Caustic cracking - a layperson's guide
TWI Bulletin, July/August 1988
Richard Pargeter, MA, CEng, MIM, is a Principal Metallurgist in the Materials Department.
A need is often felt in industry for existing knowledge to be passed on, rather than for new knowledge to be uncovered. This need is obvious to staff at The Welding Institute, through close contact with member companies and far too frequently is seen as a result of the investigation of a failure. One of the services which The Welding Institute can offer its members is education at all levels. The following words of advice are based on a presentation made to a group of oil refinery engineers at the request of a Member company: a problem of caustic cracking had been experienced, and the aim was to provide an overview of this failure mechanism and by so doing improve staff awareness. If you would like a presentation on a technical subject of specific interest to your staff, please contact us at Abington.
Cracking
Before launching into caustic cracking, let us think about cracking in more general terms. What is a crack? Most cracks that we come across in daily life, like cracks in crockery or window panes, are arrested brittle fractures, and generally result from an abuse of the material. Cups are not made to be dropped. Fortunately, this type of cracking is rare in industry, and particularly in chemical plant and refineries. Designers go to considerable length to ensure that material strength and toughness are adequate. So, if materials are strong enough, and tough enough, why do they crack?
Fatigue cracks
Well, of course there is fatigue. This is also something which we try to design against, but it is much easier to get the stresses wrong and joint shape plays an important role, through the effects of local stress concentrations. Furthermore, unexpected cyclic stressing is a much more common occurrence than unexpected static overload.
Hydrogen cracking
Then there is cracking because of local embrittlement. Fabrication hydrogen cracking would come into this category; the mechanical properties of steel are affected by the weld thermal cycle and hydrogen from the molten pool, and, because of the local embrittlement, the material fractures in that region under the influence of residual stress.
Stress corrosion cracking
Corrosive environments can also cause cracking. Most people are familiar with the way in which their cars, for instance, are eaten away by rust, but under the joint action of stress and corrosion, cracks can be formed. This stress corrosion cracking (SCC) can be a local embrittlement phenomenon, if it is what is known as hydrogen assisted SCC. This can occur in a corrosive medium such as an acid which results in the generation of hydrogen on the steel surface: steel absorbs hydrogen, rather like paper absorbs water, but whereas paper becomes softer when it absorbs water, steel becomes much more brittle when it absorbs hydrogen. So, if you have a stressed piece of steel with some small notch in it, possibly a corrosion pit, and it becomes charged with hydrogen in this way, it may crack.
Once a crack has initiated, further propagation becomes easier, as hydrogen will diffuse into the plastic zone around the crack tip. Furthermore, the yielding crack tip is a particularly active site for corrosion, and hence hydrogen generation. In this way, a high local hydrogen concentration can build up around the crack tip which can then advance through the embrittled region, and start collecting hydrogen all over again. Under these circumstances, it is fair to talk about embrittlement. However, many cases of SCC do not involve a true embrittlement process in the sense that the intrinsic material ductility and toughness are reduced. Caustic cracking is a case in point. It is often referred to as 'caustic embrittlement' but this is a misnomer. Certainly steel can suffer cracking with no evidence of gross plastic deformation, so that the cracks seem to have formed in a brittle manner, but this does not mean that the material is embrittled by the caustic environment.
Caustic cracking
Caustic cracking is a form of SCC, i.e. it is cracking caused by the conjoint action of stress and corrosion. However, unlike hydrogen assisted SCC, it does not depend on the environment changing the material mechanical properties. In fact, it is unlike all the cracking processes referred to so far in that it does not involve a fracture step. The crack is produced by localised preferential corrosion which cuts into the material.
Electrochemistry
In order to explain that, we need to consider some electrochemistry. Corrosion is not quite as simple as dissolution. If you put a piece of steel in water, it will go into solution, but you will not get steel back when you boil off the water. Figure 1 shows a simple diagram of a battery cell. On one side, the zinc corrodes; it acts as an anode and liberates electrons. On the other side, the electrons are consumed by a cathodic reaction on the copper. In fact, the external wire connecting the copper and zinc is not necessary, as electrons can pass quite happily across a copper-zinc interface, as shown in Fig.2. That is an extreme example of what happens when steel is immersed in a corrosive environment. Usually this results in surface roughening - what was a smooth shiny surface quickly becomes dull and pitted. The pits have been acting as local anodes (like the zinc), with the rest of the steel surface behaving as the cathode. Different regions may become cathodic or anodic relative to each other because of local variations in composition, or variations in corrosion product on the surface.
Fig.1. A simple battery, illustrating the process of electrochemical attack
Fig.2. A Zn:Cu couple; a battery with no connecting wire
Not all steels corrode, however, even if you put them in an acid. At least, they do corrode, but then stop almost immediately because a thin layer is formed on the surface and protects the underlying steel from further corrosion. This is known as 'passive' behaviour, the classic example being stainless steel which rapidly forms a protective oxide layer in most environmental conditions. Whether passive or active behaviour is experienced will depend on the material composition and the service environment. Ordinary steel will passivate in nitric acid, for example, while stainless steel will rust under low oxygen conditions. Passive films are usually cathodic with respect to unprotected surfaces. Under borderline conditions between active and passive behaviour, there will be a passive film with small cracks and holes in it. At these anodic sites, corrosive dissolution will take place, being driven by the relatively much larger cathode. These are just the sort of conditions under which caustic cracking takes place, as shown schematically in Fig.3.
Fig.3. Different forms of corrosion products:
3a) Protective film
3b) Small fissures, leading to rapid local attack (SCC)
3c) Large gaps, leading to rapid general corrosion
Caustic cracking arises in alkaline conditions, which tend to produce a passive layer on the steel surface. Once a crack has started, it is probable that the crack sides are successfully passivated, whereas the crack tip is not. That is why a fine crack develops rather than a blunt corrosion pit - and that is where the stress comes into play. The reason why the crack tip is special is that it is stressed; any film that is formed at the crack tip is disrupted, and the crack tip is kept anodic.
Summary of mechanism
To summarise, caustic cracking results from corrosive attack on the steel, and it depends on having the right balance of conditions, principally caustic concentration and temperature. The attack will only result in cracking in the presence of stress, and the rest of the material is undamaged. Caustic 'embrittlement' does not happen.
What to do
The questions which plant operators ask are: 'What are the conditions for cracking?' 'How do I recognise the problem when it occurs?' 'What action should I take if caustic cracking conditions are found?'
Factors influencing cracking
With regard to the environmental conditions which cause cracking, the best guide is probably the NACE chart, reproduced in
Fig.4. This tells you the conditions where you need to be careful. If you find yourself in areas B or particularly C, and need more specific advice, then The Welding Institute will be happy to help. The main points to remember are that cracking only occurs at temperatures above 50-80°C, depending on concentration (more caustic, lower temperature) and that conditions become more severe as temperature increases. It is also necessary to remember the importance of stress. Under the influence of high stresses, cracking can occur in relatively mild environments whereas at high temperatures and concentrations, little stress is needed. Residual welding and forming stresses are usually the most significant, which is why there is an intermediate band on the diagram where the problem can be alleviated by post-weld heat treatment for stress relief.
Fig.4. Caustic soda service chart published by the National Association of Corrosion Engineers (NACE)
There is an effect of material type, as indicated on the diagram by the suggestion of nickel alloys for more severe service. There is little variation in susceptibility to caustic cracking within ferritic steel types, however. The most significant compositional factor which has been highlighted is carbon content in ferritic steels - high carbon generally being found to be beneficial. With quenched structures, however, the reverse is true, so that even this guidance is of little help when dealing with a welded vessel. A beneficial effect of high Cr, and detrimental effects of high Mo and Si have also been reported while in Cr-Mo steels, phosphorus can also be detrimental. Nonetheless, even though steel composition and microstructure matter, they are much less important than the environment and the presence of tensile stress. If these are 'wrong', virtually any ferritic steel will crack, and will crack sufficiently rapidly for material differences to be insignificant from the practical viewpoint.
Identification of caustic cracking
A crack will usually be detected first as a leak. When a leak occurs, steam or water escapes and then evaporates, leaving behind a residue which will contain a high concentration of caustic. This concentration mechanism at leaky riveted seams was one of the main causes of boiler failures earlier this century. One diagnostic feature therefore is often the presence of a white fluffy deposit around a leak, which will be strongly alkaline when dissolved in distilled water. The definitive way to identify the mode of cracking is to prepare a metallographic section. Caustic is one of the few environments which cause intergranular cracking in ferritic steels. Others, such as amines and nitrates, can usually be discounted from knowledge of the process stream being handled by the plant concerned.
At lower temperatures (250°C), pitting attack with transgranular fissures may be observed, but all reported cracking has been intergranular. It is also said that caustic causes preferential attack on carbide lamellae in pearlite; if this is seen, it is highly probable that caustic cracking has occurred, but experience at The Welding Institute suggests that such carbide attack is often difficult to identify unambiguously. Of perhaps more use is the irridescent sheen on the surface of a crack which has been broken open, caused by the adherent magnetite layer, although in many service failures this is obscured by subsequent rusting.
An example of caustic cracking is shown in Fig.5 and it can be seen that the crack has many branches. This is characteristic of many types of stress corrosion cracking, and results in a distinctive type of imaging of a crack using radiography. Figure 6 shows a comparison between magnetic particle inspection (MPI), and radiographic detection of a caustic crack. Whereas MPI reveals a clearly defined pattern of cracks, the radiographic image is lacy in appearance. The location of cracking can also be a guide. As stated before, stress is necessary, and most commonly arises from forming or welding. Thus cracking is usually found on cold bends or around and in welds. The cracks do not commonly follow any feature such as a weld toe closely, in the way that a fatigue crack will.
Fig.5. Typical intergranular branched cracking in a C-Mn steel
Fig.6. Comparison between images of caustic cracking: Fig.6a) Magnetic particle
Finally, a word of caution; the concentration mechanism referred to above will occur also at a leak arising from a fatigue crack, for example, and caustic cracking may then initiate as a result rather than cause of the leak. To confirm that caustic cracking is the prime cause of failure it is really necessary to find a crack which has not leaked and examine it.
Inspection and remedial action
If caustic cracking is discovered, then it is usually an indication that the concentration of caustic has somehow gone above that intended. In most boiler systems, an alkali is added in one form or another in order to bring the pH into a range where little corrosion occurs. About pH 10.5 in water at ambient temperature and pressure is generally considered to be adequate. Below about 5% caustic additions, it is generally considered that there is no problem, and there is a lot of know-how involved in water treatment aimed at holding the pH at a sufficiently high level to 'passivate' the steel and avoid corrosion without inducing so much alkali that caustic cracking becomes a hazard.
Nonetheless, control is not always perfect, and if the alkali concentration has gone too high, cracking may be present in any piece of equipment that has contained the same fluid, and that has been in the susceptible temperature range. In this situation, a detailed inspection programme is essential. It should be realised that, although the caustic concentration has been corrected, high concentrations can still remain in occluded regions, and particularly in cracks, which may thus continue to propagate. Furthermore, if equipment is taken out of service and then restarted, there has been an opportunity for any occluded regions to dry out, possibly resulting in even higher caustic concentrations.
Thus, it is imperative to find all cracks and remove them, even if a fracture mechanics analysis indicates that they are safe. When cracks are found during such an inspection programme - and at the very least several original fabrication defects will almost certainly come to light - it is important to ascertain their cause. This can best be done metallographically, and can be done on site by The Welding Institute if it is in a large component or vessel, which the plant operator does not wish to destroy.
