi-cq uivalent n1r [?!:I ~ 7,. .mum per litre. logl, c. [Cl']. x. x'. in volt. [H']. c YO 0.00306 3.48572 0.00267 0.873 0.835 0.5162 0*000116 0.0378 0.055 0*00460 3.66276 0.00395 0.859 0.829 0.5118 0.000138 0.0300 0.057 0.00613 3.78746 0.00522 0.851 0.825 0.5080 0*0001603 0.0262 0-058 0.00920 :<.96400 0.00772 0.838 0.817 0.5037 0.0001897 0.0206 0.061 0.01585 2.2000 0.01295 0.817 0.803 0.4980 0-0002275 0.0143 0.065 0.02291 2.3600 0.01848 0.806 0.794 0.4937 0.000280 0.0122 0.067 0.03162 2-5000 0.0251 0.793 G.783 0.4900 0.000323 0.0102 0.072 0.05012 2.7000 0.0387 0.772 0.764 0.4850 0.000393 0.00783 0.078 0.1585 i.2000 0.1149 0.725 0-721 0-4710 0.000659 0.0042 0-093 0-480 i.6810 - 0.666 0.664 - - 0.00362 0.112 0-631 I4300 - - 0.646 - -1.2359 0.092 0.7550 0.611 0.607 0.4190 0.00520 0.00421 0.131 2.290 0.3598 1.310 0.572 0.564 0.3870 0.01807 0.00789 0.145 0.00360 -~ ._ .. . . Hence 0.0591 lo,al(,CH- = - 0.5037 + 0.2837 = - 0.2200 volt log c - - o'2200 __- - - 3.722 =- 4.278 lo H a - 0.0591 from which CtT. =0.0001897 h = gH = 0.0206 = 2.06 per cent. As the hydrolysis in the solutions used is less than 3.8 per cent., we may neglect ths concentrations of cations such as AI(OH),' and Al(OH)" and assume or from which it follows that [Al"'] + [AlCl"] + [AlCl,'] + [H'] = [Cl'], 3cy + 2cy + cy1+ c,. = cx, 3Y3f 2?J + ? / l = X - h XI.The values of X I (column 5) are obtained by subtracting the This number XI limits the value of y3 the maximum value of which can be - (when no other cations exist in solution in which case also yo is at a maximum). The minimal value of y3 is X I - 6 in the case when most of the AlC1" cations are formed. I n this way columns 9 10 and 11 were obtained. Considering the difficulty with which second and third ionic charges are acquired the minimal values of y3 are more probable, numbers in column 8 from those in column 4. X' HEYROVSKq ELECTRODFINITY OF ALUMINIUM. PART I. 19 'TABLE I. 10 11 12 13 14 15 16 17 18 19 20 T4qui-valent con-Maxi- Miiii- duc- (L = mum. mum. tivity. A Xti' m'l'.; y3. y3. A,. A . z'' in volt. [H']'.[He]$ y3. y2. y,. 0.278 0.168 120.2 40.9 49.0 as ?TIT. as [H'] 1 0.180 0.142 <.01 0.276 0.162 116.3 40.7 49.3 - - - 0-175* 0*153* -0.276 0.138 113.7 40.5 48.93 - - - 0*170* 0*161* -- 0.160* 0*169'" - 0.272 0.150 110.3 40.0 48.75 - -0.268 0.136 106.0 39.0 49.18 0.503 0.000187 0.811 0.146 0.178 0.009 0.265 0.127 103-0 38.1 47.88 0.4985 0.000207 0.761 0.137 0.187 0.009 0.261 0.116 100.0 36.9 46.8 0.498 0.000227 0.727 0.131 0.188 0.014 0.240 0.054 83.3 27.5 38.14 0.490 0.000323 0.418 0.075 0.2 0-10 0.221 0 - - - 0.487 0.000392 0.248 0.044 0.2 0-13 0.202 0 - - - - -0.188 0 - - - -0.255 0.097 95.3 34.6 45.18 0.4955 0.000260 0.684 0.123 0.19 -- - - - - - - - -- - - -- - - - -* These values are found by extrapolation. I n order to obtain more precise numbers conductivity results have t o be included.Denoting by iM, M, M the mobilities of cations carrying the total charge of one faraday each that is of the ions AlCl,' gAlCl" +Al*** respectively and taking the mobilities of chloridioii as 75.0 and that of hydrion as 365.0 at 2S0 the equivalent conductivity of a solution of aluminium chloride A, is A = 3y3M3 + 2?/,M + ~ l M l + x .75 + h .365, from which A is obtained (column 13) as A =3~3M3+ 2yzM,+ ?/,MI=&- X .75 -h .365. The total charge carried by the three different cations AlCl,', AlCl" and Al"' is cxf faradays and as they contribute to the conductivity the amount A the mean equivalent mobility for one cation is -. Since this value as is evident' from column 14 . approaches 49.2 as the solution becomes very dilute and all the cations become Al"' the number 49.2 has been taken as the most probable equivalent mobility of Al"'.I n order to evaluate the other unknowns more relations are necessary. These are obtained from the most dilute solutions where the third stage of hydrolysis is reached and are supposed t o be in equilibrium with solid aluminium hydroxide the values of rH in pure aluminium chloride solutions and in those saturated A X 20 HEYROVSKP ELECTROAFFINITY OF ALUMINIUM. PART I. with aluminium hydroxide (column 15) coinciding in concentrations less than 0.0lN. I n these solutions, [OH'I3 C?/3 = K,,,,, 7 where KAI,oH)3 denotes the ionic product of aluminium hydroxide. Its value lies according t o the values of y3 between 1.0 and 1.5 x 10-33 (the ionic product of water being taken as 10-14).Further in most dilute solutions yo is negligible so that Y l + Y 2 + Y 3 = 9 . Solving these equations for the three most dilute solutions we get M3=about 30, and the different values of yl y2 and y3 are given in columns 18, The values of y3 in concentrations greater than 0-01N were calculated from the ratio of hydrion concentrations [He]' in solu-tions of aluminium chloride saturated with aluminium hydroxide to [H'] in solutions of aluminium chloride alone. In column .17 a denotes the ratio of the cube of [H']' obtained from the potential 7rE' of the hydrogen electrode in these solutions to the cube of [H'] in aluminium chloride solutions. KAl(oa) = 1.06 x Mz zx 47.0. 19 20. Conclusions. Progressive Hydrolysis .-In solutions below 0.01 N the third stage of hydrolysis exists, AlCl + 3HOH = Al(OH) + 3€€Cl, Al"' + 30H' -+ Al(OH),.AlC1" + 20H' -+ AlCl(OH),, or the ionic reaction For the second stage having the ionic reaction the expression should be constant. Here l K w denotes the ionic product of water. Similarly for the first stage the expression cy1* Kn [H'] . [AIC;l,OHS should be constant. These relations cannot however be tested as the concentrations of AlCl(OH) and AlCl(OH) are not known with sufficient accuracy. Since the base AlC1,OH containing two chlorine atoms is prob EEYROVSK$ ELECTROAFFINITY OF ALUMINIUM. PART I. 21 ably weaker than the base AlCl(OH) having one more chlorine atom in place of hydroxyl the slight increase of hydrolysis in the most concentrated solutions (minimum in 0*5N-aluminium chloride solution; see Fig.2) might be interpreted as due t o first-stage hydrolysis. A similar minimum of hydrolysis in the most concentrated solu-tions was found by Kablukov and Sachanov (Zeitsch. physilal. Chem. 1909 69 419) in the hydrolysis of aluminium bromide a t 2 5 O and by Bruner (Zeitsch. physikal. Chem. 1900 32 133) in the hydrolysis of aluminium chloride a t 40°. The determination of hydrolysis by means of hydrogen electrode potentials has been made by Denham (T. 1908 93 41) who obtained for aluminium chloride solutions ranging from 0.19 to 0.024N values of the potential of the electrode about 20 millivolts lower than those observed by the author. He therefore found three times more hydrolysis.In his experiments the liquid potential was eliminated by means of a concentrated solution of ammonium nitrate the solutions being allowed twenty-four hours to come to equilibrium. The hydrolysis of aluminium salts seems to be influenced by the mode of dissolution. For example if the solution is raised to a temperature above 25O it does not necessarily return to the same condition a t 2 5 O its it was before (Jones Zoc. cit.). Moreover a slight excess of hydrochloric acid in the dry salt would cause too great an acidity and any adjustment of the equilibrium takes place very slowly. This difficulty in attaining equilibrium if once disturbed seems to be due to the presence of colloidal aluminium hydroxide which exhibits the phenomenon of ageing showing marked insolubility when not in the nascent state.The solutions must he therefore prepared and kept so far as possible at the same temperature. Kablukov and Sachanov's results (loc. cit.) for the hydrolysis (at 25O) of aluminium bromide calculated from E.M.F. measurements are very close to the values given in this paper. The Heat of Ionisation of ,4 Izcmim*um Hydroxide .-Kulgren (Zeitsch. physikal. Chem. 1913 85 466) measured very accurately the hydrolysis of aluminium chloride a t 8 5 O and looo. From his results and from the values given in Fig. 2 (curve h) the heat of ionisation of aluminium hydroxide can be obtained in the follow-ing way. The degree of hydrolysis of aluminium chloride a t the dilution v=512 is 4.7 per cent. a t 25O 34.09 per cent.a t 85O and 47.68 per cent. a t looo. Since a t this dilution the third stage of hydro-lysis exists the heat of ionisation of the reaction, Al"' + 30H'- Al(OH) solid 22 HEYROVSK~ ELECTROAFFINITY OF ALUMINIUM. PART I. can be calculated from the data given above by means of van't Hoff's isochore. Thus considering the equilibria a t two temperatures T and T', Further a t the same dilution u=512 the concentrations of the Al"' ions can be taken as equal to the non-hydrolysed portion in solution and since a t this great dilution some solid aluminium hydroxide will certainly have separated out we can write: where h denotes the fraction hydrolysed and iTAl(OH) and E'Al(oHl,, are the solubility products. Substituting in van't Hoffs formula, we get " Taking K at 1OO0=48x 10-14 a t 850=27.6 x 10-14 and a t 25O= 1 x as extrapolated from Noyes' numbers (Zeitsch.physilcal. Chem. 1910 73 1)) we obtain: Q between 25O and 100°=11970 cal., Q 7 25O 9 ) 85O= 12860 cal. Thus the heat of ionisation of one gram-molecule of solid aluminium hydroxide into the ions Al'" and OH' is about 12000 cal. When neutralised by strong acids one equivalent of aluminium hydroxide should evolve 13.700 calories less than the heat necessary to ionise the molecule that is, 13700 -4133 = 9567 cal. Thomsen found 9320 cal. which agrees with the calculated value, indicating a base of medium strength. The Diffusion Potential. Having found the ionic concentrations and the corresponding mobilities it seemed interesting to compare the potential differences observed on liquid boundaries between single aluminium chloride solutions with the values calculated from Henderson's formula (Zeitsch.physikal. Chem. 1908 63 325). The formula for the diffusion potential E is E = RT (u1-v,>-(u2-v2) ul'+v,' F ' ( Ul' + Vl') - ( U + V;) loge U T V T ' - - where V l = V G fV,G2+ . . . . - U,=u~C,+uBCa+ . . . . U1' = UICIZO1 + ZL2C2Z1l2 + 1-1 - . . . . V1' = Vu1ClWl + v2c2zu2 + . . . HEYROVSK~? ELECTROABPINITY OF ALUMINIUM. PART I. 23 c c denoting concentrations of cation and anion, u , equivalent mobilities of cation and anion, w,w , valencies of cation and anion respectively. I n the special case of aluminium chloride, Ul + Vl = K (specific conductivity) therefore Ul = K - 75 CX. Further neglecting the hydrolysis, V = V,' = ex = 75 .ex. u1- V'l =Kc- 150 C X = C ( X - 150~). 77 = c(yliwl + y2. 2M2 + y3 . 3M3) U,' = c(ylM1 + 29 . 2M2 + 3y3 . 3^3f3) Uz' - Ul = c(2y$f2 + 6y3M8) hence further Ui' + Vi' = ~ ( 2 ~ 2 M 2 + 6~3M3) + Ul+ v1 = c(& + 2y2M3 + 6y&f3)* Finally, TABLE 11. Preliminary Calcula tiolts. Concen-tration of AICI, in equiv-alent/ litre. A,!. 0.1227 85-7 0.0184 104.7 0.0092 110.3 0.00613 213.7 Ac + 2M2~3+ ~ M Y + 2. y3. ~ 1 . 6Mg3. 150~. 6M3yS. 0-730 0.1 0.2 48-4 -23.8 134.1 0.814 0.14 0.18 58-3 -17.4 163.0 0.838 0.16 0.17 63-4 -14.4 173.7 0.851 0.17 0.16 65.3 -14.0 179.0 b-Q,Z C(AC + c(&- 2M 2~ 2 + 1502). 6M3y3)--2.92 16-45 -0.38 2-99 -0-133 1.598 - 0.0868 1.097 a - a t b b - b Calculation of E = 0.059 - I log, volts.According to Henderson's paper the E.M.F. is here denoted as positive if the current passes from the first solution t o the second inside the cell. E.M.P. measured between calomel electrodes in AlCl solutions of concentration. single 0-1227 0.0184 0.019 0.1227 0.0092 0.040 0.1227 0.00613 0.046 0.0184 0.092 0.017 0*0184 0.0613 0.027 P.D. of & electrodes. E.2M.P. observed without elimin-ation of E diffusion P.D. observed. 0*0100 - 0.009 0-0290 - 0.01 1 0.0321 - 0-014 0.0153 - 0-002 0.019 - 0.008 E calcu-lated. - 0.0084 - 0.01 12 -0.0128 - 0.0022 - 0.003 24 HEYROVSK$ ELECTROAFFINITY OB ALUMINIUM. PART I. I n table 11 the potential differences calculated in this way are compared with those directly observed this being the first instance in which Henderson's forinula could be tested in the case of tervalent iona.I FIG. 2. The Actavzty of H ydrochlorac Aczd an Alumzlzzum C'hCorzcCe Solutions. The results of E.M.F. measurements are plotted in Fig. 2. The lines marked T K C l and rAlClI give the potentials of calomel elec-trodes in solutions of potassium chloride and aluminium chloride respectively showing that the activity of chlorine ions in aluminium chloride solutions is very near t o that which obtains in equivalent solutions of potassium chloride the dissociation into chloridions in the case of aluminium chloride being only about 20 per cent. less than in the case of potassium chloride HEYROVSK~ ELECTROAFFINITY OF ALUMINIUM.PART T. 25 The abscissz in Fig. 2 are in all cases logarithms of the con-The curve h = 0,. shows the change of hydrolysis with dilution. The curve 7rH represents the potential of the hydrogen electrode in aluminium chloride solutions referred to the normal calomel electrode as zero (the values to be taken as negative). The curve EAIC1 shows the variation of E.M.F. of the cell Hg I calomel AlCl solution I H, giving the activities of chlorine and hydrogen ions or the activity of hydrochloric acid according to the formula centration c expressed in gram-equivalents per litre. EAICI = RT log pE ' - = 0.2837 - 0.0591 loglo [H*) . [Cl'] [H'] . [Cl'] = 0.2837 - 0.1 182 10g,O [HCI]. Thus the curve EAIC13 shows the partial pressures of hydrochloric acid formed by the hydrolysis of aluminium chloride.If this curve is compared with the curve E,,,, expressing the partial pressure or activity of pure hydrochloric acid solutions as obtained from the results of Tolman and Ferguson (Eoc. cit.), corrected from 1 8 O to 25O i t is possible to find for every aluminium chloride solution the concentration of pure hydrochloric acid which would have the same activity of hydrogen chloride (exert the same partial pressure). Thus in 2.29 N-AlCl the HCl tension is that of 0.176 1-00 , , 0.0582 ,, N-HC'1. 99 0.368 , 9 7 , 0.0184 ,, 0.100 , 9 7 , 0*00617 ,, 0.0184 , 3 , 0*0017S )) the potential difference between calomel and hydrogen electrodes, that is the ordinate in each pair of solutions being the same. The values of E,, for dilute solutions have been obtained by extrapolating from the observed values and assuming that E, increases by 0.1182 volt for a decrease in concentration of 10 1.sponding with the theoretical increase of E.M.F. by Similarly the line EAICI was produced in the direction corre-RT log [H'] . [Cl'] = -log [Al"'Ja. c = E l o g c = $- .0.0591 volt P F €or tenfold dilution which a t the highest dilutions will be greater owing to the decrease of Al"' ions due to the precipitation as Al(OH),. As is seen in the figure these lines will meet a t a normality o about 0-000027 where practically total hydrolysis is reached so that in concentrations less than lO-5iV aluminium chloride behaves similarly to boron trichloride which is completely hydrolysed in solution; on the other hand aluminium hydroxide does not dissolve in 10 - "V-hydrochloric acid to form aluminium chloride.The value of rH a t great dilutions has been assumed t o be pro-portional to log[H']. From this the value of rH in a neutral solution is 0.70 volt. The line EKC, representing the activity of hydrochloric acid in solutions of an ideal non-hydrolysable chloride of a strong base, has been drawn in which rH is taken as equal to 0.700 volt. Similar E.M.F. curves for chlorides must all lie between the two extremes namely E,, and E,, lines the slope becoming steeper as the alkalinity of the metal decreases. Thus a chloride of a strong base like lanthanum trichloride would give a curve very similar to ITKCI whereas boron chloride would give nearly that of E,,,. This would be a precise and distinct way of expressing hydrolysis. Summary. (I) From conductivity data and E.M.F. measuremenh of aluminium chloride concentration cells the amount of ionisation and hydrolysis was determined and the gradual ionisation of cations calculated. (2) The ionisation of aluminium chloride into chloridions is about 20 per cent. less than the ionisation of potassium chloride of equivalent concentration. (3) The basic solubility product of aluminium hydroxide is (4) The heat of ionisation of aluminium hydroxide has been (5) The mobility of the ion Al"'=3 x 49.2 that of the ion (6) Henderson's formula for the diffusion potential between [Al"'] . [OH']3= 1.06 x 1 0 - 3 3 . calculated. AlCl" = 2 x 47. solutions of aluminium chloride has been found to hold good. CHEMICAL DEPARTMENT OF UNIVERSITY COLLEGE LONDOE. CHEMICAL INSTITUTE OF THE CZECH UNIVERSITY PRAGUE. [Received October 20th 1919.