ELECTRICAL CONDUCTIVITY OF DIETHYLAMMONIUM CHLORIDE 61 IV.-Elect rical Conductivity of Diethylammonium Chloride i ? ~ Aqueous Alcohol. By JAMES WALKER, Ph.D., D.Sc., and FRED. J. HAMBLY, E.1.C. IN the course of some experiments on urea formation in aqueous alcohol, we found it necessary to ascertain the effect produced on the electrical conductivity of a salt solution, as successive portions of the water used as solvent were replaced by ethylic alcohol. It was desired in particular to determine the part played by the alcohol in reducing the degree of dissociation of the salt as well as in reducing the speed of the ions, The salt selected as most suitable for our purpose was diethylam- monium chloride, which is readily soluble in ethylic alcohol. It was prepared by evaporating a mixture of diethylamine with a slight excess of hydrochloric acid t o dryness on a water bath, small quantities of water being afterwards added and the evaporation repeated.After remaining for a few days in an exhausted desiccator, the substance hardened to a dry cake, which was then powdered and kept in a desic- cator over caustic potash in order that it might lose any traces of hydro- chloric acid which still clung to it. From the chloride thus obtained62 WALKER AND HAMBLY : ELECTRICAL CONDUCTIVITY a normal aqueous solution was prepared, and this was used in all the subsequent experiments. The alcohol employed was '' absolute " alcohol treated with sodium and distilled. The apparatus in which the distillation was effected resembled that used in the operation of distilling with steam.An intermediate distilling flask was introduced between the flask con- taining the alcohol and the condenser, so that the vapour was washed in the liquid which condensed in the intermediate flask. Spirting was in this way effectually prevented. A glass condenser was used, and the connections were made with indiarubber which had previously been boiled with soda and thoroughly steamed. The specific conductivity of the alcohol thus obtained was 0-36 x in reciprocal Siemens units at 25". To obtain water of a suitable conductivity, ordinary distilled water was allowed to remain overnight after being treated with a little barium hydroxide and sodium hypobromite solution. The distilling apparatus was similar to that used in the preparation of the alcohol, the condenser-tube, however, being in this case of block tin, bent down- wards and backwards into the beak of a retort which served as the intermediate vessel.The water collected had on the averitge a conductivity of 1.5 x The conductivity of the salt in each mixture of alcohol and water was determined a t eight different dilutions, the strongest solution being decinormal. The mixture used as diluent was prepared by mixing the purified alcohol and water in certain proportions by volume, the actual composition of the mixture being then ascertained by an accurate determination of its specific gravity, To prepare the decinormal solu- tion, purified alcohol and normal diethylammonium chloride solution were mixed in the same proportions as were used in preparing the diluent, the solution thus obtained being then heated to 25.0", the temperature of experiment, and made up with the warmed diluent to 10 times the volume of the normal solution employed.This method assumes that the normal diethylammonium chloride solution contains its own volume of water, which is, strictly speaking, not thecase. The error in the conductivity introduced from this source is, however, small, and has in what follows been neglected. The mixtures used as diluents contained the following proportions of alcohol by weight corresponding to the proportions by volume at 15.5" at 25". 8.1 25.3 41.8 64.6 86.0 per cent. 10.1 30.7 49.2 72.0 90.3 per cent. An incomplete series mas also made with alcohol containing 98.3 perOF DIETHYLAMMONIUM CHLORIDE IN AQUEOUS ALCOHOL 63 cent.by weight, or 99.0 per cent. by volume. I n this case the deci- normal solution was made by dissolving a weighed quantity of the chloride in the appropriate volume of alcohol. The method adopted for the determination of the resistance of the solutions was Kohlrausch’s with induction coil and telephone. The apparatus was fitted up as described in Ostwald’s Physico-clwnical Measurements, p. 222 et seq., ,attention being given, however, to the more recent precautions and improvements suggested by Kohlrausch , (Ann. Phys. Chenz., 1893,49,225). With these precautions, the method gave excellent results even a t the greatest dilutions employed, and we are convinced that the source of any errors affecting our results is not to be sought for in the method of electrical measurement, but in the effect of small quantities of impurity, inaccuracies in the dilution, &c.The cell was of the type described by Arrhenius, the hole in the ebonite cover being closed by an indiarubber plug to prevent evaporation of the alcohol. The electrodes were well platinised several times during the course of the experiments. It is well known that after platinisa- tion it is almost impossible to mash the electrodes free from con- ducting material, a week perhaps elapsing before they are suitable for use in liquids of high resistance. We find that this difficulty may be got over very simply by dipping the electrodes after platinisation into a solution of sodium acetate, which is electrolysed in precisely the same way as the platinum chloride solution, the current being reversed several times.A few washings with water are then sufficient to render the electrodes fit for use. The measuring wire was carefully calibrated before and after the whole series of experiments, and the cell-constant was taken frequently during the progress of the measurements. The resistances in the box employed were compared against each other and found to have a negligible relative error. The temperature of experiment was 25*0°, the resistance-cell being immersed in the water contained in a thermostat of the form described by Ostwald, Zoc. cit., p. 59. I n performing the dilutions we experienced unexpected difficulty, and i t speedily became apparent that volumetric apparatus which worked quite satisfactorily with aqueous solutions gave untrustworthy results when mixtures of alcohol and water were employed as solvents, the irregularities being chiefly due t o unequal draining and the formation of “tears ” on the walls of the pipettes, &c.The plan of work finally adopted was as follows. Two 10 C.C. pipettes were taken, one of which was used for removing the solution from the cell and the other for delivering the diluent. On the stem of the delivery pipette a milli- metre scale was etched, and for each particular diluent the place on the stem was noted for which the volume delivered was equal to the64 WALKER AND HAMBLY : ELECTRfCAL CONDUCTIVITY v. A x 105. p. 100117. 10 847 84.7 78.8 20 448 89% 83'3 40 234-5 93.8 87'3 80 121'9 97'5 90.7 volume removed by the other pipette.By means of these pipettes the dilution was increased by successive doublings from w = 10 to v = 320, a reading of the conductivity being taken a t each fresh dilution. A direct dilution in one operation from v = 10 to v = 320 was then made by weight in the following manner. A tared measuring flask of 100 C.C. capacity was filled to the mark with the diluent and weighed, after which it was emptied and dried. A quantity of decinormal chloride solution equal in weight to of the quantity of the diluerit that the flask contained was then introduced, and the solution made up to the mark with the diluent. The error caused by the difference in the specific gravity of the diluent and the decinormal solution was found to be small compared with errors from other sources, and was consequently neglected. The differences between the conductivities of the solutions obtained by successive and by direct dilution were under 1 per cent.In all cases, the direct dilution was assumed to be correct, the appropriate correction being applied to lower dilu- tions when necessary. A dilution was then made by weight in the manner above indicated, from w = 320 to v = 1600, and then another in the same way from v = 1600 to v = 8000. The conductivity of the diluent was finally ascertained, and the correction applied in the manner proposed by Arrhenius. At the dilution 2r = 8000 this correc- tion assumes relatively large proportions, amounting occasionally to as much as 12 per cent. of the total conductivity. The results for that dilution are therefore probably much less accurate than those for lower dilutions, but we think that even here the total error is less than 2 per cent.of the whole value, for in several cases duplicate determiria- tions made with diluents prepared and mixed at different times showed an agreement well within that limit. I n the following tables of our results, v represents the number of litres in which one gram-molecule of the chloride is dissolved, X is the specific conductivity in reciprocal Siemens units, p the molecular con - ductivity and rn the ratio pv/p,, so that l O O r n represents the percent- age dissociation. The solvent employed is placed a t the head of each table. Water.. 2'. A x 10". p. 100n.L. ~~~~ 160 62-5 100'0 93.0 320 32-0 102'3 95.2 1600 6'60 105'6 98.2 8000 1'34 107.0 99.5 107.5 100 cc: -OF DIETHYLAMMONIUM CHLORIDE IN AQUEOUS ALCOHOL.65 33.1 34'6 36 '8 38 '7 10.1 per cent. alcohol by volume. 84.0 87 '8 93 '4 98.2 A x 105. I p. 17.0 20.0 23.3 26.7 100nz. 43.E 51.5 60'1 65.8 21. A x 105. 1001n. V. Pa 77.5 79-1 81'3 82.9 83.5 77.8 82.8 86'8 90 '3 92.8 94.7 97 '4 99 3 100 649 345'5 181.3 94 '2 64.9 69 *1 72.5 75 *4 10 20 40 80 160 320 1600 8000 a 48.4 24 -7 5 -08 1 -04 - ~~~ ~~~~~~ 30.7 per cent. alcohol by volnme. A x 105. 2'. 1 A x 105. i 'u. 10 20 40 80 405 218 114.5 59.5 40 -5 43 -6 45 '8 47'6 75.1 80.9 85 -0 88.3 160 320 1600 8OOC cc 30.7 15.7 3.21 0'665 - 49 *i 50.2 51 *3 53 '2 53.9 91.1 93 52 95 '2 98.7 100 49'2 per cent. alcohol hy volume. 10011a. I v. I A x 105. j p. 100172. A x 105.P. 160 320 1500 8000 tc 300 161.5 86.5 45 -5 23 -9 38.2 12.25 ~ 39.2 2 5 5 40.8 0'527 42'2 - 1 42.9 I 10 20 40 80 30'0 32.3 34 -6 36 '4 69 '9 75.3 80.6 84'9 I 89 *O 91 -4 95.1 98.3 100 72.0 per cent. alcohol by volume. A x 105. 100m. 100?12. pv I 29. A x 105. 9. I- 239 131.5 72 *5 38 *9 23 '9 26 -3 29'0 31 -1 160 320 1600 8000 a 20.7 10 '8 2 *30 0-484 - 60 *7 66.8 73% 78 -9 10 20 40 80 90.3 per cent. alcohol by volume. h x 105. 10 O?it. V. I- 160 320 1600 8000 tc 18 '4 10 -1 2'26 0.476 - 29 '5 32.4 36'2 38 *1 38 '8 76 0 83.5 93.3 95.2 100 10 20 40 ao 170 100 58 '3 33 '4 YOL. LS XI. F66 WALKER AND HAMBLY : ELECTRICAL CONDUCTIVITY v. 10 20 40 99.0 per cent. alcohol by volume. A x 105. p. 100??2. PIL__(-- 131 13-1 34.0 78 15'6 40.5 47'3 18'9 49.1 I-l-I- 28.2 22.6 1 58.7 16.8 26'9 j 69'9 320 9.7 31-0 80.5 cc - 38.5 1 100 The relations between the numbers obtained by us are rendered most evident by means of curves. We have therefore constructed diagrams to show the influence of dilution and of addition of alcohol on the molecular conductivity and on the degree of dissociation of the diethylammonium chloride in solution.In Fig. 1 the molecular conductivities have been plotted as ordinates ; and for abscissa we have taken v * instead of v in accordance with FIG. 1. previous practice. The curves all slope gently upwards towards the right, the total'ascent being greater in the case of pure water and of 99 per cent. alcohol than with the intermediate solvents. The distances between the various curves show that as increasirgOF DIETHYLAMMONIUM CHLORIDE IN AQUEOUS ALCOHOL.6 7 quantities of alcohol are added the effect of each increment on the conductivity decreases, especially a t the higher dilutions, In Fig. 2 we have as ordinates the percentage dissociation as measured by the ratio 1OOpL,/p,. Here also the curves all slope PIG. 2. upwards to the right, indicating an increase of dissociation with increased dilution; but in this case it will be noticed that the in- fluence of the first quantities of alcohol is comparatively trivial. So much is this the case that the curve for 10 per cent. alcohol had to be omitted, as it could scarcely be shown distinct from that of pure water on the scale of the diagram. The replacement of 10 per cent. of the solvent water by ethylic alcohol is therefore practically without effect on the degree of dissociation of the dissolved salt, the diminution being less than 1 per cent.Arrhenius (Zed. physikd. Chem., 1892, 9, 499) estimated the effect of 10 per cent. of a non-conductor such as alcohol on the degree of dissociation of a good electrolyte in strong solutions to be not greater than 1 per cent., a conclusion in accordance with our direct measurement. The slope of the curves becomes greater as more and more alcohol is added ; that is, between the limits of dilution given (v = 10 to v = 320), there is a much greater change in the dissociation with strongly alcoholic than with aqueous solutions. Results have not been given for dilutions greater than 320 Z., as the diagrams to contain them would have been inconveniently extended and as the curves at the high dilutions approach each other very closely, theoreti- cally to meet on the ordinate 100 when the dilution becomes infinitely s 268 WALKER AED HAIDIBLE’ : ELECTRICAL CONDUCTIVITY great.It should be observed that a t the dilution 320 1. the salt is more than 80 per cent. dissociated even in 99 per cent. alcohol. There is, it is true, a slight uncertainty in fixing the molecular conductivity a t infinite dilution from which the degree of dissociation is calculated, but the error from this source cannot be of great magnitude. Kohlrausch has investigated for aqueous solutions, and Vollmer for alcoholic solutions (compare Ostwald’s Lehhuch, vol. ii, part 1, section ii), the conductivity of a large number of binary salts at dilutions beyond = 8000, and from their results and the general character of our own curves we were able t o conclude with certainty that a very small addition to the molecular conductivity a t SO00 1.mould give the molecular conductivity for v = a. The addenda we used were 0.5 for water, 0.6 for 10 per cent. alcohol, and 0.7 for the other solvents, the relative effect of the addition being greater as the water is replaced by alcohol. Fig. 3 exhibits the effect on the degree of dissociation of substituting alcohol for water. The abscissze (23) are the percentages of alcohol by FIG, 3, volume in the solvent, and the ordinates represent the degree of dissociation. At infinite dilution, the effect is nil, for then the salt is wholly dissociated even in absolute alcohol.At 320 1. the effect is still comparatively small, the dissociation only falling from 95 to 80 per cent. as the solvent changes from water t o alcohol. I n the solutions of greater concentration, the influence is much more marked,O F DIETHY LAMMOXIUM CHLORIDE IS AQUEOCS ALCOHOI,. 69 the fall for 2) = 10 being from '79 to 34; that is, the dissociation is reduced to less than half. The character of each curve shows that the last additions of alcohol have a much greater effect than the first. For example, it is seen from the curve for v = 20 that 80 per cent. of the water must be replaced by alcohol before half the total fall is produced; and so it is also with the other curves. This is especially noteworthy on account of the light it throws on the curves of Fig.4, where molecular conductivities, p, are the ordinates and percentages of alcohol, p the abscissa The molecular conductivity of a substance in solution depends on70 WALKER AND HAMBLY : ELECTRICAL CONDUCTIVITY two variables-the proportion of ions in the solution, and their speed. Now we have seen from Fig. 3 that the degree of dissociation at a given dilution (that is, the proportion of ions present) is diminished as the water is replaced by alcohol. The molecular conductivity mill therefore fall owing to this cause. But the speed of the ions is also reduced by the substitution of alcohol for water, and consequently for this reason also the molecular conductivity will be diminished, This second mode of action of the alcohol is exhibited in the uppermost curve of Fig.4, which represents the molecular conductivity at infinite dilution. Change in the degree of dissociation is here eliminated, for the dissociation is throughout complete. From the shape of the curve, it is evident that the retardation of the ions is practically at an end when the solvent contains 60 per cent. of alcohol. For greater amounts of alcohol, the curve runs in what is approximately a straight line only very slightly inclined to the p-axis." The curves for finite dilutions bear a general resemblance to the curve for infinite dilution, so that we may attribute the chief effect of the alcohol in diminishing the molecular conductivity to the retardation of the ions. Up to p = 60 the curves very closely resemble the curve for v = =, but beyond that point the similarity in great measure ceases.This is owing to the appearance of the second factor -the diminution in the degree of dissociation. From Fig. 3, it may be seen that up to p = 60 the diminution is small, but it then becomes very marked, especially at the smaller dilutions. The consequence is that whilst the substitution of alcohol beyond 60 per cent. scarcely affects the speed of the ions, it now exercises a great effect in diminishing their number, and that the more as the dilution diminishes, with the result that the curves for the smaller dilutions in Fig. 4 bend away from the curve for infinite dilution and become more and more inclined to the p-axis. The curves for the molecular conductivity therefore exhibit contrary flexure : at first they are convex to the p-axis, but at the end they become concave towards it.As we have already stated, the curve for infinite dilution in Fig. 4 gives us a measure of the influence of the alcohol in diminishing the speed of the ions. A general expression which indicates the influence of the alcohol in diminishing the degree of dissociation may be found in the formula of Rudolphi (Zeits. physihcl. Ckm., 1895,17,385). Rudolphi found that for highly dissociated binary electrolytes the expression exhibited satisfactory constancy, the relation between the dilution and degree of dissociation being thus expressed with a single m2 (1 -m) JJU * This is not all brought out in the figure.OF DIETHYLAMMONIUM CHLORIDE IN AQUEOUS ALCOHOL. 71 1 constant.the various solvents and give our results in the following table. We have calculated the value of Rudolphi's expression for 2; = 10 20 40 80 160 320 Mean 0.93 0'93 0.95 0 '99 0 '98 1.06 0.86 0'89 0'90 0'94 0.95 0'94 0.97 0.91 30.7. 0 -72 0.77 0.76 0.74 0-74 0'72 49 '2. 0.51 0-51 0.53 0 -53 0 5 7 0.54 0'74 j 0-53 72.0. 0 *30 0 -30 0 '32 0 *33 0 -35 0'35 0 -33 For p = 90 the values of the expression increase very rapidly with the dilution, for which reason they have not been included in the table. For smaller values of p , the expression is fairly constant, and a mean has in each case been calculated. The divergences from the mean are not greater in the solutions containing alcohol than they are in the pure aqueous solution, and are on the whole of the same order as those given by Rudolphi.An error in the determination of m, when WE = 0.9, is magnified twelvefold in Rudolphi's constant, so that strict constancy cannot be expected. As will be seen, the value of the constant falls as alcohol takes the place of water in the solvent, indicating that the concentration a t which the salt is dissociated t o a given extent falls in like manner. I n alcoholicsolutions, there is not the same close resemblance in the behaviour of electrolytes that we find in aqueous solutions, and no doubt the results that we have obtained for diethylammonium chloride are not representative of the effect of alcohol on all binary electrolytes. Acids and bases diverge very widely in particular cases, so that no conclusions can be drawn as to their behaviour in alcoholic solutions from the known properties of their solutions in water.Salts present less irregular phenomena. We venture, therefore, to think that the curves we have given may be of use in indicating broadly the influence of alcohol on solutions of binary salts with univalent ions, whatever divergences may be found in details. Lenz (cited in Ostwald, Lehrbuch, vol. ii, part 1, 708) observed that in strong solutions of potassium iodide (v = 2 to v = 8) the influ- ence exerted by alcohol on the conductivity was independent of the dilution. In the following table we have calculated the conductivity of our alcoholic solutions in terms of the corresponding aqueous solutions, and so tabulated the results that if the regularity found by72 Lenz applied to the case in hand, the numbers in all the horizontal rows would have the same value. BROWN, MORRIS, AND MILLAR: EXAMINATION OF THE 100 7 7 49 85 29 22 17 p = o 10'1 30 -7 49.2 72.0 90'3 99.0 100 77 49 37 31 25 20 ?_r = 10. 100 77 48 35 28 20 15 20. 1 40. I s 0. 100 77 49 37 32 27 23 320. 100 77 49 38 34 32 30 For small quantities of alcohol, the relativa diminution of the conductivity does indeed appear to be independent of the degree of dilution, but i t is obvious that this regularity does not extend over the whole table. The case of potassium iodide is probably an isolated one, the influence of the different factors in the range of dilution studied being of such a nature as to bring about the observed appearance of uniformity in the action of the alcohol. UNIVERSITY COLLEGE, DUXDEE.