the mechanism of corrosion fatigue of mild...

The mechanism of corrosion fatigue of mild steel

B y U R . E v a n s a n d M. T c h o r a b d ji S im n a d

{Communicated by H. J. Gough, F.R.— Received 23 February 1946)

The object of this research was to establish the mechanism of corrosion fatigue by the aid of chemical and electrochemical measurements, with special reference to the possibility of preventing corrosion fatigue by means of cathodic currents.

Two-stage tests on steel specimens subjected to alternating stress, with a chloride solution applied during the first stage only, have indicated that as the first stage (corrosion period) is increased, the total life, at first immeasurably long, becomes extremely short and then increases again. The unexpected increase in total life produced by an extension of the corrosion period may be explained by the fact that isolated cracks produce more stress intensification than a number of neighbouring cracks. Another series of experiments has shown that the rate of passage of iron into the combined state greatly increases with the applied stress.

The application of a cathodic current diminishes the rate of production of iron compounds and the number of cracks; weak cathodic currents actually shorten the life, but still stronger ones increase itagain—which can be explained in two ways: (1) for a given depth few cracks cause more weakening than many, (2) for a given amount of corrosion, an increased number of cracks means smaller depth of cracking and hence less damage. If the current reaches a certain value, corrosion becomes undetectable, and the life becomes extremely long in neutral potassium chloride (but not in acid). The value of the current needed for this complete protection increases with the stress range, but the value of the potential corresponding to the protective current moves steadily lower, i.e. in the direction of zinc, with applied stress.

Applying the graphical methods of representing corrosion phenomena to these results, we are led to the view that at least three different factors operate in causing alternating stress to enhance the rate of corrosion and the rate of mechanical damage. These are (A) diminution of cathodic polarization, (B) diminution of anodic polarization, and (C) diminution of the resistance of the path joining anodes and cathodes. It is possible that there may also be (D) a bodily shift of the anodic polarization curve in the base metal direction.

Studies of the shift of potential with time in presence of different types of stress indicate that stressing within the elastic range only affects the potential by altering the state of repair of the film covering the surface. It is likely that stresses within the plastic range depress the potential of the metal itself—irrespective of any damage to a film—but further work will be needed definitely to establish this point.

I n tro ductio n

Earlier researches. Corrosion fatigue, discovered by Haigh (1917), was studied in great detail by Me Adam (1926-31). The experimental data were brought together by Gough (1932), who associated the enhanced damage produced by superimposing cyclic stresses on corrosive action with the continued breaking of films; Bengough (1932) attributed it to an increased supply of oxygen.

Gough & Sopwith (1932, 1934), during a microscopic and X-ray examination of aluminium, observed (1) many small pits, (2) local attack with larger pits, and (3) preferential attack on slip-bands, leading to cracking. Gough (1932, 1935) obtained an enhanced life by working in vacuo. The peripheral extension of the cracks was studied by Bacon (1935), whilst Gould’s (1934) measurements provided further

[ 372 ]

on August 24, 2018http://rspa.royalsocietypublishing.org/Downloaded from

http://rspa.royalsocietypublishing.org/

evidence that failure is due to sharp-ended cracks rather than rounded pits—a distinction well brought out in the photographs of Me Adam & Clyne (1940).

The inhibitive action of chromates was demonstrated by Speller, McCorkle & Mumma (1928), and studied quantitatively by Gould & Evans (1939). The latter also measured the increase in life obtained by cathodic action, such as is produced by contact with zinc—an effect demonstrated in earlier work by Behrens (1933), Haigh (1929), Gerrard & Sutton (1935), Krystof (1935), and later studied by Stuart & Evans (1943), and by Huddle & Evans (1944), who used a paint richly pigmented with metallic zinc powder (Mayne & Evans 1944).

Plan of new research. The investigations mentioned above provided much information on the mechanical and crystallographic aspects of corrosion fatigue, but little regarding the chemistry. Doubt remained as to whether alternating stress increased the total rate of destruction or merely redistributed attack so as to produce greater mechanical weakening. In the present research, modern photoelectric methods of microanalysis have been used to obtain an answer to this question.

The influence of alternating stress upon corrosion has been attributed by different investigators to three main groups of factors:

(I) The distortion or obliteration of the crystalline structure of the metal, which thus becomes less stable and more reactive.

(II) The rupture of protective films, or removal of corrosion products which would otherwise slow down the attack.

(III) The improvement in the supply of oxygen, which is needed for the corrosion.A further object of the present research was to decide between these three

explanations.The matter can be restated in an electrochemical form. I t has in recent years been

established that corrosion by most salt solutions follows an electrochemical course, since electric currents flowing between discrete anodic and cathodic areas have been detected on naturally corroding specimens, and have been found strong enough to account for the corrosion actually observed. Translated into electrochemical nomenclature, the factors suggested above become:*

(1) a bodily shift of the anodic potential in the ‘base-metal’ direction;(2) (a) diminished anodic polarization, or (6) diminished resistance of the electro

lyte path between anodes and cathodes;(3) diminished cathodic polarization.Since cathodic stimulation tends to raise the compromise potential of a corroding

system, and anodic stimulation to depress it, the measurement of the electrode potential of metal subjected to corrosion fatigue should serve to distinguish between factors (1) and (2) on the one hand and (3) on the other.

In applying the electrochemical methods just indicated to elucidate the mechanism of corrosion fatigue, some information has also been obtained in regard to the

* Broadly speaking (1), (2), (3), are restatements of (I), (II), and (III) respectively, but D r J. N. Agar has kindly pointed out that (I) could give rise to (2a).

The mechanism of corrosion fatigue of mild steel 373



question as to whether stressing within the elastic or plastic range alters the electrode potential of a metal. Such questions possess a fundamental importance—quite apart from corrosion fatigue.

E x pe r im e n t a l m eth o ds

Materials. The specimens were wires 15 in. long and 0-1 in. diameter, produced from a single billet of mild steel by Messrs Brunton’s, Ltd.; drawing was carried out in two passes from patented steel rod 0-144 in. diameter. The ultimate tensile strength was found to be 48 tons/sq.in., and the fatigue limit of wires coated with vaseline containing zinc chromate +21 tons/sq.in. Chemical analysis, kindly arranged by the late Dr W. H. Hatfield, F.R.S., appears in table 1; microscopical examination showed freedom from inclusions.

T a b l e 1. Ch em ical com position op spe c im e n s

374 U. R. Evans and M. Tchorabdji Simnad

element content%

carbon 0-19silicon 0-07sulphur 0-03manganese 0-78phosphorus 0-02

The air-fatigue properties of such cold-drawn wire have been studied in detail by Gill & Goodacre (1936), Shelton & Swanger (1935), and by Godfrey (1941).

Apparatus. The experiments were performed in a room thermostated at 25 + 2-5°C, since Gould (1936) had shown that the effect of larger temperature fluctuations is not negligible in corrosion fatigue. A Haigh-Robertson wire-fatigue machine was adopted for its high frequency of stressing and suitable size. A wire bent into bow form is made to rotate about its own axis of flexure, being so gripped and loaded that fracture occurs very near to mid-span, where the bending stress is maximal, namely nE0d\2L for a specimen of length L, diameter d, angle of flexure 6, and elastic modulus E.

The machine was carefully earthed to prevent interference by stray currents. A counter was fitted to the motor for recording the number of cycles, and a check was kept on the speed of the motor by means of a stroboscope disk and an electric clock. Fluctuations on the mains voltage were corrected by using a ‘Raytheon’ voltage regulator in the main circuit, whilst a ‘Variac’ auto-transformer was used to control the speed of the motor.

The feeding arrangement. None of the methods used hitherto seemed suited to this research, since they either give no well-defined margin and flow of liquid to the wetted area, or interfere with the products of corrosion. The arrangement developed for feeding the corrosive liquid and applying a cathodic current is shown in figure 1. The solution is allowed to siphon from a reservoir into the glass tube A and down the vertical glass rod Bat a rate of 100 c.c./hr. The rod B, 5 mm. diameter, is waxed



completely except for an unwaxed channel along the side nearest to the wire. From a distance of about 1 cm. above the specimen, liquid descends along this unwaxed channel and a continuous ring of liquid is formed around the wire, held by capillary action. The liquid then descends into the horizontal tube G to which is fused a platinum foil anode. When a current is to be applied to the specimen, a high external voltage is used in conjunction with a high external resistance, so that small changes of resistance in the liquid portion of the circuit do not appreciably affect the current; the distance between the specimen and the platinum anode is kept constant. The liquid, flowing out of G through an opening at one end, is collected in a beaker for estimation of iron.


corrosive solution

- specimen

-bed o f machine

F ig u r e 1. The feeding arrangement.

General procedure. The specimen was tested for straightness, ground with 000 emery over a length of 2*5 cm. at mid-span, and degreased with acetone. Then polystyrene lacquer was applied to the whole of the ground length, except for a length of 0*5 cm. at its centre, so as to confine the corrosion to this area. A period of 60 min. always elapsed between the grinding and the beginning of the test.

The specimen was placed in the machine and the speed of the motor adjusted to 6000 r.p.m., a speed chosen on account of the conditions under which the rotating specimens became dynamically unstable (large vibrations occur near the natural frequency of lateral vibration, and a compromise was necessary between the highest speed possible for steady flow of liquid around the specimen and the critical value). The knurled head of the tailstock was turned until the headstock had moved through the angle corresponding to the required stress range, which was always well below



376

the air-fatigue limit. The frequency of stressing was adjusted, and the rod B and tube A were moved towards the specimen until the liquid flowing down the rod made contact with the specimen at mid-span. The counter was then read and the time recorded. On termination of the test, the time and the number of cycles endured were noted.

Except where otherwise stated, the corrosive liquid was m /1 0 potassium chloride shaken with air before use. The pH varied from 6*06 to 7*0 ; according to Fink, Turner & Paul (1943) changes of pH within still wider limits are of no consequence in corrosion fatigue.

The iron passing into the combined state (including any rust on the specimen, which was wiped off with filter paper) was estimated by a colorimetric method based on thioglycollic acid. The use of a Spekker photoelectric absorptiometer made possible the determination of quantities of iron as small as 0*005 mg.

Special procedures. In two-stage experiments, where corrosion was to take place only during the first stage, the flow of the corrosive was stopped after the chosen period and the rust on the specimen collected by wiping with filter paper. Next, while the specimen was kept rotating at a slow speed, the corroded area was well washed with distilled water and then was swabbed repeatedly with cotton-wool soaked in molar potassium chromate. The final rubbings were carried out with dry cotton-wool and filter paper. The experiment was then continued in air until fracture occurred.

Where it was desired to measure the electrode potential during a test, a saturated calomel electrode was put into communication with the specimen by means of a filter-paper strip, supported by the waxed rod B through a bridge of decinormal potassium chloride (figure 2). The electrical connexion consisted of a small copper brush which made contact with the chuck of the motor and led, in conjunction with a connexion from the standard electrode, to a Cambridge Valve Potentiometer. The

U. R. Evans and M. Tchorabdji Simnad

saturated potassium chloride

filter paper strip

L . - - T =

m / 10 potassium chloride

1/x/I

IV4 - corrosive solution

specimen

F ig u r e 2. Arrangement for electrode-potential measurements.



potentials could be read to within 0-5 mV, and the first reading could be taken about 30 sec. after the liquid had made contact with the specimen.

Apparatus for static stresses. Where it was desired to measure the potential on non-rotating specimens subjected to static stresses, a special apparatus was designed to keep a wire specimen bent into the form of an arc, as in the Haigh-Robertson machine. The bending stress was, however, measured in terms of the deflection Y a t mid-length, being n2E Y d /2L 2, where d is the diameter of the wire, L the length of the wire, and E Young’s modulus.

The apparatus is shown in figure 3. The fixed headstock A contains a Haigh- Robertson bearing B consisting of four ball bearings. The thrust exerted by the wire is transmitted through a sleeve to a single 5 mm. ball that runs on a trio of balls of the same size. The latter are contained inside a race in A. The tailstock consists of a threaded cylinder C, held in the threaded holes of the fixed support S. At the end B' of Cis a ball bearing, identical with that in A, for holding the other end of the specimen. In order to load the specimen the tailstock is rotated until the specimen is flexed to the desired extent by the longitudinal compression.


d icmmmmtspecimen

V :

F ig u re 3. Apparatus for static stress experiments.

Specimens were ground 24 hr. before each experiment over a mid-span length 2-5 mm., and coated with polystyrene lacquer except for a central area 1 mm. wide by 5 mm. long, the longer side being parallel to the wire axis. A strip of filter paper 1 cm. wide, wetted with m /1 0 potassium chloride, made contact with this exposed surface, and was kept in position by means of the special celluloid clip D. The other end of the filter-paper strip led, through an m /1 0 potassium chloride solution and a saturated potassium chloride bridge, to a saturated calomel electrode. The electric arrangements used during the measurement of the potentials were the same as those described above.

The character of the stress in the exposed area could be instantly varied by turning the wire on its own axis. Thus when the exposed area occupied the outer arc of the bent specimen it was in tension; when it occupied the inner arc it was in compression. The influence of tensile and compressive stresses on the potential could be followed continuously by turning the specimen so as to bring the exposed area alternately into compression and into tension through a neutral position; in this way, complications caused by the general drift of potential with time were eliminated.




E x pe r im e n t a l r e su l t s

Series I . One-stage experiments with corrosive liquid applied up to fracture. The relation between stress range and lives of specimens is shown in figure 4, which indicates an obedience to the usual relation between the stress range S and the number of reversals before fracture, N, namely N = Ce~kS, where G and k are constants.

life (cycles)

F ig u r e 4. Corrosion-fatigue lives to fracture.

Series II. Two-stage experiments with corrosive liquid applied during the first stage only. In the main set of two-stage experiments, the specimens were dried at the end of the first stage, and exposed to alternating stresses in air during the second stage, the stress range being the same as in the first stage. Subsidiary experiments were carried out in which the specimen was wetted with a solution of potassium chromate, an efficient inhibitor, during the second stage. These gave practically the same results as when the second stage was conducted in air, indicating that any trace of chloride left behind after the washing was without influence, and also that the cooling action of a liquid was unimportant.

In figure 5 the total fife (the sum of the two stages) is plotted against the corrosion period, both scales being logarithmic. I t is clear that wetting with a corrosive liquid during the first stage for a time longer than a certain critical period very greatly reduces the total life, which, in the complete absence of corrosion, would be infinite, since the stress range always lies below the fatigue limit; on the other hand, the application of corrosive liquid for a still longer period may actually cause the total life once more to increase. The causes will appear later.



Series III. Measurement of iron passing into the combined state. The results shown in figure 6 make it clear that alternating stresses do increase not only the rate of mechanical damage but also the rate of chemical corrosion—a matter on which doubts had been expressed. The relationship between the stress range and the average rate of corrosion over a period of 60 min. is shown in figure 7; the rate of corrosion includes both the iron present as semi-adherent rust and that found in the liquid which has run off the specimen.


± 12-5 tons/sq. in.-

±1 5 tons/sq. in.

± 17-5 tons/sq. in.

± 2 0 tons/sq. in. ~

D5 106 1total life (cycles)

F ig u r e 5. Two-stage corrosion-fatigue experiments.

Other experiments (figure 8) were carried out in n/10 hydrochloric acid. Here no insoluble corrosion product (rust) is formed. The rate of corrosion is constant until a fairly short time before breakage, when it becomes much faster. I t is worthy of notice that, just at the final stage during which the presence of corrosive liquid hardly affects the rate of mechanical damage, the mechanical action greatly enhances the rate of corrosion.

In addition, experiments were carried out in which the corrosion product was collected at intervals of 60 min. from the single specimen. In figure 9 the results




±20 tons/sq. in.

i±17*5 tons/sq. in.

±15 tons/sq. in.

± 12-5 tons/sq. in.

zero stress

time (min.)

F ig u r e 6. Initial rates o f corrosion.

4 iron in adherent rust

s tre s s ra n g e ( ± tons/sq. in.)

F ig u r e 7. Influence o f stress range on total iron corroded in 60 min.



are compared with the curve obtained from a number of specimens each subjected to corrosion fatigue for a different time, without interruption to rub off the rust. Comparison shows that the rubbing off of rust does appreciably increase the corrosion rate. Nevertheless the effect of alternating stress cannot be merely a loosening of the rust, since the quantity of rust remaining on the corroded specimen also increases with the stress range, as shown by the broken curve of figure 7.


zero

±22-5

time (min.)

F ig u r e 8. Rates of corrosion in n /10 hydrochloric acid.Figures denote stress ranges in tons/sq.in.

Series IV . Experiments with cathodic current superimposed. The results shown in figure 10a indicate that small cathodic currents actually decrease the life of specimens, but, as the current is increased, the life is prolonged again. The number of pits and cracks is found to decrease steadily with increase in the applied cathodic current, whilst the quantity of iron compounds produced decreases also, as shown in figure 11. At a certain current, which increases with stress range, the life becomes



382extremely long, the specimens being unbroken after 20 million cycles, and und r these conditions no pits or cracks can be seen, and no iron in the combined state detected. I t would seem, therefore, that a completely protective current does exist, a t which corrosion fatigue is entirely arrested. With such a current superimposed, the corrosion-fatigue curve almost coincides with the air-fatigue curve (figure 106). (Preliminary results with acid show that here a cathodic current, although diminishing corrosion, does not greatly prolong life.)


time (min.)

F ig u r e 9. Influence of removal of rust on corrosion rate.

Series V. Measurements of movement of potential with time under alternating stress. The curves plotted in figure 12 show that there is always an initial rise in potential lasting a few minutes followed by a fall in potential which is more rapid at high stress ranges than at low ones. In figure 13 a logarithmic time-scale is employed so as to include the results of more lengthy experiments a t low stress ranges. In all cases a final potential was reached which had roughly the same value for all ranges of stress. I t is only whilst the potential is falling with time that the difference between the curves representing various stress ranges is so marked.

Series VI. Measurements of movement of potential with time under steady stress. Previous attempts to study the effect of stress on potential have been complicated by the fact that no two pieces of metal behave exactly alike, and that, quite apart from




±20 tons/sq. in.

±17*5 tons/sq. in.

±15 tons/sq. in.

±12*5 tons/sq. in.

life (cycles)

F ig u r e 10 a. Influence of applied cathodic currents on total life to fracture.

air-fatigue curve

life (cycles)

F ig u r e 106. Influence of stress range on life to fracture with applied currents. Figures denote applied currents in micro-amperes.




.±12*5 tons/sq. in. tons/sq. in.

^ ^± 17*5 tons/sq. in. ± 2 0 tons/sq. in.

Fe/time

F ig u r e 11. Average rates of corrosion with applied currents.

-0 -5 -

^ 2 0 tons/sq. i„.

time (min.)

F ig u r e 12. Influence of stress range on time-potential curves.




time (min.)

F ig u re 13. Time-potential curves to fracture. Figures denote stress range in tons/sq.in.

Table 2. Influence of static stress on electrode potential

timepotential of exposed area when in*

(min.) A T C N x — T n 2- c

i 431 435 435 435 -0 0 4 000i 438 438 438 438 000 0002 456 456 457 456 000 0014 484 478 484 484 006 0007 — — 519 507 — 012

507 504 513 506 003 00710 — — 525 515 — 010

510 510 517 514 000 00315 — — 548 526 — 022

526 526 529 525 000 004_ 528 530 533 527 -0 0 2 006

20 553 560 550 536 -0 0 7 014545 548 547 540 -0 0 3 007

25 576 573 568 556 003 012562 566 564 554 -0 0 4 010565 569 567 555 -0 0 4 012

30 586 586 583 570 000 013578 581 581 571 -0 0 3 010

35 574 574 571 551 000 020562 567 567 552 -0 0 5 015

40 587 591 590 583 -0 0 4 007589 594 594 588 -0 0 5 006

45 642 645 645 643 -0 0 3 002648 649 649 647 -0 0 1 002

50 672 675 675 675 -0 0 3 000680 680 680 680 000 000

* N, T, and C denote ‘neutral’, ‘tension’, and ‘compression’, respectively. The potentials are in millivolts on the saturated calomel scale.



386

applied stress, the potential drifts with time. The method described above allows the potential to be measured on a single specimen a t comparable instants, so as to eliminate the effect of the general drift. The measurements recorded in table 2 show that a t any time compressional stresses give higher (more noble) values than when the test area is in the neutral position; when the area is subjected to tension, the values tend to be, if anything, below those obtained without stress. As in the case of the tests under alternating stress, the same value is reached, whether the area is compressional, neutral or tensional; this same value was obtained in special experiments where there was no turning of the specimen.


I n ter pr eta tio n of r e su l t s

Effect of corrosion period on life. The unexpected form of the curves in figure 5 may be explained by the shape of the pits. After short corrosion periods, these were rounded, whilst after longer periods they became elongated into cracks. Evidently the first effect of corrosion is to produce numerous pits, mostly hemispherical (as in the absence of stress). I f one of these happens to be deeper than its neighbours, the stress concentration at the bottom will increase the e.m.f. between the anodic point (the stressed region at the bottom) and the cathode (the external surface of the specimen). The extra attack a t the bottom will deepen and sharpen this particular pit, thus producing further stress intensification and further raising the e.m.f. Thus the pit, becoming steadily deeper and sharper, will penetrate in preference to its neighbours. However, as soon as the crack has become very sharp and deep, increase in resistance between anode and cathode will outweigh further increase in the e.m.f. Thus the rate of progress of the pioneer crack will fall off, and some of the neighbouring pits will in their turn develop into cracks.

If the corrosion stage is terminated at the period when isolated pioneer cracks have extended further than their neighbours, the stress concentration will already be considerable, and the subsequent life under dry fatigue will be short. When we reach the stage at which there are families of cracks instead of isolated cracks, the rapid increase of stress concentration will have ceased (Thum & Bruder 1938); there will then be only a gradual further shortening of the air-fatigue life.

I t is now possible to explain the experimental facts. In figure 14, the corrosion- fatigue period is plotted against the logarithm of the life in the corresponding second stage in air. Provided that the first stage is shorter than a certain value (the ‘incubation period’), the life endured in the second (dry) stage is very great. I f the first stage exceeds tha t value, the duration of the second stage is extremely short; the incubation period probably represents the period during which the pioneer cracks are developing. I f the corrosion stage is further prolonged, the shortening of the second stage proceeds more slowly than the corresponding lengthening of the first stage, so that the total life now increases with the increase of the corrosion period.



Certain relations, whose full significance cannot yet be assessed, were noticed between the incubation period T, the stress range S, and the total corrosion-fatigue life N as obtained in one-stage tests.

T — ae~bS, T = aNP,

T and N being expressed in cycles. The numerical values of the constants are

a = 5 x 107, b = 0-31; a = 48 x lO"8, /? = 2-45.

At each stress range, the minimum total life is approximately half the period endured when corrosion is continued to fracture.


±12*5 tons/sq. in.

±20 tons/sq. in.

air-life (cycles)

F ig u r e 14. Relation between corrosion-fatigue periods and corresponding air-lives in second stage.

The results have potential importance to engineers—assuming that they extend to the lower frequencies common in service. In presence of alternating stresses unintermittent precautions must be taken to exclude corrosive influences. If vigilance is relaxed even for a short period, the final life of the stressed members may be shorter than if the precautions had not been resumed. The crack obtained may resemble a dry-fatigue crack, and will probably be erroneously attributed to dry fatigue.

Effect of cathodic protection on life. The fact that small cathodic currents actually decrease life, whilst stronger currents prolong it, can also be explained. The number of cracks is observed to diminish steadily as the cathodic current increases; the weakening effect of a few isolated cracks will be greater than that of a family of neighbouring cracks, owing to the greater stress intensification in the first case. Also the cathodic currents seem to diminish the number of cracks more rapidly than

Vol. 188. A. 25



388they diminish the total corrosion, so that the intensity of attack will be increased, as in the case of a typical anodic inhibitor discussed elsewhere (Evans 1936; Evans & Chyzewski 1939).

The fact that corrosion is completely arrested by the application of a sufficient cathodic current density (figure 19) suggests a means of overcoming corrosion fatigue in absence of acid. Probably the cathodic protection can most easily be


current ->

F ig u r e 15

current -»■

F ig u r e 16

obtained by the application of special paints richly pigmented with metallic zinc (Mayne & Evans 1944; Huddle & Evans 1944). By rational application of electrochemical principles to the compounding of such paints, it is hoped to obtain better results than any yet achieved.

Mechanism of cathodic protection. In deciding between the three causes suggested for corrosion fatigue, the graphical method for corrosion velocity developed in this



laboratory will be used. The current flowing between anode and cathode (figure 15) is that value I which will produce an intercept IR between the anodic and cathodic polarization curves, where R is the resistance. The corrosion velocity will then be I /F, where F is Faraday’s number. If R is assumed to be small, the abscissa of the intersection can be taken to represent the corrosion current, and at the outset this assumption will be made.

If (figure 16) the anodic and cathodic curves, in absence of stress, are A P and CP, the corrosion current will be the abscissa of P. If, as in mechanism 1 (p. 373), the stresses are considered to shift the anodic curve downwards, say to A x,th e current will be increased to the abscissa of Pv If, as in mechanism 2, the effect of stress is to diminish anodic polarization, making the anodic curve less steep ( ), the currentwill become the abscissa of P2. If, as in mechanism 3, the anodic curve is left at and cathodic polarization is reduced, giving a new cathodic curve CPZ, then the enhanced current will be the abscissa of P3. Either type of anodic stimulation depresses the potential corresponding to the intersection, whilst a decrease of cathodic polarization will raise it.

The effect of excess cathodic current from an external anode is shown in figure 17, based upon the principles developed by Hoar (1938) and by Mears & Brown (1938). I f the applied cathodic current is represented by the length YZ, the corrosion current will be reduced from WPto XY(neglecting resistance and assuming that cathodic current does not alter distribution of anodic and cathodic areas). A current sufficiently strong to depress the potential to the value of an unpolarized anode, A , will prevent corrosion altogether.


current ->

F ig u b e 17

The experiments without cathodic protection showed that corrosion velocity increases with increase of stress range. Thus the corrosion current must have increased, and yet the potential was practically unaltered. This is best explained by assuming that mechanism 3 is acting along with mechanisms 2 (6), 2 (a) or 1. Since



390

the number of cracks per unit length was found to increase with stress range, the anodic polarization must have been diminished; also the resistance must have been diminished, so that, notwithstanding the increased current, the intercept should have been brought closer to the intersection. Thus mechanisms 2 (a) and 2 (b) can


current ->

F ig u r e 18

zero stress

±15 tons/sq. in.

± 2 0 tons/sq. in.

~0*60

corrosion current (fiA)

F ig u re 19. Relation between potential and corrosion currents as calculated from corrosion rates with applied currents. Figures at individual points denote applied currents in micro- amperes.



be assumed, without ad hoc assumptions, to operate. I t is not absolutely necessary to introduce mechanism 1 in order to explain the results.

Figure 18 suggests the changes brought about by increased stress. The current is increased from a value represented by the abscissa of P to that represented by the abscissa of P ', the potential remaining much the same.

The study of the fall of potential with time under alternating stress (series V) or under static stress, whether compressional or tensional (series VI), led to the same final value irrespective of stress range, or of the nature of the applied stress. This suggests that below the elastic limit, the effect of stress on the potential of metal not covered with a quasi-protective film is negligible. A stress which merely produces elastic deformation in metal may of course rupture a film, or open more widely any discontinuities already existing; this doubtless explains why the fall is more rapid a t high ranges of stress than low ones, and why compressional stresses (which should close up apertures) tend to raise the potential.

I t is probable, but not definitely proven, that stress in the plastic range (or residual stresses due to previous straining) does affect the potential.

We would like to thank Dr J . N. Agar, Dr A. J . Gould, Dr T. P. Hoar, Dr J . E. O. Mayne and Mr A. U. Huddle for stimulating discussion; thanks are also due to the Corrosion Committee of the Iron and Steel Institute, and especially Mr W. A. D. Forbes, for assistance in connexion with apparatus. One of us (M.T.S.) would acknowledge his indebtedness to the British Council for the grant of a scholarship.

R e fe r e n c e s

Bacon, F. 1935 Trans. N.-E. Cst Instn Engrs Shipb. 51, 216.Behrens, O. 1933 M itt. Wohler-Inst. Heft 15.Bengough, G. D. 1932 J. Inst. Met. 49, 93.Evans, U. R. 1936 Trans. Electrochem. Soc. 69, 213.Evans, U. R. & Chyzewski, E. 1939 Trans. Electrochem. Soc. 76, 215.Fink, C. G., Turner, W. T. & Paul, G. T. 1943 Trans. Electrochem. Soc. 83, 377.Gerard, I. J. & Sutton, H. 1935 J. Inst. Met. 56, 33.Gill, E. T. & Goodacre, R. 1936 Carnegie Schol. Mem. 35, 139.Godfrey, H. J. 1941 Trans. Amer. Soc. Metals, 29, 133.Gough, H. J. 1932 * Inst. Miet. 49, 1.Gough, H. J. 1932 J .Inst. Met. 49, 93.Gough, H. J. 1935 Proc. Staffs Iron Steel Inst. 50, 74.Gough, H. J. & Sopwith, D. G. 1932 Proc. Roy. Soc. A, 135. 392.Gough, H. J. & Sopwith, D. G. 1934 J. Inst. Met. 54, 193.Gould, A. J. 1934 Engineering, 138, 79.Gould, A. J. 1936 Engineering, 8 May.Gould, A. J. & Evans, U. R. 1939 2nd Rep. Alloy Steels Res. Comm. p. 325.Haigh, B. P. 1917 J ■ Inst. Met. 18, 55.Haigh, B. P. 1929 Trans. Instn Chem. Engrs, 7, 29.Hoar, T. P. 1938 J. Electropl. Depos. Soc. 14, 33.Huddle, A. U. & Evans, U. R. 1944 J . Iron Steel Inst. 149, 109P.Krystof, J. 1935 Metallwirtschaft, 14, 305.Mayne, J. E. O. & Evans, U. R. 1944 Chem. Ind. p. 109.




McAdam, D. J. 1926-1931 Proc. Amer. Soc. Test. Mater. (II) 26, 224; 27, 102; 29, 250; 30, 411; 31, 259.

McAdam, D. J. & Clyne, R. W. 1940 Bur. Stand. J . Res., Wash., 29, 685.Hears, R. B. & Brown, R. H. 1938 Trans. Electrochem. Soc. 74, 519.Shelton, S. M. & Swanger, W. H. 1935 Bur. Stand. J . Res., Wash., 24, 17.Speller, F. N., McCorkle, I. B. & Mumma, P. F. 1928 Proc. Amer. Soc. Test. Mater. 28, 159.Stuart, N. & Evans, U. R. 1943 J. Inst. Met. 147, 131Thum, A. B. & Bruder, W. 1938 Dtsch. Kraftfahrtforschung, No. 11.Whiteley, J. H. & Hallimond, A. F. 1925 Carnegie Schol. Mem. 14, 163.


The electric conductivity and the activation energy of ionic migration of molten salts and their mixtures

B y H . B loom a n d E . H e y m a n n

Chemistry Department, University of Melbourne, Melbourne, Australia

{Communicated by Sir David Rivett, F.R.S.—Received 4 March 1946)

The electric conductivity (k), and its variation with temperature, of many molten salts of predominantly ionic character can be represented by a simple exponential equation k = const, x e~clRT. Deviations from this relation are sometimes found for partially covalent compounds (e.g. ZnCl2, PbCl2) where constitutional changes may be expected with change of temperature. The activation energy of ionic migration (C) is always smaller than the activation energy of viscous flow. This fact is attributed to the difference in the configurational changes that occur in the two processes. For alkali chlorides, C decreases with increasing ratio of anion to cation radius. For electrolytes involving multivalent ions, C is greater than for uni-univalent ones. Increasing amount of covalency of the bonds involved tends to lower (7.

The conductivities of a number of mixtures of electrolytes (CdCl2-CdBr2, CdCl2-PbCl2, CdCl2-NaCl, CdCl2-KCl, PbCl2-KCl) were measured over a range of compositions and temperatures. The activation energies of ionic migration and, where possible, the equivalent conductivities were calculated, and the results discussed together with those obtained in other systems by various investigators.

In no system so far investigated is the conductivity a linear functiofT of the composition expressed as mole fraction. In systems which give no evidence of complex ion formation in the mixture, the conductivity usually shows moderate negative deviations from additivity (e.g. CdCl2-CdBr2). Only one system so far shows a positive deviation from additivity (CdCl2-PbCl2).

Strong negative deviations from additivity are found in systems in which complex ions are likely to exist in the mixtures (PbCl2-KCl, CdCl2-KCl, CdCl2-NaCl). In the systems CdCl2-KCl and PbCl2-KCl, the conductivity isotherms have jninima at all temperatures investigated; in these systems, the phase diagram indicates a congruently melting compound. Additional minima in the conductivity isotherms are found near compositions at which the phase diagram indicates incongruently melting compounds, but only at low temperatures; at higher temperatures, these minima disappear.

The activation energies (C) have maximum values near compositions that correspond to unstable compounds; in this case, C contains part of the energy change involved in the transition from the complex to the simple ions.

In some cases, the activation energy (C) of the molten systems rises to very high values as the crystallization temperature is approached. This is interpreted as being due to the existence of a high degree of order in the melt just above the melting-point.

Further relations between conductivity, and its temperature coefficient, the activation energy of ionic migration and the constitution of the molten salt mixtures are discussed.



the mechanism of corrosion fatigue of mild...

Documents