Historical Theories on Corrosion
by: Ulick R. Evans, 1948
The chronological sequence of scientific discovery is rarely the logical one. To arrange the facts of metallic corrosion historically would conceal the true interconnection existing between them, and thus deprive them of significance. Nevertheless, in view of the prevailing interest in the History of Science, many readers may welcome a short narrative showing how knowledge of the subject discussed in this book has grown. The note which follows should serve to indicate some names and dates associated with the advance of understanding, but it must be remembered that the credit for any particular discovery cannot be assigned to a single year or to a particular person. If a recent investigator is cited as the discoverer, objection may fairly be raised by the quotation from older papers of passages which seem to contain the germ of the idea ; yet to allot the entire credit to early investigators may be unjust to later ones, who have established as facts what had previously been mere suggestions. (reference)
At the Dawn of History, the first metals to be used were those which were either found native, or could easily be reduced to the elementary state ; such metals do not readily pass into the combined state, and their corrosion can have raised no serious problems. But with the introduction of iron, the problem of its corrosion must have presented itself, although it is an undoubted fact that some of the iron produced in Antiquity is today more free from corrosion than much of that manufactured in later years. This may have been due partly to the fact that iron reduced with charcoal contained less sulphur than modern steel, but it may also be connected with the absence of sulphur compounds from the air in the days before coal was adopted as a fuel ; for it is often the conditions of early exposure which determine the life of metal-work. Whatever the cause, ancient iron has in some cases remained in surprisingly good condition for many centuries ; the Delhi Pillar is the example which has excited most interest, but others could be quoted.
For many centuries there seems to have been little curiosity regarding the causes of corrosion, although a few significant observations were made. As early as 1788, Austin noticed that water, originally neutral, tends to become alkaline when it acts on iron. He attributed the alkalinity to the compound now called ammonia ; this was probably an error, since the alkaline reaction produced by most saline waters is due to sodium hydroxide, the cathodic product of the electrochemical corrosion process.
The belief that corrosion is an electrochemical phenomenon was expressed in a paper published in 1819 by an anonymous French writer, thought to be Thenard, and in 1830 his compatriot, de la Rive, attributed the fact that acid attacks impure zinc more rapidly than the relatively pure varieties to an electric effect set up between zinc and the impurities present. Faraday's researches, especially those conducted between 1834 and 1840, afforded evidence of the essential connection between chemical action and the generation of electric currents ; indeed (since his Laws of Electrochemical Action apply as much to anodic as to cathodic processes) he provided a quantitative basis for the observations of later investigators. One of the most interesting chapters of Faraday's work was concerned with the study of passivity-the subject of a famous correspondence in 1836 with Schonbein, then Professor of Chemistry at the Swiss University of Bale (Basle). The experiments described suggested that the strange inactive condition which Schonbein had observed on iron was favoured by anodic action, but was often dispelled by cathodic treatment ; this was all the more remarkable in that normally anodic action favours corrosion whilst cathodic action tends to prevent it.
After Faraday's time, interest in the electrical mechanism of corrosion processes seems to have waned. This may have been due to the fact that electrochemistry was hardly ready to be applied to the detailed elucidation of corrosion, until the conception of Single Electrode Potentials had been made familiar by publications emerging from the schools of Ostwald and Nernst and culminating in an important paper by Wilsmore (1900). But in Great Britain, at least, attention may have been diverted by alternative suggestions ascribing corrosion to the presence or formation of certain substances. Some of these suggestions are now seen to possess a -modicum of truth, but the underlying ideas are themselves consistent with an electrochemical mechanism.
Between 1888 and 1908 the view was frequently advanced that acids were the agents mainly responsible for corrosion ; particularly it was held that the rusting of iron would only take place if carbonic acid was present. It was shown, however, by Dunstan, Jowett and Goulding in 1905, by Tilden in 1908, and by Heyn and Bauer in the same year, that iron exposed to water and oxygen, with exclusion of carbon dioxide, underwent rusting. Acid is not needed for the corrosion of iron, and indeed some of the most intense forms of attack take place in salt solutions which have been made definitely alkaline. There is, however, one important type of corrosion that associated with the liberation of hydrogen-which occurs much more rapidly in acid than in neutral water ; in atmospheric corrosion also, acid impurities in the air greatly stimulate attack upon iron, although they are not necessary for it. In such circumstances, therefore, the view that acid is an arch-promoter of corrosion may be held to be justified, but in atmospheric attack-as shown by Vernon-it is not carbonic but sulphurous acid which is mainly responsible ; indeed Vernon (1935) demonstrated that under certain circumstances carbon dioxide can retard the attack upon iron.
Another theory of corrosion arose about 1900 as a result of the detection of hydrogen peroxide during the corrosion of many metals ; this gave rise to the idea that it acted as intermediary in corrosion processes. It is now known that hydrogen peroxide may be formed in electrolytic cells where oxygen has access to the cathode, Consequently. if the electrochemical mechanism of corrosion is correct, the presence of traces of hydrogen peroxide among the corrosion products requires no further explanation.
Whilst British chemists were disputing the claims of carbonic acid and hydrogen peroxide to be regarded as the agents responsible for corrosion, the electrochemical standpoint was being developed in Sweden and elsewhere, especially in Arrhenius' laboratory by Ericson-Auren and Palmaer, working separately and in collaboration from 1901 onwards. Their work dealt mainly with corrosion by those acids which act on metals eliminating hydrogen, and the results provided strong evidence that this attack was connected with the formation of microscopic cells, which they called " local elements ". Somewhat similar views were expressed by the American investigators Whitney (1903) and Cushman (1907) in their discussion of corrosion by neutral liquids.
The electrochemical views put forward by these writers overlooked, to some extent, the part played by oxygen, and broader theories of corrosion by neutral liquids, which recognized the function of oxygen as cathodic stimulator, were provided by the Americans Walker, Cederholm and Bent in 1907, and the Englishman Tilden in 1908. The most recent research, however, rather represents a return to earlier views, by indicating that oxygen is not absolutely necessary for the attack of water on iron. Experiments by the American investigators, Corey and Finnegan (1939) and de Kay Thompson (1940), suggest that iron is slowly attacked by oxygen-free water, with liberation of hydrogen, although in this case the final product is not yellow rust (ferric hydroxide) but black granular magnetite or sometimes ferrous hydroxide. The fact that at the boiling point pure water, or even dilute alkali, could attack iron with liberation of hydrogen, was clearly demonstrated by the Germans, Thiel and Luckmann, in 1928, and was probably known at an earlier date.
Extensive researches into the facts of corrosion were carried out in Germany about 1908-10 by Heyn and Bauer, whose writings contain the germ of many ideas developed later. They were probably the first to carry out really comprehensive measurements of corrosion velocity in numerous liquids, embracing iron and steel in different conditions both alone and in contact with other metals. Amongst other things they quantitatively established the fact that attack upon iron was stimulated by contact with a nobler metal, whilst contact with a baser metal often conferred partial or complete protection (qualitatively this had long been believed to be the case, and indeed as early as 1824 Sir Humphrey Davy had proposed connection with iron or zinc as a means of protecting copper against sea-water). The behaviour of iron in contact with other metals was further studied at a later date. In 1924, Whitman and Russell demonstrated that when three-quarters of the surface of a steel specimen is coated with copper, the total corrosion by running water may remain the same as that of a similar specimen completely bare, but is now concentrated on the bare quarter. This experiment provided an illustration of the manner in which corrosion may be intensified when a small anode is connected to a large cathode, a matter also brought out by the author's experiments in 1928, and by the extensive work of Akimow, with Clark and others, published in Russia about 1935.
Despite the importance of the so-called " galvanic currents " set up by dissimilar metals, the conception-prevalent up to about 1923-that corrosion currents were generally due to dissimilar metals in contact, e.g. the dominant metal and the impurities present in it, led to some erroneous conclusions. It spread the belief (definitely held by Palmaer) that perfectly pure and uniform metal-if it could be obtained-would be uncorrodible. Now in the hydrogen-evolution type of attack (which Palmaer studied in his laboratory) impure metal really is attacked much more quickly than the purer varieties, and the varying influence of different impurities was afterwards brought out by the researches of the Czech, Vondraoek, and the Latvian, Straumanis. But this is not true of all types of attack ; as pointed out in 1931 by the German scientist, Tammann, impurities do not increase the rate of corrosion of zinc by a persulphate solution. Actually, electric currents can arise in other ways than by the contact of dissimilar metals ; they can, for instance, be generated by differences in the liquid wetting different parts of a metallic surface, notably by differences in the oxygen-content.
The fact that variations in oxygen concentration can set up electric currents should have been clear from the experiments of the Italian scientist, Marianini, performed as early as 1830, or from those of the German, Warburg (1889), and the Russian, Kistiakowsky (1908) ; but the significance of these researches was largely overlooked. In 1916, however, the American, Aston, laid emphasis on the part played by local
Experiments carried out in the author's laboratory at Cambridge from 1923 onwards indicated that " differential aeration currents " (set up by differences in oxygen concentration) played an important part in the corrosion of many metals, although, as emphasized by Bengough; oxygen concentration is only one of many factors determining the potential existing at different points on a metallic surface. Considerable discussion-as to the relative importance of different factors took place between the two groups of investigators associated with Bengough and the author respectively, and an agreed statement was published in 1938.
In the years 1931-9 the electric currents flowing over the surface of metal corroding in salt solutions were detected and measured by several of the author's collaborators at Cambridge, including Bannister, Hoar, Thornhill and Agar ; also, after his return to America, by Mears, in collaboration with Brown. The currents were found to be strong enough to account for the whole of the corrosion actually measured, in the sense of Faraday's Law, and thus the electrochemical mechanism of corrosion by salt solutions may be said to rest upon a quantitative experimental basis.
Nevertheless, the increasing evidence for the electrochemical nature of corrosion by salts and acids led in some quarters to the exaggerated view that the mechanism of every corrosion process is necessarily electrochemical. No doubt any chemical change involves the displacement of electrons, but if we reserve the term "electrochemical corrosion" for those processes where current flows between anodic and cathodic areas situated at different parts of the metallic surface (with electrons flowing over paths long compared to the interatomic distance), then it is clear that many true corrosion processes are not electrochemical, since they occur in solutions where electrical conductivity is almost absent. The work at Cambridge has suggested that the reason why electrochemical action is so destructive is that it often leads to the formation of soluble compounds as the immediate corrosion products, whereas direct combination with oxygen often produces a sparingly soluble body in physical contact with the metal, thus stifling further attack. Frequently the electrochemical mechanism leads in the end to a sparingly soluble body, such as rust ; but if this is formed as a secondary product through interaction between the anodic and cathodic products at a perceptible distance from the point of attack, there is no stifling and attack continues.
Actually in those cases where electrochemical action would produce a sparingly soluble body either at the cathode or at the anode, it stifles itself just like any other form of attack. Consequently it is possible to classify substances which stop corrosion by coating the anodic or cathodic parts with layers of sparingly soluble compounds as cathodic and anodic inhibitors. The Polish investigator, Chyzewski, working in the author's laboratory in 1938, made an experimental classification of inhibitors into these two groups. The anodic inhibitors are the most efficient, but are apt to be dangerous, since when added in insufficient amount they often reduce the area undergoing corrosion more rapidly than they diminish the total amount of corrosion, so that the intensity of the local attack is actually increased by the addition. It had been shown by the author in 1936 that this intensification was to be expected on theoretical grounds, and in recent years much work has been devoted at Cambridge to the search for an inhibitive system which will be both safe and efficient.
Many of the substances which form useful pigments in anti-corrosive paints probably act in rather the same way as the soluble inhibitors just discussed. This was indicated by the early work of Cushman and Gardner published in America about 1910. The distinction between inhibitive paints, which will prevent rusting even when the iron is left bare at a discontinuity in the coat, and the mechanically-excluding paints, which only give protection in so far as they prevent access of moisture to the metal, became increasingly clear in a series of field and laboratory tests, organized from Cambridge and carried out between 1930 and 1945, mainly by Britton, Lewis, Thornhill and Mayne. Another type of anticorrosive paints which will give protection at gaps in the coats are those richly pigmented with zinc dust ; here the exposed iron escapes attack largely because it is the cathode of the cell zinc/iron.
Although electrochemical corrosion will proceed where direct oxidation would stifle itself, circumstances can arise in which this stifling does not occur, and in such cases, corrosion need not take an electrochemical course. For instance, water containing oxygen may produce rusting, if the conditions are such as to produce at the metallic surface the appreciably soluble ferrous hydroxide, and to permit of its conversion to the less soluble ferric hydroxide (yellow rust) at a slight distance from the point of attack, so that the rust is non-protective. Forrest, Roetheli and Brown (1930-1), in a series of instructive researches carried out at the Massachusetts Institute of Technology, showed that rust coats varied in protective character according to the rate of supply of oxygen to the surface, and the same principle was found to operate in statistical experiments with oxygen-nitrogen mixtures carried out by Mears (1935) during his work in the author's laboratory. These indicated that oxygen can depress the probability of corrosion starting within a given area, although, where once attack has set in, oxygen accelerates the corrosion rate ; this distinction between corrosion velocity and corrosion probability has served to clear up several of the apparent contradictions which puzzled earlier investigators.
The view has been advanced by several investigators that oxides of iron can in effect act as oxygen-carriers. In 1921 Friend, who had already carried out an extensive series of researches on the action of numerous salt solutions on iron, suggested that a colloidal solution of ferric hydroxide behaves as a carrier for oxygen, passing alternately between the ferrous and ferric conditions. Later (1936-8) Herzog attributed a rather similar role to "solid" rust, but combined this with an electrochemical mechanism. After long immersion in stagnant water, iron frequently becomes covered with an inner layer of magnetite, overlaid with an outer layer of ferric hydroxide. The magnetite is supposed to act as cathode towards the iron as anode, and the ferric hydroxide just above suffers cathodic reduction to hydrated magnetite. This may either lose water, reinforcing the magnetite, or may absorb oxygen from the air, returning to the state of ferric hydroxide. It seems very likely that under certain circumstances some such mechanism does operate. In any case, it is fairly certain that in the acid corrosion of iron, ferric salts can act as oxygen-carriers, being reduced to the ferrous state by the metallic iron, and then reoxidized by atmospheric oxygen, as indicated in the researches of Bryan (1934-5) ; cupric salts can perform a similar role in the attack of acids upon copper, as shown by Glauner in 1933.
The electrochemical study of passivity has been the subject of a long series of experiments started about 1927 in W. J. Muller's laboratory at Vienna, which has placed the mechanism of anodic passivation on a mathematical basis ; he was aided by numerous collaborators, of whom Konopicky and Machu deserve special mention.
Most of the laboratory experiments mentioned above have been of short duration, and have been concerned mainly with the opening stages of corrosion. In this respect, they were not entirely representative of natural corrosion processes, which continue for long periods. A feature of the extensive and very accurate work of Bengough, started about 1927, first in collaboration with Stuart, then with Lee and later with Wormwell, was the performance of experiments under strictly controlled conditions, extending over many years.
Whilst the mechanism of low-temperature corrosion was being investigated, high-temperature oxidation was also under study and here it was found possible to establish simple laws of growth. The scientific study of the oxidation process may be said to have begun with the classical work of Pilling and Bedworth published in 1923. The careful observations of Pfeil (1929-31), based on refined analytical work, showed that the growth of oxide consisted in the movement of metal outwards rather than the penetration of oxygen inwards through the scale. Further valuable information regarding the oxidation of iron and its alloys was obtained by the detailed metallurgical observations of the French investigators, Portevin, Pretet and Jolivet, published in 1934. About the same time, the German physical chemist, Wagner, was producing a number of papers, practical and theoretical, which showed that high-temperature oxidation is connected with the passage of cations and electrons outward through the growing scale ; low-temperature tarnishing, due to sulphur compounds in the air, follows a similar mechanism. A mathematical relationship was established between the oxidation rate and the electrical properties of the film substance ; this shows that good resistance to oxidation may generally be expected where the electrical resistance of the oxide formed is high. A new and instructive derivation of Wagner's equation was published by Hoar and Price in 1938, whilst in the same year Price and Thomas, working at Cambridge, applied these new theoretical ideas to the development of a process known as selective oxidation, which produces enhanced resistance of copper alloys to high-temperature oxidation, and similarly improves the behaviour of silver alloys towards low temperature tarnishing. Shortly before this, Miley, also working in the author's laboratory, developed an electrometric method for measuring the thickness of oxide films, depending on the quantity of electricity needed for their reduction, and was thus able to obtain curves showing the early stages of the rapid growth of invisible films on iron and copper exposed to air at ordinary temperatures. A very complete study of the oxidation of zinc over a range of temperature by Vernon, Akeroyd and Stroud was published in 1939.
Our knowledge of oxidation has been aided by the examination of the oxide films. Here X-rays and, more particularly, electron diffraction methods, have been most helpful ; the work of Finch and Quarrell (1933-4) on electron diffraction methods deserves special mention. Less well known, but holding great promise, are the optical methods depending on the changes in the ellipticity of polarized light reflected at a metallic surface-changes which are themselves modified when a film is present on the metal. This method, which in suitable cases provides information both of the thickness and refractive index of the film, was first used to study oxide films in Freundlich's laboratory in Berlin (1927), was elaborated by the Norwegian physicist, Tronstad (1929-34), and has been further improved by Winterbottom ; in a different form it was used by Leberknight and Lustman in Mehl's laboratory at Pittsburgh about 1939. Other optical methods for obtaining the thickness of films depend on interference. The simplest of these methods was used by Tammann in his classical studies of the laws of film-growth in the years 1920-6, whilst later a more accurate spectroscopic method was employed at Cambridge by Constable (1927-9).
Further information has been obtained by the author alone (1927), and with J. Stockdale (1929), by stripping the films from metal, or, as in his later process (1938), by their transfer to transparent plastic ; films which produce interference tints on the metal usually display complementary colours after transfer. (The method of Stockdale and the author has recently been used by Phelps, Gulbransen and Hickman in their application of the electron microscope to the study of films stripped from several metals.) A form of film-stripping suitable for the chemical estimation of the constituents was developed by Vernon, Wormwell and Nurse in 1939, who established the fact that the polishing of the surface of stainless steel leads to an enrichment of chromium, the protective element, in the surface film ; evidence of analogous cases of enrichment had been obtained in 1938 by the Polish investigator, Dobinski, using the electron diffraction method.
Another field of research which has been developed concurrently is that of atmospheric corrosion. Vernon's extensive and accurate studies of the behaviour of metals exposed to the atmosphere, published at intervals from 1923 onwards, established simple laws connecting corrosion and time. One of the important results attributable to Vernon, Hudson and Patterson was the Principle of Critical Humidity ; it was found that frequently corrosion only became rapid in air when the humidity exceeded a certain value.
Very extensive atmospheric corrosion tests have been carried out for various Committees on both sides of the Atlantic. The American Society for Testing Materials were pioneers in this field, and have accumulated a quantity of valuable data. Particularly impressive are the results obtained by J. C. Hudson (1929-45), working first on non-ferrous metals for the British Non-Ferrous Metals Research Association, and then on ferrous metals for the (British) Iron and Steel Institute. Hudson's extensive researches have established the relative resistance of different materials in numerous different atmospheric conditions, and in the case of iron and steel have indicated how best protection can be achieved by coatings of paint or non-ferrous metal.
Up to about the time of the war of 1914-18, most of the experimental work on corrosion had been conducted by pure scientists. In later years, a serious increase in the frequency of breakdowns due to corrosion led to the inception of a number of technical researches on industrial problems. In Great Britain some were started by individual firms, but most by research organizations receiving support both from the Government and the interests concerned. A good example is the work on the corrosion of condenser tubes sponsored first by the Corrosion Committee of the Institute of Metals, whose first report appeared in 1911, and later by the British Non-Ferrous Metals Research Association. The experimental work was carried out in the early years by Bengough and several colleagues, one of whom, May, took over the leadership when Bengough passed on to other branches of corrosion research. Condenser trouble had been so serious in the war of 1914-18 that at one time it was causing the British Admiralty more anxiety than the enemy's fleet. Largely as a result of the work of Bengough, May, and their colleagues, along with the industrial development of new corrosion-resisting alloys, the position so much improved that in the war of 1939-45, condenser trouble, although not unknown, caused little anxiety.
Naturally researches dealing with any technical problem have tended to be carried out in countries where the problem in question has arisen. Thus marine corrosion has been extensively investigated in Great Britain, corrosion troubles in hydroelectric installations have received attention in Switzerland and Northern Italy, the bacterial attack upon pipes buried in clay soils was first studied in detail by the Dutch bacteriologist, von Wolzogen Kuhr, whilst special methods of protecting long pipe-lines have been worked out in the United States.
Industrial necessity has led to the investigation of many points which the pure scientist would probably have overlooked. It is somewhat doubtful whether a programme of pure science research would have established the fact that chemical and mechanical stress operating together can cause more weakening than when they operate separately. The mechanical stresses may be internal or applied. An early example of corrosion favoured by internal stresses was the so-called season-cracking of cold-worked brass in an atmosphere containing ammonia. This received study in many countries, but the extensive work of Moore and Beckinsale in 1920-3 not only established the conditions favourable to intergranular cracking, but also indicated the means of avoiding the trouble. Types of intergranular cracking connected with stresses occur on steel exposed to hot alkaline water (as in some power boilers) or in hot nitrate solutions ; these have also received study in many countries rather especially in the United States. Very important, in view of the development of aircraft, is the stress-corrosion cracking of certain aluminium alloys. The views held to-day owe much to the Russian investigator, Akimow, the American group associated with Dix, the British group led by Sutton, the Swiss metallurgists, von Zeerleder and Bosshard, and the German investigator, Bollenrath.
Especially dangerous is the combination of alternating stress and corrosion, producing the trouble known as corrosion-fatigue. This was observed during the war of 1914-18 on paravane towing ropes, and Haigh was the first to ascribe their rapid failure to conjoint chemical-mechanical action. About 1926 the American investigator, McAdam, commenced an extensive set of elaborate tests on corrosion-fatigue ; his earlier results were brought into a convenient compass by Gough in a lecture delivered in 1932. The scientific aspect of corrosion-fatigue has been under investigation for some years at Cambridge by Gould (1934-9), and later by Huddle and Tchorabdji Simnad.
Great importance attaches to the intergranular attack on austenitic stainless steel, met with most frequently in the neighbourhood of welds. Here an intergranular network deficient iii chromium (and therefore anodic to the main part of the metal) appears to be the cause of the trouble ; evidence was supplied by the photomicrographic work of Bain, the electrical measurements of Mears, Fink and Brown, and the analytical observations of Schafmeister and of Sy.
The researches on corrosion undertaken for technical reasons have often advanced the scientific understanding of the mechanism, and indeed Committees which have sponsored the technical researches have also arranged generous support for work in universities and elsewhere devoted to the scientific causes of the phenomena. The activities of Hatfield, for instance, were equally fruitful in promoting investigations of immediate industrial value and in supporting researches of a more academic character. Nevertheless, the prominence given to the technical aspect of corrosion has tended in recent years to spread the idea that the subject is simply a commercial one, and, perhaps for that reason, interest in the mechanism of the reactions between metals and non-metals has considerably waned among pure scientists during the last thirty years. Thus, although the kinetics of such reactions are in many cases well established, and obey relatively simple laws for which a logical interpretation can usually be provided, it is a strange fact that to-day such laws receive no attention in most textbooks of chemical kinetics, and often remain unmentioned in lectures on physical chemistry given to students.
Possibly it is really the strangeness of corrosion reactions which causes the orthodox physical chemist to regard the whole subject with suspicion. The type of procedure which he has come to apply almost automatically to each new problem-the measurement of a temperature coefficient followed by the calculation of some energy of activation-is not of immediate service in a subject where Geometry, and not Energetics, controls the situation, and where the velocity of a reaction often diminishes as temperature rises. This does not mean, of course, that the principles which govern the rest of chemistry become invalid in corrosion, but there is clearly need for some hard thinking before their significance becomes apparent.
The artificial exclusion of certain types of chemical reactions from the purview of the scientific chemist is not only discouraging to those working in the subject excluded, but is unfortunate for the science of chemistry, since it leads to a false picture of the whole situation. To some extent, however, this neglect may be due to the fact that the established facts of corrosion processes cannot readily be found collected in a book of convenient size. If this little volume should succeed in bringing the subject to the attention of chemists and physicists who have hitherto taken no interest in it, it will not have been written in vain.