THE HYDROLS WILDER D. BANCROFT
AND
LAWRENCE P. GOULD
Department of Chemistry, Cornell University, Ithaca, N e w York Received October $1, 19%
Liquid water is recognized as being more or less polymerized (7), but very little use is made of this except to account for the temperature of maximum density of water and for its not boiling a t 80-193°K. as calculated by van’t Hoff, Walden, Tammann, and Sidgwick. In many thermodynamical reasonings the assumption is made tacitly that the degree of polymerization does not change. Very few people are interested in the probability of the different forms of liquid water having different properties. This is the more remarkable because we are quite used to marked changes with different metamers and polymers. We have red NO2 and colorless Nz04; soluble and insoluble sulfur; anthracene and dianthracene; red phosphorus and white phosphorus; diamond and graphite; cyanic acid, cyanuric acid, and cyamelid ; acetaldehyde and paraldehyde; oxygen and ozone; ordinary and nascent hydrogen, oxygen, and nitrogen. The enol and keto forms can be distinguished. Polymerized alcohol peptizes pyroxylin while depolymerized alcohol does not. The two forms of chromic chloride and of benzaldoxime each lower the melting point of the other. The dielectric constant of solid nitrobenzene I is nearly four times that of solid nitrobenzene 11. The lyotropic or Hofmeister series of chloride, bromide, and iodide is found to hold both for true solutions and for colloidal solutions. In the case of true solutions the series holds for reaction velocity experiments (neutral salt effect), electromotive forces, boiling points, and displacements of the temperature of maximum density. Sodium chloride increases the dissociation constant (18) of carbonic acid one thousandfold. Sodium chloride increases the apparent strength of boric acid (16). The apparent molecular weights of sodium chloride, bromide, and iodide vary in that order. The potential difference between hydrogen and hydrochloric acid differs from that between hydrogen and hydrobromic acid (26). Richards and Rowe (19) found that the heats of neutralization of caustic soda and caustic potash decrease as the neutralizing acid changes from hydrochloric acid to hydrobromic acid and then to hydriodic acid. The solubility of gases in salt solutions, the heat capacity of water, the displacement of the temperature of maximum density of water, and the quenching of fluorescence in certain solutions by salts show the same general phenomenon. 197 THE J O C R N A L OF PHYSICAL CHESIISTRY, VOL. XXXYIII, X O . 2
198
WILDER D. BANCROFT AND LAWRENCE P. GOULD
The Hofmeister series holds in colloidal solutions for the peptization and precipitation of proteins, cellulose, and many other colloids. We do not at present consider adsorption as occurring in true solutions, though the conception of a water sheath on ions is practically adsorption. So long as we adopt the ordinary point of view, adsorption is not a necessary cause of the Hofmeister series. The Hofmeister series cannot be due to electrostatic effects because nitrates are very different from chlorides and bromides, though coming between them in molecular weights. It cannot be due to hydrated or hydrous ions because the boiling points are more abnormal than the freezing points and also because on that hypothesis hydrous or hydrated iodides should raise the setting points of gelatin jellies and should increase the rate of saponification of esters. The neutral salt effect cannot be due to an increased dissociation of the acid, because it is easy to get effects corresponding to many times 100 per cent dissociation. Many people have suggested an increased dissociation of water, but no one has maintained it seriously for any length of time. Sodium chloride alone does not invert sugar solution a t an appreciable rate. Sodium chloride, bromide, and iodide decrease the rate of saponification of methyl acetate. The only thing that is left is that the Hofmeister series is due in part to a change in the degree of polymerization of water (1). For the cases in which one can eliminate adsorption, the neutral salt effect is due exclusively to a change in the water equilibrium. For other cases selective adsorption may play the important part. With gelatin the order of effectiveness of the anions is the same on both sides of the isoelectric point and the amounts of chloride or iodide adsorbed are so small that nobody has ever measured them in aqueous solutions. With gelatin the change in the water equilibrium is the important one, and the effect due to adsorption of the anion is practically negligible for chlorides, bromides, and iodides, though it does exist. With albumin the order of the effectiveness of the anions reverses as one changes from acid to alkaline solutions. Here the adsorption is the important factor and the displacement of the water equilibrium is of much less importance. Gelatin and albumin may be taken as representing the two extremes. Hofmeister (11) deduced the Hofmeister series from experiments on proteins. Since adsorption is selective, the order of the anions will not necessarily be the same in all cases and colloid chemists know that the Hofmeister series varies somewhat with different adsorbents, nitrate jumping around in an especially disturbing way. It has not yet been shown that all these cases are due to differences in selective adsorption, but that will come in time. We have already proved it for the two cases that we have studied. The boiling point curve for sodium thiocyanate is more nearly normal than that for sodium iodide, whereas sodium thiocyanate peptizes gelatin more strongly than does sodium iodide. By using 80 per cent and 95 per cent
THE HYDROLS
199
alcohol it was possible to show that sodium thiocyanate is adsorbed more strongly than sodium iodide. A similar explanation holds for sodium nitrate and sodium chloride. The nitrate has the more normal boiling point curve, peptizes gelatin a little better, and is adsorbed by gelatin more strongly. The usual order of the anions in the Hofmeister series when adsorption is negligible is I-, SCN-, Br-, NOa-, C1-, 504--. When selective adsorption occurs, a common order is SCN-, I-, Br-, NO3-, C1-, SO,--. We have introduced a set of arbitrary assumptions for the effects of these six ions on water. This will make the discussion more definite and should be helpful. 1. Under ordinary conditions liquid water may be considered as essentially a mixture of trihydrol (or polyhydrols), dihydrol, and monohydrol coexisting in reversible equilibrium. 2. Chloride, bromide, thiocyanate, and iodide ions tend to change dihydrol into monohydrol and have no especial effect on trihydrol beyond that which is a necessary consequence of the displacement of the equilibrium between dihydrol and monohydrol. The order of increasing effectiveness is chloride, bromide, thiocyanate, and iodide. 3. The nitrate ion tends to convert trihydrol into dihydrol, and dihydrol into monohydrol. For equivalent concentrations the amount of monohydro1 is less with nitrate ion than with chloride ion. 4. The sulfate ion tends to convert trihydrol and monohydrol into dihydrol. 5 . The electrolytic solution pressure of hydrogen is greater in dihydrol than in monohydrol, and that of oxygen is less. Many years ago Kahlenberg (14) showed that potassium iodide gave a greater rise of boiling point than potassium bromide, and potassium bromide a greater rise than potassium chloride. He also showed that these results were distinctly more abnormal than those with potassium nitrate. We have confirmed and extended his results. The experiments were made a t 760 mm. with the arrangement of the Davis-Brandt apparatus (8), which permits working a t constant pressure irrespective of fluctuations in barometric pressure. The data obtained with this apparatus are given in table 1. The concentration of the solutions is expressed in moles of the solute per kilogram of water. By plotting molal concentrations against rise of boiling points, we eliminate all question of the applicability of any particular formula. One reason why Kahlenberg’s results were ignored was because he calculated molecular weights and people said that the formula did not hold for such concentrations. The solutions of sodium chloride, sodium bromide, sodium iodide, and sodium thiocyanate were titrated with standard silver nitrate solution, using potassium chromate as an indicator. The solutions of sodium nitrate were analyzed gas-
200
WILDER D. BANCROFT AND LAWRENCE P. GOULD
volumetrically with a Dennis nitrometer. The sodium sulfate solutions were determined by precipitating and weighing as barium sulfate. TABLE 1 Boiling point data at 760 mm. MOLES P E R KILOGRAM OF
SALT
lNaCl.
. \
/
NaBr , . . . . . . . . . . . . , . . . . . . . . . , , . . , . . ,
(1[I
NaI . . . . . . . . .
NaNOa , . . . . . . . . . . . . . . . . . . . . . . . . . . , , I
NaSCN. . . . . . . . . . . .
Na2SOc,. , , . . . . . . . . , . . . . . . . . . . . . . .
1 .(
\
RISE O F BOILING POINT
BOILING POINT R I S E P E R MOLE
WATER
degrees C.
degrees C .
4.404 2.070 1.456 0.889 0.382
5.16 2.04 1.38 0.79 0.34
1.17 0,986 0.948 0,888 0,895
3.896 1.932 0.687 0.272
4.73 2.01 0.63 0.26
1.21 1.04 0.918 0.954
3.875 2.398 2.043 1.910 1.010 0.588
5.15 2.73 2.21 2.04 1.01 0.55
1.33 1.14 1.08 1.C7 1.00 0.935
4.04 3.01 1.81 0,926
3.65 2.70 1.69 0.85 0.53
0.903 0.897 0.935 0,918 0.970
4.28 3.01
5.25 3.38 2.07 0.92 0.73
1.22 1.12 1.06 0.980 0.974
2.70 2.09 1.37 0.892 0.476 0.232
2.55 1.99 1.34 0.90 0.50 0.26
0.945 0.952 1.01 1.01 1.05 1.12
In dilute solutions the results for all salts fall practically on the same line, as they should. I n more concentrated solutions the curves diverge. These results do not check those of Johnston (13) for the sodium bromide
THE HYDROLS
20 1
solutions. His error is probably due to superheating, giving too high a boiling point. There is a very real difference between such similar salts as sodium. chloride and sodium iodide. At about 4 molal the rise of boiling point for sodium chloride is about 85 per cent of that for sodium iodide. This is a difference of about 0.7"C., which is far beyond the estimated experimental error of 0.03"C. There cannot be any such difference in dissociation. The difference cannot be due to adsorption in the ordinary sense of the word, because the salts are in true solution. Even a t 3 normal the difference between sodium iodide and sodium nitrate is about 1°C. Sidgwick and Ewbank (22) deduce from freezing point data that the lowering of the freezing point varies with the hydration of the ions. A similar conclusion has been drawn by Fajans and Karagunis (9). The difficulty with this is that in general the boiling points are more abnormal than the freezing points, whereas the degree of hydration must be greater at 0" than a t 1OO"C., because the formation of hydrates is an exothermal process. Arbitrary assumption No. 2 was based on the boiling point data. Iodides have the highest boiling points and then come thiocyanates, bromides, and chlorides. The differences between sodium thiocyanate and sodium bromide are very small. If iodides are the most effective in converting dihydrol into monohydrol, they are more soluble in monohydrol than in dihydrol and are consequently lowering the vapor pressure of the more volatile constituent. Therefore the rise of boiling point is abnormally large. The temperature of maximum density of water is lowered by most salts. That means that these salts depolymerize water to some extent, thereby making it necessary to cool more before the expansion due to increased polymerization overbalances the contraction due to temperature contraction. On this basis iodide should have more effect than bromide and bromide more effect than chloride. This is the case. Unfortunately nitrate is more effective than iodide and sulfate more so than nitrate. Arbitrary assumptions Nos. 3 and 4 take care of the apparent exceptions. Arbitrary assumption No. 5 is made to account for the apparent increase in concentration of hydrogen ion on addition of sodium chloride, bromide, or iodide to a solution of an acid. There can be no such change in concentration as there appears to be, and consequently the observed change in electromotive force must be due to a change in the so-called solution pressure. It would be much better if people would evaluate singly as many of the disturbing factors as possible instead of lumping them all under the purely formal term of activity. As has been stated, the boiling points for sodium iodide are more abnormal than those for sodium thiocyanate, and the boiling points for sodium
202
WILDER D. BANCROFT AND LAWRENCE P. GOULD
chloride are more abnormal than those for sodium nitrate. On the other hand, sodium thiocyanate is a better peptizing agent for gelatin than sodium iodide, and sodium nitrate than sodium chloride. It was therefore necessary to show that gelatin adsorbs thiocyanate somewhat more Strongly than sodium iodide, and sodium nitrate more strongly than sodium chloride. Previous experiments at Cornell and elsewhere had shown the practical impossibility of getting the necessary data with water as solvent. The experiments were therefore made in alcoholic solutions in which the peptization of gelatin and the adsorption of alcohol by gelatin are practically negligible. TABLE 2 Adsorption by gelatin SALT
SOLYEN'I
MOLE0 P E R LITER I N SOLUTION
MILLIMOLES ADSORBED P E R GRAM OF G E L A T I N
NaSCN
95 per cent alcohol
0.0592 0.040 0.0173 0.0059
0.171 0.123 0.059 0.020
NaI
95 per cent alcohol
0.0805 0.0504 0.0240
O.OO86
0.193 0.139 0.063 0.030
NaCl
80 per cent alcohol
0.0554 0.0410 0.0236 0.0060
0.291 0.215 0.115 0.023
NaN03
80 per cent alcohol
0.0558 0.0385 0.0225 0.0058
0.303 0.212 0.121 0.028
Three grams of Eastman's de-ashed gelatin was used as obtained, adding 35 cc. of the alcoholic solutions. Sodium thiocyanate and sodium iodide were dissolved in 95 per cent alcohol; since sodium nitrate and sodium chloride are less soluble, 80 per cent alcohol was used for them. Trial runs showed that it was necessary t o wait thirty-six hours in order to be sure of reaching equilibrium. At the end of that time 25 cc. of the supernatant liquid was titrated with a known silver nitrate solution, using potassium chromate as indicator. The data are given in table 2. The results show that sodium thiocyanate is adsorbed more strongly by gelatin than sodium iodide is, and that sodium nitrate is adsorbed more
THE HYDROLS
203
strongly than sodium chloride. There is therefore no necessary reason why the order of peptization should be identical with the order of boiling points. Bowe ( 5 ) has used the theory of the water equilibrium to explain the neutral salt effect. If neutral salts cause a shift in the water equilibrium, we then have different solvents in which solubilities, reaction velocities, and degrees of ionization may be different from those in the original solvent. He showed that the equation for the electromotive forces postulates the constancy of the solution pressure of hydrogen. This is not true experimentally for hydrochloric acid, hydrobromic acid, and hydriodic acid. Bowe showed that the addition of sodium chloride, sodium bromide, and sodium iodide to 0.1 N hydrochloric acid increased the apparent concentration of hydrogen ion in each case, the increase being greatest for sodium iodide and least for sodium chloride. Making the solution 4 normal with respect to sodium iodide increased the apparent concentration of hydrogen ion tenfold. To account for the facts we postulate that the solution pressure of hydrogen is less in monohydrol than in dihydrol. All determinations of hydrogen-ion concentration in halide solutions are therefore in error to some extent. Since all determinations of pH in biological systems are made in the presence of chlorides, all these determinations are in error to some extent, unless some other unsuspected error counteracted the first one. These experiments suggested an even more striking one. The cell Hz j HC1 I KC1 -HgzCIz I Hg should have a higher electromotive force than the cell Hz1 HBr I KC1 HgzClz -- I Hg, provided the molal concentrations of hydrochloric acid and hydrobromic acid are the same. The corresponding cell with hydriodic acid should have the lowest electromotive force of the three. The electromotive forces of these three cells were measured, using a saturated calomel half-cell. After the electromotive forces had been determined, the solutions were analyzed volumetrically by titration with silver nitrate, using potassium chromate as indicator. It was not possible to run up to high concentrations of hydriodic acid, because under ordinary conditions iodine is set free and poisons the electrode. The results are given in table 3. With strong acids of the same type, such as these, the percentage dissociation should not differ appreciably. The migration velocities are practically identical. At 8 normal concentration the difference between hydrochloric acid and hydrobromic acid is about 40 millivolts, which is far beyond the experimental error and which must therefore be due to a difference in solution pressure. These results agree with those of Wilke and Schrankler (26), who measured the cell Hz ] HCI [ HX I Hz.
204
WILDER D. BANCROFT AND LAWRENCE P. GOULD
From a consideration of the boiling points of salt solutions and from the displacement of the water equilibrium by salts, we deduced (arbitrary assumption No. 4)that the sulfate ion tends to convert trihydrol and monohydro1 into dihydrol. I n this reaction heat is evolved, which accounts for the fact that the solubility curve for anhydrous sodium sulfate shows a decreasing solubility with rising temperature along the first part of this curve. The heat of formation of sodium sulfate decahydrate is sufficient to change the sign of the heat of solution, and consequently this salt shows an increasing solubility with rising temperature. At higher temperatures the effect of sodium sulfate in displacing the water equilibrium will probably become less, and therefore there may come a temperature a t which this heat effect will just equal the algebraic sum of the heat of fusion plus the heat of dilution not due to displacement of the water equilibrium. At this temperature anhydrous sodium sulfate will have a minimum solubility. Experimentally the solubility curve for anhydrous sodium sulfate does TABLE 3 Electromotive forces Hz I HX I I g ,HgzCla I Hg NORMALITY
ACID
--___
'
HC1.. . . . . . . . . . . . . . . HI.................
(
1.148 2.900 4.636 6.415 8.136 0.800
1.430
E.M.F.
____
0.237 0,215 0.197 0.181 0.162 0,240 0.217
NOR-
ACID
E.M.F
MALITY
'
H B r . . . . . . . . . . . . .,