Multi-Part Report

Corrosion at Cabo da Roca – 2 (a)

The second of a four part series

In conducting this work I have been greatly assisted by Luis Fernandes Silva and Rui Rosado who have provided multiple samples and photos for analysis. Luis has spent a great deal of time measuring spot-sulphate levels, and the pH profiles of bolts he has extracted during route refurbishment. We now know a lot more about why the Cabo da Roca crags eat stainless steel.


In Part 1 of this series, we identified the main types of corrosion occurring at the sea-cliffs of Cabo da Roca as sulphide stress cracking (SSC) and hydrogen embrittlement (HE). Here, in Part 2, we will address SSC in greater depth.

Sulphide Stress Cracking (SSC):

For a well developed, and well understood alloy like 304 grade stainless steel, it should be possible to design a product that never fails through the development of stress cracks. If we eliminate human shortcomings such as poor material control, and poor process control, this should certainly be the case. The popularity of this alloy comes down to the fact that it is very forgiving with regard to manufacturing requirements. It is a tough, robust material.

However, there are environments where this toughness can be seen to be totally compromised, or perhaps, where this steel is not so forgiving of poor manufacturing processes. There are forbidden environments for sure, and for many such, it is only in hindsight that we become aware of them. The multiple failures of 304 bolts at just certain specific sea cliffs around the world are a sure sign we have strayed from 304’s comfort zone.

But …. in what way have we strayed?

SSC is not SCC – What? Say again?

Most climbers are familiar with the term SCC, which stands for stress corrosion cracking. To be more exact, we should use the term chloride-mediated stress corrosion cracking. There are other mechanisms that attack other materials, but here we are referring to a failure mode of austenitic stainless steels in chloride-rich environments.

For the climbing community, this bogey-man, lurking in the shadows for well over a decade, has been ever ready to step into the spotlight to take the blame for the corrosion mischief de jour.

Initially, there was considerable opposition to allowing SCC up onto the stage of climbers’ affairs. Metallurgists pointed to the fact that SCC was a creature of higher temperature environments, and not the sort a mere sea cliff could provide. However, as the years went by, people began to point to exceptions where ambient temperature SCC was demonstrably a real possibility, and these exceptions began to be accepted as what must be happening on the corrosive sea cliffs.

Section of an extensively cracked hanger. This was the landmark study that put SCC into the climbing vocabulary. But is it SCC? From Sjong and Eiselstein

The currently accepted theory is that seawater encrustations can harbor high chloride concentrations at the metal surface. See Lieberzeit et al 2019 . There is solid physical chemistry to support this view, provided the ambient relative humidity is in the range 30% to 40%. Add to this, the fact that a cliff facing the sun can get very hot, and then indeed, the initiation of SCC might be a possibility.

However, there are both general and specific objections to this theory.

An urban myth has arisen that all 304 fails when installed on sea cliffs. If you live in one of the few specific locations where this is the case, you endorse the myth. However, if you live in one of the very many regions of the world where 304 has given long and robust service on sea cliffs, you tend to ignore it, but you, nevertheless, wonder if your bolts are about to fail. If the salt-crust theory is to be considered valid, it needs to explain in general terms the distribution of corrosive locations. Thus far, it has not.

A more specific, and in my opinion, overwhelming argument is that the thermodynamics of maintaining the elevated water potential of the concentrated chloride solution adjacent to the metal makes it an unlikely occurrence in the sea cliff environment. For a location a few metres from the sea, under what conditions are you likely to maintain a local relative humidity of say 35%? I would say none. Yes, I know that with much hand-waving you can describe local eddies of warm air moving off sun heated rocks, but, is this a solid generalisation? A lot (the vast majority?) of bolted routes are shaded from the sun, and are to be found under overhangs. I can easily point to failed bolts within caves at Railay where the sun never shines. And what about the fact that many failures of expansion bolts are below the surface, not exposed to the atmosphere, and in contact with the rock? There is zero chance of maintaining a high concentration of electrolyte at a local point in a porous rock. Chloride located at points of high concentration, if free to diffuse to points of lower concentration, will do just that. In all of this, basic thermodynamics is against the salt-crust theory.

We can summarise the key characteristics of SCC as follows –

  1. the part must be either stressed, or must contain residual manufacturing stresses.
  2. the process is an anodic corrosion process involving the removal of metallic iron into solution via its solvated cation, and thereby forming pits on the metal surface.
  3. the geometry of the pit formed at 2., interacting with the applied stress field at 1., must be such that the stress intensity at the root of the pit is sufficient to initiate a fracture.
  4. the fracture will propagate across the existing stress field at some specific rate.
  5. the rate of formation of pits in 2. and the rate of propagation of the fracture at 4., depend upon the temperature and concentration of chloride ion. Moderate chloride concentrations require temperatures above 60ÂșC before SCC can be shown to occur.
How does SSC differ?

SSC stands for sulphide stress cracking. The first thing to notice is the lack of the word corrosion. Whilst SCC is an anodic process necessarily initiated by the removal of metal to form a pit, SSC is a cathodic process and no removal of metal is involved. This is not to say you won’t see anodic processes operating in a sulphide environment. You do, and serious amounts of metal may be removed. However, as far as the initiation and propagation of stress cracks is involved, no metal is removed. What happens can be attributed to atomic hydrogen diffusing into the metal lattice.

SSC in the wall of a 304 SS pipe. Although pitting of the inner surface is evident, pitting is not a necessary requirement for the initiation of a stress crack. From Roffey & Davies

Where a section of metal presents a cathodic potential of sufficient magnitude, hydrogen ions at the surface accept electrons from the metal to form hydrogen atoms. As the process proceeds, hydrogen atoms pair up to form hydrogen gas. Because the paired atoms are more stable than the individual atoms on the surface of the cathode, the concentration of single atoms never rises to a significant level.

However, if we add sulphide into the mix, the equilibrium of the above cathodic reaction is disturbed. The pairing of atoms to form hydrogen gas is inhibited. For this reason, sulphide is called a hydrogen poison. With pairing suppressed, the atomic hydrogen concentration of the surface increases, and hydrogen begins to diffuse into the metal itself.

Whilst most authors agree that the inward diffusion of atomic hydrogen is a root cause of embrittlement, there is considerable debate as to the mechanism. It could well be there are several operating at once. It is also clear that the diffusion rate of hydrogen is very variable, with work-hardening, strain history and applied stress all impacting local hydrogen concentrations and flows through the metal.

SSC occurs when, for whatever reason, hydrogen makes its way to the stressed zone at the tip of an existing flaw, and once there, reduces the fracture toughness of the material in advance of a developing crack.

We can summarise the key characteristics of SSC as follows –

  1. the part must be either stressed, or contain residual manufacturing stresses.
  2. the process is cathodic, and non-corrosive, in that metal is not moved into solution.
  3. although cracks may initiate at pits, pits are not a necessary condition if sufficient cathodic potential exists.
  4. the fracture moves across the stress field applied in 1. but the direction is greatly modulated on the micro-scale by factors impacting hydrogen diffusion such as the presence of strain-induced martensite, and strain-induced dislocations.
  5. the propagation rate depends upon there being sufficient atomic hydrogen, which in turn, depends upon sufficient sulphide and sufficient cathodic potential.
  6. the process occurs at ambient temperatures.

Take a look here for a recent quick review of the anodic/cathodic, pit initiation requirement debate.

This is what SSC looks like at the cliff

Assuming there is sulphate available then, sulphate reducing bacteria (SRB) can colonise any point adjacent to a steel part where the oxygen level is sufficiently low. See Part 1 of this series.

Here we see a typical fracture just where the bolt enters the rock.

From Espinhaco: A total SSC failure just below the rock surface. In some locations it is common to fit a rubber gasket between hanger and rock in the belief that it will prevent moisture incursion. Perhaps it helps in that regard, but almost certainly it contributes to the low oxygen environment in which SRB can thrive.

Notice the uneven fracture surface and the dark iron sulphide deposit coating it. Notice also that the rubber gasket, which creates a seal between rock and hanger, and thus helps maintain the anoxic conditions favoured by the bacteria.

One of the ironies of the sulphate story is that whilst sulphate feeds SRB within hidden spaces, on the exterior surfaces sulphate has the effect of electrochemically protecting stainless steel from the chloride pitting agency of sea water. Thus, provided the internal sulphide chemistry never ruptures into the outer world, the effect of sulphate is to maintain stainless steel surfaces that look new, belying the dangerous corrosion within.

Provided the internal sulphide process doesn’t leak to the outside, the effect of freely available sulphate is to suppress chloride-enhanced surface pitting by seawater. Old bolts that are dangerously corroded internally can present as new and shiny.

SRB will colonise other anoxic regions besides the space where the bolt enters the rock. For instance, it is quite common to get stress cracking of the nut of an expansion bolt as SRB take up residence in the anoxic zone between the threads. Expansion bolts are typically tightened into yield, so it is not surprising that the stresses exist to bring about this phenomenon.

Furthermore, because both the nut and bolt of the assembly are under stress, and given that the assembly is home to SRB, it is no surprise to find examples where the bolt fails before the nut, as we see in the pictures below.

It is also possible for anoxic spaces to form between the hanger and the rock, and between the washer and the hanger. Fixed hangers are rarely annealed, and thus contain residual stresses. They are cold-formed and can be as much as 50% work-hardened with substantial quantities of martensite transformed from the original austenite. All this bears upon their stress cracking behavior, should a location on the surface be found that favours SRB colonisation.

It is common to find evidence of SRB activity at points on the back of a fixed hanger where it contacts the rock, and nearly always, at some point between washer and fixed hanger, but for reasons unknown, perhaps oxygen levels are too high, perhaps residual stress levels are too low, we rarely see stress cracking propagate from these regions. However, when it does occur, extensive fracturing both across and within the plane of the hanger is observed. Maybe there is some hydrogen blistering involved (to be considered in the next part of this series).

This fixed hanger from Promontorio shows cracking both within and normal to the plane. Fixed hangers are substantially work-hardened with relatively high levels of strain-induced martensite. There is a big difference between the hydrogen diffusion rate in martensite compared with austenite. This fact, as much as the residual stress field, likely explains the complex cracking pattern we see in fixed hangers. The red arrow indicates what may be a “hydrogen blister” rather than a stress crack. The pH paper will be explained down page.

One would think that glue-in ring bolts would be ideal because the only stresses they carry are residual manufacturing stresses, and the glue completely fills the space below the rock surface. This is certainly true, but there is another weakness that makes ring bolts vulnerable.

Although the overall strength of the glue to metal bond is high, it is nevertheless a weak bond from a chemical point of view. The polymer will release the metal, section by infinitesimal section, as water, and the dissolved salts it carries, migrate into the joint. It is a very small space, but SRB are likewise very small, and capable of colonising this growing, low-oxygen void.

Thus, is is not uncommon to encounter typical SSC fractures just below the glue-line.

These ring bolts from Promontorio have cracked at a point just below the glue-line. On the upper left we see that the crack extended only part way through the thickness of the bolt. The remaining bright section is a brittle fracture. For a glue-in ring bolt, the region just at the glue-line is the only point at which a low oxygen environment is available for the SRB to colonise. However, for many ring bolts it is also a point of residual internal stress associated with the weld. Thus it is not surprising that this is the commonest presentation of SSC ring bolt failure.
How do we know SRB are involved?

The answer to this is pretty straightforward. The environmental production of sulphide is always biogenic. Thus far, we have amply demonstrated the widespread existence of sulphate, which acts as the precursor to biogenic sulphide production by SRB.

If we can further demonstrate that the dark coloured deposits we see in the anoxic zones are sulphide, then there can be little doubt we are seeing the consequences of SRB activity.

Fortunately there is classical test that is very sensitive and specific for sulphide called the iodine-azide test. It works very well with metal sulphides like those that result from the action of SRB on steel. It also has the advantage of being easily applied to samples being examined under the microscope.

The process is illustrated below. Here we examine the dark, blue/black deposit below the washer of a fixed hanger. It’s likely that this is the iron sulphide, mackinawite, which is a common SRB corrosion product. It is a hard, almost glassy substance, but tiny amounts can be removed with a scalpel. When the iodine-azide reagent is introduced we observe a slow but steady evolution of bubbles of nitrogen gas which identifies this substance as a sulphide.

The genesis of metal sulphides has been extensively studied because of the latter’s economic significance to a number of fields. Thus there exists considerable published literature which we can draw upon to confirm the observations we make when we examine bolts suspected of having undergone attack by SRB. Let’s look at what is known about the sulphides that are produced.

The mono-sulphide, mackinawite (FeS), is always the first sulphide to be deposited when sulphide ions react with metallic iron or steel. It is a somewhat unstable compound, and can exist only within environments where oxygen levels are low, and an excess of sulphide ions is available. At ambient temperatures, and in the presence of even small amounts of oxygen, it transitions to the stable, mixed-valency-sulphide, greigite (Fe3S4). In its turn, this compound can transition via poly-sulphide intermediates to form pyrite (FeS2) . It is known that anaerobic cultures of SRB produce both mackinawite and greigite, but not pyrite. Thus we are dealing with a situation where we might be observing greigite that was formed as a direct result of SRB activity, or greigite that has transitioned from mackinawite, subsequent to a cessation of SRB activity due to, say, a rise in oxygen tension.

Mackinawite forms as nano-crystals that are tabular in form and aggregate by stacking to form a multi-layed coating over any available substrate such as a metal surface. It is not magnetic, and readily dissolves in strong acids. It commonly co-crystalises with nickel, should such be present.

Greigite forms iridescent crystals, and is the sulpur analog of the common iron oxide, magnetite (Fe3O4). Like magnetite it forms octahedral crystals that are strongly magnetic. It does not dissolve in strong acids.

Naturally occurring druse of greigite crystals. Credit here.

Armed with this information let’s lake a closer look at a deposit found on a washer from an anchor retrieved from Espinhaco.

The deposit has formed on the washer in the oxygen deprived space where it presses against the surface of the fixed hanger. Evidently it is quite brittle and most of it has been chipped off this specimen by the act of removal of the bolt from the rock. Where the deposit has been removed, we see broad pitting corrosion has occurred. It seems likely that the metal ions thus removed end up as constituents of the immediately adjacent sulphide deposit.

Taking a closer look at the deposit we see that it exhibits some crystal structure with similarities to the tabular forms of mackinawite. However, given the known instability of mackinawite, it would be expected that at least some transition to greigite would have occurred, and the observed crystal forms are, at best, pseudo-morphs after the original mackinawite.

Close up view of the deposit reveals some crystal structure reminiscent of the forms of mackinawite, but given the latter’s instability in an oxygen environment, these are more likely greigite pseudo-morphs after the original mackinawite.

Analysis of a chip taken from this deposit, supports the above supposition, ie we are looking at a deposit that, at formation was an iron, nickel mackinawite, the iron components of which have subsequently partially transformed to greigite.

iodine-azide test for sulphidepositive
soluble in strong acidonly partial
spot test acid-soluble component for ironpositive
spot test acid-soluble component for nickelpositive

The pattern of deposition and corrosion on the washer is fairly typical of what we find at these external, oxygen-deprived, metal-to-metal contact points. However, for bolts that show high levels of corrosion and cracking at zones just below the rock surface, other sulphide deposition patterns arise.

Taken from the same anchor that provided the washer in the example above, we find, just at the fracture zone below the rock surface, the iridescent mineral pictured below. It occurs not only on the open surfaces of the flanks of the threads, but also within voids of the bulk metal itself. It is a sulphide, forming what looks like octahedral crystals, and it is magnetic. It is almost certainly greigite, and, given the size of the individual crystals, it looks like it has been formed in situ, during SRB activity, rather than transformed from a layer of mackinawite after exposure to the air.

The picture below is from a different bolt retrieved from Espinhaco. In this particular case we see patches of mackinawite formation, and separate patches of greigite formation. It seems that the oxygen-depleted, free-space at the point just below the rock surface provides an environment for greigite formation that is independent of prior mackinawite deposition and the latter’s subsequent oxidation.

Bolt from Espinhaco showing both mackinawite and greigite deposits on the thread surfaces at a point close to the fracture just below the rock surface.

If sulphate is present, then we normally can find evidence of SRB activity in the form of sulphide deposits at any location where oxygen levels are suppressed. It seems that many of the colonies are benign, and it is only when they align with a stressed region do we see problems with the integrity of the component.

For example, the hanger in the picture below failed due to hydrogen embrittlement of the bolt at a point below the rock surface. Certainly SRB occupying the gap between rock and bolt are responsible for this failure, but if we take a closer look, we can find evidence of SRB activity at other points where the hanger was pressed to the rock surface.

Taking a further look at the same hanger, we see the outer surface presents that dangerously, misleading “new look” of bolts in a sulphate environment. However, there is a small leakage of rusty material from the anoxic thread zone of the nut that should stand as a warning that all is not well. Although this is not a typical, hard, sulphide deposit, it still tests as sulphide the iodine-azide reagent.

A second line of evidence that can be used to indicate that we are dealing with an active biological process is the marked differentiation in pH that occurs across areas of putative SRB colonization. It is not always possible to be sure of the geometry of the colony given their irregular three-dimensional nature, but as a general rule the margins are alkaline and the internal parts acidic.

The measurement is made by laying a strip of pH paper across a corroded zone, and observing the colour change.

What is striking is the degree to which the pH can deviate from neutrality. It can be as low as pH 2 (fully red) and as high as pH 12 (purple).

Sometimes the pH differentiation is not present, yet there are clear signs of the activity of SRB. My interpretation is that the pH differentiation across the surface is inherently unstable, and will only persist for as long as the SRB are active. If, for some reason, they become inactive, for example the supply of sulphate fails, then the differences in pH will fade, and the entire surface will approach neutrality.

This fixed hanger from Promontorio lacks pH differentiation across the surface yet shows signs of SRB activity. It seems likely that the SRB were inactive when this bolt was retrieved, perhaps due to a shortage of sulphate.

Based on what is known about the chemistry of SRB, a marked increase in pH around the margins of an active colony is easily understood. However, it is more difficult to explain the very acidic conditions we are seeing toward the centre of the colony. There are just a few studies where definitive methods have been used to investigate the process, and the results of those don’t align with what we are seeing here. For example Enning et al describe a well characterised system that results in iron carbonate (siderite) accretion with no pH differentiation. Complete understanding will need to await a time when more exact characterisation of the corrosion products is available.

Given our observations to date, it seems likely that SRB are early adopters of the anoxic zones in and around the installed bolt. However, the supply of sulphate needed for them to thrive is likely to be variable, depending, as it does, on the flow of ground-water through fissures in the rock. Certainly, sulphate spot tests often show surface sulphate levels to be anything but consistent. There is no clear association between highly corroded bolts and surface sulphate levels.

Further to the above, there is no reason to believe that the supply of sulphate into the ground-water is consistent over time. If the source is marine elemental sulphur, as I suspect is case for every sulphur crag I’ve investigated, then that supply might be very irregular, depending upon patterns of volcanism.

Given the above considerations, it is easy to understand that local start/stop conditions with respect to SRB activity could be quite randomly distributed, both across a crag, and over a span of many years. Thus it is very difficult to draw conclusions that one type of bolt is better than another, without long term tests involving the installation of bolts across a large number of locations.

Just when we think we understand ….

There is abundant evidence that SRB, and the hydrogen sulphide they produce, are the main drivers of the corrosion of stainless steel at Cabo da Roca.

It seems reasonable to conclude that an SSC mechanism is operating, with SRB feeding on the metal surface to produce the elevated sulphide environment, which in turn promotes atomic hydrogen diffusion into the bulk metal, ultimately enabling the stress cracking process.

This process requires little in the way of bulk metal removal for it to cause significant damage by way of stress cracking.

But, this explanation does not sit well with the picture below. This is a transverse section through a bolt immediately next to the fracture zone. Here we see a honeycomb of tunnels threading through the bulk metal. We also see crystals of greigite deposited within the bulk metal adjacent to the tunnels. It is becoming increasingly clear that sulphide, and quite possibly the bacteria are not located superficially, but have a mechanism by which they can deeply penetrate the metal. It appears that substantial amounts of metal are being removed.

Bolt from Espinhaco. Transverse section adjacent to the fracture zone following polishing and etching. The black features are tunnels through the bulk metal, the iridescent crystals are presumed to be greigite.

This is a story for another day. It looks to be quite technical and involved.

Coming Next:

I’ll be taking a look at the evidence for sulphide migrating through the metal of the bolt to conclude this part on SSC.

Then in Part 3, I look at the phenomenon of hydrogen embrittlement and see what we can learn from the published literature to guide future materials choices.

In Part 4, I will consider why it is we seem to be seeing no failures for 316, and consider what this means for future material choices.

Lieberzeit, J., Prošek, T., Jarvis, A. & Kiener, L. Atmospheric Stress Corrosion Cracking of Stainless Steel Rock Climbing Anchors, Part 1. CORROSION 75, 1255–1271 (2019).
Benning, L. G., Wilkin, R. T. & Barnes, H. L. Reaction pathways in the Fe–S system below 100°C. Chemical Geology 167, 25–51 (2000).
Sjong, A. & Eiselstein, L. Marine Atmospheric SCC of Unsensitized Stainless Steel Rock Climbing Protection. J Fail. Anal. and Preven. 8, 410–418 (2008).
Sherar, B. W. A. et al. Characterizing the effect of carbon steel exposure in sulfide containing solutions to microbially induced corrosion. Corrosion Science 53, 955–960 (2011).
Roffey, P. & Davies, E. H. The generation of corrosion under insulation and stress corrosion cracking due to sulphide stress cracking in an austenitic stainless steel hydrocarbon gas pipeline. Engineering Failure Analysis 44, 148–157 (2014).
Enning, D. et al. Marine sulfate-reducing bacteria cause serious corrosion of iron under electroconductive biogenic mineral crust: Microbial iron corrosion under electroconductive crust. Environmental Microbiology 14, 1772–1787 (2012).

Leave a Reply

Your email address will not be published. Required fields are marked *