Corrosion of Iron

When exposed to the combined action of air and liquid water iron readily oxidises or "corrodes," being converted into a brown, porous mass of ferric oxide in various stages of hydration. This oxide is popularly termed "rust." Any attempt to explain the changes taking place during the corrosion of iron must take into account the following facts: -

1. Dry air or oxygen has no effect upon iron at the ordinary temperature. If, however, the metal is heated in either of the gases, superficial oxidation takes place, the action being visible at about 220° C., when the metal acquires a pale yellow tint. This, as the temperature rises, gives place to a straw colour, and ultimately to purple and blue; but no rust is formed.
2. Liquid water alone, at ordinary temperatures, has no appreciable action on iron. A piece of polished iron may be kept for an indefinite time in a hermetically sealed tube in contact with air-free distilled water without undergoing any appreciable change.
   
   Pure Swedish iron has been kept by the author under these conditions for twelve years, during which time, apart from the merest trace of tarnishing, which could only be detected under a bright light, the metal appeared to undergo no change whatever. Even at its boiling- point water appears to have no action on compact iron, although the finely divided metal decomposes it with evolution of hydrogen gas.
3. A mixture of air or oxygen and water vapour has no action on iron, provided the temperature is not allowed to fall to that at which liquid water begins to be deposited.

On allowing the temperature to fluctuate, however, so that liquid water forms upon the iron, corrosion readily takes place. This disposes of the possibility that rusting is a simple case of direct oxidation, such as occurs when the metal is heated in air.

The fact thus established that the presence of liquid water is essential to iron corrosion points to the conclusion that before iron can rust it must pass into solution, presumably first in the ferrous state from which it is then precipitated by atmospheric oxygen in the form of hydrated ferric oxide or rust. In other words, the reaction is essentially ionic.

In the majority of cases of aerial and aqueous rusting there can be little doubt that one of the most active and important agents is carbonic acid. This unites with the iron, forming ferrous carbonate, FeCO₃, or perhaps the soluble bicarbonate, FeH₂(CO₃)₂, according to the equation: -

Fe + 2CO₂ + 2H₂O = FeH₂(CO₃)₂ + 2H,

the nascent hydrogen uniting with any dissolved oxygen to form water. The ferrous carbonate thus produced is next converted into rust by the oxygen of the air with the simultaneous liberation of carbon dioxide, which is now free to attack more iron. Thus a small quantity of carbon dioxide is able to assist catalytically the conversion of an infinite quantity of iron into rust.

In the presence of excess of carbon dioxide free oxygen is not, according to Paul, essential to the corrosion of iron. As the result of several series of experiments carried out with iron in contact with air-free water and carbon dioxide, this investigator concludes that under these special conditions the reactions involved are as follow: -

1. Solution of some iron and evolution of hydrogen gas: -
   
   Fe + CO₂ + H₂O = FeCO₃ + 2H.
2. The ferrous carbonate is decomposed by water into ferric oxide, carbonic acid, and formic acid: -
   
   2FeCO₃ + 2H₂O = Fe₂O₃ + H.COOH + H₂CO₃.
   
   The carbonic acid is then free to dissolve more iron, but after each neutralisation its amount is reduced by some 50 per cent., in consequence of the formation of formic acid.
3. The formic acid attacks the free metal, yielding ferrous formate: -
   
   Fe + 2H.COOH = Fe(CHO₂)₂ + 2H.
4. A portion of the formate is reduced by nascent hydrogen, yielding some formaldehyde and ferric oxide: -
   
   2Fe(CHO₂)₂ + 4H = Fe₂O₃ + 3H.CHO + H.COOH.
   
   The formaldehyde, being inert, passes out of the system, whilst the liberated formic acid is free to attack more metal.

From the foregoing it is evident that the ultimate fate of the carbon dioxide is conversion into formaldehyde, so that a trace of carbon dioxide cannot be expected to catalytically assist the oxidation of an infinite quantity of iron, as, theoretically, it should be capable of doing according to the simple cycle first described as occurring in the presence of air. Paul suggests that these reactions probably take place during the ordinary atmospheric corrosion of iron. Very possibly such is the case to a minute extent, but a great deal more evidence would be required before accepting these complicated cycles as representing the main course of iron corrosion. In any case, however, the theory is very suggestive, and is worthy of further investigation, for reduction of carbon dioxide in aqueous solution to formaldehyde is not an entirely new idea, Fenton having effected it by the nascent hydrogen generated from amalgamated magnesium, and observed that the presence of ferric hydroxide assists the reaction.

The problem as to whether or not a pure iron will rust in the presence of pure water and pure oxygen alone is one which has been the subject of considerable discussion. It is well known that many reactions which take place readily when the reagents are of mere commercial purity proceed only with diffidence when the substances are carefully purified. Thus, for example, commercial zinc dissolves with great rapidity in dilute sulphuric acid, whilst the pure metal is highly resistant to solution. Again, commercial hydrogen peroxide, which usually contains small quantities of dissolved salts, readily attacks iron, yielding a voluminous precipitate of rust. But the metal remains untarnished in a solution of Merck's pure hydrogen peroxide in pure distilled water, despite the fact that its surface is swept by a continuous stream of oxygen bubbles from the peroxide which it is catalytically assisting to decompose.

The question, therefore, as to whether or not pure liquid water and pure oxygen are alone sufficient to induce iron corrosion is one of considerable theoretical interest; and opinions are divided.

Berzelius was aware that iron does not rust when immersed in aqueous solutions of alkali hydroxides, and during the greater part of last century this fact was regarded as definitely proving that the presence of an acid or negative radicle is essential to the corrosion of iron. In 1903, however, Whitney suggested that, water being an electrolyte and split up, albeit to only a minute extent, into hydrogen and hydroxyl ions, renders it unnecessary to assume that any substance other than oxygen is required to effect the corrosion of iron in water. The condition of equilibrium between the whole and the ionised molecules of water may be represented as follows: -

nH₂O ⇔ (n-1)H₂O + H^• + OH',

where n represents a large whole number, the precise value for which is uncertain. If, now, a piece of iron is placed in the water, a minute portion will pass into solution in the ionic condition, free, gaseous, molecular hydrogen being deposited upon the surface of the undissolved metal. Thus: -

Fe + 2H^• + 2OH' ⇔ Fe^•• + H₂ + 2OH'.

Some of the ferrous and hydroxyl ions unite to form un-ionised ferrous hydroxide, until equilibrium is attained, according to the equation: -

Fe^•• + 2OH' ⇔ Fe(HO)₂.

The admission of oxygen to the system serves the double purpose of oxidising the liberated molecular hydrogen and converting the ferrous hydroxide into the basic ferric derivative, popularly known as rust. These two changes disturb the equilibrium, more iron passing into solution to be converted, in turn, into a further quantity of rust.

According to this theory, therefore, the presence of an acid is unnecessary; and the inhibiting power of alkalies is attributed to the suppression of aqueous ionisation by the presence of so many additional hydroxyl ions in accordance with the law of Mass Action.

Theoretically, it should be a simple matter to decide whether or not pure water, pure oxygen, and pure iron are alone sufficient to induce iron corrosion. In practice, however, the problem is one of extreme difficulty, owing to the elaborate precautions that must be taken to obtain each of the three substances in a pure condition, for any small trace of impurity, acting as a catalyst, may be sufficient to effect the oxidation of an indefinitely large quantity of the metal.

The extreme difficulty of removing traces of carbon dioxide from the walls of any apparatus was probably first realised by Moody. This investigator placed a piece of pure Swedish iron (99.8 per cent. Fe) in a tube containing a 1 per cent, solution of chromic acid to clean its surface, and a current of pure carbon dioxide-free air was passed through the apparatus for several weeks in order to remove all traces of foreign gases. Water was now distilled on to the metal, washing the chromic acid away, and the passage of air continued. The metal did not rust.

Similar results may be obtained in a much simpler manner by means of the apparatus. A is a small iron cylinder closed at one end, the other end being fitted with a stopper and tubes so arranged as to allow a constant stream of water to circulate through the cylinder to keep it cool. The whole is inserted in a flask containing a twice-normal solution of alkali hydroxide, and well shaken to remove all traces of carbon dioxide. Upon inserting in a water bath pure carbon dioxide-free water condenses on A in a continuous stream, washing away the alkali. The metal may be kept for an indefinitely long time in contact with the air and pure condensed water without corroding.

These experiments appear to prove fairly conclusively that pure oxygen and pure water are not sufficient to effect the corrosion of the purer forms of iron.

Unless the surface of the iron is cleaned and freed from condensed or occluded gases by a preliminary washing with chromic acid or alkali hydroxide, as in the experiments described, contact with pure water and oxygen invariably leads to the corrosion of iron, as is to be expected. It has therefore been urged that the cleansing reagents referred to above render the iron passive, so that the results are misleading. In 1910, however, Lambert and Thomson described their experiments with iron of an exceptionally pure character, prepared by reduction of highly purified ferric nitrate. The metal was found to be remarkably inert or passive, and could be exposed to the combined action of water and oxygen for several months without undergoing any visible oxidation. Passivity is thus one of the characteristic properties of pure iron, and it is very justly claimed that any passivity induced in the metals used by Moody and by Friend is simply due to the thorough cleansing of the metallic surfaces, and is, to that extent, an indication of their purity.

An attractive theory of the mechanism of corrosion has been outlined by Aitchison. Compact iron, when examined under the microscope, is seen to consist of crystals of ferrite separated from each other by an amorphous cement. It is reasonable to suppose that the solution pressure of this cement differs from that of the ferrite, for differences of this kind invariably occur between amorphous and crystalline varieties of substances. Upon immersion in an electrolyte, therefore, such as ordinary tap water or aqueous solutions of inorganic salts, a difference of potential exists leading to the corrosion of iron. If the cement is positive to the ferrite, it is the cement that will oxidise away; and vice versa. In a perfectly annealed specimen, in which there is but little mechanical strain, the action will, in the main, be confined to that between the cement and ferrite. If, however, there is any appreciable potential difference between the crystals of ferrite themselves, this will increase the effect, the total observed corrosion being the sum of the two actions.

The solution pressure of the cement might, conceivably, lie between that of the more positive and the more negative ferrite crystals, in which case the cement would function cathodically towards the one and anodically towards the other.

Taking the simplest case, however, in which the ferrite is at practically the same potential throughout, the corrosion will proceed at the junctions of the crystals and the cement. The action will not of necessity be confined to one face of contact, but may be expected to proceed at a maximum rate on that plane, resulting in the formation of a pit.

Corrosion is further accelerated by the presence of impurities such as oxides, sulphides, carbides, phosphides, and silicates, since these are invariably at a lower potential than the ferrite. The influence of alloying elements is particularly interesting. With carbon, for example, cementite or iron carbide, Fe₃C, is formed, and as this is electro-negative to ferrite, the latter corrodes at the points of contact. Addition of carbon, therefore, to iron tends to enhance its corrodibility. If a third element is added to the system, its influence upon iron corrosion is determined largely by the manner in which it distributes itself.5 If it dissolves in the ferrite, reducing its solution pressure, it reduces the potential difference between the ferrite and cementite, and thus enhances the resistance of the whole to corrosion. Nickel behaves in this manner, the whole of the metal passing into solid solution with the ferrite until the steel contains more than 8 per cent, of nickel. Such steels, therefore, do not readily corrode.

If, on the other hand, the third element is associated entirely with the carbon and forms part of the carbide, the corrodibility of the alloy will not be appreciably affected. Vanadium, tungsten, and molybdenum are cases in point, their saturation percentages in the carbide being approximately as follows: -

Vanadium	5.4 %
Tungsten	11.7 %
Molybdenum	19.0 %

Below these concentrations the metals do not pass into solid solution, but remain entirely associated with the carbide.

The behaviour of chromium is interesting, for this metal distributes itself between the ferrite and carbide, and tends to reduce the corrodibility of the alloy to an extent determined by the portion that passes into solid solution.

An effective method of showing that differences of potential exist between different parts of a piece of iron consists in utilising the ferroxyl indicator devised jointly by Cushman and Walker. A 1.5 per cent, solution of agar-agar jelly is prepared, a few drops of phenolphthalein added, and the whole rendered perfectly neutral whilst hot by titration either with alkali or acid as occasion requires. A small quantity of potassium ferricyanide solution is now added, and the solution poured into a shallow dish to cool. A clean sample of iron is placed on the solidified jelly and covered with a layer of warm solution, and the whole allowed to cool. After a few hours some very beautiful colour effects will have developed, and may be preserved for several months by keeping the surface of the agar covered with alcohol.

Where the iron remains bright the agar assumes a pink colour indicative of the presence of hydroxyl ions. At those points where the iron passes into solution the familiar colour of Turnbull's blue compound makes its appearance.

The presence of salts dissolved in water may greatly influence the manner in which the iron is attacked. Many salts exert a distinct chemical action on the metal. Thus, for example, when immersed in solutions of copper sulphate, iron readily dissolves, an equivalent amount of copper being precipitated in accordance with the equation: -

Fe + CuSO₄ = FeSO₄ + Cu.

On the other hand, iron remains perfectly bright and free from all traces of corrosion when immersed in solutions of the chromates or bichromates of the alkali metals, unless, indeed, the solutions are excessively dilute. This is generally attributed to the formation of a thin film of oxide on the surface of the metal which shields the underlying portions from attack, but this is not the only explanation, as has been seen.

Ammonium salts are very corrosive, particularly in warm or hot solutions, probably on account of the ease with which they undergo hydrolysis. For example, a saturated solution of ammonium sulphate is less corrosive than distilled water at 6° C., but at 18° C. it is much more corrosive. Again, on boiling iron drillings in a concentrated solution of ammonium chloride, hydrogen and ammonia are evolved, ferrous chloride passing into solution. Thus: -

Fe + 2NH₄Cl = FeCl₂ + 2NH₃ + H₂.

When ammonium nitrate solutions are heated in contact with iron, an analogous reaction occurs; ammonia is evolved, whilst the iron suffers appreciable iron corrosion under the action of the liberated nitric acid.

The effect of exposing iron plates to the action of salt solutions

It is instructive to examine the effect of exposing iron plates to the action of salt solutions of varying concentrations by determining the losses in weight consequent upon corrosion. Curve ACLS shows diagrammatically the usual type of results obtained. The point A indicates the loss in weight of the plate immersed in distilled water to which no salt has been added.

The presence of small quantities of the dissolved salt effects an increase in the corrosion of the iron, a maximum being reached at C, known as the critical concentration. Further increase in the quantity of the salt reduces the iron corrosion to nil at L, which point is termed the limiting concentration, and from this point onwards, until saturation is reached, the liquid is non-corrosive.

There are several salts that behave in this way at atmospheric temperatures, the more important being ammonium acetate; potassium bromate, carbonate, cyanide, ferricyanide, ferrocyanide, iodate, and permanganate; disodium hydrogen phosphate; and sodium borate and carbonate. In the case of potassium chlorate the points L and S appear to be practically coincident, whilst for the majority of salts the point S lies somewhere to the left of L, namely at S' - that is to say, saturation occurs before the limiting concentration is reached. Generally speaking, at the ordinary temperature, concentrated solutions of salts are less corrosive than distilled water - that is, the point S' lies below the level of A, exceptions being ammonium sulphate, aluminium sulphate, ferrous sulphate, and (at temperatures in the neighbourhood of 6° C.) sodium sulphate, potassium nitrate, and barium chloride.

Solubility of oxygen in the solution

Sufficient data have not as yet been accumulated to allow of a complete explanation of the form of the curve ACLS. The initial rise from A to С is probably connected with the number of ions introduced into the solution with rise of salt concentration. As the latter increases, however, another factor begins to make itself felt, namely, the decreased solubility of oxygen in the solution. This acts in the opposite direction by retarding the corrosion of iron. This is shown in fig. where the relative corrodibilities of Kahlbaum's pure iron foil in various concentrations of sodium chloride solution are depicted at 10° C. and 23.5° C., the relative solubility curve for oxygen being represented by the broken line.

On raising the temperature above that of the atmosphere, the tendency is for the critical concentration to fall - in other words, the point C (previous Fig) is pushed toward the left. Hence a solution that is more corrosive than fresh water at the ordinary temperature may prove to be less corrosive than fresh water at higher temperatures.

For example, a 3 per cent, solution of common salt at 10° C. is much more corrosive than tap water at the same temperature; but as the temperature rises the relative corrosivity falls, so much so that at 21° C. the salt solution is the less corrosive of the two. Since sea water contains some 3 per cent, of sodium chloride, it is of interest to inquire into the effect of temperature upon its corrosive powers. The few laboratory tests that have been carried out on the subject indicate that at temperatures below 13° C. sea water is more corrosive than tap water, whilst at all higher temperatures it is less so. Now, in the western part of the tropical Pacific Ocean a temperature of 32° C. is sometimes attained, and in the Red Sea and Persian Gulf temperatures of 34.4° C. and 35.5° C. respectively have been registered. Such waters should therefore prove less corrosive than river waters at the same temperatures.

In the Arctic Ocean, on the other hand, where the temperature lies in the neighbourhood of 0° C., the sea water is more corrosive than fresh.

The foregoing observations are of great importance to marine engineers, and further research on the subject is eminently desirable.

Mention has already been made of the fact that iron may be preserved from iron corrosion by immersion in dilute solutions of the alkali hydroxides.

Alkaline and Corrosion

An interesting case arises when iron is immersed in alkaline solutions containing inorganic salts. For example, iron will remain bright in a 1 per cent, solution of caustic potash for an indefinite time, but upon addition of potassium chloride corrosion readily takes place. It is possible, however, to increase the alkali to such an extent that corrosion is entirely prevented, no matter how concentrated the solution of chloride. The minimum amount of alkali required rises with the percentage of chloride until saturation of the latter is arrived at. This is indicated in fig. AK represents the solubility curve of potassium chloride in aqueous solutions of potassium hydroxide, and CE the maximum concentration of the chloride that may be present in the corresponding alkali solution without causing corrosion. Within the area CEA, therefore, corrosion readily takes place, but outside this area corrosion is impossible.

A characteristic feature of the corrosion under these conditions, however, is the tendency to " pit." This is a form of localised corrosion, the rust eating deeply into the metal at small isolated areas. Undoubtedly the rusting is originated by some irregularity - chemical or physical - in the metal, each pit being started at some point between the crystal grains. Even Kahlbaum's pure electrolytic foil readily pits in this manner, although when immersed in neutral solutions it usually corrodes fairly uniformly over its entire surface. The masses of rust formed during pitting are rich in ferrous oxide. Sometimes filaments of rust spread out in hair-like growths, brown in colour, which float or sink according to the density of the solution in which they are produced.

In practice such iron corrosion may prove disastrous. For example, an iron boiler might lose several pounds in weight through uniform superficial corrosion and yet not be much the worse. But a single ounce removed through pitting might be sufficient to perforate the metal and lead to serious consequences. The employment of weakly alkaline feed waters containing dissolved salts in ordinary boilers is a dangerous procedure, for the foregoing reason. The remedy would appear to lie in the addition of sufficient alkali to render the liquid non-corrosive.

Iron will remain untarnished for indefinite periods in the presence of concentrated solutions of the carbonates of the alkali metals, even in the presence of small quantities of other salts. If, however, the alkali carbonate is very dilute, it cannot entirely inhibit the corrosion of iron. Now, the minimum quantities of alkali carbonate required to inhibit the corrosive actions of a given concentration of various other salts of the same alkali metal have been determined. The results show that, if the added salts are arranged in order according to the amount of alkali carbonate required to inhibit corrosion, they are also not merely in the order of the relative strengths of their acid radicles, but the relative quantities of carbonate bear a general relationship to the numerical values found for the strengths of the acids by electrical conductivity methods. This is well illustrated in the following table: -

Sodium Salts (Concentration N/20).	Relative Concentrations of Na2CO3 inhibiting Corrosion.	Relative Strengths of the Free Acids
Chloride	100	100
Iodide	89	98
Bromide	72	98
Nitrate	54	98
Sulphate	52	70
Fluoride	39	10.5
Acetate	9	2

The close connection between ionisation and corrosion in dilute solution thus receives interesting confirmation.

Both the physical condition and chemical composition of rust vary considerably according to the conditions under which the corrosion of iron has taken place. Dunstan gives the result of analysing (A) rust collected from iron apparatus rusted in the laboratory, and (B) rust from an iron railing exposed for thirty years to .the air within twenty yards of the sea. His data are as follow: -

	A	B	Calculated for Fe2O3.H2O.
Ferrous iron	0.13	3.25
Ferric iron	60.93	59.25	62.8
Hydrogen	1.01	1.17	1.3
Oxygen	34.35	35.70	35.9
Carbon dioxide	0.15
Ammonia	0.006
Insoluble matter	3.20	0.60

From this it is evident that ordinary rust produced by exposure of iron to air corresponds very closely to the formula Fe₃O₂.H₂O.

When iron is completely immersed in distilled water there is usually no pitting, and the metal becomes covered with a loosely adherent cover of brown oxide, in which there may or may not be a trace of greenish ferrous oxide. When the metal is only partially immersed in water, a particularly vigorous oxidation takes place at the surface of the liquid, for at this point the latter is relatively rich in dissolved oxygen. The resulting mass of rust is frequently high in ferrous oxide. A somewhat similar result obtains when iron is exposed alternately to the action of water and air, the proportion of ferrous oxide in the mass depending upon the difficulty experienced by the atmospheric oxygen in penetrating its surface. This is clearly shown by the following analyses of samples of rust obtained from the unpainted interiors of iron flushing tanks in constant use. Several of the tanks had been unscraped for years, and the sides were blistered with masses of rust, brown without but black within.