DECOMPOSITION OF SODIUM HYPOCHLORITE :

Can. J. Chem. Downloaded from www.nrcresearchpress.com by Scientific Library of Lomonosov Moscow State Univ on 02/18/14

THE CATALYZED REACTION1

ABSTRACT
T h e catalyzed decomposition of sodium hypochlorite has been examined; the
catalysts tried were manganese, iron, cobalt, nickel, and copper oxides. I t was
shown that in no case was the decomposition t o chlorate and chloride accelerated,
only the reaction t o chloride and oxygen. Manganese and iron did not catalyze
even the latter reaction, or only t o a very small extent; this was in fairly concen-
trated sodium hypochlorite containing some sodium hydroxide. T h e manganese
and iron are largely oxidized t o permanganate and ferrate under these conditions.
I t was found that copper could catalyze the formation of perrnanganate and
ferrate, and niclcel the formation of permanganate. Cobalt catalyzed the reaction
going to oxygen, and the rate was proportional t o the cobalt added, but little
dependent on the hypochlorite concentration; the same is true of nicltel. Copper
(as reported earlier) gives a catalyzed reaction not far from first order in hypo-
chlorite. T h e activation energies were measured, and were consistent with the
relative catalytic activity of these metals. T h e mechanism of the reaction is
briefly discussed.

The decomposition of sodium hypochlorite is catalyzed by the oxides oE

transition elements such as cobalt. Qualitatively this was first ltnown a long
time ago, but quantitative measurements on the catalysis are much more
For personal use only.

recent. Howell (3) made a careful study of the catalysis by cobalt oxide. He
added the cobalt oxide as a fine, almost colloidal, suspension in water, and
measured the rate of oxygen evolution from the vigorously stirred mixture.
At fairly low catalyst concentrations (about 2 mgm. cobalt in 60 ml. of solu-
tion), he found the rate to be proportional to the catalyst and to the hypo-
chlorite concentrations. T h e activation energy of the catalyzed reaction was
16.6 ltcal./gm-mol. i\/loelwyn-Hughes (10) was able to calculate that Howell's
rate constants are fairly near what one would expect if the catalyst were present
as particles of 5X10-7 cm. diam., and i f every collision with a hypochlorite ion
with sufficient energy were effective. Chirnoaga made somewhat similar
measurements for cobalt and nicltel oxides (1). He found the rate to be inde-
pendent of the stirring speed, provided this was moderately rapid. His rates
were not proportional to the hypochlorite concentration: with nickel oxide
they were proportional t o (C10-)0." a t low coilcentrations (0.02 to 0.01 M ) .
With cobalt oxide the rate was proportioilal to (C10-)0.9. The activation energy
was 15.8 ltcal. for cobalt oxide, and 16.6 ltcal. for nickel oxide. Both Howell
and Chirnoaga reported that if the catalyst was washed for a long time till
all alkali was removed, it was more active. This may be a peptizing effect due
to removal of electrolyte. However all the hypochlorite solutions used con-
tained free alltali. Chirnoaga reported that the order of catalytic activity was
Ni > Co > Cu > Fe > Mil, although iiicltel and cobalt might be expected
t o be in the reverse order from the activation energies.
Dixon and White (2) examined the conversion of Mn(I1) to permanganate
by hypochlorite, aitd noted that copper could catalyze this. Lewis (5, 6)
examined cobalt, copper, and iron oxides as catalysts. A t fairly high concen-
lManzrscript relzeiued November 14, 1955.
Contribution from the Department of Chemistry, Utziversity of Toronto, Toronto, Ontario.
 480 C A N A D I A N J O U R N A L O F CHEMISTRY. VOL. 3 4 , 195ri

trations he found the rate to be independent of the hypochlorite concentration,

and it seems difficult to reconcile his results with those of Howell. He was also
Can. J. Chem. Downloaded from www.nrcresearchpress.com by Scientific Library of Lomonosov Moscow State Univ on 02/18/14

particularly interested in the effect of adding substances, ordinarily inert, as

promoters.
The present author (8) made measurements on the catalysis by copper,
during some investigations of the complex copper tellurates and periodates.
As these results were obtained ~ ~ n d conditions
er comparable to the measure-
ments reported in this paper, they will be discussed briefly later.
One of the difficulties in interpreting the results of these measurements
(either the earlier or the present work) is the uncertainty in the size of the
catal>.tic particles. T h e apparent gross size is, of course, no certain indication
of the true surface area, as pores are probably present. Comparisons with
previous rate measurements always contain this element of uncertainty.
Consequently the present work was confined t o some more limited matters
about the reaction. Firstly, although the decomposition to sodium chloride
and oxygen is certainly catalyzed, it is not clear whether the decompositio~l
to sodium chloride and chlorate is similarly catalyzed. Secondly, it has some-
times been suggested that qualitatively the mechanism of catalysis is through
unstable higher oxides, which are formed, decompose, and are reformed by
further oxidation, and so on. However manganese and iron do not have such
For personal use only.

unstable higher oxides, so their degree of catalytic activity was investigated.

Thirdly, the energies of activation of all the catalyzed reactions were measured
again. Finally, some further evidence on the dependence on hypochlorite
concentration was obtained, which supports I,ewisls views a t least over a
certain range of conditions; but there are several factors which can affect the
rate, and a complete stud). of the matter would require a great many experi-
ments.
EXPERIMEN'I'AL METHOD
T h e materials used were the same as in the previous paper, with the addition
of the salts of various metals used as catalysts. Reagent grade samples of
manganese chloride, ferric nitrate, cobalt chloride, nickel chloride, and copper
chloride were used. Stoclr solutions of these salts were made LIP,and various
amounts were pipetted out, and added to well-stirred alkaline s o d i ~ ~hypo-
m
chlorite. Precipitates of the oxides immediately formed, except with copper.
These precipitates were finely divided, but not colloiclal; as far as possible the
conditions of addition were the same each time. The solutions were then
brought t o the desired temperature in a thermostat, with continual stirring,
and the rate of gas evolution was measured. Samples were taken for analysis
from time to time. The apparatus was the same as described in the first paper
on the uncatalyzed reaction (9). The analytical methods have been described
in another previous paper (7).
T h e sodium hypochlorite was the same as in the investigation of the un-
catalyzed reaction; its ionic strength was 3.79, and it was 0.32 $1 in sodium
hydroxide.
EXPERIMEK'I'AL RESIiLTS A N D CONCLUSIONS
Before describing the quantitative runs, a few qualitative observations
might be included. Dixon and White (2) observed that copper catalyzed the
 LISTER: CATALYZED REACTION 48 1

formation of permanganate. Manganese dioxide is, of course, first formed,

starting with Mn(I1) ; and even without copper, permanganate is formed slowly
Can. J. Chem. Downloaded from www.nrcresearchpress.com by Scientific Library of Lomonosov Moscow State Univ on 02/18/14

a t room temperature. I t was noticed that nickel also catalyzed the formation
of permanganate, though less efficiently than copper. Sodium chlorite does not
oxidize Mn(I1) beyond manganese dioxide.
Iron is similarly slowly oxidized to ferrate in the cold, and quite rapidly a t
80' C.; as would be expected the reaction goes faster in strongly alkaline solu-
tion. Although the ferrate is unstable on boiling, it decomposes only very
slowly a t 60' C. Copper also accelerates the formation of ferrate, but nickel
seems to be unable to do this. The qualitative behavior of copper was described
in the paper mentioned earlier (8). As is well known, cobalt and nickel are
oxidized to higher oxides by sodium hypochlorite, which are almost completely
insoluble in the alkaline solution. Sodium chlorite will not accomplish any of
these oxidations.
In the quantitative runs, it is probably simplest to consider each element
separately; the results will be given and any conclusions drawn from them
before going on to the next element.

The d a t a on manganese as a catalyst are given in Table I. The conclusion

For personal use only.

t h a t must be drawn from these results is that manganese does not catalyze

TABLE I

'Temp., M n added, Solution Time, (CIO-), (CIO2-), (C103-),

Run O C. mgm. volume, ml. mln. M M fif
1 50 3.32 706 0 1.285 0.009 0.122
286 1.261 0.008 0.128

Gas rate,
Temp., Meat1 (CIO-), ml. N.T.P./min.
Run " C. Af per liter soln.

the reactioil a t all. If we use the same equation that was applied to the un-
catalyzed runs (9),
- d x / d t = $klx2+ kox,
where x = (C10-),
kl = rate coi~stantfor first stage of reaction t o chlorate,
k o = rate constailt for reaction to oxygen,
then the results above give for the constants:
Temp. 50 60 " C .
k3 4.0 10.7 X10-6 min.-I
kt 3.25 10.75 X 10-& (gm-mol./l.)-I n~in.-'
 4 82 CANADIAN JOURNAL O F CHEMISTRY. VOL. 34. 1956

By "reaction to chlorate" is meant the reaction

+ 2C1-
Can. J. Chem. Downloaded from www.nrcresearchpress.com by Scientific Library of Lomonosov Moscow State Univ on 02/18/14

30C1- + c103-
of which the first step is
20C1- + C102- + Cl-.
This is the slow rate determining step, and k 1 is its rate constant. By "reaction
to oxygen" is meant the reaction
20C1- + 2C1- + 02.

This was found to be a first order reaction, and ko is its rate constant. These
reactions are the ones meant in all references in later sections to "reaction to
chlorate" or "reaction to oxygen1'.
The ionic strength of the solution after addition of catalyst was 3.78. T h e
ko values are just about right for this ionic strength, and make the energy of
activation 21 kcal., in agreement with results obtained earlier. The k l values
are a little low, but certainly show that manganese does not catalyze the re-
action under these conditions. The reason for the difference between these
results and those of Chirnoaga probably lies in the higher concentration and
sodium hydroxide content of the present solutions. Chirnoaga's sodium hypo-
chlorite was only 0.042 M, and his manganese dioxide was 0.0061 11.1, so an
appreciable fraction of the hypochlorite could be used up in oxidizing the
For personal use only.

manganese; indeed over half of the decomposition that he observed could be

explained in this way. However the present results do show that manganese is
not catalytic in concentrated solutions of hypochlorite.
Iron
The data on iron as a catalyst are given in Table 11. Again we can treat
these results by the method for uncatalyzed runs, which gives values for the
rate constants as shown below. Values of k 1 from the chlorate analyses are
also included, as these help to distinguish the catalytic effect (if any) on the
two reactions. I t may be added that constants obtained from the chlorate
analyses have the advantage of not needing a value of the gas rate for their

TABLE I1

Temp., F e added, Time, (C10-1, (ClOz-), (ClO?-),

Run "c. mgm. ~III. nl nr AI

Gas rat:,
Temp., Mean (CIO-), ml. N.T.P./min.
Run " C. M per liter soln.
 LISTER: CATALYZED REACTION 483

calculation; but against this must be set the fact that the chlorate concen-
trations are obtained by the difference of two analyses, one of which does not
Can. J. Chem. Downloaded from www.nrcresearchpress.com by Scientific Library of Lomonosov Moscow State Univ on 02/18/14

have a very easily observed end point. The values of the constants are:
Temp. 50 60 C.
ko 5.2 12.0X10-G min.-l
kI 3.59 11.6 X (gm-mol./l )-I min.-I
k~ 3.70 11.5 X 10-"gm-mol./l.)-I min.-I
(from chlorate)

There may be some catalytic effect on the reaction to oxygen, but it is very
slight. As compared with, for instance, cobalt, the iron is a t most 1/200 times
as active as a catalyst. There is no sign of any catalysis of the chlorate reaction:
the uncatalyzed rate constants a t this ionic strength (3.79) are 3.61 X10-5 a t
50" C., and 11.4X10-5 a t 60" C. We must add, of course, that these conclusions
only apply t o the conditions of these experiments; that is to say in fairly high
hypochlorite concentrations, moderate alkalinity, and not too high tempera-
tures; but under these conditions iron does not catalyze the chlorate reaction,
and it catalyzes the oxygen reaction very slightly, if a t all.
Cobalt
The data obtained are given in Table 111. Cobalt, of course, catalyzes the
oxygen reaction, but its effect on the chlorate reaction is not so obvious. T h e
For personal use only.

rigorous determination of k1 from these data is not easy because of the high

TABLE 111

Toemp., Co added, Time, (C10-), (C102-), (ClOs-),

Run C. mgm. mln. M M M

Temp., i\/Iean (C10-), Gas rate, Co added,

Run " C. M ml. N.T.P./min. mgm.
 484 C A X A D I A N J O U R N A L OF CHEMISTRY. VOL. 34, 1956

gas rate. However a reasonable approximation can be obtained by observing

the total rate of decomposition between any two analyses; we then put:
Can. J. Chem. Downloaded from www.nrcresearchpress.com by Scientific Library of Lomonosov Moscow State Univ on 02/18/14

total rate = rate to oxygen + rate to chlorate

and :
rate to chlorate = iklf2

where 3 is now the meail hypochlorite concentration between the two analyses.
Hence k l can be calculated. T h e error from taking this approximate equation,
and not an integrated equation, is small. This was checked by integrating the
above equation for various plausible expressions for the dependence of the
rate to oxygen on x ; then the approximate k l could be expressed in terms of the
true k l and known quantities such as x or the time. This enables an estimate
of the error in the approximate k l to be made, and in the actual runs this never
exceeded 1%. Although k l could have been obtained from these integrated
equations, they were rather cumbersome expressions; and it was felt that the
approximate equation allowed for the actual observed gas rate in a more direct
manner.
Considering first runs 5 and 6, we get the following values of k l , averaged
over the entire run (ionic strength 3.79):
For personal use only.

Ruri kl from C10- kl from C103-

5 3.74 X10-j 3.68 X10-5 (gm-mol./l.)-I min.-I
6 3.66 3.61

These values are near enough to the uncatalyzed rate constant (3.61 X10-5) to
be within the experimental error. Hence cobalt does not catalyze the chlorate
reaction to a detectable extent.
If we look a t the oxygen reaction, a comparison of runs 5 and 6 shows that
a t the same hypochlorite concentration, the rate is proportional to the cobalt
added. This agrees with all previous work. T h e dependence on hypochlorite is
more obscure, but a few tentative remarks can be made. Firstly, it is difficult
to distinguish, during any one run, an 'aging' effect in the catalyst from a
dependence of the rate on the hypochlorite concentration. Howell avoided this
difficulty in some of his runs by filtering off a catalyst a t the end of a run and
re-using it; he found its activity had fallen a few per cent. T h e 'aging' is
presumably due to agglomeration of the catalyst particles, thereby reducing
their surface area; or to poisoning of the catalyst by impurities. Howell found
that typical catalyst poisons had no effect, so probably changes were due to
agglomeration. T h e following observation supports this view. I t was found that
if a catalyst was used for some time, then left overnight without stirring, and
stirring was started next day a t a fairly low temperature, then the rate of gas
evolution was low (run 9a a t 30' C.). If the temperature was increased in
steps, then the rate rose towards what was found for fresh catalyst, and a t
50" C. the catalyst had recovered (runs 9b, 9c). T h e simplest explanation is
that the vigorous gas evolution a t 50' C. had broken up the agglomerated
particles. If the temperature was subsequently decreased, the fresh catalyst
rate was more or less maintained (compare runs 8a, 9g), a t least for long eno~igh
 LISTER: C.4TALYZ EII KISACTION 485

t o observe the gas rate. Hence it is impossible t o say with a n y confidence

whether the fall in rate during runs 5 and 6 is due t o aging of catalyst or fall in
Can. J. Chem. Downloaded from www.nrcresearchpress.com by Scientific Library of Lomonosov Moscow State Univ on 02/18/14

(C10-). However run 7 does provide some evidence on this, a s the two rates
in this run were measured on the same solution t o which water was added
between the measurements. T h e rate only fell 7y0 for a decrease in hypochlorite
concentration of 45%. Similarly a comparisoil of runs 5, 8c, and 9c shows
nearly the same rates over a moderate range of concentration. On the whole
it seems fair t o conclude t h a t , under the conditions of these experiments, the
rate is very little dependent on hypochlorite, in agreement with Lewis.
This in no sense contradicts Howell's results for his experimental conditions.
T h e difference does not lie apparently in the concentrations used, which were
much the same as in the present work. T h e difference is probably in the
method of preparation of his catalyst: per milligram his catalyst was about
three times as active a s t h a t in Lewis's or the present work. Even so the degree
of difference seems surprising, and t h e matter cannot be regarded a s settled.
Runs 8 and 9 give t h e temperature dependence of the rate; they make the
energy of activation 15.9 ( 5 0 . 5 ) kcal./gm-mol. This is t o be compared with
15.8 kcal. found b y Chirnoaga, 16.6 kcal. b y Howell, and (in effect) 13.3 kcal.
b y Lewis.
TABLE IV
For personal use only.

Temp., Xi added, Time, (CIO-), (ClOz-), (C103-),

Run " C. mgm. n1111. 111 M M

13 50 0.85 0 1.315 0.008 0.116

230 1.280 0.008 0.123
-

Temp., Mean (CIO-), Gas rate, Ni added,

Iiun " C. A[ I . N.'.P./mi~. mgm.
10 40 1.36 0.47 3.39
1.33 0.52
1.27 0.50
 486 CANADIAN JOURNAL O F CHEMISTRY. VOL. 34, 1956

Nickel
Can. J. Chem. Downloaded from www.nrcresearchpress.com by Scientific Library of Lomonosov Moscow State Univ on 02/18/14

The data on nickel are given in Table IV. The first point to settle is whether
nickel catalyzes the chlorate reaction. If we calculate the rate constant, kl, as
for cobalt, the result is:
Run Temp., " C. kl from CIO- kl from C103-
10 40 1.35 X10-6 1.29 X1O-S(gm-mol./l.)-lmin.-~
11 50 3.52 3.62
12 60 10.5 13.0

The experimental error a t 40' C. is fairly large, as an error of 0.001 in the hypo-
chlorite concentration would alter kl by nearly 2$%. However it seems possible
to conclude that within the experimental error, nickel does not catalyze the
chlorate reaction. Runs 11 and 13 show that the rate is proportional to the
nickel added, in agreement with all previous work.
The dependence of the catalyzed rate on hypochlorite concentration is best
seen from runs 11, 12, and 14. Over the range of concentrations examined
(0.81 to 1.23 M) there was no observable change in the catalyzed rate with
hypochlorite concentration. Again this does not mean that under other con-
ditions such a dependence might not be found. Chirnoaga found the rate to
be proportional to (C10-)0.47,a t low concentrations; this is a lesser degree of
dependence on concentration than he found with cobalt.
For personal use only.

The mean gas rates a t about I M hypochlorite, with 3.39 mgm. of nickel,
a t various temperatures are:
Temp. 40 50 60 ' C.
Gas rate 0.50 1.17 2.64 ml./min.

These make the activation energy 17.2 kcal./gm-mol. A little (about 4%) of
the total rate can be attributed to the uilcatalyzed reaction, and if allowance
is made for this, the activation energy is reduced to 17.1 kcal.
Copper
In another paper (8), the writer has given details of the effect of copper as
a catalyst. The only point that will be dealt with here is whether copper
catalyzes the reaction to chlorate. Table V gives the experimental results.
From these results we can calculate kl, using interpolated values of the rate of
oxygen evolution a t the appropriate hypochlorite concentration. The values
so obtained are:
Run 15 16 17(0-1162') 17(1162-2921') 18 19
k, from (CIO-) 3.88 3.61 3.35 3.65 3.82 3.44 X10-6
kl from (C103-) 4.08 - 3.82 3.60 - - X

The long run (17) was arbitrarily divided illto two parts for this calculation
as the range of concentration is large, but similar results would be obtained
from other time intervals in this run. The accuracy of kl, as obtained above,
is not very great, as small changes in the ailalytical results would give rela-
tively large changes in kl. However it seems reasoilably certain from these
results that copper does not catalyze the reaction to chlorate.
As mentioned above, the dependence of rate on concentration has been
reported by the writer in an earlier paper on another aspect of the chemistry
 LISTER: CATALYZED REACTION

TABLE V
Can. J. Chem. Downloaded from www.nrcresearchpress.com by Scientific Library of Lomonosov Moscow State Univ on 02/18/14

Cu added, Time, (ClO-), (C103-),

Run mgm. mln. M M

19 1.96 0 0.982 -
493 0.839

Mean (C10-), Gas rate,

Run M ml. N.T.P./min.
15 1.23 6.39
For personal use only.

of copper. Briefly, it was found that the rate could best be represented by an
equation of the form:
rate = (Cu) {a+ b(C10-) }
where a and b are constants, and b is larger than a. The over-all energy of
activation was 15.5 kcal./gm-mol.

GENERAL REMARKS
Comparison of the results with various metals enables us t o make a few
general comments on these reactions. In the first place, it was found that i11
no case was the chlorate reaction catalyzed (or only t o a very small extent).
This suggests that the mechanism of the catalysis is through oxidation of the
metal t o a higher oxide, which then loses oxygen, and is subsequently re-
oxidized, and so on. The inertness of iron and manganese where there is no
unstable higher oxide supports this view. Unfortunately, if we calculate the
rate of decomposition of the higher oxide which would provide the observed
gas evolution, it seems t o be impossibly high. For instance in run 5, 5.4X10-5
gm-atoms of cobalt gave 2.2X10-4 gm-atoms of oxygen per minute. If the
 488 CANADIAN JOURNAL OF CHEMISTRY. VOL. 34, IYX

change is from Co(II1) to Co(II), this means that the half life of the oxidized
cobalt atom is only 5.2 sec., even if all the Co(I1) is instantaneously re-oxidized.
Can. J. Chem. Downloaded from www.nrcresearchpress.com by Scientific Library of Lomonosov Moscow State Univ on 02/18/14

As Coz03.HzO(or CoO.OH) is reported (4) to have been isolated, and to be

stable on drying a t 150" C., the above theory of the reaction seems to be un-
tenable. Similar remarks apply, though with somewhat less force, t o nickel
and even to copper. Of course, we cannot entirely dismiss the possibility t h a t
the reaction is due to some still higher oxide, b u t this is hypothetical.
The oilly alternative seems t o be that the reaction goes in the stages (M is,
any catalytic metal):
(i) 2 M 0
++C10-
(ii) M?Oa C10-
-+ M?Oo
+ M203.CIO-
+ CI-
(iii) M?O3.ClO- +2M0 + CI- (adsorbed)
+ O:!
This, of course, was realized by earlier workers on the reaction. If the reaction
rate is independent of the hypochlorite concentration, then (iii) must be the
slow stage, and much the same amount of hypochlorite must be adsorbed in
(ii) over a range of concentration. While there is really no direct evidence of
this, the simplest picture is one in which the entire catalyst surface is kept
covered with adsorbed hypochlorite ions a t the rather high concentrations
actually used. If the catalyst surface is not completely covered, it is impossible
to predict what the reaction kinetics would be without knowing the dependence
For personal use only.

of adsorptioil on concentration. Only if the adsorption were proportional t o

the hypochlorite concentration could we expect a first order reaction; this
might happen in dilute solutions. Thus cobalt and nickel, a t the concentrations
used, behave as though the catalyst were fairly completely covered with
adsorbed ions. Copper is entirely, or nearly entirely, in solution; and the re-
action, as might be expected, is not far from being first order in both copper
and hypochlorite. b'Ianganese and iron do not catalyze the reaction, and this
is a t least in part attributable to their oxidation to stable states.
I t is interesting to compare the rates with the activation energies. At 50" C.,
lop4 gm-atoms of metal in 1 111 sodium hypochlorite would give gas rates a s
follows:
Metal Co i Cu
Rate 4.0 2.0 7.4 ml./inin.
Activation energy 15.9 17.1 15.5 kcal./gin-n-01

Thus the rate increases as the activatioil energy decreases. T h e rate with nickel
seems somewhat large compared with the others, as the activation energy is.
over 1 kcal. larger.
REFERENCES
1. .CHIRNOAGA, E. Trans. Chem. Soc. 1693. 1926.
2. DIXON,B. E. and \&'HIT&, J . L. J . Chem. Soc. 1469. 1927.
3. HOWELL, 0. R . Proc. Roy. Soc., A, 104: 134. 1923.
4. HUTTIG,G. F . and I~ASSLER, R. Z. anorg. u. allgem. Chern. 184: 284. 1920.
5. LEWIS,J . R . J . Phys. Chem. 32:243. 1928.
(j. LEWIS,J . R. J . Phys. Chem. 32: 1808. 1928.
7. LISTER,NI. 14'. Call. J . Cheni. 30: 879. 1852.
8. LISTER,M. 1V. Can. J. C h e ~ n 31:
. 638. 1953.
9. LISTER,M. W. Can. J. Chem. 34: 465. 1956.
10. MOELWYN-HUGHES, E. rZ. T h e kinetics of reactions in solution. 2nd ed. T h e Clarendon
Press, Osford. 1933. p. 360.