CORROSION, the BILLION- DOLLAR THIEF - ACS Publications

DOLLAR THIEF. 11. The Facfors Afecting Corrosion. FREDERICK A. ROHRMAN. Michigan College of Mining and Technology, Hoqhton, Michigan. Mwh of the ...
8 downloads 0 Views 7MB Size
CORROSION, the BILLIONDOLLAR THIEF 11. The Facfors Afecting Corrosion FREDERICK A. ROHRMAN Michigan College of Mining and Technology, Hoqhton, Michigan

M w h of the misunderstanding in corrosion studies and observations is due to failure to comprehend the nature and intensity of the many factors involved. A natural classij5cation into influences inherent within the metal, in the solution, and on the outside seems to present itself. Oftentwo or more influences act tocether either to neutraliie or to intensijy Lhe inditiduul effects. An attempt has been made to segregab and discuss each possible factor in a unijorm manner so as to chrify the subject. Short explanations are giner of the values of these studies in understanding metal solution rates, industrial boiler design, automob& manufacture, milk pasteurizing, . . . and even cramznology.

+ + + + + + Do not look for anything behind phewmena. They are themselves their own lesson. Goethe CLASSIFICATION

fi might affect the corrosion of a metal makes it be grouped under three obvious that heads: the metal, the corroding medium, and external influences. The classification may be as follows: The Meld Electromotive position Purity Nature of impurities Physical state Oxide or surface behavior Salt solubility

The Corroding Medium Nature of cation Concentration of cation Conductivity Temperature Diusibility Viscosity Circulation Nature of anion Nat- of corrosion produds

{

External Influences Oxygen concentration Light Colloids Bacteria Cathodic metals Stray nurents

In the discussion of each factor it will be necessary to assume that none of the other factors interferes. Actually such a state never occurs, but it must be posited for the purpose of clarifying each phenomenon involved. THE METAL

1. Electromotive Position.-From the previous paper* it should be plain that the metals in the upper portion of the electrode potential series are more subject to corrosion than those in the lower portion. Thus, in nature, gold, silver, copper, and mercury occur in the free state, while aluminum, zinc, iron (other than meteoritic), A d lead are nev& uncombined. One expects to find, and does find, that salt water, acid solutions, or copper sulfate solutions will attack clean aluminum more readily than they will attack zinc, and zinc more readily than copper, and copper more readily than gold. The series, then, gives one the first impression of the corrodibility or corrosion

an element could be freedbf al En E.

+

the solc factors involved in acid corrosion arc the potential of the metal, Em, the potential of hydrogen, E,, and the hydrogen overvoltage, E.. The only function of the more noble constituent is to act as a focal point for the deposition of hydrogen. In other words, the only r81e the noble constituent has to play is that of its hydrogen overvoltage. C a s ~ a r has i ~ ~found that amalgamated zinc is only very slowly dissolved in acid. The reason for this becomes evident upon noting the high hydrogen overvoltage on mercury, a factor that makes for a low value in the equation just given. Because of the low hydrogen overvoltage on iron or platmnm, additions of these metals to zinc result in rapid solution and consequently lead to a high value.s8 4. Physical State.-The physical state of a metal has an important bearing upon its electrode potential and rate of solution. If the potential of any piece of metal is measured and a stress or strain is then imparted to it, the potential will become more negative, showing t h t there is a greater tendency for the strained metal to lose electrons or go into solution. This difference in potential can be observed in very pure metals as weSl as in impure metals. I t may sometimes amount to several tenths of a volt.

A stress or strain may be imposed on a metal by three means : 1. Physical disturbance at a temperature below the annealing point. 2. Rapid quenching from temperatures above the annealing point. 3. Electrodeposition a t high cnrrent densities. Usually these three cases manifest themselves by a change in crystal strncture. Yet, it is also known that energy is absorbed in the processes. Too often metallic specimens are tested for resistance to corrosion without regard to their physical state. A metal may be sawed or cut for a sample, drilled for a holder, filed, or polished; each one of these operations imparts to the metal a strained area which differs from that of the homogeneous metal. The result is that corrosion will be more pronounced at the edges and around the hole, and that its immediate rate of solution will be higher than its later rate of solution. The problem of the steel spring is an interesting one, and in this case the crystal structure does not seem to change. What happens to the energy of a wound spring as it is dissolved in an acid? The answer is that it dissolves with the evolution of more heat than an unwound spring. How could this energy have been stored in the wound spring? It may be that the atoms, if distorted by compression or tension, have a greater solution potential. X-ray evidence does show a slight lattice distortion; and if one considers the energies involved within the atomic lattice it is evident that the shift, though slight, would be representative of a great amount of energy. When the crystals of a metal are broken up it is easy to see that a greater surface results and that the solution rate should increase. Mere rate of solution without any increase in heat does not connote a potential increase. Thus, two small pieces of metal may have the same potentials as one larger piece, hut they will dissolve faster because of their greater surface. It is also known that the different sides of a crystal grow and dissolve a t different rates. As a crystal is broken up, more of the more soluble sides are exposed, and these may have a greater potential than the less soluble sides. Straumanis3' has shown that the different sides of a zinc crystal have different potentials. These potentials are, however, in the order of millivolts, as compared to tenths of a volt. As a metallic body is changed from a large crystal structure to a small crystal structure the grain boundaries become focal points for local action because of the greater number of impurity segregations at these points and because of the existence of oxygen concentration cells (to be referred to later). These factors all tend to hasten solution. It is then possible that a stress or strain will hasten corrosion by virtue of atomic distortion, smaller crystal structure, presentation of favorable crystal faces, uncovering of impurities, and crevices for the existence of oxygen concentration cells. STRAUMANIS, ibid.. 147, 161 (1930).

In industry it is known that corrosion will be more acute at rivet points and the machined portions of various equipment. I t is becoming good practice to anneal even large fabricated parts after their manufacture, in order to reduce the localized strained areas. It is known that after the electrodeposition of metals on certain objects, corrosion takes place first at the edges and corners because at these points the current 'density is the highest and smaller crystals are deposited. Of general interest is the importance of stress-strain corrosion in criminology. It is known that revolvers and pistols are numbered by steel dies and that such a number would often lead to the identification of the owner. Owners of such weapons for illegitimate purposes file or grind out the numbers, thus trying to prevent identification. When a number is stamped in a metal, however, a strained facsimile of each number is outlined into the body of the metal, so that even if the number is obliterated on the surface it is potentially present below the surface. Criminologists have found that acids will etch into these strained areas and will soon reveal the outlines of the numbers. 5. Oxide or Surface Behavior.-Most metals are readily attacked by oxygen to form oxides. Some of these are quite soluble, and some are not. If a metal forms an oxide which is insoluble and tends to cover the surface thoroughly, then this metal will be protected from further attack. Oxide films which tend to protect the underlying metal from corroding media are called passive films, and the phenomenon is known as passivity. This protective film of oxide may be induced to form by atmospheric oxygen or oxidizing agents such as nitric acid, dichromates, chromic acid, etc. In the case of arsenic passive films, produced by soluble arsenic salts, the mechanism of formation is different. The arsenic deposits out of solution over the less noble metal and forms an arsenic film which has a high hydrogen o v e r v ~ l t a g e . ~Phosphate ~ coatings are sometimes produced on ferrous metals, the best known process being the "Parkerizing Process," which is a modification of the "Coslett P r o c e s ~ . " ~These ~ iron phosphate films, which are very insoluble, are produced by immersing the metal in a phosphate solution. As previously mentioned, the reason why aluminum and chromium resist corrosion so well is found in the insoluble oxide 6Ims which cover these metals so thoroughly. It is noteworthy that most of the metals below hydrogen, such as copper and silver, do not form protective oxides; in fact, before these metals can be made to dissolve they must he attacked by oxidizing agents. Schonbein and Faraday" were the fust to explain the passivity of iron in this way. Vernons8and Evans3B have actually isolated and determined the thickness of such films. The author has produced dense passive

films upon nickel. These films were recovered by cleaning one side of the metal, allowing the metal to dissolve as an anode, and then recovering the film as it floated on the solution. 6. Salt Solubility.-The topic of salt solubility is closely related to topic 12, nature of corrosion products, under which it will be more completely treated. As a metal is attacked it combines with the ions of the corroding medium. If the resulting salt or compound is quite insoluble and tends to cover the surface, further corrosion will be hindered. Good examples are the action of dilute sulfuric acid on lead and of dilute hydrochloric acid on antimony. In the former case the slightly soluble and adherent lead sulfate is formed; in the latter case antimony chloride is formed which hydrolyzes to form the insoluble and adherent oxychloride. If the salts which are formed are very soluble, corrosion proceeds rapidly. A clear example is furnished by the lactates. Lactic acid, though a weak acid, forms very soluble salts, and, as any one familiar with its action on metals realizes, attacks them with avidity. Thorpe40 has stated, "The salts of lactic acid are all soluble in water, many so readily that they are difficult to obtain crystalline."

"WATT~, Trans. Am. Electrochem. Soc.. 21, 337 (1912). POLLITT, "The causes and prevention of corrosion," Ernest Benn, Ltd., tondon, 1924, p. 140. SCH~NBE~N AND FARADAY, Phil. Mag., 9, 53 (1836). a 8 V ~ n ~3.o Chen. ~ , Soc., 129, 2273 (1926). EVANS.ibid., 130, 1020 (1927).

TKORPE,"Dictionary of applied chemistry," Longmans, Green & Co.,New York City. 1925, vol. IV,p. 8. " CALCOTT,W ~ T Z E L LAND , W m r c m ~ ,''COrmsion tests and materials of ~0nst~Ction for chemical engineering apparatus," D. Van Nostrand Co., New York City, 1923 p. 4. FINKAND ROHRMAN. 3. Dairy Science, 15,73 (1932).

-

"

THE CORRODING MEDIUM

7, 8. Nature and Concentration of Cation.-According to Calcott4' 95% of corrosion is acid corrosionthat is, corrosion due to the substitution of metal for hydrogen in the corroding medium. I t is just as possible for metallic salt solutions to react upon metals as for acids to do so, the only difference being the supplanting of metal by hydrogen. Copper sulfate solution attacks iron, aluminum, zinc, etc., with facility, the copper precipitating on the metal and the metal going into solution. The corrosion of metals by mine water is often caused or accentuated by the presence of copper salts. Brass is sometimes corroded severely because of the reprecipitation of the copper constituent upon the zinc, a phenomenon known as dezincification. Fink and the author,42studying the corrosion of metals by milk, attributed the excessive corrosion of nickel pasteurizing equipment to the small, natural copper content of milk. Raw milk in contact with nickel causes the nickel to acquire a dark coat of precipitated copper; prolonged contact results in the complete destruction of the nickel apparatus. It was demonstrated (Part One) that the influence of the cation concentration is important in determining the potential of an electrode. It suffices to repeat that as the concentration increases for a particular electrode, its potential becomes more noble, and as the concentration decreases, the potential becomes less noble.

-

"

Water, the most common corroding medium, has a very low hydrogen-ion concentration, pure water having a concentration which may be expressed as 10W7 gram ions per liter. Its potential at 2 5 T . will then be Ex+ = Eo 0.059log 10-1

+

En+ = 0 - 0.413 En+ = -0.413

I t is, then, quite unlikely that this concentration of hydrogen ions will attack metals having potentials more positive than -0.413. Thus, nickel, if in a normal solution of its ions and if the solution has a hydrogen-ion concentration of would not be affected.' If, however, the concentration of nickel ions were lo-', corrosion could take place because

+

EX++= -0.23 0.059 lag lo-' 2 Exit+ = -0.437 The nickel potential is now less noble than the hydrogen potential by 0.024 volt. When a metal is placed in a medium having an unknown or insignificant content of its own ions it is impossible to calculate what the potential relationships should be. Morewer, since the potential depends upon the ionic concentrations at the faces of the electrodes, one is not always right in basing his calculations on the concentrations in the body of the liquid. If a negative electrode is placed in a solution very low in concentration of its own ions the potential will invariably be more positive than one would expect or than calculations would indicate. This condition is obtained because the metal-ion concentrations at the anodic areas are always higher at the interface metal-medium on account of the slow rate of diffusion of the dissolved metal into the body of the liquid. At equilibrium, however, the concentrations must be the same. If nickel or almost any other metal is placed in a solution of sodium chloride its potential, of come, will be more negative than if it had been placed in a solution of its own ions. It would immediately go into solution and its ions would diffuse into the body of the liqnid. Soon its potential would be as high as the hydrogen potential, and attack would cease.

dv

F,

Equation 7, -;; = - it (I ac can be seen that the conductivity of the corroding medium has a direct effect upon the solution velocity of the attacked metal. The medium must be conductive to local-action currents, for otherwise local action would cease. Kajander43has showed that there is a direct relationship between the conductivity of the acids attacking magnesium, and the solution velocity. The increase of hydrogen-ion concentration with the conductivity must not be considered as the prime reason why the solution velocity increases. The effect of the hydrogen-ion increase would make the value of the reaction velocity change according to the log H+, and not directly,. as the conductivity does.

9.

Conductiuity.-From

-

'%JANDBR,

1.RUSS.Phys.-Chem. Soc.. 13, 457 (1881).

From the point of view of acting conductors and not from the point of view of ion concentrations, the factor of conductivity is of greater importance than has generally been realized. Palmaer has constantly pointed this out in his researches and writings, and it cannot be too thoroughly emphasized. 10. Diffusibility and Temperature.-The concentration c6anges which might alter the electrode potentials are dependent upon the rates at which diffusion will proceed. Fink44and the author, studying the diffusion rates of copper and nickel electrolytes, demonstrated the dependence of diffusionupon viscosity, temperature, and circulation. I t can be said that diffusion increases with an increase in temperature or circulation and with a decrease in viscosity, while viscosity decreases with an increase in temperature. The rate at which dissolved oxygen is delivered to the electrodes is also dependent upon the rate at which the solution is circulated. Besides affecting diffusion in the manner outlined, temperature affects the electrode potential according

--

X1

to the Nernst relation, - log C. The hydrogen overnF ..voltage on metals is also reduced by an increase in temperature. Salt solubility generally increases and oxygen solubility decreases with a rise in temperature. 11. Nature of Anion.-It is a well-known fact that the metals which form protective oxide films are only slowly dissolved in sulfuric add, nitric acid, and many other media. Since most of these metals form soluble sulfates and nitrates, their insolubility must result, not from salt insolubility, but from oxide insolubility. The halogen acids react with these oxides, however, so that traces of chlorides, fluorides, or bromides in corroding media cause these metals to go into solution much more readily than if the halides were absent. I t is believed by some that the halogen ions, being smaller, have the ability to penetrate the small openings of the oxide film and attack the metal. Yet one must reason that mere accessibility of anions to the metal does not necessarily result in corrosion; with metals of the passive film type corrosion is an immediate consequence of hydrogen substitution, and hydrogen is a much smaller ion than any of the halogens. It must follow that the action of the haloeen is secondary, and it is quite plausible that its r8leis one of combining with the oxide to form a soluble halide. Some practical aspects of this influence are illuminating. During the electrodeposition of nickel it is necessary that some chloride be present in the sulfate bath in order to facilitate anode solution. Without the chloride, or other halide, very little nickel goes into solution and the bath soon bccomes dcpleted of nickel, a condition to be avoided. In some sections of the country it is the practice to apply calcium chloride to roads in order to lay the dust. This practice proves rather tragic to automobile parts plated with chromium or nickel, for when the roads become wet the parts " FINK AND R O E R ~ A Trans. N , Am. Electrockam. Soc., 57, 325

-

(1930).

are splashed with this very active medium and rapidly has been ordinary water. The iron goes into solution corrode. For certain experiments it is sometimes neces- as ferrous iron (Fe++) and forms ferrous hydroxide, sary to dissolve in acids large quantities of the metals Fe(OH)%,with the hydroxyl ions of the water. Now just mentioned. The reagent that the author has ferrous hydroxide has a whitish green color, while rust found to be most effective is concentrated nitric acid as we see it, is red; another reaction must therefore with about one per cent. of hydrochloric. This medium enter in. What happens is that the oxygen of the is much more active than aqua regia or other mixtures atmosphere enters into the reaction a t this point and the ferrous ions are oxidized to ferric ions, the result of hydrochloric and nitric acids. Metallurgists have constantly been attempting to being ferric hydroxide, Fe(OH)3. These hydroxides perfect a ferrous alloy which would resist hydrochloric undergo transformation with the formation of the acid. All the "stainless" ferrous alloys resist corro- oxides FeO and Fez03 with waters of crystallization. sion because of their homogeneity and their possession Sturnper," on analyzing 23 samples of rust, found the of a non-permeable oxide film. Since their resistance composition to vary within the limits indicated by 1 depends chiefly upon maintaining this oxide film which to 3 Fe0.4 to 20FezO5.22H~0. The outer rust portions combines with the halogens, they are resistant to all were found to contain the greatest percentage of ferric media except those containing halides. It is, then, iron. Of course, under the surface oxides one will find quite unlikely that a ferrous alloy will ever be developed Fe(OH)3, and under the Fe(OH)3 there will be unoxiwhich will resist such media unless some other pro- dized Fe(OH)%;so that a cross section of a rust coating will reveal a gradation of these compounds, and a mass tective surface is developed. topic is analysis will show a mixture. The presence of carbon 12. Nature of Corrosion Products.-This closely related to topics 6 and 11. Let us assume that dioxide will complicate the compound formation further the metal to he corroded is zinc and that the corroding with the formation of carbonates. medium is sodium chloride; the following reactions EXTERNAL INFLUENCES can be said to take place. 13. Oxygen Concentration.-The importance of oxyZn +Zn++ - 2. (at anodic ooints) gen concentration cannot be overemphasized. Too Hf f r + H " (at cathodic points) often scientists speak of the presence or absence of oxygen without realizing that the important factor is ~ ~ ~ - ~ z , " $ HinIbody J of solution concentration. In those cases of corrosion which are of the hydrogenThe formation of zinc chloride or zinc hydroxide will evolution type, oxygen plays little or no part. But depend upon the hydrogen-ion concentration and the where hydrogen polarization asserts itself and a curchloride concentration of the solution, a high hydrogen- rent does not flow, oxygen concentration is most imion concentration and high chloride-ion concentration portant. With those metals which form protective favoring the formation of zinc chloride. Complex films two tendencies oppose each other-one for a compounds may also form, but their formation will protective oxide film to form and prevent further atnot be discussed here. tack, and one for the oxygen to be steadily consumed On the basis of their physical nature and behavior, by the polarized hydrogen and promote attack. With possible corrosion products may be placed in these those metals which do not form protective oxide films groups : or in media which have a solvent action on the protective film, attack is proportional to the oxygen concen1. Those which are soluble. 2. Those which adhere to the metal, are impervious tration. If a piece of iron is placed vertically in a dilute sohto ions, and are insoluble. 3. Those which prevent passive film formation (act tion of sodium chloride the attack first takes place deep in the solution, while the portion near the air-water as oxygen screens). line is unattacked. Again, if a drop of water is placed If the product is only slightly soluble and has a ten- on a piece of iron the attack centers itself a t the center dency to adhere to the metal surface, forming a pro- of the drop, while the outside ring is unaffected. Deep tective film, corrosion will not take place, as in the recesses and crevices in the metal will he more subject previously cited example of lead sulfate. Thme com- to attack than will the surface of the metal. Areas pounds which act as screens to oxygen will cause pre- protected by physical objects or by corrosion products. ferred attack, a topic to be taken up under the next as mentioned above, also prevent the rapid diffusion heading. A recent paper by Britton and Evans's has of oxygen and its attendant protective-film formation. elucidated m;uiy oi the points discussed under the All these cases are dependent upon the existing condipresent topic. tions, which may or may not be amenable to the inEvery one is familiar with the appearance of iron fluence of oxygen. (Where access of oxygen is not hinrust. The mechanism of its formation is quite simple. dered a protective film may be encouraged to form or reMost iron rust is evidenced where the corrodingmedium main; where access of oxygen is inhibited a profec&film BRITTON ANDEVANS, Trans. Am. Electrochem. Soc.. 61, 123 cannot form and the oxygen acts as the depolarizer of the

:g' 1

-

(1932).

-

STUMPER. Chimie & industrie, 13, 906 (1925).

hydrogen.) Figure 1 shows several illustrations of this differential akation. The knowledge of oxygen influence is largely due to work and publications by Evans, Bannister, Britton, Bengough, Stuart, Lee, Wormwell, Forrest, Roethelli, and Brown. Reference to these valuable contributions will clear up many vague points on the topic. 14. Light.-All substances have the property of light absorption to a greater or lesser extent. Since light is a form of energy the absorbing material must make use of this energy in one way or another. Metals are affected in such a way that their solution potentials are increased, this being one of the explanations for the weU-known phenomenon of photo-electricity. A number of investigators have found that the solution rates and potentials of metals increase in the presence of light. Bengough and Hudson4' showed that under such conditions copper tarnished and went into solution more quickly, Crihh and showed that iron corroded more quickly, Liverseege and Knapp4* found the same true of lead, and Benedicks and Sundberg60the same of "stainless steel." The potentials due tolthe photo-electric effect may be quite small,51but

had been added to the acids. Friends4and his associates have done considerable work on this same topic. 16. Bacteria.-Certain microorganisms are known to influence corrosion, chiefly of iron, considerably. These organisms are known as lefitothrix ochracia, beggiatoa alba, crenothrix polyspora, zallionellu ferruginia, etc. They may affect corrosion in the following ways: 1. By changing oxygen concentrations (aiirobic forms). 2. By the production of active end-products. 3. By the formation of incrustation.

Any aiirobic form of life in a medium will naturally decrease the oxygen concentration of that medium. This change may influence corrosion as discussed under 13. Active end-products may be formed, such as humic acids or ~ulfides.~b These products will affect corrosion as previously outlined. Other organisms may collect a t the metal interface and form innustations or films of themselves or their end-products. These films will influencethe oxygen concentration snfficiently to create oxygen concentration cells.

Corrosion most evident at A (anodic). Corrosion least evident at P (passive).

with time their influence, in a corrosion reaction, may be appreciable. The importance of wave-length has also been emphasized, it being obvious that a shorter wave-length possessing a greater energy will cause greater corrosion. CoblentzS2has reported remarkable cases of corrosion which were accelerated by ultraviolet light. 15. Calla&.-The influence of colloids on corrosion should not he mysterious. Colloids may a d in such a way that they are adsorbed on the surface of the metal, or they may act to increase the viscosity of the solution. In the former case an adherent, protective film may form; in the latter case the flow of ions will naturally be hindered. It is interesting to note that the addition of almost any colloid inhibits corrosion. Many times, in fact, such additions are called inhibitors. Sieverts and LuegSahave observed decided decreases in the solution rates of zinc, iron, and aluminum in hydrochloric and sulfuric acids after various alkaloids

-

" B~iscoucrgAS"

Hr.oson. J. Inrl. 3fetolr. 21, 93 (1919 Cnrnn .AND A R N A ~Annljrl, . 30, 225 (1005,. 1 . r v ~ ~ s ~.AND r c sKSAW.J. Soc. Clzem. l a d . . 39. 201' 11020) 5 0 B ~AND~SUNDBERG. ~ ~ ~ 3. Iron ~ & ~ s ~ A s t . .l i 4 121, 117 1,A"-",. 101R> 6 L A ~Compt. ~ rend., ~ ~178,~386 ~ (1924). ~ ~ . COBLENTZ. Science, 40,64 (1924). " SIEVERTS AND LUEG,2.anorg. chcm., 126, 193 (1923) '8

It is also believed that certain bacteria actually consume metals, hut it is more likely that this process is a secondary action. The metal is probably converted to an oxide, hydroxide, or sulfide and then metabolized by the organism. 17. Cathodic Metals.-When more noble metals, other than impurities or precipitated metals, are present with another metal in a medium, corrosion takes place at the expense of the less noble metal exactly as in the case of local action. The difference between local action and the presence of two dissimilar metals resides only in the greater electrical resistance between the two bodies in the latter case. The conductivity of the medium is, then, the factor determining how great this influence shall be. It is obvious that if an iron and a copper strip are separated by a few inches the iron will he more subject to corrosion than if they had been separated by several feet. It is a very poor practice, though a common one, to fabricate equipment out of dissimilar metals provided that the equipment is to be exposed to corroding media or fumes. Only with time and experience will people 64 FRIEND AND TmMus, 3. Inrt. M~tals,33, 19 (1925). 6r FRIEND, "Corrosion of iron and steel," Longmans, Green & Co., New York City, 1911, p. 104. 6e GAINES. J. I n d . Eng. Chem., 2 , 1928 (1910).

realize that such practice is most uneconomical and perhaps even dangerous. The case is well known of the ship, constructed on the Eastern coast, which had a Monel metal hull and various steel fixtures such as the rudder. Exposure to the salt water made short work of the less noble steel, and the ship was never sailed." In England and on the Continent it is sometimes the practice to introduce aluminum strips into milk pasteurizing equipment made of nickel, tinned iron, etc., so that the less noble aluminum strips are corroded rather than the equipment itself. 18. Stray Currents.-Corrosion sometimes results from leakage of industrial alternating or direct currents. These currents are caUed stray currents (vagabundierende Strme). When, through poor insulation or for other reasons, electric currents have leaked from power lines such as electric railways, they constitute a potential menace to buried metallic bodies, like gas and water mains, as well as to the power circuit itself. If the current is direct current the flow of electrons will be from anode to cathode; that is, corrosion will take lace where the electrons are leavinz. Corrosion will &en take place where the electrons &e leaving the main power circuit and where they leave the buried metallic conductors. Figure 2 demonstrates this state. In the case of stray alternating currents the situation is more complicated, but the theory is the same. The reversal of current tends to re-deposit the dissolved metal; hence corrosion is not always so severe as in direct-current electrolysis. Theoretically, all the metal which has gone into solution should be re-deposited on the next reversal of current; actually, however, this does not take place. Electroplating and electrorefining experts are well acquainted with the

WATTS

AND

6'

(1921).

KNAPP,Trans. Am. Eleclrochmn. Soc.. 39, 160

fact that with metals above hydrogen, normal anodic reactions are generally more complete, according to Faraday's laws, than normal cathodic reactions. Smce this is true, each time a point is anodic more metal will go into solution (corrode) than will re-deposit when the metal is cathodic. THE CORROSION PROBLEM

Too often a corrosion problem is complicated by the influence of so many factors that it is difficult even to isolate the factor of greatest weight. An alloy may "test" well against certain media under certain controlled conditions and yet when applied to practical use under almost identical influences may prove worthless. The frequency of such results has led to many doubts concerning the value of laboratory tests in relation to the future, applied uses of the metal tested. In the next paper, therefore, the author proposes to present the methods of testing and the methods of corrosion protection. (Pa76 III will a$pcar in the May isnrc.)

REPRINTS OF "THE DISCOVERY OF THE ELEMENTS" In response to numerous inquiries the editorial office is pleased to announce that an agreement has been mched edition of bound with Miss Weeks whereby we are permitted to invite subscriptions for a

collected reprints of her series of articles on "The Discovery of the Elements." Those to whom this announcement is of interest will find details set forth on page XI.