Go to National Library of New Zealand Te Puna Mātauranga o Aotearoa
Volume 77, 1948-49
This text is also available in PDF
(604 KB) Opens in new window
– 1 –

The Constitution of Cobalt Chloride in Aqueous Solution.

[Received by Editor, November 1, 1946; issued separately, April, 1948.]


Vapour pressure measurements have been made on solutions of cobalt chloride and cobalt nitrate. Measurements have also been made on solutions containing added lithium chloride and calcium chloride as well as cobalt chloride under such conditions that the rose colour of cobalt chloride is changed to blue. It is shown that the amount of blue compound formed under these conditions is small. The constitution of the blue compound has been studied by spectrophotometric measurements and the blue colour shown to be due to the formation of undissociated cobalt chloride molecules accompanied by a decrease in hydration. The mechanism suggested is:

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

(Co 6H2O) ++ + 2Cl − ⇌ (CoCl2.4H2O) + 2H2O


Whilst there are a number of complex ions whose constitution is well defined, such as the ferrocyanide ion, [Fe(CN)6], and the triethylenediamino-zinc ion, [Zn3(NH2CH2, CH2NH2)]++, there are a number of cases where neither the structure nor, indeed, even the existence of complex ions is at all authenticated. For example, cobaltous chloride dissolves in water to form a solution which is rose-coloured at low temperature; provided that the solution is not too dilute, it becomes violet or blue on heating and the rose colour can be restored either by cooling or by dilution at the higher temperature. The blue colour can be produced even at room temperature by the addition of high concentrations of other chlorides; a simple but effective lecture experiment consists of adding a small amount of concentrated hydrochloric acid to a cobalt chloride solution, when the colour change from rose to blue is demonstrated in a convincing manner. Two theories have been advanced to account for this colour change, one of which postulates the formation of a complex cobaltanion:

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

2CoCl2 ⇌ Co [CoCl4] ⇌ Co++ + CoCl4 =

or possibly:

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

3CoCl2 ⇌ Co [CoCl3]2 ⇌ Co++ + 2CoCl3

It is further postulated that the complex ion, CoCl4 or CoCl3, is blue whilst the Co++ ion derived from the simple CoCl2 molecule is rose coloured. Temperature and concentration favour the formation of the complex ion, which is also assisted by the addition of chloride ions from other electrolytes such as hydrochloric acid, calcium chloride or lithium chloride. The experimental fact that zinc chloride does not produce the colour change is ascribed to the greater power of this salt to form its own complex ion, ZnCl4=, so that a concentrated zinc chloride solution contains only a negligible number of chloride ions.

– 2 –

On the other hand, it may not be necessary to postulate any complex-ion formation, since there is available an alternative theory of hydration according to which the rose colour is due to a heavily hydrated ion, such as [Co6H2O]++, which can undergo dehydration with change in colour.

A number of experiments have now been made to investigate the cause of this colour change, which lead to the conclusion that it is not necessary to postulate complex ion formation but that the blue colour is due to formation of the undissociated cobalt chloride molecule accompanied by dehydration.


Measurements of the vapour pressures of cobalt chloride solutions were made by Robinson (1938) in a preliminary investigation of this problem, but these measurements extended only up to a concentration of 2M. We began, therefore, by carrying these vapour-pressure measurements, using the isopiestic technique with calcium chloride as reference salt, up to higher concentrations. Cobalt chloride of A.R. quality was recrystallized twice from water acidulated with hydrochloric acid and then made up as a stock solution whose strength was determined by gravimetric silver chloride analysis.

The calcium chloride solution was part of that used by Stokes (1945) in his vapour-pressure measurements. Pairs of calcium chloride and cobalt chloride solutions were equilibrated until they had the same vapour pressure and the molalities of these solutions are recorded in Table I.

Table I.—Molalities of Isopiestic Solutions of Cobalt Chloride and Calcium Chloride at 25⇌.
CoCl2 CaCl2
0.3351 0.3390
1.235 1.264
2.170 2.187
2.898 2.845
3.878 3.602
0.3492 0.3535
1.543 1.578
2.473 2.463
3.335 3.198
4.064 3.733
0.8398 0.8556
1.970 1.997
2.754 2.725
3.512 3.333
1.033 1.056
2.081 2.105
2.794 2.764
3.687 3.475

From these results the water activity, osmotic coefficients, activity coefficients and molal-vapour-pressure lowerings of cobalt chloride solutions have been calculated in the standard way. (See Appendix I.) An important feature of these results is illustrated by a graph (Fig. 1) of the molal-vapour-pressure lowering against the molality. In this graph are also plotted the data for calcium chloride to illustrate the behaviour of a normal salt and for zinc chloride which has been selected to exemplify the influence of complex-ion formation (ZnCl4=) to a marked extent. Up to 2 m. the curve for cobalt chloride is slightly higher than that of calcium chloride; at 2·5m. the curves cross, and at high concentrations the cobalt chloride curve is lower than the calcium chloride curve. Although it cannot be claimed that such behaviour never occurs with normally behaved salts, sufficient data have been accumulated with salts of this type to indicate that this reversal in the order of the curves is peculiar; moreover, it should be noted that the peculiarity is encountered in the concentration range in which the rose colour of the cobalt chloride solution is changing to violet. This experiment cannot distingush beween the complex-ion

– 3 –

and the hydration theories, although perhaps it might be urged against the latter that, judging from the water of crystallisation in the solid salts, both calcium and cobalt ions should be hydrated to the same extent and therefore that, on the basis of a hydration theory, the behaviour of the curves in Fig. 1 would not be anticipated. Granting, however, that complex-ions are formed, a comparison with the zinc chloride curve of Fig. 1 shows that the extent of any complex-ion formation must be small even in concentrated solutions of cobalt chloride.

Picture icon

Fig. 1.—Relative Vapour Pressure Lowerings of Calcium Chloride, Cobalt Chloride, Cobalt Nitrate and Zinc Chloride.

– 4 –

A parallel study of cobalt nitrate was made. The salt was recrystallised twice from water and the vapour pressure of its solutions measured relative to solutions of calcium chloride. Cobalt nitrate has already been studied by Robinson, Wilson and Ayling (1942), but measurements have now been extended to 5·6 m. The experimental data are given in Table II. The water activities and related thermo-dynamic quantities have been calculated and are given in Appendix I, whilst the molal-vapour-pressure lowering of this salt has been plotted in Fig. 1. At low molalities the curve lies just below that of cobalt chloride; this would be anticipated by analogy with other bivalent metal chlorides and nitrates, the influence of a chloride in lowering the vapour pressure being somewhat greater than that of a nitrate.

Table II.—Molalities of Isopiestic Solutions of Cobalt Nitrate and Calcium Chloride at 25°.
Co (NO3)2 CaCl2
1.624 1.602
2.846 2.715
4.802 4.393
2.144 2.079
3.067 2.899
5.552 5.027
2.483 2.390
3.729 3.473
5.790 5.233
2.673 2.574
4.039 3.741

At all concentrations the curve for cobalt nitrate is normal and the peculiarity of cobalt chloride is emphasized by the fact that at high concentrations (above 4m.) the chloride has less effect than the nitrate in lowering the vapour pressures. This experiment with cobalt nitrate bears against the hydration theory because if varying hydration of the cobalt ion occurs, it is difficult to understand why the same effect is not observed with both the chloride and nitrate. On the other hand, the normal behaviour of the nitrate would seem to support the complex-ion theory for the chloride, since nitrates do not form complexes as easily as chlorides.

The third experiment consisted of measuring the vapour pressures of solutions made up to contain two mols of lithium chloride to one mol of cobalt chloride, i.e., corresponding to the complex salt Li2CoCl4. It has already been shown (Robinson and Stokes, 1945) that the vapour pressures of solutions of mixed salts, such as 2KCl + MgCl2, where no complex-ions are formed, can be calculated with considerable accuracy by assuming additivity of the vapour-pressure lowerings of the component salts separately. Parallel measurements were made on solutions of Li2Co(NO3)4 with the following results.

V.P. lowering in mm. Hg.
mLi2CoCl4 obs. calc. % diff.
1.104 4.482 4.535 +1.2
1.505 6.773 6.802 +1.0
1.639* 7.524 7.570 +0.6
2.046* 9.716 9.761 +0.5
2.232* 10.640 10.641 0
mLi2Co(NO3)1 obs. calc. % diff.
1.065 3.889 3.815 —1.9
1.360 5.230 5.125 —2.0
1.812 7.352 7.177 —2.4
2.188 9.082 8.865 —2.4
2.414 10.065 9.824 —2.4

The observed vapour pressure and those calculated on the assumption that no complex-ion formation occurs are similar; it would be unsafe to base any hypothesis of complex-ion formation on the small differences found, but it can be concluded that any complex-ion formation which may occur must be small in extent.

In a fourth experiment measurements were made of the vapour pressures of solutions containing calcium and cobalt chloride in equimolecular amounts, the data being compared with values calculated

[Footnote] * These solutions were blue.

– 5 –

from the vapour pressures of solutions of the two salts separately. Parallel measurements were made on solutions of CaCl2 + MgCl2 where complex-ion formation is unlikely. The following results were obtained:

V.P. lowering in mm. Hg.
mCaCoCl4 obs. calc. % diff.
1.138 3.983 3.962 —0.5
1.209 4.318 4.308 —0.3
1.310 4.822 4.845 +0.5
1.342 4.997 5.017 +0.4
1.515* 5.906 5.947 +0.7
1.920* 7.982 8.068 +1.1
1.821 8.386 8.316 +0.8
2.186 10.704 10.666 +0.4
3.110 15.909 15.968 —0.4

Again it must be concluded that any complex-ion formation must be small in extent.

It was evident that some more delicate measurement was necessary for the constituent causing the blue colour of these solutions.

On making some preliminary measurements of the absorption of light by these solutions, using a Coleman No. 11 spectrophotometer, it was found that the rose-coloured solutions gave a strong absorption band with a maximum at 525 m.μ. The addition of small quantities of lithium chloride enhanced the absorption in this region, whilst a second band appeared at a higher wave-length. The effect of lithium chloride in increasing the normal absorption of the rose-coloured cobalt chloride solutions is difficult to understand; the phenomenon, however, is genuine, for it has also been observed by Weyl (1946). The second region of absorption is undoubtedly due to the blue-coloured compound because it increases in intensity with increasing concentration of added chloride, the strongest absorption being at 685 m.μ. Brode (1928), using an instrument of higher resolving power, was able to show that seven bands were to be found in this region, the two most intense being at 695 and 679 m.μ. In the most concentrated lithium chloride solutions there was no absorption in the region of 525 m.μ. In all further work, therefore, measurements were made at 685 m.μ, a wave-length which was taken to be characteristic of the blue compound.

It was then shown that Beer's Law relating the intensity of absorption to concentration:

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

D = log Io/I = kC

where D is the “optical density” of the solution defined in terms of the intensity of light, I transmitted and k is a constant, was approximately true. The following values were obtained with a solution of 18 m. lithium chloride at 23° and varying normalities (mols per litre) of cobalt chloride:

C D k
0.0005 0.255 510
0.001 0.490 490
0.002 0.890 445
0.0025 1.075 430

In order to take advantage of the most sensitive range of the spectrophotometer, it was found advisable to make further measurements at a cobalt concentration of 0·002 N.

Temperature was found to have considerable effect on the optical density, as the following figures for 0·002 N cobalt chloride in 6·44 m. calcium chloride show:

[Footnote] * These solutions were blue.

– 6 –
Temp. D
20° 0.440
23 0.474
32 0.552
39 0.587
57° 0.668
70 0.705
85 0.730

Measurements were therefore made under the following standardised conditions: a wave-length of 685 m.μ, a cobalt concentration of 0·002 mols per litre and a temperature of 23°. A series of measurements was made with increasing quantities of lithium chloride; in the absence of lithium chloride the optical density was 0·006 and this increased to a flat maximum at 20 m. lithium chloride, where the density was 0·953. The fraction, α, of cobalt present in the form of the blue compound was then calculated by the equation:

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

α = (D—0·006)/0·953,

D being the optical density at intermediate lithium chloride concentrations. Fig. 2 shows the variation of 100 α with concentration of lithium chloride and also of hydrochloric acid and calcium chloride. Separate experiments showed that the addition of dehydrating agents, such as sulphuric acid and lithium nitrate, even at high concentration, gave no absorption in this region. It was evident, therefore, that the chloride ion is essential to the production of the blue colour. In the experiments with these three chloride solutions, however, we have

Picture icon

Fig. 2.—Percentage Formation of Blue Compound of Cobalt Chloride in Chloride Solution. 1. HCl. 2. LiCl. 3. CaCl2.

– 7 –

been varying not only the chloride concentration, but also the activity coefficient of the chloride and the activity of the water. An experiment was then devised in which changes in the chloride concentration could be made while keeping the activity coefficent of the chloride and the water activity at least approximately constant. This was effected by making mixtures of constant cobalt chloride content but varying ratios of hydrochloric acid and perchloric acid. Table 3 gives the optical density, D, and the fractional degree of formation, α, of the blue compound, in six mixtures of hydrochloric and perchloric acid. The water activities recorded were calculated from the data given by Åkerlöf and Teare (1937) and by Robinson and Baker (1946). At least four equilibria are possible:

Co++ + Cl ⇌ CoCl+

Co++ + 2Cl ⇌ CoCl2

Co++ + 3Cl ⇌ CoCl3

Co++ + 4Cl ⇌ CoCl4

corresponding to which there are equilibrium constants, K1, K2, K3, K4, defined as shown at the foot of Table 3. On substituting values

Table III.—Equilibria of Cobalt Chloride in HCl–HClO4 Mixtures.
CoCl2 concentration = 0.002 N.
mHClO4 mHCl D α aH2O —log K1 —log K2 —log K3 —log K4
0 10.40 0.620 0.644 0.4024 0.760 1.777 2.794 3.811
1.96 8.12 0.541 0.561 0.3994 0.803 1.713 2.622 3.532
3.32 6.54 0.456 0.472 0.3975 0.864 1.680 2.496 3.311
5.09 4.46 0.292 0.300 0.3970 1.017 1.667 2.316 2.965
5.93 3.47 0.192 0.195 0.3961 1.156 1.696 2.237 2.777
6.99 2.25 0.086 0.084 0.3952 1.390 1.742 2.096 2.448
(1—α) [Cl] K1 = α
(1—α) [Cl]2 K2 = α
(1—α) [Cl]3 K3 = α
(1—α) [Cl]4 K4 = α

of the chloride concentrations and the measured values of α, it was found that only K2 was to any degree constant, K1 increasing and K3 and K4 decreasing with increasing chloride-ion concentration. It follows that the formation of the undissociated cobalt chloride molecule, CoCl2, must be a predominant factor in the production of the blue colour.

The accurate expression for the equilibrium constant is:

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

(1—α) γCo γ2Cl [Cl]2 K2 = γu K2

where γu is the activity coefficient of the undissociated molecule and γCl, γCl the activity coefficients of the ions. These should have been approximately constant in the experiments shown in Table III, but this would not be the case in the measurements portrayed in Fig. 2. With varying concentration of lithium chloride or hydrochloric acid, it may be assumed that the change in γu will be small, but there is no way of determining the change in γCo and γCl, although we should expect it to be considerable. Fortunately, in the case of the calcium chloride solutions, we can, from the known resemblance between the activity coefficients of calcium chloride and cobalt chloride, make the reasonable assumption that the activity coefficient of cobalt chloride at low concentration in concentrated calcium chloride solution is equal to the activity coefficient of calcium chloride at the same concentration, i.e., γCo γCl2 = γCa γCl2, which, for brevity, we

– 8 –

write γ3CaCl2. Hence we may use the activity coefficients of calcium chloride given by Stokes (1945) to substitute in the expression:

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

(1—α) γ3CaCl2 [Cl]2 K2 = α

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

Table IV.—Equilibria of Cobalt Chloride in Calcium Chloride Solution.
CoCl2 concentration = 0.002 N.
mCaCl2 D α aH2O γ CaCl2 K2 × 106 K × 104
First Series of Measurements.
4.02 0.032 0.0273 0.6206 3.044 15.4 1.09
4.57 0.079 0.0766 0.5499 4.505 10.8 1.18
4.75 0.100 0.0990 0.5274 5.10 9.3 1.21
5.28 0.190 0.193 0.4649 7.34 5.3 1.13
5.54 0.233 0.238 0.4362 8.67 3.9 1.08
5.92 0.333 0.343 0.3977 10.93 2.9 1.14
6.02 0.347 0.358 0.3881 11.60 2.5 1.09
6.32 0.423 0.437 0.3615 13.65 1.9 1.12
6.77 0.521 0.540 0.3265 16.94 1.3 1.16
Mean 1.13
Second Series of Measurements.
4.77 0.084 0.0818 0.5250 5.17 7.1 0.93
5.21 0.161 0.163 0.4728 7.00 5.2 1.05
5.56 0.247 0.253 0.4341 8.74 4.1 1.16
6.00 0.363 0.375 0.3899 11.48 2.8 1.19
6.44 0.474 0.491 0.3515 14.48 1.9 1.25
6.91 0.552 0.573 0.3168 17.99 1.2 1.20
Mean 1.13

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

(1—α) γ3CaCl2 [Cl]2 K2 = α (1—α) γ3CaCl2 [Cl]2 a4H2O K = α

Table 4 records two series of measurements from which K2 may be calculated. It will be seen that K2 is by no means constant. This would appear to contradict the conclusions at which we arrived from a study of Table 3. However, in these hydrochloric-perchloric acid experiments we studied the variation in the formation of undissociated cobalt chloride molecules with varying chloride-ion concentration keeping the activity coefficient of the chloride and the water activity constant. In the experiments of Table 4 with calcium chloride we have varied the chloride concentration by known amounts, the activity coefficient by an amount which we have attempted to allow for in calculating K2 and also the water activity, aH2O, of the solution, for which no allowance has yet been made. These water activities for calcium chloride solutions, have also been measured by Stokes and on substituting in the equation:

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

(1—α) γ3CaCl2 [Cl]2 a4H2O K = α

we find we obtain very constant values of K, shown in the last column of Table 4. It follows that not only does the chloride-ion concentration enter into the equilibrium to the second power, but also the water activity to the fourth power. This is consistent with the equilibrium:

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

(Co mH2O)++ + 2Cl ⇌ (Co (m—4) H2O, Cl2) + 2H2O

of which a particular case would be:

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

(Co 6H2O)++ + 2Cl ⇌ (CoCl2.4H2O + 2H2O

which would preserve the co-ordination number of cobalt at six.

– 9 –

Appendix I.

Water Activities, Osmotic Coefficients, Activity Coefficients and Relative Molal Vapour Pressure Lowerings of Cobalt Chloride and Nitrate Solutions at 25°.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

CoCl2 Co(NO3)2
m aw φ γ P°–P/mp° aw φ γ P°–P/mp°
0.1 0.99538 0.857 0.522 0.04620 0.99540 0.854 0.518 0.04600
0.2 0.99066 0.869 0.479 0.04670 0.99074 0.861 0.471 0.04630
0.3 0.98575 0.886 0.463 0.04750 0.98592 0.875 0.452 0.04693
0.4 0.98059 0.907 0.459 0.04853 0.98090 0.892 0.445 0.04775
0.5 0.97511 0.932 0.462 0.04978 0.97560 0.914 0.445 0.04880
0.6 0.96938 0.959 0.470 0.05103 0.97010 0.936 0.448 0.04983
0.7 0.96351 0.982 0.479 0.05213 0.96440 0.958 0.455 0.05086
0.8 0.95723 1.011 0.492 0.05346 0.95848 0.981 0.463 0.05190
0.9 0.95055 1.043 0.511 0.05494 0.95222 1.007 0.473 0.05309
1.0 0.94354 1.075 0.531 0.05646 0.94572 1.033 0.488 0.05428
1.2 0.92865 1.141 0.578 0.05946 0.93194 1.087 0.521 0.05672
1.4 0.91266 1.208 0.634 0.06239 0.91716 1.143 0.561 0.05917
1.6 0.8957 1.274 0.699 0.06521 0.90151 1.199 0.607 0.06156
1.8 0.8779 1.339 0.773 0.06786 0.8848 1.258 0.661 0.06400
2.0 0.8590 1.406 0.860 0.07050 0.8674 1.317 0.723 0.06630
2.5 0.8095 1.564 1.120 0.07620 0.8201 1.468 0.918 0.07196
3.0 0.7578 1.711 1.458 0.08073 0.7690 1.620 1.178 0.07700
3.5 0.7086 1.821 1.832 0.08326 0.7156 1.769 1.522 0.08126
4.0 0.6637 1.896 2.215 0.08408 0.6613 1.913 1.966 0.08468
4.5 0.6069 2.053 2.538 0.08736
5.0 0.5525 2.196 3.298 0.08950
5.5 0.5013 2.323 4.228 0.09067


Åkerlöf, G., and Teare, J. W., 1937. J. Amer. Chem. Soc., 59, 1855.

Brode, W. R., 1928. Proc. Roy. Soc., 118A, 286.

Robinson, R. A., 1938. Trans. Faraday Soc., 34, 1142.

—— and Baker, O. J., 1946. Trans. Roy. Soc. N.Z., 76, 250.

—— and Stokes, R. H., 1945. Trans. Faraday Soc., 41, 752.

—— Wilson, J. M., and Ayling, H. S., 1942. J. Amer. Chem. Soc., 64, 1469.

Stokes, R. H., 1945. Trans. Faraday Soc., 41, 637.

Weyl, W. A., 1946. J. Applied Physics, 17, 628.