Thanks to Luis Fernandes Silva for providing the sample.
Which is it, sulphide stress cracking (SSC) or chloride-induced stress corrosion cracking (SCC)?
I raised this question in one of my earlier posts where I described the aggressive corrosion of stainless steel at Cabo da Roca. In that particular post I highlighted the main differences between the two mechanisms. In my opinion, the evidence is steadily stacking-up in favour of the SSC as the dominant mechanism prevailing at those corrosive sea cliffs where exceptional levels of sulphate are present.
In this post I’ll report on an example of stress cracking I found while working through some brittle failure samples from Praia da Ursa. It is a bit of a side issue for me right now, but I thought it worth posting given it illustrates what I see to be the main indicators of a sulphide stress failure.
For SCC, the mechanism is understood as the action of a locally high chloride concentration initiating pitting corrosion at the surface of the metal. If the metal habours pre-existing stresses, then a pit could possibly present a stress-riser of sufficient magnitude to initiate a crack. A whole bunch of stuff such as ambient temperature and metal composition, dictate the speed of propagation of a crack once initiated. It might propagate dangerously fast, or it might prove harmless. This is the basis of the salt-crust theory of anchor failure at sea cliffs .
On the other hand, SSC is thought most likely to be a hydrogen embrittlement phenomenon courtesy of cathodic charging of the metal in the presence of sulphide. Once again, there has to be a pre-existing stress field within the metal, but, unlike SCC, there is no requirement for a corrosion pit to serve as an initiator. Thus cracks can be initiated within the bulk metal itself. There are a number of theories concerning initiation mechanisms . Unlike SCC, the propagating crack need not be connected to the exterior surface.
Praia da Ursa is a granite/syenite sea cliff, so there is plenty of sea salt to lend credence to the SSC theory. The only catch is that in this particular case you need to explain why the cracking occurred at a point within the nut/bolt contact zone. Is it feasible that the high chloride concentration necessary to initiate the crack could be provided by evaporation at a point not exposed to the atmosphere? Surely, if such a mechanism were to occur, it would be more likely that we’d see it at points on the outside surface.
And, if this cliff has chloride in abundance, the same can be said for the calcium sulphate it oozes from cracks and fissures. There is no shortage of food for sulphate reducing bacteria (SRB), a major driver of SSC, and an organism quite capable of colonizing the anoxic space between nut and bolt.
Thus we have the prerequisites for both SSC and SCC, but which is it?
Bolts at this crag, and elsewhere at Cabo da Roca, have an unnerving tendency to embrittle over a relatively short number of years, and can be removed quite easily with a few hammer blows. One such bolt is illustrated below. I will be addressing the phenomenon of brittle fracture in a later post. For now I’d like to show a few pictures of stress cracking I located during this investigation.
The bolt in question is a 10mm, 304 expansion bolt that had been installed for about 15 years before being removed from the rock. I’d judge that the nut just enclosed the point where the stress cracks were observed.
It has been a consistent observation that SRB are somehow capable of invading the solid metal. Whilst there are sound reasons to suspect that hydrogen will penetrate the metal over a 15 year time span (see this post), it is hard to explain the occurrence of sulphide deep inside the material of the bolt except by assuming that the bacteria are capable of opening channels of a size sufficient to facilitate their entry. This is a subject for another day. For now, it is enough to observe the presence of sulphide, and of cavities of a size large enough to accommodate the 2um to 3um dimensions of the bacteria.
When tested by means of the iodine-azide spot test, the fractured surface of the bolt reveals significant quantities of sulphide even at the very centre.
And more surprisingly, at least to me, is the fact that a longitudinal section of the bolt reveals that sulphide is distributed right to the exterior end.
If we test a sample of 304 that has been installed for many years on a non-corrosive cliff, there is no observable response in the iodine-azide test, and nor would we expect one given the upper limit for sulphur in 304 steel is 0.03%. The skeptic might argue that we are doing little more than observing a rogue piece of high-sulphur steel. It is not possible to disprove this assertion because we don’t have a pre-installation test to guide us, but it would seem freakish if all the poor quality steel in the world has ended up on the high sulphate crags, but not elsewhere.
The picture below is taken from the polished, but unetched surface. The feature at the bottom of the picture is the thread root. The threads on this bolt are rolled, not cut, and thus we can anticipate substantial residual stresses to be present in the vicinity of the thread root, that is, in addition to the axial stress associated with tightening of the bolt.
It makes sense that, if there is to be stress cracking then we will see branched cracks radiating from the thread root as the point of highest stress in the overall field. To my mind, this observation is an indicator of the nature of the stress field, and not of the mechanism driving the cracking. A radiating set of cracks could be indicative of either SCC or SSC.
However, the real giveaway that we are dealing with SSC and not SCC is the observation that bulk metal removal is occurring not on the exterior surface at the thread root, but well within the bolt itself. The features labelled A and B cannot be explained by the normal chloride induced pitting mechanism for stainless steel. It is far more likely these features are associated with invasive colonisation of the bulk metal by SRB. The deeply creviced margins we see for the corrosion feature labelled B are typical for these internal structures. Note how the exterior surface of the thread root is relatively free from pits. For now, I’m guessing that a switch over in metabolism occurs as the bacteria moves into progressively more anoxic environments.
If we take a closer look at the corrosion structure labelled C, we see stress cracks radiating from its inner surfaces. Either this tiny structure is acting as a stress riser and initiating those cracks, or cathodic charging is elevating the hydrogen concentration within the local area, or perhaps both of these things. Whatever the truth of the matter, we can be fairly sure this is not chloride induced SCC.
I ground the sample back a small amount, re-polished it and applied 1/10th strength Beraha’s reagent as a contrast agent. I’m not at all sure of the chemistry we have happening here. It differs to that I’ve seen published for austenitic stainless steel, and I’m guessing we are observing something indicative of profound underlying changes wrought by the bacteria.
The idea that stress cracks are propagating as clean breaks through otherwise homogeneous metal does not stand up at this level of examination. The lack of homogeneity is striking, and whilst classic fracture mechanics might be operating at the micro level, I really can’t see how estimates of macroscopic propagation rate could be made for such an uncertain matrix.
If we take an even closer look, cracks can be seen to involve regions of metal removal. Notice that at this level of magnification, many of these “voids” are too small to be occupied by bacteria.
As nothing more than a side-note at this stage it is worth remembering that Desulfovibrio vulgaris is known to form direct electrical connections with the metal by means of conductive pili. Thus it could extend its corrosive influence (electron extraction) into spaces much smaller than its primary dimensions.
One final experiment I did was to repeat the grind-back and re-polish, but this time I electrolytically etched the surface in 50% nitric acid to reveal the grain boundaries.
This preparation seems to confirm the observation that cracks are not clean fractures through homogeneous metal but involve significant levels of corrosive activity. It maybe that propagating cracks are followed by bacterial colonisation and subsequent corrosion.
At higher magnification it seems the fracture follows a mainly inter-granular path with the occasional trans-granular fracture.
- Fairly Certain: The damage to stainless steel bolts that we are seeing is the result of the activities of sulphate reducing bacteria (SRB).
- Fairly Certain: The related stress cracking processes we observe are mechanistically sulphide stress cracking (SSC) rather than chloride-induced stress corrosion cracking (SCC).
- Speculative: The damage mechanism proceeds as follows –
- cathodic charging facilitated by sulphide leads to episodes of hydrogen diffusion into the metal.
- during such episodes stress cracks are initiated and propagated to some minor extent.
- bacteria move into the cracks favouring the anoxic conditions they provide, and seeking direct electrical connection to fresh metal surfaces to drive their metabolism.
- metal ions are released from the surface in proportion to the electrons withdrawn, thus leading to corrosion cavities forming within the initial stress cracks, and the establishment of bacterial colonies within the metal itself.
- repeat from a. above …. for as long as substantial levels of exogenous sulphate are available then this process will run. It will start and stop as sulphate levels fluctuate.