A paper presented at the Eighty-first Gen- eral Meeting, held at Nashville, Tenn., April 18, 1942, R. B. Mears presiding.

SOME CORROSION CHARACTERISTICS OF HIGH PURITY MAGNE.51UM ALLOYS 1

By R. E. McNuLTY 2 AND J. D. HANAWALT 3

ABSTRACT The corrosion behavior of high purity magnesium alloys has been

followed by means of the hydrogen evolution method. These data, combined with anodic and cathodic polarization curves, suggest an elec- trochemical mechanism for the process. Although trace impurities are still effective in the steady state, the encroachment of the anodic film over the cathodic particles causes a high "pore resistance" which tends to stifle the reaction. The characteristics of the hydrogen evolution curves are thus explained.

INTRODUCTION A previous paper 4 recorded a study of the salt water corrosion be-

havior of some common magnesium alloys as a function of composi- tion and of per cent of additional elements present such as Fe, Ni, Cu, Si, Pb, etc. Corrosion behavior was measured in terms of mg./cm.2/day average weight loss over a four months' period of exposure to salt water.

It was observed that for magnesium and certain alloys there exist "tolerance .limits" for certain elements. The meaning of "tolerance limits" is that, when the amount of the element present exceeds a par- ticular critical amount, there results a large increase in corrosion rate. Data were also presented from measurements of solution potential and hydrogen overvoltage whi'ch served to show which elements are detri- mental to corrosion resistance of magnesium.

The assumptions made were that these elements, which are dis- tributed throughout the alloy as discrete particles, a c t ed as cathodic electrodes of local cells and that the effect of any anodic polarization was negligible. The explanation proposed for the existence of those "tolerance limits" which do not correspond to solubility limits as ob- served was that, when the element was below the critical amount, the corrosion which takes place does not expose new particles, but that above this amount the particles are numerous enough and close enough together so that the corrosion caused by the particles continually exposes new particles to the corroding medium. Whether this concept should

1 Manuscript received January 15, 1942. z The Dow Chemical Co., Midland, Mich. a The Dow Chemical Co., Midland, Mich. *J. D. Hanawalt, C. E. Nelson and J. A. Peloubet, Metals Teeh. 8, 6 (1941).

423

424 1~. E. MCNULTY AND J . D. HANAWALT

be applied to the individual particles themselves or rather to individual groups or segregations of particles is not clear at the moment.

The emphasis of the earlier paper was on the point that the inherent or true corrosion stability of magnesium and its alloys in salt solution had no chance to show itself as long as certain elements exceeded their tolerance limits, because of the drastic effects of the local cell action in destroying film formation.

It is the purpose of the present paper to study the corrosion charac- teristics of some common magnesium alloys of high purity by observing the corrosion rate through the period covering the initial contact to the steady state in an effort to discover what factors determine the initial behavior and the steady state. These experiments are supplemented by measurements of the anodic and cathodic polarization curves at various times within the period above described.

METHOD

A. Hydrogen Evolution. It is generally accepted that the corrosion of magnesium and its alloys in aqueous chloride solutions is quantita- tively represented by the amount of hydrogen gas liberated in the reac- tion. 5 In the present investigation samples 20 cm. 2 in area were ground down to #320 aloxite paper, brushed free of particles and immersed without handling in neutral 3% NaCI solution. Since the volume of the solution was very large with respect to the sample, the overall pH remained approximately neutral during the 'course of the reaction. The process of corrosion was continuously observed by the determination of the amount of hydrogen evolved. The results are the mean of several observations at room temperature which was 25 ~ • 3 ~ C. Additional samples which received metallographic polishes yielded corrosion curves of a similar nature.

B. Anodic Polarization. Because in magnesium the active cathodic electrodes are microscopically small particles spaced 0.001 in. (25 t~) or less apart, the isolation of these areas as accomplished by Brown and Mears for aluminum 6 is not experimentally possible. Mears 7 has reported experiments on magnesium corrosion in dilute chloride solu- tions inhibited with NaF2, in which he artificially created anodic areas by scratching off the fluoride film. This technic is not applicable to the present study of magnesium corrosion in salt water.

Anodic current density values are chosen on the assumption that the area of the specimen is substantially all anodic and the current is equivalent to the hydrogen evolved. Anodic polarization values were obtained by passing current from an external source through a neutral chloride solution with a magnesium anode and a platinum cathode. Potentials were measured with a Leeds & Northrup special wide-range type K potentiometer using a Weston unsaturated cell as standard. The reference electrode was the saturated calomel type. Precautions were taken to prevent diffusion at the liquid boundary and frequent checks of the working calomel cell against a reference gave differences never

L. Whltby, Trans. Faraday Soc. 28, 353 (1932). 6R. H. Brown and R. B. Mears, Trans. Electrochem. Soc. 74, 495 (1938).

L. J. Benson, R. H. Brown and R. B. Mears, Trans. Electrochem. Soc. 76. 260 (1939).

CORROSION CHARACTERISTICS OF MAGNESIUM ALLOYS 425

exceeding 0.0002 v. Data were taken corresponding to three points on the corrosion curves: upon initial immersion, after 24 hours and after 96 hours.

C. Hydrogen OvervoItage. The hydrogen overvoltages, defined as the negative potential which must be applied to liberate hydrogen gas at a specified current density, have been determined for the common cathodic impurities in magnesium alloys and for highest purity mag- nesium by passing a current through the cell, metal cathode [ 3% NaC1 solution [ platinum anode, and measuring the increase in potential above the zero current value with respect to the saturated calomel cell. The values for a current density of 5 milliamp./cm. 2 are taken as the normal overvoltages. The magnesium base for the experiments was purified by sublimation. Table I gives a typical composition of the specimens studied.

TABL~ I Typical Analysis of the Magnesium Metal Tested

A1 as shown* Mn as shown Fe as shown Ni 0.0001% Cu 0.0001 SI 0.001

Pb 0.01% Sn 0.0001 Cd 0.0001 Zn 0.001 Na 0.002 Ca 0.001

* See Legends of figures.

GC. H 2 EVOLVED PER CM~

1.4 ~B

1.2

I.O

0.8

0.4

0.2

0 0 4 8 12 16

TIME- DAYS

FZG. I. Progress of corrosion as shown by hydrogen evolution curves. Composition of Alloys.

A. Mg-0.032% Fe E. Mg-6% Al-0.001% Fe IL Mg-6% AI-0.006% l~e F. Mg-0.010% Fe C. Mg-6% AI-0.002% Fe G. Mg-1.5% Mn-0.006% Fe D. Mg-0.0006% Fe H. Mg-6% AI-0.3% Mn-0.001% Fe

20

426 R . E . MCNULTY AND J . D. t I A N A W A L T

RESULTS

The results of the H2 evolution experiments are presented in Fig. 1. The initial rate of I-I2 evolution over the first few hours for the high purity alloys corresponds to an anodic current density of about 100 microamp./cm. 2, while the rate after reaching the steady state is of the order of 1 microamp./cm3 Anodic polarization characteristics initi- ally, after 24 hours and after 96 hours are shown in Fig. 2, 3, and 4 respectively. Hydrogen overvoltage measurements are given in Fig. 5.

POTENTIAL 1.80

1.70

1.60 ~ ~'G

1 .50 " ' 0 I0 2 0 3 0 4 0 5 0 6 0

CURRENT DENSITY-/uA/CM. 2 FIc. 2. Anodic polarization characteristics of high purity mag-

nesium alloys upon initial immersion in 3% aq. NaC1. Compositions.

A. Mg-0.0006% Fe E. Mg-6% AI-0.3% lXIn20.001% Fe B. Mg-6% AI-0.001% Fe F. Mg-l.5% Mn-0.006% Fe C. Mg-6% A1-0.005% Fe G. Mg-0.010% Fe D. Mg-6% A1-0.002% Fe

DISCUSSION

While not decisive on all points, the above experimental results ap- pear to give a reasonable picture of the mechanism of magnesium corro- sion when interpreted from the point of view of the electrochemical theory of corrosion, s

For the specimens with impurities over the tolerance amounts, corro- sion starts rapidly and continues at a high rate. This is a result of the continued presence of impurities acting as local cathodes. Film forma- tion is not effective in covering them over and they are so numerous

s See for example, R. M. Burns and A. E. Schuh, "Protective Coatings for Metals," Reinhold Publishing Corporation, New York (1939).

CORROSION C H A R A C T E R I S T I C S OF M A G N E S I U M ALLOYS 427

that corrosion continually exposes new particles. The more interesting question is whether for the specimens of highest purity the shape of the H2 evolution curves obtained is characteristic of metal which is entirely free of metallic impurities or whether it is still determined by the re- sidual traces of metallic impurities in the specimens. It is known from long-time alternate immersion studies of highest purity metal that Mg- A1 alloys have a lower steady state corrosion rate than Mg or Mg-Mn

POTENTIAL f 9 0

,8ot

1.70

\

~ u--.-..-.,-, o B

1.5C 0 tO 20 30 40 50 60

CURRENT D E N S I T Y - f f A / C M . 2

FIG. 3. Anodic polarization characteristics of high purity alloys after 24 hours in 3% aq. NaC1 solution.

Compositions. A. Mg-0.0006% Fe E. Mg-6% AI-0.006% Fe B. Mg-0.010% Fe F, 1%1g-1.5% 1%{n-0.006% Fe C, Mg-6% A1-0.001% Fe G. Mg-6% A1-0.3% Mn-0.001% Fe D. Mg-6% A1-0.002% Fe

alloys without Al. While not clearly revealed in Fig. 1, similar results have been obtained in this investigation.

One would expect that for absolutely pure metal the change from the high rate of corrosion of about 0.4 cc. of H2/cm.~/day to a steady state value of about 0.01 cc./cm.2/day in the ~ourse of 7 or 8 days a sob- served would be due to the formation of a protective film during this period of time. However, from the data on anodie polarization one sees that the anodic film action seems to reach a steady condition in about twenty-four hours or less instead of in 7 or 8 days. Also to be noted

428 R . E . MCNULTY AND .]. D. HANAWALT

is that the anodic polarization in any case is small compared with the potential differences of the local cells.

It therefore seems more reasonable to the authors that the time to reach the steady state is determined by the presence of the cathodic impurities, and that the mechanism taking place is one, as discussed in the previous paper, 9 of cathodic particles being gradually under- mined and bodily removed or of being covered over and "strangled"

POTENTIAL

t 80 ;~

1.7C ~ =B

.60

! .4C 0 IO 20 3 0 4 0 5 0 6 0 CURRENT DENSITY-yA/CM. 2

Fzc. 4. Anodic polarization eaharacteristies of high purity magnesium alloys after 96 hours in 3% aq. NaCI solution.

Compositions. A. Mg-0.006% Fe E. Mg-6% AI-0.001% Fe B. Mg-0.010% Fe F. Mg-l.5% Mn-0.006% Fe C. Mg-6% A1-0.002% Fe G. Mg-6% A1-0.3% Mn-0.001% Fe D. Mg-6% A1-0.006% Fe H. Mg-6% A1-0.3% Mn-0.001% Fe (After 2 years'

alternate immeesion.)

by the encroachment of the anodic film. Probably both mechanisms take place. When corrosion is violent, the particles are removed. This is the ease for impure specimens or at the impurity segregation regions of specimens with average impurities below the tolerance limits. Only where corrosion is less violent does the encroachment of the anodie film over the microscopically small cathodic particles take place. If all of

P See footnote 4.

CORROSION CHARACTERISTICS OF MAGNESIUM ALLOYS 4 2 9

the exposed particles have been removed at the steady state, then ,steady state corrosion rate should be representative of specimens of absolute purity; while, if the particles have been covered over, the .impurities would stiII be a factor in determining the steady state corrosion.rate, although the effect might become negligibly small, Since the amount of H2 evolved to reach the steady state was about the same for the 0.002% Fe-containing specimens as for the 0.001~ Fe.containing specimens and the corrosion rate at the steady state was greater for

POTENTIAL 16

1 4 .......- A

OA ~ ?

0 0 LO 20 30 40 ~0 60

CURRENT DENSITY-MA/CM 2

F~o. 5. I"tydroger~ overvoltag'es, Compositions.

A. Cu D. Ni B. Fe E. Ca C, Mg~Cu F. FeAI=

the 0.002% Fe, i t appears that at least some of the particles, are cow ered over and not removed. The conclusion follows that the :cbrrosion behavior of highest purity specimens studied is.still a function Of'the tracelmpurities c0nt~ined. . . . .

A , , , , , . suggested mechamsm m pictured m Fig:. 6 and further amphOed in Fig. 7. In the initial state the potential difference between, e'.::q:j the Fe cathode particie and the magnesium anode is balanced by ti~e i~,' overvoltage of the Fe, the polarization Of the anode film a n d t h e 'iR.'~, drop in the salt solution. As corrosion proceeds and thel an0dic:'fii~ encroaches on the cathode, a new resistance to current fl0w Rp comes

430 R.E. MENULTY AND J. D. HANAWALT

into action; also, the area of the cathodic particle is decreased. The relative magnitudes of these effects as indicated in Fig. 7 is only a guess since the authors can think of no means of experimental measurement of these quantities.

Under the conditions as represented in Fig. 6 and 7, the sum of the H2 overvoltage and IRp gives the cathodic polarization and the specimen would be said to be under cathodic control. For metal absolutely pure o r with all surface impurities removed or with impurities effectively entirely covered over by film formation, the possibility, of course, exists that the specimen is under anodic control. No experimental evidence for such a condition appears available at the moment.

He- H + OH" ANODIC FILM ,' /

~ , \ \ \ \ \ \ \ \ \ \ \ N N ~ \ \ \ \ \ \ \ \ \ \ \ \ \ \ X x ~ \ \ \ \ \ X \ \ \ \ \ \ X \ \ \ \ ~

) CATHODIC ~ O- METAL PARTICLE

INITIAL STATE

He,b . H ~' OH" ANO/DIC FILM

/ PO s / , - / / / v METAL --

CATHODIC PARTIGLE

STEADY STATE

FIG. 6. Suggested mechanism for corrosion behavior of magnesium and its alloys.

Initial State---Slight film formation with cathodic particle f~ncfioning unimpaired.

Steady State---Anodie film encroachment over cathodic particle reduces its effectiveness by large pore resistance.

It has been assumed that H2 gas evolution is the only cathodic reac- tion, whereas under the conditions at steady state for high purity speci- mens it is not known that oxygen depolarization may not be a factor.

The approach to the steady state and the corrosion rate at the steady state are naturally functions of the film growing and spreading charac- teristic and as seen in Fig. 1 differ according to the composition of the alloy. Electron diffraction and chemical analysis serve to show that the structure and composition of the film depend upon the composition of the alloy.

Whitby 1~ studied the H2 gas evolution of high purity magnesium and published a curve in substantial agreement with the present paper. Whitby states that, since the impurities are present in small amounts, the effect of these impurities is small and the potential differences due to film ennoblement govern the corrosion behavior. He concludes that

t~ Whitby, Trans. Faraday Soc. 29, 415, 853, 1318 (1933).

CORROSION CHARACTERISTICS OF M A G N E S I U M ALLOYS 431

the initial mechanism is one of preferential hydroxyl ion discharge and that dissolution of magnesium is anodically controlled.

Very early Boyer n advanced a mechanism for the corrosion of mag- nesium and its alloys which was based on the assumption that corrosion differences are due to the addition of alloying elements which change overvoltage relations. It has since been shown, however, that these phenomena are "due to the shifting of the "tolerance limits" for the ira- purities.

POTENTIAL I~R;~

Po - - - ~ _

Pc

+

IARA

IsRsoL

r / ~ O.V

!

i I b CORROSION CURRENT o

Pie. 7. Electrochemical relations in the initial and steady states.

(a) In{tlal State. I ~ The corrosio~ current. Pa = Anode potential. Pc = Cathode potential. Ha o.v. = Hydrogen overvoltage. TAR A = Anode polarization. IsRsoi. = Potential drop through solution, the effective

driving force.

(b) Steady State. Pa, Pc, H2 overvoltage = as in initial state. I ' = The corrosion current. I ' aR A = The anodic polarization. I 'sRso L ~ Potential drop through solution. I'pRp ~ Potential drop resulting from large pore resistance.

i s R SOL.

H 20.V.

Catty and Spooner 12 assume that the cathodic area on magnesium is a hydride film and the anodic areas are the points at the base of the pores of this film. This assumption appears to be due to their dismissal of the effect of impurities because of their small area and a belief that the hydrogen overvoltage on magnesium is low.

A recent paper by Beerwald 13 presents curves showing the H2 gas evolution of magnesium and magnesium alloys in salt solution. AI-

n j . A. Boyer, Natl. Bureau of Aero., Report 248 (1926). See also footnote 4. O. Gatty and E. C. R. Spooner, "The Electrode Potential Behaviour of Corroding Metals

in Aqueous Solutions," Oxford University Press (1938). A. Beerwald, Z. Metallkunde 3,% s (1941).

432 DISCUSSION

though showing the ex t reme effect of impur i t ies on corros ion rate, his curves do not show the same t rend to a s teady s tate condi t ion fo r high pur i ty specimens as found in the present paper .

CONCLUSION

I t has been shown that the corrosion behavior of highest pur i ty magnes ium and magnes ium alloys is de te rmined not only by the charac- te r i s t ics of the na tura l film fo rmat ion but by the t races of impur i t ies still p resen t in the metal. A mechanism for this corros ion behavior is suggested.

Resumen del artlculo: "A]gunas Caracterlst[cas de la Corrosi6n de Aleacionlm Pura, de Magneslo"

Algunas impurezas que pueden hal larse en el magnesio o sus aleaciones aumentan sobremanera la velocidad de corros i6n si est/m en cant idad m a y o r que cierto l imite de tolerancia. En la Fig . 1 se ve el efecto repent ino al va r i a r la cant idad de la impureza hierro. Cuando hay poco de la impureza la pelicula an6dica sobre el magnesio se ext iende tambi6n sobre las parf iculas extraf ias , d i sminuyendo la cor ros i6n a una velocidad minima. (F ig . 6 ) . S e midi6 la cor ros i6n por la evoluci6n de hidr6geno.

DISCUSSION W. P. FISHr, t)4: I should like to make one statement in criticism, and that is,

this "tolerance limit" is evidently the saturation value or solubility limit. Why create a new term for it?

J. D. HANAWALa': We have given a different definition to the term "tolerance limit," and to see why we have, one must consider the facts brought out in a previous paper. ~ It was there shown that, if the steady-state corrosion rate (as determined over a 4-months' period alternate immersion in NaC1 solution), is plotted against the per cent iron in the metal, one will find that this corrosion rate does not appreciably change as the per cent of iron increases, until a certain percentage iron content is reached at which point there will be a dis- continuity in the corrosion rate graph. The corrosion rate for specimens with iron even very slightly above this critical amount will increase by a factor of many fold. We have called this critical amount of iron at which the discontinuity in corrosion rate occurs the "tolerance limit" of magnesium for iron.

Unless the number of iron particles is below this critical amount, the initial high corrosion rate will be maintained and the approach to a low steady-state rate as described in the present paper will not be observed.

H. H. Urn,rot6: What is the effect of heat treatment on the "tolerance limit?" J. D. HANAWAI,T: The magnitude of the tolerance limit is the same whether

the metal is cast or wrought and whether it is heat-treated or not. This is understandable from the metallographic examination which shows that the iron remains in its random distribution of discrete particles throughout these various conditions of treatment.

W. P. F I S H E L : Accordingly, the "tolerance limit," as you use it, may be defined as that amount of iron present, as a separate phase, which renders the alloy so corrosive that it can no longer be tolerated.

R. LANDAU 17 (Communicated) : Some of the European writers such as Strau- manis TM have discussed the effect of impurities on the corrosion of aluminum,

xt Dept. of Chemistry, Vanderbilt University, Nashville, Tenn. ~ See footnote 4. lo Research Laboratory, General Electric Company, Schenectady, N. Y, ~7 M. W. Kellogg Company, New York City. is Korrosion u. Metallschutz 14, 67-77, 81-83 (1938).

CORROSION CHARACTERISTICS oF MAGNESIUM ALLOYS 433

and their observations would seem to have application to the present paper on magnesium. From their discussion it appears possible to explain the initially in- creasing corrosion rate as due to the re-plating of the nobler impurities at cathodic spots, lowering the I-I2 overvoltage, and increasing the cathode area. At times the film covering most of the surface can also act as cathode. However, the leveling off in the corrosion rate after a time is the result of the formation of protective films, probably because of the alkalinity resulting at the cathodes due to impoverishment of hydrogen ions. Resistance changes may also be involved. This explanation gives a more graphic picture of the mechanism underlying the authors' Fig. 7.

The authors apparently have indicated in Fig. 6 that Mg ions dissolve by passing through an anodic film. This is not in agreement with the extensive work on the local cell theory of corrosion, and it is more probable that solution of magnesium occurs at gaps in the film, and that the cathodic areas of low overvoltage where the impurities deposit can be plugged up and corrosion stifled, as the pH there changes sufficiently to permit precipitation of basic compounds. If, however, there is a sufficiently large number of impurities, the rate of cathode formation exceeds that of film deposition, and no diminution in the corrosion rate is observed.

The authors state that their anodic polarization studies were conducted at cur- rent density values chosen on the assumption that the area of the specimen is substantially all anodic, and the current is equivalent to the hydrogen evolved. But the work of W. J. Miiller 1~ and others has indicated that, even for metals corroding severely, the surface films cover most of the apparent area, and the anodic sections (gaps in the film) are only a few per cent, at most, of the total. Consequently their results shown in Fig. 2, 3, and 4 may not be completely repre- sentative since the current density at the local anodes may be much higher. Per- haps the authors can clarify what they did in obtaining these curves.

R. E. McNuLTY (Communicated): The authors are glad to receive Dr. Landau's suggestions relative to the corrosion mechanism. There are several factors to be cited in answer to each of the points he has raised.

The rate of corrosion as shown in Fig. 1 is initially high for the pure alloys, but is decreasing rather than increasing throughout the approach to the steady state. Hence the re-plating of nobler impurities, which caused an acceleration in the work of Straumanis, does not seem likely to occur here. We have stressed that development of the anodic film is responsible for the removal of the effect of local cathodes and the low steady-state rate. We have not been able to detect that the film ever acts as a cathode initially or in the approach to the steady state. As pointed out on page 430, in the low corrosion steady state, other factors (about which we have no information as yet) may be involved.

The exact method of dissolution of magnesium in salt-water corrosion seems in doubt. Whitby ~ presented a series of three papers which developed a theory of anodic control of magnesium corrosion, based on primary preferential hydroxyl ion discharge. The authors have concluded that the reaction is cathodically con- trolled in the range covered here. Fig. 6 was offered as a schematic (but not exact) representation of the over-all picture on a magnesium surface, but we are unable to assert a rigorous method of dissolution at present.

The current density values given in Fig. 2, 3. and 4 are based on the assump- tion that the area of the microscopic cathodes of detrimental elements (as Ni, Fe, Cu) is negligible, and the area of the specimen substantially all anodic. Trans- lation of the amount of hydrogen gas evolved into current gives values in agree- ment with those used. We would agree that, if Dr. Landau's assumptions that anodic areas exist only at gaps in the film were true, the anodic current density values would be higher. However, our studies of the importance of discrete particles of detrimental elements as active cathodes lead us to believe that the assumption that the area is almost completely anodic is more satisfactory.

~g Stahl u. 1~isen 61, 535 (1941). Loc. tit.