. .‘
O N THE R E L A T I V E VELOCITIES O F THE IONS I N SOLUTIOKS O F S I L V E R N I T R A T E IN PYRIDINE AND ACETONITRILE’ BY HERMAN SCHLUNDT
Although the velocities of the ions have been extensively investigated in aqueous solutions, the relative ionic velocities in non-aqueous solutions is a subject that has been given but limited study. Hittorf made some determinations of transference numbers in alcoholic solutions. H e investigated solutions of zinc chloride, zinc iodide, and cadmium iodide in absolute ethyl alcohol, and alsomade a determination with cadmium iodide dissolved in amyl alcohol. Lenz3 also worked with solutions of cadmium iodide in ethyl alcohol, and extended his investigations to solutions of this salt in various mixtures of alcohol and water. Determinations of transference numbers with solutions of potassium iodide and potassium chromate in mixtures of alcohol and water were also made by Lenz. More recently Campetti4 studied the influence of the solvent upon the migration velocities of the ions. He investigated solutions of lithium chloride and silver nitrate in water, methyl alcohol, and ethyl alcohol, and found that the solvent often exercises considerable influence upon the migration velocities. T h e results of Hittorf and Lenz likewise show this. Mather’s5 determination of the relative velocities of the ions in a tenth-normal solution of silver nitrate in ethyl alcohol concludes the list of investigations on transference numbers in non-aqueous solutions. This paper was presented before the \Visconsin Academy of Letters, Sciences and Arts, in a preliminary form, December 27, 1901. Pogg. Ann. 106,551 (1859). Also Ostwald’s Klassiker der exacten Wissenschaften No, 23, p. 105. M6m. Ac. Imp. St. Petersb. 30, No. 29, p. 34 ( I S S Z ) . Also Ostwald. Lehrbuch d. allgemeinen Chemie, 11 Auflage, p. 6 1 8 . Nuovo Cimento, 35, 226 (1894). Reference. Zeit. ph>s. Chem. 16,165 ( 1895). Am. Chem. Jour. 26,473 (190;).
I 60
Hernza IZ SchZu ndt
Apparently no attempts have heretofore been made to determine transference numbers in solvents other than water and the alcohols. This work was accordingly undertaken with the view of determining transference numbers in other solvents that yield solutions which conduct fairly well and at the same time to ascertain, in a general way, the effect of increasing dilution upon the relative velocities of the ions. For several reasons solutions of silver nitrate in pyridine and methyl cyanide were chosen for this investigation. These solutions conduct well, and the volumetric estimation of sil\-er by Volhard’s excellent inethod is rapid and accurate. T o rile the silver nitrate solutions in pyridine appeared of special interest. Pyridine, as is well known, forins definite compounds with silver nitrate., i. e., silver nitrate crystallizes out with pyridine of crystallization froin its solutions forming the compounds AgN03.2C5H5N and AgK03.3C-H5N. In perusing Hittorf’s classical researches on migration velocities, i t was noted that the relative velocity of the cation increased with increasing dilution, as a rule, in solutions of such salts as show a marked affinity for water, i. e., salts which, in the anhydrons state, are strongly hygroscopic, and which, for the iiiost part, crystallize out with water of crystallization. Copper‘ sulphate and ferric chloride are good examples. On the other hand, salts showing relatively weak affinity for water, yield transference numbers for the cation, which vary but little, or decrease with increasing dilution. Potassiini chloride and silver nitrate will serve as examples. Since pyridine and silver nitrate have a strong mutual affinity, we should expect the transference number of the cation in solutions of silver nitrate in pyridine to increase with increasing dilution, analogous to the behavior of solutions of salts which have a marked affinity for water. T h e experimental results wliich follow will skow this to be the case. Apparatus and method
Several preliminary tests were made with various simple forms of apparatus, such as had been successfully used by other
P
Velocilz'es of the Ioizs
161
investigators in their researches on ionic velocities in aqueous solutions. A tenth-normal solution of silver nitrate in pyridine served for this preliminary work. These tests showed that such forms of apparatus in which the solution remains in contact with rubber stoppers or rubber connections during the test were not well suited owing to the reducing effect of the rubber. T h e portion of the solution near the rubber stoppers gradually becomes turbid and, upon standing, finely-divided silver settles. T h e solvent effect of pyridine upon rubber is also harmful. T h e form of apparatus devised by Loeb and Nernst' is free from this objection, and since good results were obtained with it, this apparatus was used for most of the subsequent tests. Two sizes of this apparatus were used, -one of about 30 cc capacity for the concentrated solutions electrolyzed, and another of about 60 cc capacity for the more dilute solutions, -tenth- and fortiethnorm al. A disk of Kahlbaum's pure silver served as anode, and a bar of silver as cathode. For further details regarding the apparatus the reader is referred to the article of Nernst and Loeb. T h e apparatus was connected in series with a silver voltameter, a Weston milliammeter, and a resistance for regulating the current strength. A set of twelve storage batteries or a I I O volt dynamo furnished the current. During electrolysis the apparatus was kept in a large reservoir of water in which the temperature did not vary more than half a degree as indicated by a thermometer strapped to the apparatus. T h e amount of solution used in the several tests was determ'ined by weighing the apparatus with fittings, then introducing the solution, and weighing again. T h e weighings were made to the nearest milligram for solutions stronger than tenthnormal, and to the nearest centigram for the tenth-normal and more dilute solutions. Before beginning electrolysis the cell was placed in the bath for about fifteen minutes to equalize temperatures. After electrolysis the solution was transferred to three tared flasks proZeit. phys. Chem.
2,
948 (1888).
I6 2
Nermun Sclzlundt
vided with well fitting stoppers, and weighings of each made. T h e differencesgave the quantities of solution taken for the anode portion, the unchanged middle layer, and the cathode portion respectively, exclusive of the residue remaining in the apparatus. T h e weight of the residue was obtained by weighing the apparatus containing it and subtracting froin the weight thus obtained the weight of the complete, empty apparatus previously ascertained. T h e weight of the residue plus the weight of the cathode portion in the weighing flask gave the total weight of the portion of the solution about the cathode. T h e sum of the weights of the three samples and the residue should equal the amount of solution originally weighed out in the apparatus. T h e check was never perfect, the sum being several milligrams less, even for solutions electrolyzed at o o , and in the solutions transferredat room temperatures the differences amounted to a few centigrams. This loss, which was evidently due to evaporation during the transfer of the solution from the apparatus to the weighing flasks, was divided in proportion to the weights of the respective samples, and each sample was allotted its share of the loss. T h e samples were then analyzed. T h e middle layer, which should show no change in concentration, was subjected to analysis first. When it differed more than one percent from the original solution in strength, the experiment was entirely rejected, and no further analyses were made. I n cases where the change in concentration of the middle layer (which is comparatively small) was one percent or less, it was considered a part of the anode portion of the solution, when stronger than the original, and a part of the cathode solution when weaker. In the several experiments the middle layer generally showed no change in concentration. T h e duration of the electrolysis, the current strength employed, the temperature, etc., are noted in the tabulated statement of results. Preparation and analysis of solutions T h e silver nitrate used was of the c. p. variety. T o obtain
it perfectly dry and neutral, it was fused. After powdering, it was kept dry in a desiccator. T h e pyridine used was twice dehydrated with fused caustic potash and distilled. T h e sample used distilled between 1 1 4 ~ - 1 1 under 6~ a pressure of 742 mm. I am indebted to Prof. Kahlenberg for the sample of acetonitrile used. It had been carefully dehydrated by him with phosphorus pentoxide, and distilled from phosphorus pentoxide. Its boiling-point was 81.0' under a pressure of 739 mm. T h e various solutions electrolyzed were prepared the day before usiiig. m7hen silver nitrate is dissolved in methyl cyanide a slight reddish-brown precipitate was obtained, which, when filtered off and dissolved in nitric acid, shows the presence of silver, but the amount of precipitate was too small to warrant * further investigation. T h e strength of the solutions was determined volumetrically by Volhard's method, using standard solutions of ammonium sulphocyanate. Different strengths of ammonium sulphocyanate solutions were employed. They were standardized by titrating weighed quantities of metallic silver dissolved in nitric acid, and also by titration of weighed samples of silver nitrate in aqueous solution. Comparison of the solutions generally employed fortieth- and fiftieth-normal -with a hundredth-normal silver nitrate solution, and titration of the silver deposits obtained in the voltameter during electrolysis, showed that the cyanate solutions remained of the same strength for a long time. T h e solutions of silver nitrate in acetonitrile were analyzed by distilling off the nitrile from a weighed sample, as completely as possible, on a water-bath.' T h e residue was diluted with water and titrated. Whenever the so-called middle layer differed from the original solution in strength by one percent or more, a second determination of the strength of the original solution was made. Generally the strength of the middle layer checked well with that of the original solution. T o guard against mistakes in the observation of the end of the reaction, the scheme of Not all the acetonitrile distils at this temperature, a syrupy liquid remaining.
Nkrnst and Loeb was,followed. One cc of the ~ Z / I O O AgNO, solution was added after the titration was believed to be cotnplete, and cyanate solution again added till the color of the indicator appeared. In the analysis of the solutions of silver nitrate in pyridine a somewhat different course had to be pursued. During the distillation of these solutions some decomposition of the -pyridine takes place, and violent decomposition -combustion -occurs toward the end of the operation. These decomposition, products color the aqueous solution of the residue to such a degree as to interfere somewhat with the sharpness of the end reaction in titration. But the silver may be accurately determined by adding a n excess of nitric acid to the original pyridine solution, diluting with water, and titraking as in aqueons solutions. T h e pyridine nitrate in the dilute solutfons does not interfere in the least with the titration, as the following comparative tests show. Five samples of silver nitrate were weighed out ; three of these were dissolved in pyridine and two in water. T h e solution of ammonium thiocyanate used in these titrations was approximately twenty-fifth-normal. T h e results obtained are recorded in the following table : TABLE I. NH,CNS required
i
-I
-
.____
NH,CKS per gram AgNO,
28.98 29.12
T h e pyridine was recovered by concentrating the solution of pyridine nitrate, and then adding a large excess of crude caustic potash. T h e alkali layer was removed by means of a separatory funnel, and the pyridine purified as previously described. In analyzing the three portions into which the original solution in the apparatus was divided after electrolysis, it was not deemed necessary to take the entire samples for the analysis,
'65
IfeZocities of the lons
in the case of most of the solutions of silver nitrate in pyridine. Since the middle layer is comparatively small, the whole of it was generally titrated. T h e anode and cathode solutions were weighed in tared flasks to get the weight of the total solution of each. In working with the normal and half-normal solutions, a 2 cc sample was removed from the weighing flask containing the anode solution, and after weighing the flask with its remaining contents, a second sample of two cc was removed, and another weighing made. T h e procedure with the cathode solution was the sanie. T h e two pairs of samples thus obtained were titrated as before described, and from the results obtained the number of cubic centinieters of ammonium sulphocyanate required for the total solution was calculated. With the tenthnormal solutions one sample of I O cc was taken from each, the anode and cathode solutions, and its weight and strength determined as before. With the fortieth-normal solutions the entire solutions were used for the analysis. T h e method of calculating the transference numbers and some other details will appear in the following example.
An example T h e relative velocities of the ions were calculated by Hittorf's well-known method.' T h e following example illustrates the inetliod. An approximately tenth-normal solution of silver nitrate in pyridine was prepared, and its strength accurately determined by titrating two samples : Sample
I I 2
NH,CNS required
Weight
~
9.923 10.580
I
1
g. solution required
I
I
40.70cc 43.38
4.101 CC NH,CNS 4.100 " ( '
One cc of the NHICNS solution* used is equivalent to 4.191 mg of A g N 0 3 or 2.6624 m g Ag. From these data the strength of the original solution was calculated as follows : Pogg. Ann. 98, 6 (1856). Also Ostwald's Klassiker No. It is about fortieth-normal.
21,
p. 3s.
Her man SchZu ndt
I 66
The quantity of AgNO, in
g of the solution = 4.1 X 4.191 = 17.18 nig. The quantity of Ag in I g of the solution = 4.1 X 2.6624 = 10.92 mg. Hence the amount of pyridine in I g of solution = I - 0.017182 098282 g. And the silver with
I
I
IO
92
gram of pyridine = A = I I . I I mg. 0.98282
T h e amount of the above solution taken for electrolysis was T h e solution was electrolyzed for 240 minutes, a t o o , with a current of 6 milliamperes as indicated by the ammeter, the potential difference between the electrodes being 39 volts. T h e silver deposited in the voltameter amounted to 96.4 mg. The anode portion of the solution weighed 19.710 g The middle layer of the solution weighed 5.701 g The cathode portion of the solution weighed 37.836 g The residue in the apparatus weighed 0.460 g
63.715 grams.
~-
Total weight 63.707 g T h e loss is so small that it is negligible. T h e following tabulated data show (I) the amounts of each portion taken for analysis, (2) the number of cubic centimeters of NH4CNS required, (3) the number of cubic centimeters calculated for the entire sample, and (4) in the last colamn the actual number of cubic centimeters of KH4CNS used in titrating the entire samples. I
Solution
I
Sample
Middle 5.701 g Anode 1 1 0 . 1 2 0 Cathode 9.891 Residue I 0.460
I ,
NH,CNS required
NH,CNS per gram
i I
1
Total NH,CNS calculated
Total' NH,CNS found
I
~
I
2 3 . 3 3 ~ ~4.O9CC j.182 52.45 3.568 35.28 1-15
I
23.33CC
' 102.14
' 13j 00
- I
1.15
23.33CC 102. I O
134.85 1.15
T h e whole amount of solution taken when multiplied by the NH4CNS equivalent of one gram gives a check on the
-
17
I ~
18
~
19
20
21
-
I 22
,
Cathode portion of solution
A g migrated Contains Ag conin tent in nig. original Solvent Ag ~ _ _ _
14.865
1.6802
I .5660
-
-
-
55.7 -
8.236 1.361 8.195 1.321
16.875 0.8646 0.9509 16.877 0.8392 0.9512
44.5 j6.8
6.185 0. 502 8.296 0.570 1.22 0.1278 1.75 0 . I $8j
35.683 37.726 41.09 41.60
0.3189 0.3963 0.3619 0.4191 0.0812 0.1136 0.0943 i 0. I I52
48.8
__
--
23
-
-
39.2
25.4 16.8
24
Cation
I z anode in cathode
32.6 34.1 34.3 33.5 40.4 38.7 40.2 43,2
~
~
32.7 -
34.0 33.6
-
44.7
38.7 40.6 43.9 44.5
38.5 42.4 45.2
38.2 42.0 44.5
~
I
4,434 2.494 4,470 0,709 2 , 2 0 2 0.574 4 * 670 0.140
11.940 13.761 31.628 34.530
I
1.5840 ' 1.7281 0.4506 0.5233 0.3644 ~0:432g 0.0887510.1290 I
89.1
52.6 55.0
36. I j
I 68
Herman Schlundt
readings observed on the ammeter used, which was graduated to milliamperes, and must be taken to indicate the approxiinate current strength.' Column 5 gives the potential differences between the electrodes. These were measured by closing for an instant a shunt circuit containing tlie voltmeter. Columns I 5 and Z T give the amounts of silver in the silver nitrate which, before electrolysis of the solution, was dissolved in the quantities of solvent indicated in coluiiins 13 and 19. These figures are obtained by multiplying the number of grams of solvent by the amount of silver with one gram of solvent found in column IO. T h e last two colunins- 23 and 24-give the transference number of the cation as calculated from the portions of the solution about tlie anode and cathode respectively. I n the experiments numbered 2 and 5 an apparatus resembling one of Bein's simpler forms2 was used. A fortieth-normal solution of thiocyanate was used for most of the titrations; but for the analysis of tlie dilute solutions employed in experiiiients 8 and 13,a hundredth-normal solution oE thiocyanate was wed. I n experiment I O - normal silver nitrate in acetonitrile - the analyses of the solutions were made with a tenth-1iorii:al solution of thiocyanate. Table 111. gives tlie relative velocities of the silver ion, a t TABLE 111. Transference number of the cation multiplied by IOO Volume in liters containing one gram-molecule of AgNO,, Solvent I
2
1
4
~ . I O
16
35
Water 50.0 Acetonitrile 38 3 P y ri di ti e 32.6 Methyl alcohol Ethy! alcohol different concentrations in solutions of silver nitrate in water, methyl alcohol, ethyl alcohol, acetonitrile, and pyridine. T h e 1 From the amount of silver deposited in t h e voltameter the exact current strength in amperes could of course be deduced. See Fig. 5 . Zeit. phys. Chern. 27, 25 (1895).
Velocities of the ZOIZS
169
trarisfereiice numbers in aqneous solutions are taken from the work of Hittorf, and those in the alcoholic solutions are from the determinations of Canipettil and Mather". Carnpctti's original article was not available, and as none of the various abstracts of his paper mention the concentration a t which he worked, the position of his results in the table is uncertain. Concluding remarks From Table 111. it appears that there is coiisiderable difference in the relative velocities of the ions, in solutions of silver nitrate of the same concentration, in different solvents. T h e results also indicate that with increasing dilution the relative velocities apparently converge to the same value, about 0.47 for the cation, in the different solvents. But the experimental evidence at hand on this point is still too meager to make the general statement that with increasing dilution the relative velocities of the ions of a substance dissolved in different solvents approach the same value. In aqueous solutions of silver nitrate we see that the velocity of the cation decreases with increasing dilution, while in the solutions of pyridine and acetonitrile the opposite holds true. In this respect, these non-aqueous solutions of silver nitrate behave like certain salts in aqueous solutions, copper sulphate, for example. Moreover, when the transference numbers of the various salts studied in aqueous solutions at different concentrations are classified, it appears that those salts which show a iiiarked affinity for the solvent, as a rule, yield transference numbers for the cation which increase with increasing dilution ; and again, solutions of salts which show but slight variation, or a decrease in velocity of the cation with increasing dilution, have a relatively slight affinity for the solvent. Silver nitrate has a pronounced affinity for acetonitrile and pyridine, the rise in temperature being especially marked when silver nitrate is dissolved in pyridine.3 From the analogous bel
Campetti. abstract Zeit. phys. Chem. 16, 165 (1895). Mather. Am. Chem. Jour. 26,473 (1901 ), The thermal effects are nuw being investigated in this laboratory.
havior of these solutions with respect to the variation of the transference numbers with the dilution to aqueous solutions of salts which show a marked affinity for water, and from the contrast of behavior of silver nitrate solutions in water on the one hand, and in pyridine,and acetonitrile on the other, it appears that the affinity of the solvent for the dissolved substance must receive due consideration in any explanation of these phenomena. These variations of the transference numbers with the dilution may be explained by assuming the presence of various complex ions in the solutions ; thus, for example, A. A. Noyes’ cites the case of barium chloride,-a salt in which the velocity of the cation increases with increasing dilution, -and states in explanation of this change in velocity with the dilution that it is necessary to assume that there are present a considerable quantity of complex negative ions formed by the union of one or more chlorine ions with one or more chloride molecules, e. g. BaC1’3 or BaC1”4, and that these dissociate with increasing dilution. For the opposite behavior of the cation, i. e., decrease in relative velocity with increase of dilution, as is the case of aqueous solutions of silver nitrate, we must on this basis assume the existence of complex positive ions in the solution, and that these dissociate with increasing dilution. For solutions of such salts as show little or no change in the velocities of the ions with the dilution, we should have to say that but few complex ions exist ; or, assunling their existence, the number of each kind, positive and negative, is about the same, and that they dissociate at about the same rate with increasing dilution. To explain the opposite behavior of the solutions of silver nitrate pointed out above, in different solvents, on this basis, it is necessary to assume the existence of complex positive ions in aqueous solutions, and complex negative ions in the pyridine and acetonitrile solutions. But the assumed existence of these various complex, free ions in the different solutions of silver nitrate under consideration must be ascribed to the influence of the solvent. I n fact the results in Table 111. illustrate well the Jour. Am. Chem. SOC. 23,37 (1901).
Velocities of the Ions
171
important r61e of the solvent in ionization phenomena. Moreover in view of the parallelism here pointed out between the affinity of solvent and dissolved substance and the change of the transference numbers of the ions with the dilution, it seems to me that the true explanation for these variations in the ionic velocities lies in a complex union of solvent and dissolved substance, and not in the assumption of several kinds of complex, free ions of the solute. During the progress of this work Professor Kahlenberg offered helpful suggestions for which I desire to express my thanks. Laboratory of Physical Chemistry, Universily of Wisconsin, Madison, Wis., Feb. r4, zgor.
