Thailand Crags – 1

Chasing Some Demons

The first Part of a multi-Part Series

Introduction:

Tonsai/Railay is where stainless-steel anchors go to die.

So, this is where it all began.

Actually, I tell a lie, it began elsewhere, but, as far as notoriety within the climbing community is concerned, our story begins many years ago on the tropical beaches of Tonsai and Railay.

These magical crags cast a spell over even the most casual of visitors. It is not surprising, therefore, to find magic infusing the explanations as to why such crags eat stainless-steel. It’s almost as if scientific rigor is shunned as plain bad manners in such a beautiful location.

But… time is a ruthless predator of casual myth, and if you climb often enough, delusions will be shattered.

So, it was for me. One day, reaching up to clip a bolt, the magic deserted me as the fixed hanger casually detached from the rock. Both fascinated and appalled in equal measure, I vowed to take a close look at what was happening here should I survive the back-climb to the beach. Over the years that followed, I, with the help of others, was able to show the problem was specific to Tonsai/Railay rather than Thailand in general. Furthermore, I was able to elucidate that it likely involved attack by sulphate reducing bacteria (SRB).

As time moved on, I shifted my attention to an analogous SRB corrosion problem occurring on the Portuguese coast. My reason for doing so was that I was keen to simplify the problem space by leaving the “tropical magic” chatter behind. Unfortunately, such a shift in direction left a body of work unpublished. So, let’s fix that omission right now, and at the same time let’s recognize the hard work of the folks that helped me with the Thailand crag sampling.

I am indebted to the following people who have helped out with sampling over the past many years – Martin Roberts, Francis Haden, Cheang Qing Xin (QX), Josh Lyons, Spenser Gray, Stephen Gladieux, Elke.

And, yeah, I guess I’m going to have to confront that chatter.

Crags are all good until suddenly they are all bad:

It really is like that. It is very hard to “unsee” a bad bolt once you have encountered it. One begins to wonder if in fact all bolts pose death traps for the unwary – whatever happened to those not-so-distant halcyon days when all bolts were bomber? And so, feeding on such fears, we find the demons of dubious scientific rigour rearing their heads in the public discourse.

But, against that tide of unreason, let’s get something important said right at the get-go –

For climbing purposes 304 (A2) is a brilliant engineering material. Its toughness, its general corrosion resistance, its low cost and its tolerance to inexact manufacturing marks it out as an outstanding material for climbing anchors.

There, I’ve gone and said the unsayable! Already I sense the drawn daggers of the keyboard-warriors of the internet. Ouch! But… patience good people. Holdoff with the daggers, and instead come forward with solid evidence.

So, 304 is absolutely all good until… until what? Let’s put some demons on trial. Let’s work out the real reason for catastrophic failure of this grade of stainless steel.

There can be no doubt that when it is bad, it is terrible…. but why?

The “tropical karst” demon

The “tropical karst” vibe took quite a hold on the collective imagination, with talk of stainless-steel bolts being attacked by “acid jungle juice” running down the cliff faces. Such fanciful explanations cropped up in places where you would think chemistry 101 should have prevailed.

Certainly, it is a tenet of basic chemistry that water in contact with both limestone and the atmosphere will be alkaline, no ifs, no buts, no magic.

The fundamental equations of this process include the following. Taken together they explain both the dissolution and the precipitation of calcium carbonate as a function of carbon dioxide availability.

CO_2(gas)  \rightleftharpoons CO_2(soln)

CO_2(soln) + H_2O \rightleftharpoons HCO_3^- + H^+

CaCO_3{^0} + H^+ \rightleftharpoons CaHCO_3{^+}

CaHCO_3{^+} \rightleftharpoons Ca^{++} + HCO_3^-

The diagram below illustrates the interplay of the various equilibria associated with this process. Note that atmospheric CO₂ concentrations result in an equilibrium hydrogen ion concentration somewhat over pH8.

As an aside, I would point out that the pH of sea water is similar because it is dominated by the same equilibria. Yes, there is a lot of sodium plus a lot of chloride in play, but, because of the lack of an ion in common with those of the calcium bicarbonate system, the sodium and chloride don’t contribute much to the observed pH.

As a sanity check, we can look at the pH values of the wall-wash samples in our extensive data base. We see that across many rock types, and differing climates, it is the limestone crags that are associated with alkaline pH, whilst the non-limestone ones are slightly acidic. Yes, I admit that a wall-wash sample is unlikely to have perfectly equilibrated with the rock surface during sampling, but nevertheless I’d argue that the pH will be indicative of the equilibrium value.

We sampled five karst crags: the benign inland crags of Chiang Mai, Khao Lin Lae and Khao Yoi and two corrosive beachfront locations at Railay and Tonsai. I have less data for the granite sea crag of Ko Tao but will comment on it later in this post.

The results are presented in the diagram below. As anticipated, the analysis of the electrolytes present on the walls of these crags confirms the dominance of the alkaline CaHCO₃⁺/CO₂ system.

All of the limestone crags are dominated by calcium bicarbonate, and it’s only those at the beachside where things look different. Here we see the addition of sodium, magnesium and chloride as would be expected given the close proximity of sea water. None of this is anomalous, and it fails to provide any evidence that “tropical karst” is in any way special.

However, there is one anomaly that jumps out from the data, and that is the fact that all highly corrosive sites are associated with the high levels of sulphate, maybe as much as x100 higher. I will go on to argue that these are the true colors of our corrosion demon; it is neither tropical, nor karst, but enabled by available sulphate.

But… but! I hear you cry, you can’t just ignore the sea water electrolytes showing in the Railay and Tonsai samples. These are the corrosive locations, right? Surely there is a correlation right there! Well maybe, maybe not – we’re starting to drift into the province of another demon here. Let’s look at it next.

The SCC demon

The next cab off the rank of climbing-anchor bogeymen is one called stress corrosion cracking or SCC. It is now many years since it began to feature in climbing horror stories and, in so doing, strike fear into the hearts of the uninitiated. To this day, it still holds sway over the popular narrative.

Unlike the “tropical karst” demon, this one is not pure fantasy. It is a phenomenon well established in the scientific literature. However, such formal positioning does mean that there can be nothing vague about how it manifests. Although SCC may take many forms in many different materials, we are referencing a very specific category of SCC, namely that of chloride-mediated stress corrosion cracking that occurs in austenitic stainless steel.

So, let’s put away the magic wand and start applying the exacting notions of the world of physical chemistry. Firstly, we need to ask two fundamental questions to position the context.

• Is failure mediated via stress crack propagation in stainless steel? Yes… yes, it is. The anchor material is cold-worked stainless steel and thus contains a residual stress field. Crack propagation is observed concordant with this field.
• Is failure associated with the presence of high chloride concentrations? Yes, this also is true. In the case of highly corrosive crags, the ocean is always right next door.

The fact that the observed failures involve stress cracking in the presence of high levels of chloride means we cannot ignore SCC as a possible explanation. Indeed, it is a distinct possibility.

However, suspicion is one thing, proof is quite another. I am about to buck the prevailing consensus and argue that such suspicions are misfounded within the context of the crag environment. Let’s see if I can cut a path through the complexity.

It is trivial to reason that there is real risk of SCC failure simply because the causative chemistry, although not well understood, must necessarily exist at any temperature above absolute zero. While such is true, the engineer in me resiles from such fanciful extrapolation to far horizons. It is surely the stuff of bogey men, and we can do better. Let’s work out what we can say about the real-world risk to climbers of an SCC event becoming manifest.

If the scientific literature points to just one thing, it is the fact that there is a thermodynamic barrier to the SCC process. The question of whether this barrier presents just at the initiation of a crack, or is a constant feature of crack propagation, need not concern us. The fact remains that a substantial barrier exists, and elevated temperatures and chloride activity levels are necessary to overcome it. Indeed, the history of chloride-mediated SCC marks it out as a creature of high temperature processes and, for this reason alone, many metallurgists are reluctant to believe stories of catastrophic failure of austenitic stainless steels at ambient temperature.

However, as the old saying goes, the absence of evidence is not evidence of absence. When researchers began to look more closely, it became clear that the thermodynamic barrier could be overcome at ambient temperatures, assuming sufficient chloride was in place.

The diagram below, taken from , bundles together a number of the factors we need to consider if any comparison between observations is to be meaningful. At the very least we must consider material, temperature and chloride activity.

Straight-up we can see that nickel content matters. The rule is quite simple: if there is sufficient nickel to stabilise an austenite phase, then SCC is a possibility.

Given that the vast majority of climbing anchors have been, and still are, fabricated from austenitic stainless steel containing nickel in the range of 8% to 12% – shown in red on the graph above – it is clear that the choice of material could not be much worse.

But is it really such a bad choice? How often do crags run with 45% MgCl₂ at 155ºC? Surely this caution is irrelevant? The answer is both a yes and a no. If we insist on considering such extremities then yes, but in reality no, because chemistry is chemistry, and whilst we might suspect the event unlikely at ambient temperatures, we can draw no firm conclusion without proper consideration of the role of both chloride activity and temperature. We need to look further.

It turns out that the research has been done to allow us to get close to the answer. If we accept that there is a critical chloride concentration, c_crit , which must be exceeded for SCC attack to proceed, then this parameter can be shown to depend upon temperature in the following way.

c_{crit} = k_0 \mathrm{e}^{k_1 \above{1pt} T} \phantom{content}eqn 1

where the constants k₀ and k₁ are characteristic of the material type.

In the diagram below, taken from Prosek et al (2014), we can see eqn.1 in action. As the temperature decreases (i.e. 1/T increases) the chloride concentration required to initiate SCC becomes larger. For 304, k₀ is estimated as 0.1 and k₁ as 1333. Note that 316 is a bit more resistant to SCC, but not hugely so.

The scaling of the graph is hardly intuitive, so I have marked in the situation for 25ºC.

Now we can say something definitive –

Failure Proposition 1:

For a 304 bolt on a sea cliff at typical ambient temperatures, the chloride concentration has to exceed 8.7M before we need to entertain the possibility of SCC mediated failure.

Given the fact that saturated sodium chloride at 25ºC is 5.2M you’d think that there is no way that sea water, which is more dilute, could ever reach the critical chloride concentration. Even if a splashed droplet evaporated on the bolt surface, it can’t do better than saturated, right?

Wrong! Or at least way too simplistic. Sea water contains more than just sodium and chloride, and if it is evaporated by placing it into a dry enough environment, it deposits a sequence of mineral species according to the thermodynamics of the crystallization process. Thus, during the process of evaporation, individual ionic concentrations can vary over a substantial range as the different species drop out of solution.

Eugster et al (1980) provide the following list of minerals that are deposited during the evaporation process.

They enumerate the deposition sequence as follows –

As sea water is evaporated, initially all ionic species increase in concentration until some begin to be incorporated into the expanding mineral crust. The authors provide the following diagram of the process. The x-axis scaling is in units only a physical chemist could love, but don’t panic, it is simply the amount of water left in solution, running from pure sea water on the left to dryness on the right plotted on a log scale.

I’ll admit that it came as a surprise to me to see that the ultimate effect of evaporation is to produce a solution of almost pure magnesium chloride rather than sodium chloride. Such a phenomenon begs the question of whether a saturated magnesium chloride solution could approach the chloride c_crit for 304 at that temperature. Doing some calculations shows a value of 10.0M is likely, and therefore we could well be in trouble.

In the light of the foregoing data, we need to modify our failure proposition as follows –

Failure Proposition 2:

For a 304 bolt installed in a sea cliff at typical ambient temperatures, evaporation of sea water on the surface of the metal has the potential to exceed the critical chloride concentration of 8.7M. Thus the probability of failure via SCC can be interpreted as the probability that sufficient evaporation of sea spray occurs.

Let’s look at the final phase of evaporation a bit more closely. We can identify a point, labelled X in the diagram below, where the solution begins to saturate with MgCl₂ (Kieserite) and the chloride concentration reaches its peak value. Converting from the authors’ y-axis scale I estimate this as being 10.2M which is close to my estimate for a saturated solution of MgCl₂ . The saturated solution phase immediately prior to this transition I estimate as 8.4M. Thus the significance of the evaporation point I’ve labelled X is that it represents the maximum evaporation that can occur before c_crit is exceeded.

The progress of any evaporation process is dependent upon the surrounding air being drier than the solution itself. Yes, the solution can be “dry” from a physical chemistry point of view because as it concentrates it contains less and less water and more and more of the dissolved salt. The evaporation will proceed only until such point as the decreasing water activity (a measure of “dryness”) is matched by the relative humidity (RH) of the surrounding atmosphere.

Eugster et al (1980) provide the water activity data for the evaporative process we are considering. Thus we can fit our evaporative progress point, X, we identified above, and deduce that it corresponds to a relative humidity of 45%.

Armed with this information we now arrive at the following intriguing failure proposition –

Failure Proposition 3:

For a 304 bolt installed in a sea cliff at typical ambient temperatures, the probability of failure via SCC can be reduced to the probability that the ambient humidity is less than 45% for extended periods of time.

Now here comes the problem.

Given that considerations of physical chemistry set the equilibrium RH of the atmosphere in contact with the sea to approximately 80%, and given that many crags at Railay/Tonsai are right on the hightide mark, it is difficult to see how the conditions necessary to promote SCC could ever occur. There is a clash of physical circumstance operating here; on the one hand we need to be physically close to the sea to get exposure to sea spray droplets, but on the other we need to be far enough away to permit a RH below 45%.

I believe this paradox exposes a clear gap in our understanding. We need to admit that the conditions necessary to induce cracking in the laboratory don’t seem to exist in the field. The problem is that, when all you have is the hammer of SCC, everything looks like a nail. Thus there is need to invent fanciful environmental conditions that permit both sea spray and 45% RH to coexist.

For example, we might propose that air close to the sea surface at 80% RH could be drawn across sunbaked rock to increase its temperature, and thereby decrease RH to the critical 45%. There is nothing wrong with the physics of this proposition, but without direct evidence drawn from instrumentation it has to be treated as pure speculation. Surely, this is little better than hand-waving. Furthermore, I can point to locations where such an explanation cannot be supported by the terrain, the one pictured below for instance.

Being situated in a small cave just above the sea, this wall never sees the sun, and has zero exposure to airflow from off sun-heated rock. It is inconceivable that the RH could ever approach the 45% necessary to promote SCC, yet we observe stress cracking of 304.

The simplest conclusion is that there has to be a stress-cracking mechanism other than SCC in play, and, if such is so for this location, surely it could be so across all Railay/Tonsai crags.

Enough of demons – it’s time to reveal the culprit:

Over time, it became clear that every aggressively corrosive sea cliff around the world could be shown to be associated with high sulphate levels. If there was no sulphate, there was no exceptional corrosion, despite the abundant chloride. This fact raised the possibility that sulphur bacteria might be involved.

And indeed, the culprit turned out to be a miniscule living thing. The picture below, although not from a bolt on a Railay crag, is representative of the sulphate reducing bacterium (SRB) that is causing our grief.

This little guy extracts electrons from the metal surface it sits on via the electrical connections you see in the micrograph above, and in so doing, facilitates the reduction of environmental sulphate to sulphide. In that sense, it ‘burns’ iron using sulphate as an oxidant, analogous to how we climbers burn glucose using atmospheric oxygen. We breathe out carbon dioxide and it exhales hydrogen sulphide.

Thus for every corrosive crag we find an excess of sulphate within the immediate environment, and critically, we find the product(s) of metabolism, iron sulphide(s), at the point of corrosion.

Although some metallic iron is dissolved, the real mischief of this organism is that it creates an acidic environment rich in hydrogen sulphide against the metal surface. The sulphide acts to poison the catalytic recombination of atomic hydrogen to gaseous hydrogen at the cathode surface, leading to build-up and eventual migration of atomic hydrogen into the bulk metal. For cold worked 304 this process is rapid, and hydrogen embrittlement and ultimately stress-cracking can occur in a few years.

Thus we can now point to the involvement of an alternative mode of stress-cracking known as sulphide stress cracking (SSC). Most importantly, we have dispensed with the need to conjure up mechanisms by which the extreme chloride concentrations required for SCC might arise.

Coming Next:

I will take a closer look at the SSC failures we are seeing, and compare them with the phenomenon as it manifests on the Portuguese coast.