ADSORPTION OF SALTS ON CELLULOSE
195
STUDIES ON T H E LYOTROPIC SERIES. I11 THE ADSORPTION OF SALTSON CELLULOSE E. HEYMANN
AND
G. C. McI I- > IO; > Br- > NOT > C1- > tartrate = CH,COO- > SOT-
197
ADSORPTION OF SALTS ON CELLULOSE
This order is, in general, that anticipated from Kats’ theory (18) and the previous papers on the subject. Thiocyanate ion, iodide ion, bromide ion, and nitrate ion all show strong positive adsorption in the order given. With chloride ion the apparent adsorption is negative a t low concentration,
IO 5 Q
O -5 I.o
0.5 C’
FIG.2. Potaasiuni salts: plot of apparent adsorption (a = e per liter against equilibrium concentration in moles per liter.
- c’) in millimoles
a
0.I
0.3
0.5
C‘ FIG.3. Potassium salts (large-scale): plot of apparent adsorption (a = c in millimoles per liter against equilibrium concentration in moles per liter.
- c’)
but becomes positive as the concentration ia increased. Acetate, tartrate, and sulfate ions show negative apparent adsorption, which is most marked with sulfate, as anticipated. This means that the true adsorption of thiocyanate and iodide ions is appreciable, whereas that of sulfate and acetate ions is very small. The same order is shown by the sodium salts, as far as they have been investigated (figure 4).
198
E. HEYMANN AND 0 . C. MCKILLOP
Adsorption effects are small with the salts of the alkaline-earth metals; nevertheless the order is the same as that found for gelatin, namely
Ba++ > Ca++
> Sr++ > Mg++
Moreover, it is seen that barium thiocyanate is more strongly adsorbed than barium chloride. Magnesium chloride is negatively adsorbed a t low concentrations, but the apparent adiorption becomes positive a t higher ones (figure 5 ) . This behavior is similar to that found by van der Hoeve (17) in his experiments on the adsorption of salts on starch. Magnesium sulfate, like all sulfates, shows negative adsorption a t all concentrations. It should be noted that'many curves have an unusual concave upward
a
a
, 0.5
I
012
04 0.6 0.8
C' FIQ.4
I
C'
FIG.5
-
FIG.4. Sodium salts: plot of apparent adsorption (a = c c') in millimoles per liter against equilibrium concentration in moles per liter. FIQ.5 . Divalent cations: plot of apparent adsorption (a = c c') in millimoles per liter against equilibrium concentration in moles per liter.
-
shape (cf. also the adsorption on methylcellulose), whereas the adsorption curves on gelatin have the usual convex shape with respect to the concentration axis. Copper salts, the chloride and sulfate, were also investigated. They are interesting on account of Schweitzer's reagent, a long-known solvent for cellulose (figure 6). The chloride curve follows the same form as that shown by potassium chloride, first negative and then positive apparent adsorption, while the sulfate is negatively adsorbed a t all concentrations. I n ammoniacal solutions, however, both salts show appreciable positive adsorption, supporting the conclusion of many investigators (e.g., 1, 13, 14,20, 22, 25) that chemical combination occurs between the cellulose and the copper ammonia complex. In the experiments the ratio Cu:NHa was
ADSORPTION OF SALTS ON CELLULOSE
199
kept constant (approximately 1:7), except in the very dilute solutions (0.02 M ) , where a much greater excess of ammonia was necessary in order to prevent hydrolysis. The values were corrected for the amount of cellulose that had gone into solution. The series of experiments in ammoniacal solution was carried out by Lucy F. Kerley. The univalent cations do not differ very much with regard to adsorption on cellulose (figure 7). The curves for the sulfates are almost identical, whereas those for the chlorides show a certain amount of gradation, the series being N H t > K+ > Li+'> Na+
5
a
0
-5 I
02 04 0.6 0.8 C' FIG.7 PIG.6. Copper salts: plot of apparent adsorption (a = c - c') in millimoles 1,c.r liter against equilibrium concentration in moles per liter. FIG.7. Univalent cations: plot of apparent adsorption (e = c - c') in millimoles per liter against equilibrium concentration in moles per liter.
which is obscure from a theoretical viewpoint and does not agree with thc experiments on gelatin. The reversibility of adsorption was tested in a number of cases in which the apparent adsorption is appreciably positive (potassium thiocyanate, potassium bromide). The cotton wool was left in the adsorption vessels in the usual manner for the usual time, and then washed with distilled water till no trace of the respective salt could be detected in the wash water. The cotton wool was dried in the oven and ashed carefully; the ash was weighed as sulfate. The results are given in table 1. It is evident that the adsorption is reversible. This applies, however, only t o the untreated cotton wool, which contains a considerable amount of calcium. The purified cotton wool behaves differently also in this respect (21). Experiments on the adsorptive property of cellulose have been carried
200
E. HESMANN AND G. C. MCKILLOP
out previously (7, 19). Comparison is, however, difficult, because all previous investigators have used cellulose which contained water. Evans used air-dry filter paper, while Kolthoff used, as a standard, filter paper saturated with water vapor.f With undried cellulose the apparent adsorption is higher than with dried; hence it is not surprising that Evans has found no evidence for negative adsorption. A few of our experiments were carried out with undried cellulose, and on correcting for the water content the adsorption was found to agree with that of the dried cellulose. The possible influence of a Donnan effect on the experimental data has been fully discussed in the previous paper. The possibility of a falsification of the experimental results by such an effect is small in the system under investigation, because the bulk of the water imbibed by the cellulose is free. The volume of solution in the immediate neighborhood of the surface of the fibers, the salt concentration of which may be influenced by the existence of a hypothetical cellulose-salt-electrolyte, must be \ * e n
BALl
(Original cellulose)
CONC=NTBbRON
...................
KCNS. ............................... KBr. ................................ CuClr ................................
1
1 1
A B E COJTBNT OF CEI.CDU)BE AFTER WASEINQ
0.171 0.170
0.166
small. Nevertheless, its bearing on the experimental results was tested in two cases in which the adsorption is comparatively high. The two salt solutions, 1.25 M potassium iodide and 1.5 111 potassium thiocyanate, were put through the adsorption process in the usual manner, and as much liquid as possible was recovered from each by gentle pressure with a glass rod. The wet pad of cellulose was then taken from the bottle and put between two highly polished heavy brass plates attached to the two faces of a clamp. High pressure was applied by means of an elongated handle and the pad squeezed as dry as possible, the expressed solution being caught in a glass dish under the clamp. .The solution recovered in the usual way, and that recovered under pressure, showed no difference in concentration on analysis. In the procedure described above, a total of about, 90 per cent of the solution is recovered. Eyen if the concentration of the salt solution in the remaining 10 per cent should differ from the bulk, 1 Filter paper gencrally contains from 7 to 10 per cent moisturc and cotton wool from 8 to 12 per cent, depending on conditions of moisture and temperature. .4t high humidity, thcg may contain up to 20 per cent.
.IDSORPTION O F SALTS ON CELLULOSE
20 1
as a result of n 1)onnrln cffert, the adsorption values would not be materially affected. IY. THE HTDKi'TION O F THE CELLCLOSE
Assuming t1i:it the tiue adsorption of the sulfate is zero, the hydration of the cellulose mag be roughly estimated from the values of negative adsorption in dilute solutions. The percentage adsorption of 0.1 to 0.2 111 solutions of potassium sulfate, sodium sulfate, and lithium sulfate is of the order of - 0 . i to -0.8 per cent, when 6 g. of rellulosc ha1.e come to equilibrium with 100 ml. of solution. From this the minimum valuc for thr hydration of cellulose is estimated as of the order of 0.12 to 0.14 g. to water per gram of cellulose, which is considerably lower than that of methylcellulose (0.25 g. of water per gram of methylcellulose). This agrces with the generally more hydrophilic nature of the latter (cf. 15). V. ADSORPTION AND LYOTROPIC SERIES
The results of this iiwestigation, as well as those of the previous ones, rhow that there is a close connection between the order of adsorption and the lyotropic series as generally found. Anions like thiocyanate ion and iodide ion, which are known for their liquefying properties with regard to all the systems under investigation, show appreciable positive adsorption. I t may be mentioned that cellulose also is soluble in salt solutions, although only a t high concentrations and high temperatures, thc order of effcrtireness being CSS-
> I- > Br- > C1- > SO;-
(12, 23)
which agrees very well with the adsorption series of this investigation. I t is noteworthy that the thiocyanates, which show th(3 strongest adsorption, are among the best peptizing agents for cellulose (5, 23, 24). On the other hand, the sulfate ion, which is a well-known precipitant for gelatin and methylcellulose (cf. l5), shows strong negative ndGorption in all c'ases. Similar conditions are found with the alkaliiie-rarth rations, the order of adsorption agreeing well with the lyotropic scrim, thc only exception being that the position of calcium ion and strontium ion is rewrscd. Here, also, increasing adsorption runs parallel to more pronounrrd liqiirfying properties. With the alkali cations, gradation is not w r y pronoimcrd with regard either to lyotropic properties or adsorption. V I . ADSORPTION AND ENERGY O F HYDRATION
While the relation between lyotropic properties and adsorption can generally be explained on the basis of Iiatz' theory (cf. 18),there is still considerable difficulty in undeigtanding the reasons for the order of adsorption itself. I t has been pointed out in the first paper (16) that the
202
E. HEYhlANN AND G . C. MCKILLOP
smallest adsorption may be expected with ions which have the strongest nffinity for the solvent. Values for the free energy of hydration which, according to Gurney (lo), differ very little (about 2 per cent for univalent ions) from the heat of hydration, are now available for the elementary ions. The calculation of the heat of hydration for individual ions rests on the knowledge of the lattice heat (calculated by the Born-Haber cycle (3, 11)) and the heat of solution of the crystalline salts (Fajans (S)), and on the additional assumption that ions of the same charge and radius have the same heat of hydration (Gurney; cf. also Bernal and Fowler (2)). For all colloids under investigation the adsorption series is
CNS-(?)
> I-(2.1) > Br(2.5) > NO;(?) > C1-(2.8) > SO;-(?)
The values in parentheses are the energies of hydration in electron volts, according to Gurney (10). It is seen that the adsorption decreases with increasing energy of hydration. Although the numerical values for sulfate ion and thiocyanate ion are not known, there is evidence from other source? (surfacc tension and surface potential of aqueous solutions ; cf. Freundlich (9)) that sulfate ion is strongly and thiocyanate ion weakly hydrated. The same applies t o the alkaline-earth ions. Magnesium, possessing the largest heat of hydration (21 electron volts), shows the weakest adsorption, while barium, with the smallest heat of hydration (15 electron volts), exhibits the strongest adsorption. So far, there is qualitative agreement of the adsorption data with the scanty information of the respective energies of hydration. The behavior of the alkali ions is, however, irregular. The series of the heats of hydration is Lif(6.1)
> Ta+(5.0) > 1
