Corrosion in Industry.
Introduction. The deterioration of metal structures by corrosion may be regarded as one phase in the world-wide process whereby all natural materials tend to revert to states of lowest energy level. Mam by his technical skill and ingenuity is wresting hundreds of millions of tons of iron from the earth every year, and the implacable forces of corrosion are returning millions of tons of iron to its original form each year because it has not been given sufficient protection to withstand the conditions under which it is being used.
In fact, the process of recovery of the metal from its ore is so completely reversed by corrosion that often the products formed are very similar to the minerals occurring in the earth's crust, and in some cases they are crystallographically identical.
Although many writers have attempted to estimate the monetary losses involved for the world in replacing corroded metal and in taking protective measures, the human aspect is of even greater importance. Who can estimate in terms of money the psychological effect on industrial staffs of the ever-present menace of corrosion; and of the exasperation caused by a sudden breakdown of a smoothly running system due to rapid localised corrosion? The first step in attempting to prevent this colossal wastage of energy due to corrosion is to reach an understanding of its mechanism and to determine the major factors influencing its progress.
The Mechanism of Corrosion. Except for a comparatively few more-or-less specialised conditions, most instances of corrosion occur in the presence of a liquid having an appreciable (though often low) electrical conductivity. Therefore, the electrochemical theory of the corrosion of a metal in contact with an electrolyte has very general application.
Let us consider the action of oxygen in dry air, free from solid particles, on the surface of a metal which is normally unstable in the pure state, e.g., iron or zinc. No visible change occurs, but it has been shown, largely by the fine work of U. R. Evans and his co-workers of Cambridge, that air invisible protective oxide film does form. If kept in a perfectly dry atmosphere this film will continue to give protection. If the metal is placed in a moist atmosphere, water will condense on the surface and the conditions will become favourable for the establishment of electrochemical corrosion. If by means of direct chemical action,
products are formed on the metal surface in the right physical form and the right geometrical position to protect the underlying metal, the reaction does not proceed and the corrosion is “self-stifling.” It is typical of an electrochemical type of attack that the products are deposited away from the region where solution of the metal is occurring; no self-stifling is possible, and the attack continues, e.g., zinc in sodium chloride solution.
Text Fig. 1 (U. R. Evans).—Electrochemical Corrosion of Zinc in Sodium Chloride Solution (Schematic).
The final result is the same as that when zinc combines directly with moist oxygen, hydrated zinc oxide being formed. When an electrochemical action is involved the stifling which is characteristic of the direct attack does not occur. The potential difference between anodic and cathodic areas necessary to establish the cell may arise from a large number of different conditions—in fact, any condition which will cause a significant change in equilibrium-electrode potential of one part with respect to another. This may be brought about by the presence of some material on the surface of the metal, whether another metal, a corrosion product of some form, or a foreign inert material, or again by cell conditions in the electrolyte itself such as differences in concentration of salts, or of dissolved gases. Once the potential is established, the rate of attack is governed by other factors such as the availability of oxygen to depolarise hydrogen liberated at the cathode, and is directly proportional to the magnitude of the current flowing in the circuit.
In industrial problems bi-metallic galvanic action often occurs and it is essential for the industrial chemist to know what metals may be safely joined together when in contact with a common electrolyte.
Metals are corroded by several different conditions as listed below:
2. Forms of corrosion.
Visual inspection or mechanical examination of the corroded materials usually indicates that the corrosion has proceeded according to the following forms:
2. Rate Factors.
Possibly the most important feature of corrosion is the rate at which It proceeds. This rate is determined or at least affected in most cases by the following more or less measurable factors which exist during the progress of the corrosion.
4. Metal Corrosion Properties.
The attack of corrosion on the metals depends somewhat on certain properties of the metals some of which are typical of the metals themselves and some of which depend on its particular form while corroding.
Corrosion takes place on metals and for completeness the classification of metals according to their corroding properties or tendencies is given below:
Text Figure 2 is taken from McKay and Worthington's work on Corrosion Resistance of Metals and Alloys and lists all the commoner industrial metals and alloys. Metals from separate groups in contact will cause corrosion of the one higher in the series. Metals grouped together have little tendency to produce galvanic action. This series has been found from laboratory measurements to be correct for many common dilute aqueous solutions, e.g., sea water, dilute acids and alkalies. The actual values of the electrode potentials are not given as these vary widely according to the conditions of immersion.
Classification of Corrosion. The numerous and complex factors involved in corrosion problems are well shown by another table from McKay and Worthington.
Corroded end (anodic)
Chromium iron (active)
Protected end (cathodic)
Here an attempt has been made to classify corrosion under five headings—corrosives, forms of corrosion, rate factors, metal corrosion properties, and metals. Each of these groups is further subdivided. The total number of possible combinations of these factors representing different materials and conditions of corrosion make the task of dealing with a complete range, even of representative types, one for a full-length text book.
Here are given some instances of a number of types of corrosion encountered at the Dominion Laboratory in recent years.
Corrosion in Boilers. One constant concern of every industrial plant of any size which uses steam-raising equipment is the prevention of corrosion in boilers, feed-water, and pre-heater tubes. The composition of a boiler water which will neither deposit excessive scale upon the hot surfaces of the boiler nor cause excessive corrosion of the metal parts is rather critical and requires close chemical control.
By far the largest damage to boiler installations is caused by the presence in the feed-water of gases dissolved from the atmosphere. Of these, dissolved oxygen is the most destructive. Carbon dioxide also assists considerably by giving the water an acid reaction. The most common sources of dissolved gases reaching the boiler are in make-up water from the mains or from condensed steam by contact with the atmosphere on its way back to the boiler. In both cases the result is the same—rapid attack on iron and steel fittings with the formation of various mixtures of the oxides of iron.
The amount of dissolved oxygen necessary to cause noticeable corrosion varies from 0.03 to 1.0 ml. per litre of water, depending on the design and operating conditions of the boiler. The bulk of the dissolved oxygen may be removed by de-aeration or pre-heating, and the final traces removed by the addition of a reducing agent such as sodium sulphite, or by passing the water over iron turnings or some other form of scrap iron having a large surface area.
Plate 2, Fig. 1, shows the inner surface of a feed-water tube pitted by dissolved oxygen. The light-hand section shows the nature and depth of penetration of the pits which are formed in a matter of months. In this case the issue was complicated by the presence of an adherent scale over a large portion of the tube—illustrated in the left-hand section. In some cases deep pits had been completely sealed off by it. Where this had not occurred the attack continued.
The scale was found to be mainly silicate and sulphate of calcium. The majority of the pits were filled with oxides of iron—very largely magnetite Fe3O4. The water passing through the pipes was a mixture of natural make-up water and return-condensate from the boiler, the latter in a highly aerated state.
It seems likely that the pipe when originally installed carried a certain amount of rust or patches of grease or oil on its inner surface. Scale from the make-up water would form readily on this surface except where these regions of foreign matter occurred. Now natural scales such as these tend to be cathodic to iron and steel; the regions where oil or rust had retarded the formation of an adherent protective scale would thus become anodic, and galvanic action and rapid solution of the metal from the anodic areas would occur. In this case the cathodic areas were considerably larger than the anodes and so the corrosion progressed by rapid pitting.
The dissolved oxygen present would tend to increase the rate of attack by acting as a depolarizer. The remedy here was to ensure that the pipes when installed were free of grease and rust and to de-aerate the water fed to the boiler.
Plate 2, Fig. 2, illustrates a different type of pitting in which the characteristics of galvanic attack are even more marked—note the deep cup-shaped pits with under-cut edges. In this case the corrosion product has been largely retained in solution and so removed, leaving the surface of the metal and the interior of the pits comparatively free of rust.
It was known that copper piping had been used in other parts of the installation, and this, combined with the unusual colour of the deposit on the part of the tube not pitted, which was reminiscent of cuprous oxide, suggested examination for copper. The deposit was found to contain considerable quantities of this metal. The cause of the pitting was now obvious—copper had been dissolved from the copper tubes elsewhere in the system by the hot condensed steam, and deposited as metal over a considerable proportion of the steel nipple. The chances of obtaining a perfectly homogeneous continuous film of copper over the whole surface under these conditions are very remote. Accordingly, those areas not coated became the relatively small anodes of a copper-iron couple and intense pitting resulted.
Here the remedy would be to use as far as possible throughout the installation the one type of material or at least materials not far removed in the galvanic series referred to earlier.
Corrosion due to the Poor Quality of an Iron Casting. The next three figures illustrate very well the way in which variations in the microstructure and the presence of segregated impurities in an alloy may affect its resistance to corrosion.
A firm manufacturing hydrochloric acid found that the large cast-iron pots from which the acid was distilled were being perforated in a very short time, failure in some cases occurred after only 30–50 charges instead of a normal life of 80–100 charges of the acid-salt mixture used in preparing the hydrochloric acid. After checking that there had been no change in operating conditions the composition and structure of the iron from the faulty pots was examined.
Plate 3, Fig. 1, shows the appearance of a portion of cast iron from the vicinity of a hole in the pot, after deep etching with hydrochloric acid. The upper edge of the specimen was the edge of the hole. The pits and lines are due to the selective attack of the etchant on non-metallic inclusions in the iron, probably slag. This gives an indication of what would happen, during the manufacture of the acid, on the surface of an acid pot where numbers of such inclusions were present. Analysis of the iron showed that there were considerable variations in total-carbon content, the figures for three regions, two of which were the outer and inner portions of the same section, were 2.7, 2.9, and 3.3 per cent. In addition, considerable variations in micro-structure were found.
Plate 3, Fig. 2, shows the structure of the iron on the inner surface of the pot (3.3 per cent. total carbon) consisting of graphite flakes, and approximately equal areas of pearlite and ferrite.
The next photograph (Plate 4, Fig. 1) is from the outer surface (2.9 per cent, total carbon) and has slightly more graphite, which is embedded in a matrix of ferrite almost completely devoid of pearlite.
Non-uniformities such as these are indications of faulty melting practice and irregular heat-treatment, and both the presence of slag inclusions and the non-uniformity of micro-structure could contribute to a reduction in corrosion resistance and a much shorter working life.
A number of other materials are available with a higher resistance to concentrated acids than grey cast iron—e.g., high and low silicon irons and alloy cast irons. However, because of certain disadvantages of each of these alternatives, such as poorer mechanical properties, difficulty of machining, increased precautions in casting and heat-treating, or higher cost of alloying elements, it is doubtful whether they could compete economically with a good quality grey cast iron. The remedy in the case described is to ensure that the pots used are of a good-quality iron.
Corrosion of Stainless Steels. The use of stainless steels for industrial and domestic purposes has increased tremendously in the last 15 years, and notable advances have been made in developing new alloys for special conditions. However, stainless steel is not by any means a cure-all for corrosion troubles. There are many conditions of usage where it is unsuitable and some in fact where failure would occur more rapidly than if some of the older better-known alloys were used.
Having had occasion to consider the suitability of stainless steels for use at high temperatures in the presence of oxidising sulphur gases, it was of particular interest to find that there was in the Laboratory an excellent example of the performance of such a steel under these conditions.
The next three illustrations show the nature of attack on stainless steel pins used to support the combustion capsule in an oxygen bomb used by the Coal Survey Laboratory for sulphur determination on coal. Plate 4, Fig. 2 shows the pitting of the collar of the pin where it passes through the top of the bomb.
Plate 5, Fig. 1 shows similar pitting in another portion of the same pin. This pin is electrically insulated from the metal of the bomb and from the second pin and it is therefore likely that it would reach a higher temperature during combustion of the sample than would the second pin which has direct metallic contact with the top of the bomb. The lesser degree of attack on the second pin due to the lower temperature is shown in Plate 5, Fig. 2.
This type of corrosion by pitting is very common with stainless steels. Their resistance to corrosion is due to the formation of a very thin protective oxide film on the surface of the metal. This film, normally resistant to attack
under a wide range of corrosive conditions is broken down by hot sulphur gases and pitting such as shown occurs.
Corrosion due to Dezincification. The next photograph (Plate 6, Fig. 1) shows the extent to which zinc has been removed from a brass condenser tube used in a steam-turbine installation.
In this case the cooling medium was sea-water. The darker regions, occupying at this point more than half the circumference of the tube, are masses of spongy copper from which the zinc has been removed. The remainder of the tube is the original brass which had the composition copper 69 per cent., zinc 29.5 per cent., and tin 1.5 per cent.
Plate 6, Fig. 2 shows one of these dezincified plugs under a higher magnification; note the porous irregular nature of the copper left in the pit. There has been a certain amount of controversy as to whether in this process the zinc is preferentially dissolved leaving the copper, or whether the brass as a whole goes into solution and the copper redeposits as a result of concentration cell action near the surface of the metal. Recent work by W. D. Clark on a “dezincification” type of attack on phosphor-bronze seems to confirm the theory that solution and redeposition of copper takes place.
The remedy for this form of attack is to use a material not subject to dezincification; aluminium bronze is considerably better than a straight brass and a 70/30 copper-nickel alloy is better still.
Corrosion due to Bi-metallic Couples. The next four illustrations depict corrosion due to the formation of a bi-metallic couple.
The first (Plate 7, Fig. 1) is a general view of three tubes from a heat exchange unit. The tubes were of mild steel, copper-coated, and were of rather interesting construction. They were fabricated by double rolling a copper-clad steel sheet. The effect of this was to introduce a layer of copper parallel to the surface of the tube and located half-way through the wall.
Plate 7, Fig. 2 shows a cross section of the portion of the tube where rolling began and ended. In Plate 7, Fig. 1, the uppermost tube carrying the T joint was obviously more severely corroded than the other two; and the severity of attack diminished with distance from the joint. The other two pipes, although superficially pitted, still have a long period of usefulness.
Plate 8, Fig. 1 is a cross section of the top tube etched to show the central copper layer. The outer and inner copper coatings were too thin to show up in the sections. In one part the pitting has penetrated to the copper layer, has been arrested for a while and has then carried on as pores in this layer became uncovered. The copper in the bottom of these pits, with smaller pits developing in them, could be seen on the unsectioned tube, and in shown in section at this point.
Plate 8, Fig. 2, shows an enlarged view of one of these pits, which has not yet reached the central copper layer. Note the deep rounded shape of the pit and the overhanging lip characteristic of galvanic attack. The piece overhanging on the right of the section is probably a portion of the original copper coating which would be the cathode for the initial attack.
The cause of the corrosion in this instance was initially the poor quality of the original copper coating allowing superficial pits to develop through its pores. In the case of the tube carrying the T joint this action has been greatly accentuated by the joint forming the cathode and the exposed steel in the pits, the anode of a galvanic cell.
A heavy tin or galvanised coating would have been of considerably more value in this case than the copper coating. The electrolyte necessary for the action to proceed was derived from condensation of water vapour on the tubes and the leaching of small quantities of salts from the insulating packing surrounding the unit.
Another instance of corrosion due to galvanic couples formed by the breakdown of protective coatings was the failure in a very short period of tinned copper steam-heated vessel. Normally pure tin is anodic to pure copper and would be expected, if the coating was perforated, to offer sacrificial protection as does zinc on iron. In this case the tin coating had broken down and deep pits had penetrated the copper sheet. An attempt to determine the cause of the trouble was made by sectioning the tin coating on the perforated sheet and comparing its structure with that of a good-quality coating.
The average tin coating is much too thin to reveal details of the alloy layers in a vertical section with the microscope available at the time. Sections were therefore prepared by forming a strip of the sheet to be examined (previously copper plated to protect the tin coating) into an arc of a circle of predetermined radius, mounting it in bakelite and preparing an oblique section of the coating by grinding the convex curved surface.
In this way the following micrographs were obtained.
The first one (Plate 9, Fig. 1) shows the structure of a good-quality hot-dipped tin coating on copper. In looking at such a section we are really examining a phase diagram in pictorial form for the full range of alloys of tin and copper. In practice the lines of demarcation between the phases are not ideally sharp. From the outer surface of the tin in towards the base copper the major constituents are—
A layer of a solid solution of copper in tin.
Next to this, and scattered throughout it as isolated grains, is a region consisting of a eutectic mixture of tin-rich alloy, Cu6 Sn5, and the solid solution (1); the eutectic containing approximately 0.75 per cent. of copper.
This is followed by a relatively thick layer of the tin-rich alloy Cu6Sn5.
Then follows a thinner layer of copper-rich alloy Cu3Sn.
Finally the base copper itself.
In normal hot-dipping practice excess tin is drained or wiped off before solidifying so that the outer layer of a good coating will consist mainly of eutectic mixed with a certain amount of solid solution. The resistance to corrosion of tinned copper is dependent on the continuity of this eutectic layer. If it is porous or absent in places, galvanic action will occur; since it has been found that under a fair range of conditions the alloy layer, Cu8Sn5, is cathodic to the eutectic layer.
Now this alloy layer is harder and more brittle than the eutectic and as solution of the latter progresses and more and more alloy is exposed, mechanical stresses are bound to find flaws in it. Once this occurs galvanic action will continue between the copper-rich alloy, Cu3Sn, as anode and the tin-rich alloy, Cu6 Sn5, as cathode. The attack will take the form of deep and rapid pitting, since the cathodic layer is uppermost and has by far the larger electrode area. Now alloy Cu6 Sn5 is also cathode to copper. This means that when the Cu3Sn layer has been penetrated the pitting will not stop there but will continue on through the copper until perforation of the sheet occurs.
In the case of failure under consideration the structure of an unattacked portion of the pitted coating (Plate 9, Fig. 2) was found to be very different from that of the section (Plate 9, Fig. 1) just described. Instead the eutectic layer was missing entirely and the alloy layer, Cu6 Sn5, was made up of very large grains which did not form a continuous layer.
The next illustration (Plate 10) shows an attempt to reproduce this type of structure by overheating a specimen of the good coating shown in Plate 9, Fig. 1. This section was heated for five minutes at 400°C with the result seen in this slide. An enrichment of all the layers occurred due to the migration of copper from the base sheet throughout the coating.
The coating now consisted of—
Remnants of the tin-rich alloy, Cu6Sn5,
A very thick copper-rich alloy layer, Cu3Sn,
Note the entire absence of the protective eutectic layer.
The conclusion drawn from this work is that the protective properties of a tin coating on copper can be very easily destroyed by overheating for a very short time at a comparatively low temperature. Such an effect could easily be produced by allowing a steam-heating vessel to boil dry.
Discussion on Paper on Corrosion in Industry.
Mr. L. Wilkinson, after expressing his appreciation of the lecture, referred to work he had carried out several years ago with Mr. S. H. Wilson on the solution of copper from tinned and untinned hot-water clyinders. It was found
that with untinned cylinders the copper content of the water decreased with time to a low figure. With tinned cylinders, however, the copper continued to build up to a relatively high value. He considered that Mr. Hunt's work on the metalography of tin coatings offered an explanation of this phenomenon; that a tin coating of poor quality could be responsible for pitting of the under-lying copper and thus cause a continuous build up of copper in the water.
Dr. G. Moir referred to trouble which he had experienced some time ago with pitting along each side of welds in stainless-steel vessels used in brine solutions.
Mr. G. S. Lambert explained how an unstabilised stainless steel would undergo this type of “weld decay” under corrosive conditions. This difficulty has been overcome by using alloy steels stabilised by the addition of small amounts of titanium and columbium. This prevents precipitation of chromium carbide at the grain boundaries.
Mr. I. S. Hunt, in replying to Mr. Wilkinson's remarks, confirmed that a poor-quality tin coating in which the protective eutectic layer was missing or discontinuous could cause excessive solution of the underlying copper such as was described.
In reply to Dr. Moir's remarks Mr. Hunt referred to tests being carried out at the Dominion Laboratory which were showing that even welded stabilized stainless steels were not completely resistant to chloride solutions. In the tests carried out so far, pitting developed in the welds in both acid and alkaline oxygen-saturated brines. This pitting did not develop when the brines were saturated with nitrogen. The work had not yet been carried far enough to determine what were the conditions favouring the development of pits in the weld regions.