628 t. p. hoar. - proceedings of the royal...

628 T. P. Hoar.

first integral on the right-hand side of (31) cannot vanish since the integrand is essentially positive and hence

X j* (V — c) u . u dy < 0 ;

it follows from (30) that p, is zero.Therefore no real values of Roc exist corresponding to real values of c and all

profiles Y(y) are stable for disturbances of this degenerate type.

Summary.The stability of the flow of a viscous fluid between parallel walls for three

dimensional disturbances is discussed. A fourth-order differential equation is derived and it is shown that, if any velocity profile is unstable for a particular value of Reynolds’ number, it will be unstable at a lower value of Reynolds’ number for two-dimensional disturbances. Further, all profiles are shown to be stable for disturbances of a certain degenerate type.

The Mechanism of the Oxygen Electrode.

By T. P. H o ar , University Chemical Laboratory, Cambridge.

(Communicated by Eric K. Rideal, F.R.S.—Received July 15, 1933.)

1. Introduction.

I t is well known that the so-called “ oxygen electrode ” does not behave in a thermodynamically reversible manner. The decomposition voltage of water has been calculated thermodynamically from various calorimetric and solubility data by Lewis,* Nernst and von Wartenberg,f Bronsted^ and Lewis and Randall.§ The final critical value given by the last-named authors is 1-227 volt at 25° C., which should therefore be the e.m.f. of a cell consisting of a reversible hydrogen electrode and a reversible oxygen electrode immersed in the same electrolyte, both gases being at 760 mm. pressure. In practice this value has never been obtained. Smale|| found that the e.m.f. of the

* ‘ J. Amer. Chem. Soc.,’ vol. 28, p. 158 (1906). f * Z. phys. Chem.,’ vol. 56, p. 534 (1906). t Ibid., vol. 65, pp. 84, 744 (1909).§ ‘ J. Amer. Chem. Soc.,’ vol. 36, p. 1969 (1914).|| ‘ Z. phys. Chem.,’ vol. 14, p. 577 (1894).

on May 28, 2018http://rspa.royalsocietypublishing.org/Downloaded from

http://rspa.royalsocietypublishing.org/

M echanism o f the Oxygen Electrode. 629

hydrogen-oxygen cell, though independent of the of the electrolyte, was only 1*07-1-08 volt. Wilsmore* obtained a value of 1*07 volt, rising to 1 • 12 volt if the cell were allowed to stand for some days, while a similar result, 1*06 volt, was obtained by Crotogino.f More recently, Richards;}: has reported 0*979 volt, and Furman§ also obtains a value of about 0*98 volt.

Since it is well established tha t the hydrogen electrode behaves in a perfectly reversible manner in accord with thermodynamic laws, the discrepancy between the “ theoretical ” and experimental e.m.f. of the hydrogen-oxygen cell must have its origin in the oxygen electrode. I t is in fact experimentally found that oxygen electrodes, whether set up with bright or platinized platinum, (a) tend to be irreproducible, (b) do not obey the thermodynamic relation between electrode potential and partial pressure of oxygen, and (c) are readily polarized even by minute currents, thus failing to conform with any of the criteria of reversibility.

Various explanations have been put forward for the cause of the incomplete reversibility of the oxygen electrode. Tartar and Wellman|| suggest that it is due to the formation of hydrogen peroxide, for addition of hydrogen peroxide depresses the potential still further below the thermodynamic value. However, the formation of hydrogen peroxide appears to be very doubtful,** and the most widely held view is that oxides of platinum are formed on the electrode, which never becomes saturated with oxygen and thus never attains the reversible oxygen potential. Lorenz and his co-workersft and GrubeJJ found that the various platinum oxides gave electrode potentials very similar to that of a platinum electrode surrounded by oxygen. Again, by working at high temperatures where the oxides are unstable, using glass, porcelain, or fused alkali as electrolyte, Haber and his co-workers§§ obtained e.m.f’s of the hydrogen-oxygen cell very close to the calculated values. Richards]||| considers * * * § ** * * §§

* Ibid., vol. 35, p. 291 (1900).t 4 Z. anorg. Chem.,’ vol. 24, p. 258 (1900).t ‘ J. Phys. Chem.,’ vol. 32, p. 990 (1928).§ ‘ J. Amer. Chem. Soc.,* vol. 44, p. 2685 (1922).|| 4 J. Phys. Chem.,’ vol. 32, p. 1171 (1928).If Brislee, * Trans. Faraday Soc.,’ vol. 1, p. 65 (1905); Wilsmore, * Z. phys. Chem.,’

vol. 35, p. 291 (1900); Fisher and Kronig, ‘ Z. anorg. Chem.,’ vol. 135, p. 169 (1924).** Bornemann, ibid., vol. 34, p. 1 (1903).t t ‘ Z. anorg. Chem.,’ vol. 51, p. 81 (1906) ; ‘ Z. Electrochem .,’ vol. 14, p. 781 (1908);

vol. 15, pp. 157, 206, 293, 349, 661 (1909).XX 4 Z. Electrochem.,’ vol. 16, p. 621 (1910).§§ Ibid., vol. 12, p. 415 (1906); 4 Z. anorg. Chem.,’ vol. 51, pp. 245, 289, 356 (1906).1111 4 J. Phys. Chem.,’ vol. 32, p. 990 (1928).



630 T. P. Hoar.

his results are best explained on the oxide theory ; furthermore, Bowden* has put forward evidence for the formation of platinum oxides at a platinum anode, although his view is questioned by Butler and Armstrong, f

Besides having considerable theoretical interest, the oxygen electrode has been used for electrotitration, J and it plays an important part in the corrosion of metals§ ; it therefore appeared desirable to have more definite evidence as to the nature of the reactions involved. In the present investigation, the

kinetics of the dissolution and deposition of oxygen at several so-called “ inert” electrodes have been examined, and a tentative mechanism of the irreversibility is put forward. I t has also been found possible to deduce an approximate value for the e.m.f. of the hydrogen- oxygen cell from electrical measurements, and the result, 1 -20 ± 0-03 volt at 25° C., appears to be in closer agreement with the theoretical value of 1 • 227 volt than any previously obtained.

2. Apparatus.

The electrode vessel, fig. 1, consisted of a cylinder 15 cm. high by 4 cm. diameter. The electrode E under test was arranged centrally and horizontally, with the face on the underside, the back and edges being coated with an insulating

layer of resistant wax, to ensure that the current-density was as uniform as possible. Polarizing current could be supplied to the test electrode via a halfcell making liquid contact in the cup C1? while the electrode potential could be measured against a half-cell connected to C2. Calomel or quinhydrone

* 2 * 4 Proc. Roy. Soc.,’ A, vol. 125, p. 446 (1929). t Ibid., vol. 137, p. 604 (1932).J Furman, ‘ J. Amer. Chem. Soc.,’ vol. 44, p. 2685 (1922); Goard and Rideal, 4 Trans.

Faraday Soc.,’ vol. 19, p. 740 (1924); Britton, 4 J. Chem. Soc.,’ vol. 127, p. 1896 (1925).§ Aston, 4 Trans. Amer. Electrochem. Soc.,’ vol. 29, p. 449 (1916); Evans, 4 J. Inst.

Met.,’ vol. 30, p. 239 (1923); 44 Corrosion of Metals ” (Arnold) (1926); Evans, Bannister and Britton, 4 Proc. Roy. Soc.,’ A, vol. 131, p. 355 (1931); Evans and Hoar, ibid., A, vol. 137, p. 343 (1932).



Mechanism of the Oxygen Electrode. 631

electrodes, being only slightly polarizable, were used indiscriminately for the current half-cell. The half-cell used for potential measurement consisted, in experiments where the electrolyte was of pK < 7, of a quinhydrone electrode in the same electrolyte. In experiments in alkaline solution, where the quinhydrone electrode becomes inaccurate, a mercury/mercuric oxide alkali half-cell was used, the alkali being of the same* concentration as the main electrolyte ; this half-cell was standardized against a hydrogen electrode in the same alkali. All liquid junction P.D’s were thus avoided, and the potential of the test oxygen electrode could be immediately found with reference to a hydrogen electrode in the same solution.

The apparatus was contained in a brass box immersed in a water thermostat, to avoid risk of electrical leakage which might occur if the apparatus were directly immersed in water. The electrical leads were brought out through insulators in the lid of the box. The lid was heavily lagged with asbestos, to prevent loss of heat by radiation and convection, and a satisfactory temperature control of ± 0 -1 ° C. within the box was obtained.

Before a run, the electrode vessel and connecting tubes were swept out with purified oxygen admitted through tap Tv Electrolyte, which had been boiled out under reflux and saturated with oxygen at the temperature of the experiment, was then introduced through tap T2, until the electrode was immersed to a depth of 3-0 cm. ; preliminary experiments showed that this depth was immaterial. The connecting tubes and the cups C± and C2 were also filled with electrolyte, as shown, and the half-cells were then inserted so that they made liquid junctions in C± and C2. The arrangement of the liquid levels and the length of the connecting tubes effectively prevented any appreciable siphoning or diffusion in the system, and as an additional precaution all taps were kept closed.

The electrical lay-out is shown schematically in fig. 2. E.m.f. was measured by means of the valve electrometer of Evans and Hoar* which takes less than 5 X 10~12 ampere from the cell under measurement and is accurate to ± 0*001 v o lt; the voltmeter on which readings are obtained was checked against a standard Weston cell and a Cambridge Instrument Company potentiometer.

Current was supplied to the cell from a potential divider through a resistance box containing resistances ranging from 10,000 ohms to 10 megohms. This was constructed from radio resistances sealed into paraffin wax and was

* ‘ Proc. Roy. Soc.,’ A, vol. 137, p. 343 (1932).



632 T. P. Hoar.

frequently calibrated ; the resistances, though often 10-15% off the rated value, showed a satisfactory constancy. Current was measured by obtaining the potential drop across the resistance carrying it by means of the valve electrometer; the larger currents were sometimes measured on a Weston type 440 galvanometer or a Weston type 301 milliammeter.

EMF. HALF-CELL

VALUEELECmonETER

TEST ELECTRODE

CURRENT HftLFCELL

Fia. 2.

3. Preliminary Work.In the first experiments, the electrode was a circular piece of bright platinum

foil 1*10 cm. diameter, prepared by alternate heating in a spirit flame and immersion while hot in concentrated hydrochloric acid. The electrolyte was N/10 sulphuric acid, saturated with oxygen at 25*0° C. The potential difference between the electrode and the solution, referred to a hydrogen electrode in the same solution, which we shall henceforward shortly call the “ potential,” was initially about +0*85 volt, well below the theoretical value of + 1-227 volt.

The unused electrode was made slightly cathodic, and a current-potential curve obtained over a current-density range of 2 X 10~8 — 1 X 10~6 ampere/cm.2. The steady value of the potential for each applied current was reached rather slowly, but the reproducibility was moderately good. When, however, the anodic polarization curve was investigated, equilibrium was reached only after many hours even at fairly high (c. 5 X 10” 7 ampere/cm.2) current-density, while equilibrium points at lower anodic current-densities could not be obtained owing to the slow rise of the potential with time. These preliminary results made it clear that an irreversible process was occurring which prevented the equilibrium potential for a given anodic current-density being reached, and it was decided to investigate this in greater detail.



M echanism o f the Oxygen Electrode. 633

4. Experiments at Constant Electrode Potential.

Since the “ reversible ” oxygen potential is somewhat more noble than the open-circuit potential adopted by an oxygen electrode, such an electrode must be polarized anodically in order to bring it to the reversible potential. A bright platinum electrode 1 • 10 cm. diameter prepared as in Section 3 was set up in oxygen-saturated N/10 sulphuric acid at 25*0° C., and the anodic current-density required to maintain it at the reversible potential, viz., -J- 1-227 volt, was measured at intervals. I t was found that the necessary current- density decreased with time, at first rapidly and finally very slowly, becoming after some hours of the order of 1 — 2 X 10~8 ampere/cm.2. Similar results were obtained in N/10 sodium hydroxide (carbonate-free), though here the necessary current was, at the corresponding time, about half that required in N/10 sulphuric acid. After a run of 400 minutes the polarizing current was removed, and the potential returned slowly to about + 0 -9 8 volt for the acid and to about + 1 -0 1 volt for the alkali. If the electrode were allowed to stand overnight on open circuit, and a similar current-time experiment carried out on the following day, a more rapid fall of the current was observed. A third repetition gave a still more rapid fall.

A platinum electrode prepared by flaming in the manner described almost certainly possesses an oxide-film on the surface some molecules thick.* Such a film may be expected to contain cracks and pores, pervious to the electrolyte, which therefore also comes into contact with exposed platinum metal. The adsorbed oxygen on the film surface will render the film cathodic towards the exposed metal,*j* and current will therefore flow between film and metal, with cathodic dissolution of oxygen at the film surface and a consequent fall of its potential. The equivalent anodic reaction at the metal surface is probably oxygen-deposition, with subsequent formation of platinum oxide, though some platinum dissolution may perhaps occur, particularly in acid solutions. Any platinum ions formed will, however, at once be precipitated by the cathodically produced hydroxyl ions migrating inwards through the pores. The net result will be the formation within the pores of platinum oxide or hydroxide at a rate equivalent to the current flowing.

If the reversible oxygen potential is to be maintained at the film surface, the current necessary for this irreversible oxide-formation within the pores

* Jacobs and Whalley, ‘ Proc. Roy. Soc.,5 A, vol. 140, p. 489 (1933).t Evans, ‘ J. Chem. Soc.,’ p. 92 (1929); Bannister and Evans, ibid., p. 1361 (1930);

Muller, ‘ Mhft. Chem.,’ vol. 52, p. 221 (1929).

VOL. CXLII.— A. 2 T



634 T. P. Hoar.

must be supplied from elsewhere than the film, that is, from a separate cathode. Thus, an anodic current-density, of value such that the consequent fall of potential along the resistance of the pores is equal to the difference of potential between the film surface and the exposed metal surface, must be applied, and this is the current th^,t has been measured.

Since the film surface is maintained at the reversible oxygen potential, and the exposed metal surface may be expected to show approximately the platinum/platinum oxide potential, not much affected by current polarization, the drop of potential E volts along the pores may be regarded as roughly constant. Thus if the applied current-density is i amperes/cm.2 at time t seconds, and the electrolytic resistance of the pores is R ohms per cm.2 of film,

. iR = E = constant.

But the pore-resistance, say initially R0, must continuously increase as platinum oxide is deposited within the pores, and we shall assume that to a first approximation the increase of pore-resistance is proportional to the amount of oxide

formed by the process, that is, to [ i dt. ThenJo

i ( R0 + A | i dzj = E, A being a constant,

thereforea P % d = ? — :

Jo %or

A* - - ii* ' dt

Thus if the initial current-density is i0 = E/R0E r _ diA J ;0 i3

= A/1 _ 1 \2A ' i2 i02/ ( 1 )

an equation rather similar to one obtained by Muller and Konopicky* for somewhat different conditions. It has been shown experimentally that i0 is large compared with values of i obtained even after a few minutes, and the equation thus reduces to

\ — kt, (1a )

where k is a constant.* ‘ Mhft. Chem.,’ vol. 50, p. 861 (1928).



635

In fig. 3, 1/i2 is plotted against t, for bright platinum in N/10 sulphuric acid and N/10 sodium hydroxide. Excellent straight lines are obtained, the value of k being 7*65 X 1010 in the acid and 2-50 X 1011 in the alkali. When,

however, the 1 j t relation for the second experiment on an electrode is plotted,

the gradient of the curve obtained is at first greater than k (since i falls more

Mechanism of the Oxygen Electrode.

- 08

EXP. I .

20 0 10 TI ME (SECONDSx/03)

F ig. 3.

rapidly with t than with a fresh electrode), but eventually the curve approaches asymptotically to a straight line of gradient k which can be written

~ = kt + l, (1 b )i*

Here l represents the constant term of equation (1), which cannot now be neglected, since the treatment of the electrode in the first experiment has greatly increased the initial pore-resistance. The fact that the second experiments show early values of i much greater than those deducible from the straight

part of the \ j t curves probably indicates that the film has undergone a certain

2 T 2



636 T. P. Hoar.

amount of spontaneous cracking while standing overnight between the two experiments.

The third experiments, not shown in the figure, gave results very similar to the second experiments.

The curves given have been selected as typical of three concordant sets of experiments.

It is of interest to consider the extent of the irreversible process, which will

be equivalent to j i dt coulombs per/cm.2 at time t seconds. Jo

400-minute run,Number of coulombs passed —

r24,000i dt.

, o

Thus in a

But 1/i2 = kt, i.e., i = 1/kh*, and for N/10 sulphuric acid k = 7*65 X 1010. Therefore

c24,000 f i - 0 6 V 1 0 23Number of OH' ions discharged = id t X -------------- .

8 J o 96,500

_ r24’000 dt w 6-06 X 1023Jo (7-65 X 1010) ¥ X 96,500 5

= 7-0 X 1015.

Similarly in N/10 sodium hydroxide, k = 2-50 X 1011, and 3*9 X 1015 OH' ions are discharged in a 400-minute run. These amounts would produce one or two molecular layers of oxide if distributed over the whole of an “ apparent ” cm.2 of the electrode surface, which for bright platinum has a “ true ” surface area of 2-3 cm.2.* But, if the initial pore-area of the oxide-film is only some 1/10 of the whole, it will be seen that the extra amount of oxide produced in a run is sufficient to close up the pores to a considerable extent, and cause a much increased pore-resistance, in harmony with the hypothesis developed above.

Even if the above explanation of the irreversible process is incorrect or incomplete, it has been shown that the applied anodic current-density to which the process is equivalent rapidly becomes very small, being of the order of 1 — 5 X 10~8 ampere per apparent cm.2 at the reversible potential after a few minutes. I t therefore appeared desirable to study the main cathodic and anodic reversible reactions at considerably higher current densities where the irreversible reaction would be of relatively small extent.

* Bowden, 4 Proc. Roy. Soc.,5 A, vol. 125, p. 446 (1929).




5. The Relation between Current-density and Electrode Potential.

Cathodic and anodic polarization curves, for several electrodes in various electrolytes, with current-densities between about 10-7 and 10~4 ampere per apparent cm.2 of electrode surface, were obtained as follows. The electrode system was set up and allowed to stand for about 1 hour, when the potential had become nearly constant. The cathodic polarization of the electrode was then investigated over the range of current-density given above; the equilibrium potential at each fixed current-density was usually attained in 5-15 minutes. After this cathodic polarization, an anodic current-density of c. 10~6 ampere/cm.3 was applied for some 18 hours, by which time the rate of the irreversible process had, as expected, become very small, for the potential and current- density wrere very nearly constant in time. Further points on the anodic polarization curve could then be obtained, the equilibrium potential at each current-density being attained in 5-15 minutes. A satisfactory agreement between points obtained with successively increasing and decreasing current- densities was achieved.

The following systems were studied ; each electrolyte was saturated with oxygen and the temperature was 25*0° ± 0-1° C. :—

(1) Bright platinum in N/10 sodium hydroxide.(2) Bright platinum in M/15 sodium phosphate buffer, p n = 7-0.(3) Bright platinum in N/10 sulphuric acid.(4) Black platinum in N/10 sodium hydroxide.(5) Bright gold in N/10 sodium hydroxide.

The results are shown in figs. 4 and 5 as plots of electrode potential in volts referred to hydrogen in the same solution, V, against log10 (current-density, i§ or ia, in amperes per apparent cm.2). I t will be seen that both the log10 iJ Y and log10 iJ Y curves, representing the cathodic and anodic processes respectively, are linear over a considerable range,* but begin to depart from linearity in opposite senses at lowr current-densities comparable to the rate of the irreversible process. The constants of the straight portions, and of two repeats of system (1) not shown in the figures for the sake of clarity, are given in Table I, columns 4-7.

* Cf. Tafel, 4 Z. phys. Chem.,’ vol. 50, p. 641 (1905); Bowden, 4 Proc. Roy. Soc.,’ A, vol. 126, p. 107 (1930).



638 T. P. Hoar.

6. Deduction of the Reversible Oxygen Potential.The reversible oxygen potential is obtained thermodynamically on the

assumption that the net electrode process a t an oxygen electrode is

0 2 + 2H20 + 4s ^ 40H'.

It must, however, be remembered th a t this process undoubtedly takes place

Loq io (Cathodic Current-Density in Ah peres/ cm?)P ig. 4.



639

in a series of steps. Thus, it is very possible that molecular oxygen is first adsorbed on the oxide-film by Van der Waals’ forces, becoming subsequently chemi-adsorbed either as molecules or atom s; reaction with cations or discharged cations then takes place, the equivalent of hydroxyl ion eventually

Mechanism of the Oxygen Electrode.

Loc- io (Anodic Current-Density in /\mperes/cm.z)F ig. 5.

being formed. The rate of the reaction will, of course, be limited by that of the slowest stage.

We have seen in Section 5 that the rates of both the cathodic and the anodic processes vary exponentially with the potential over a considerable range. It has been theoretically shown by Gurney* that in accordance with quantum

* ‘ Proc. Roy. Soc.,’ A, vol. 134, p. 137 (1932).



640 T. P. Hoar.

Table I.—Summary of Polarization Curves.The linear portions of the curves may be represented as V = Ac log10 ic + Bc an

V = Att log10 ia + Ba. Values of Ac, Bc, Aa and Ba are given for V in volts agains hydrogen and ici ia in amperes/cm.2. The last two columns give the simultaneous solution of the cathodic and anodic equations for V and log10 i.

Electrode system.

1

No.of

system.

1Open-circuit

potential(volts

against H 2)

1Cathodic

constants.Anodic

constants.[logio »]b. Vb.

A(». Bc. Aa. Ba.

Pt, bright, N/10 NaOH ............... la +0-988 -0 -0 4 8 +0-592 +0-0641

+2-01 13-34 + 1-26

lb +0-982 -0 -0 5 5 +0-513 +0-074 +2-10 13-78 + 1-19

1 c +0-973 -0 -0 4 6 +0-615 +0-068 + 2-03 13-52 + 1*19

Pi, bright, M/15 phosphate buffer 2 +0-936 -0 -1 1 8 -0 -129 +0-101 +2-33 12-53 + 1-22

Pt, bright, N/10 H ,S04 ............... 3 +0-828 -0 -0 9 7 +0-111 +0-086 +2-13 12-98 + 1*19

Pt, black, N/10 NaOH ............... 4 + 1-057 -0 -0 4 6 +0-690 +0-078 + 1-97 IT-68 + M 7

Au, bright, N/10 NaOH ............... 5 + 1-099 -0 -0 4 8 +0-586 +0-078 +2-12 13-73 + 1-17

Mean (omitting (4) and ( 5 ) ) .......... . +1-20

mechanical principles, the rate of discharge of an ion at an inert electrode is an exponential function of the electrode potential. It would appear from his discussion that the rate of the reverse process, namely, the formation of an ion from a neutral molecule or atom, is also an exponential function of the potential, of course in the opposite sense. It is therefore very probable that the stage limiting the reaction rate is one involving electron transfer, quite possibly the chemi-adsorption stage.

In any case, if the rates of the charge and discharge processes (whatever they may be) are px and p2 gram-equivalents per second per cm.2 of electrode surface, then

total cathodic current-density on oxide-film surface = (px — p2) F,

where F is Faraday’s number. This is the sum of the applied cathodic current- density ic and that due to the irreversible current flow through the pores of the oxide-film, say ip. Therefore,

h + i9 = (pi — P2) F . (2)




p2 quickly becomes quite negligible even at very small values of ie, and at somewhat higher values ip is also negligible. Here

*. = PiF, (2a)

and the straight portion of an experimental log10 iJ Y curve, fig. 4. is therefore equivalent to the exponential relationship between p5 and V. For smaller values of ic, where ip cannot be neglected,

ic = PiF — ip, (2b)

in harmony with the fact that the log10 iJ Y curve here falls off from the straight line in the sense that ic is too small for a given value of V.

Similarlytotal anodic current-density on oxide-film surface = (p2 — Pi)F.

The applied anodic current-density is equal to the sum of this and the anodic current-density supplied through the pores for the irreversible process, ip.

Thereforeia ip — (p2 Pi)®1- (3)

At sufficiently high values of ia, px and ip can be neglected. Hence

ia = ?2̂ > (^A)and the straight part of a log10 iJ Y curve, fig. 5, therefore gives the exponential relation between p2 and V.

For smaller values of ia,ia — p2F + ip, (3b)

in agreement with the fact that the log10 iJ Y curve here departs from linearity in the sense that ia is too large for a given value of V.

At the thermodynamically reversible oxygen potential, the total reversible electrode reaction must be in dynamic equilibrium, and hence the rates of the charge and discharge processes must be equal, i.e

Pi — ?2*The linear portions of the cathodic polarization curves of fig. 4 are log10 p^F/V curves, while the linear parts of the anodic curves of fig. 5 give the log10 p2F/V relation. The reversible oxygen potential VR is therefore obtained by putting Pi — P2 and solving the simultaneous equations representing the linear parts of corresponding cathodic and anodic curves for V. This is done in Table I, column 9, and graphically by extrapolating corresponding cathodic and anodic curves till they intersect, fig. 6. In all cases the result is in satisfactory agree-



642 T. P. Hoar.

ment with the thermodynamically calculated value of + 1 -227 volt. Systems (4) and (5) are less experimentally reliable than the remainder, and have been omitted in calculating the mean value of + 1 -2 0 volt, with a calculated experimental error of _-t O'03 volt, at 25• 0° C.

— la

sj IS

REV. 02. POTENTIAL

PREVIOUS aC POTENTIALS

-10 - q -8 • -7Lot lo f tF and LoQIOf£>zF

F ig. 6.

7. Discussion.Since the value for the reversible oxygen potential just obtained is in agree

ment within the experimental error with that calculated thermodynamically, there appears to be good justification for the theoretical extrapolation involved, and for the assumption that the net electrode reaction is in fact

0 2 + 2H20 + 4s 40H'.




It may also be noted that we have tacitly assumed that the cathodic and anodic polarization curves obtained on any one electrode system are measures of the forward and reverse reaction velocities under identical catalytic conditions, although the electrode surface must have been somewhat changed by the irreversible process taking place during the two runs. But if, as has been suggested, the surface is always covered almost completely with an oxide-film, and the irreversible process occurs merely at the small cracks in this, producing more oxide, the assumption is justifiable ; for the cathodic and anodic processes both take place on a nearly complete oxide surface, which is sensibly the same for both. Furthermore, although the reaction rates found in the triplicate experiment on system (1) are not very closely reproducible owing no doubt to the general irreproducibility of a platinum surface, nevertheless the equilibrium potentials deduced are in very close agreement—alteration of the catalyst may change the rate but not the equilibrium.

The range of potentials found by previous investigators for experimental oxygen electrodes on open-circuit is indicated in fig. 6. The points marked with circles on the extrapolated log10 pxF/V curves at the potentials given by the various electrodes on open-circuit (Table I, column 3), accord with the previous work. The corresponding open-circuit values of pxF are of about the same order of magnitude, 10~8 ampere/cm.2, as the applied anodic current- density necessary to maintain the electrode at the reversible potential as in Section 4. Since on open-circuit the applied current is zero, these values of pxF represent the current-flow through the pores in the oxide-film from the cathodic film to the anodic base of the pores (see equation (2 b ) ). An oxygen electrode on open-circuit is thus “ self-polarized,” and is undergoing a slow irreversible oxidation, the rate of which can be found by extrapolating the straight part of the cathodic polarization curve to the open-circuit potential.

Furthermore, the equal values of pxF and p2F at the reversible potential, viz., about 10~13-10~12 ampere/cm.2 for bright platinum (Table I, column 8, and fig. 6) are several orders of magnitude smaller than the open-circuit values of pxF and the smallest value of i observed in the experiments of Section 4, viz., about 10~8 ampere/cm.2 after some hours. Even if the relation there found, 1/i2 = kt, holds up to very large values of t, which is unlikely, owing to spontaneous film-breakdown, calculation shows that it would take at least 10,000,000 years for i to descend even to the same order magnitude as pxF and p2F at the reversible potential, so it is not surprising that an oxygen electrode showing the reversible potential on open-circuit has never been prepared.



644 T. P. Hoar.

It appears that the low potentials given by the oxygen electrode are due not so much to the presence of the oxide-film as to its permeability to the electrolyte, which gives rise to self-polarization. A perfectly impermeable oxide-film should give the reversible oxygen potential, whereas a very porous film should give the lower metal/metal oxide potential.

Much of the previous experimental work falls into line with this hypothesis. The rise of the open-circuit potential with time, noted by many previous workers and confirmed in this investigation, is due to the gradual closing up of the pores by the self-polarizing process, the rate of which therefore continually decreases. Lorenz’s observation* that silver, nickel, copper, iron and zinc, when used as “ inert ” bases for oxygen electrodes, give potentials almost identical with those given by their oxides, is readily explained, since the oxide- films of these metals are known to be easily penetrated by electrolytes.*)* Again, Richards J found that increase of the concentration of the electrolyte produces a lowering of the potential of the ordinary platinum oxygen electrode ; it would seem that the higher conductivity reduces the pore-resistance and so increases the self-polarization. Smale,§ studying the effect of different electrolytes on the hydrogen-oxygen e.m.f., obtained lowest values (+ 0*88— f- 0-96 volt) in hydrochloric acid and the alkali chlorides, and highest (+ 1 -08 + 1*09 volt) in the alkali hydroxides. Now chloride ion is particularly effective in promoting oxide-film breakdown, while hydroxyl ion effects repair|| ; presence of chloride ion should therefore cause increase, and of hydroxyl ion decrease, of the self-polarization, which accords with Smale’s results. Similarly, the depressing effect of hydrogen peroxide on the potential of the oxygen electrode may be due to decomposition of the peroxide at the platinum surface, with consequent mechanical disruption of the film by the oxygen gas produced.^

There is thus a dual difficulty involved in the preparation of a reversible oxygen electrode. In the first place, the electrode reaction must perforce take place on an oxide surface. I t is therefore not surprising that it is slow

* ‘ Z. Electrochem.,’ vol. 14, p. 781 (1908).t Cf. Evans and Stockdale, ‘ J. Chem. Soc.,’ p. 2651 (1929); Bannister and Evans,

ibid., p. 1361 (1930); Muller, ‘ Mhft. Chem.,’ vol. 52, p. 221 (1929); Hoar and Evans, ‘ J. Chem. Soc.,’ p. 2476 (1932); Hoar and Evans, * J. Iron and Steel Inst.,’ vol. 126, p .3 7 9 (1932); Bengough, Lee and Wormwell, ‘ Proc. Roy. Soc.,’ A, vol. 134, p. 308 (1931).

f ‘ J. Phys. Chem.,’ vol. 32, p. 990 (1928).§ * Z. phys. Chem.,’ vol. 14, p. 577 (1894).|| Evans, ‘ J. Chem. Soc.,’ p. 1020 (1927); p. 92 (1929); Britton and Evans, ibid.,

p. 1773 (1930).If Cf. Hoar and Evans, * J. Chem. Soc.,’ p. 2476 (1932), who found that hydrogen peroxide

intensified the breakdown of the oxide-film on iron in chloride solution.




compared with, for example, the corresponding reaction at a hydrogen electrode, which takes place on the much more catalytically active metallic surface.* This slow reaction rate would not of itself depress the oxygen electrode potential, were it not that the porosity of the oxide-film allows the self-polarizing current to flow between the film and the underlying metal. Although the current is very small, it is yet large compared with the reaction rate at the reversible potential, and thus produces considerable self-polarization.. The criterion which a surface must satisfy in order to function as an inert basis for a reversible oxygen electrode is that it shall catalyse the reversible reaction at the reversible potential at a rate relatively rapid to the rate at which it allows electromotively active material to be removed by self-polarization, and with this criterion it appears at present impossible to comply.

The formation of an oxide surface, and the consequent very slow velocity of the cathodic dissolution of oxygen, at any metallic surface exposed to oxygen- containing electrolyte, has an important consequence in determining the velocity of metallic corrosion. As has been shown in previous work,*)* most metals on exposure to oxygen-containing electrolytes become anodic at those points where their initial air-formed oxide-film is weakest, and relatively cathodic at other points, particularly those best supplied with oxygen. Current therefore flows, with oxygen dissolution at the cathodes and metallic dissolution at the anodes. The basic potential at the anodes is determined by the nature of the metal and the solution, and is not much affected by current- polarization ; but the cathodic potential is greatly affected, and tends to approach the anodic value. Even at such low values of cathodic potential, the rate of oxygen dissolution is very slow, and this limits the rate of corrosion.

Thus, while the slowness of the oxygen hydroxyl ion reaction has the unfortunate academic consequence that the reversible oxygen electrode cannot be prepared, it nevertheless has the much more important practical effect of rendering the corrosion of metals in aerated electrolytes comparatively slow.

I am very grateful to the Master and Fellows of Sidney Sussex College, Cambridge, for a Research Studentship, and to the British Non-Ferrous Metals Research Association for a grant, which have made this investigation financially possible.

* Cf. Bowden, 4 Proc. Roy. Soc.,’ A, vol. 125, p. 446 (1929).t Evans, 44 Corrosion of Metals ” Arnold (1926); Evans, 4 J. Franklin Inst.,’ vol. 208,

p. 45 (1929); Evans, Bannister and Britton, 4 Proc. Roy. Soc.,’ A, vol. 131, p. 355 (1931); Evans and Hoar, ibid., A, vol. 137, p. 343 (1932).



646 Mechanism of the Oxygen Electrode.

My thanks are also due to Mr. W. J. Palmer for the loan of a hydrogen electrode, and especially to Dr. U. R. Evans for his kindly interest and stimulating suggestions throughout the progress of the work.

Finally, I am greatly indebted to Professor E. K. Rideal for his valuable advice and for communicating this paper.

Summary.The kinetics of the processes which occur at the so-called “ oxygen-electrode ”

have been investigated. The results are consistent with the hypothesis that the total electrode process can be represented by the reversible reaction.

0 2 + 2H20 + 4s 40H',

which takes place, no doubt in stages, on the surface of the oxide-film with which the inert electrode is covered. Since the oxide-film contains pores pervious to the electrolyte, current flows between the film surface and the relatively anodic metal of the base of the pores, causing an irreversible removal of electromotively active material from the film surface, and a lowering of the potential. . To maintain the reversible potential a small anodic current- density % is needed ; this is shown to decrease with time t according to the relation

1 /i2 = kt,

which is theoretically justified on the assumption of further oxide formation within the pores.

Cathodic and anodic polarization curves have been obtained for oxygen electrodes formed by platinum and gold in various oxygen-saturated electrolytes, and the logarithm of the current-density is shown to bear a linear relation to the electrode potential except at very low current-densities. By an extrapolation of the anodic and cathodic curves, involving certain reasonable assumptions, a value of + 1*20 i 0-03 volt at 25-0° C. is obtained for the reversible oxygen potential referred to hydrogen in the same solution, in good agreement with the value of -f- 1*227 volt calculated from thermal data by Lewis and others.

The rate of the reversible process at the reversible potential is extremely slow, and reasons are given for the belief that it is consequently impossible to prepare a truly reversible oxygen electrode.

The importance of the sluggishness of the oxygen hydroxyl ion reaction in limiting the velocity of metallic corrosion is pointed out.



628 t. p. hoar. - proceedings of the royal...

Documents