Effect of pearlite on stress corrosion cracking of carbon steel in fuel-grade ethanol

2.1 External surfaces

Selective dissolution of ferrite phase from the pearlite was observed on the external surfaces of the SSRT and NCTLT specimens (Torkkeli et al., 2013). As ferrite dissolves selectively, cementite lamellae are left on the steel surface. This can be clearly seen in Figure 2A showing the bands of cementite lamellae (light phase) left on the SA-106 steel surface after SSRT testing in FGE for 2 days. Higher magnification of the cementite lamellae structure left on the St35 steel surface from the selective dissolution of a single pearlite nodule during SSRT testing in FGE for 2 days is shown in Figure 2B.

Figure 2:

(A) Bands of cementite lamellae left on the SA-106 steel surface due to selective dissolution of ferrite phase from the pearlite during SSRT testing in FGE. Note also that transgranular SCC initiation sites are related to the pearlite bands. (B) High magnification of the cementite lamellae left on steel surface from selective dissolution of ferrite from a single pearlite nodule of St35 steel during SSRT testing in FGE.

It was also shown in the previous study (Torkkeli et al., 2014) that in FGE with chloride content below the analysis limit, SCC initiates as transgranular cracking in the pearlite and intergranular cracking in the ferrite phase. This can be seen from Figure 3A showing transgranular SCC cracking initiated at the pearlite of SA-106 steel surface during SSRT in 20% ethanol-gasoline blend and Figure 3B showing intergranular SCC cracking initiated at the ferrite phase of the same steel.

Figure 3:

(A) Transgranular SCC cracking of SA-106 steel initiated at pearlite during SSRT in 20% FGE and gasoline blend. (B) Intergranular SCC cracking of SA-106 steel initiated at the ferrite phase in the same test.

The cementite structures have also been observed on the surfaces of the failed pipeline samples taken from FGE service. These samples always have a thick oxide layer on the surface, which was studied with scanning electron microscope (SEM) by making cross-section specimens. There were cementite structures inside the oxide layer, as shown in Figure 4.

Figure 4:

(A) Cross-section of oxide layer on the pipeline steel surface from FGE service. (B) Some cementite left on the steel surface and empty sites inside the oxide layer are most likely due to the cementite based on their dimensions.

2.2 Fracture surfaces

The first findings from the fracture surfaces were made from a failed process pipeline due to SCC in FGE service (Hirsi et al., 2011). Lamellar cementite structures were observed on the fracture surface of the failed pipeline steel sample, as shown in Figure 5. In this case, the cementite structure was covered by a thick oxide layer.

Figure 5:

Lamellar cementite structure on SCC fracture surface of failed St35 pipeline steel sample from FGE service.

After the first pipeline failures, SSRT and NCTLT testing was done, and selective dissolution of ferrite phase from the pearlite was observed on the fracture surfaces of the tested SSRT and NCTLT specimens, as shown in Figure 6 (Torkkeli et al., 2013). It should be noted that the chloride concentration was higher during the NCTLT testing than what is expected in the field samples due to the chloride leakage from the reference electrode.

Figure 6:

Selective dissolution at the pearlite phase on SCC fracture surface of SA-106 steel NCTLT specimen in FGE with some chlorides.

The study using SSRT testing was made with various ethanol-gasoline blends, and selective dissolution was observed on the fracture surfaces of the SSRT specimens (Torkkeli et al., 2014). In this case, most of the tests were made without reference electrode, and differences were observed between chloride containing conditions and conditions with negligible chloride concentrations below analysis limit. Selective dissolution of the pearlite on SCC fracture surface in SSRT testing in 85% ethanol-gasoline blend with chloride concentrations below the analysis limit is shown in Figure 7A and in 85% ethanol-gasoline blend with chlorides, in Figure 7B. It is obvious that selective dissolution is much more aggressive in conditions with chlorides.

Figure 7:

Selective dissolution of the pearlite on SCC fracture surface in SSRT testing in (A) 85% ethanol-gasoline blend with chloride concentration below analysis limit and (B) 85% ethanol-gasoline blend with some chlorides.

It was also observed that in ethanol-gasoline blends with chloride concentrations below the analysis limit, the SCC fracture mode changes from fully intergranular cracking to transgranular cracking at the pearlite. This effect is clearly seen with the SA-106 steel specimen with banded pearlite microstructure in Figure 8, on the SSRT SCC fracture surface formed in 25% ethanol-gasoline blend with chloride concentrations below the analysis limit. Fully intergranular SCC cracking initiated at the ferrite phase changes to transgranular SCC cracking in the first pearlite band perpendicular to the direction of crack propagation.

Figure 8:

Fracture surface of SCC crack of SA-106 steel SSRT specimen tested in 25% ethanol-gasoline blend with chloride concentration below the analysis limit. Fully intergranular SCC cracking changes fracture mode to transgranular cracking inside the pearlite band perpendicular to the direction of crack propagation.

Another failed field pipeline sample was characterized when dissolution of MnS inclusions was previously studied (Torkkeli et al., 2015). Selective dissolution of pearlite was observed inside the SCC cracks. It was observed that the propagation of the intergranular SCC cracks stopped when the pearlite nodules of the carbon steel were reached, as shown in the cross-section SEM image in Figure 9A. Then the pearlite nodules were dissolved, selectively leaving empty spaces inside the crack, as seen from Figure 9B. Partially dissolved pearlite nodule on the fracture surface of the same sample with cementite lamellae visible on the fracture surface is shown in Figure 10.

Figure 9:

Intergranular SCC crack in SA-106 carbon steel pipe sample from FGE service. (A) The intergranular crack propagation stops at the pearlite nodule. (B) Partially dissolved pearlite nodules.

Figure 10:

Partially dissolved pearlite nodule on SCC fracture surface of SA-106 carbon steel pipe sample from FGE service.

Finally, a newly failed SA-106 pipeline sample from the FGE service was characterized in this study. This pipeline was taken into FGE service in 2013, and it failed in 2015. It was observed that the propagation of the intergranular SCC stopped at the pearlite nodules and selective dissolution of the pearlite nodule started leaving empty spaces filled with oxide inside the intergranular SCC crack. The fact that the crack propagation stopped at the pearlite nodules is shown in Figure 11A, and a partially dissolved pearlite nodule, in Figure 11B.

Figure 11:

Intergranular SCC cracking in a carbon steel pipe sample from FGE service. (A) The intergranular SCC crack propagation stops at the pearlite nodules. (B) A partially dissolved pearlite nodule inside the crack.

2.3 Discussion of the findings

Conclusions can be drawn on how cementite phase affects the SCC cracking mechanism of carbon steel in FGE service by considering the observations together. The most important observations are below:

• The selective dissolution of ferrite phase from the pearlite occurs in FGE even with negligible chloride concentrations, but the conditions become more aggressive with higher chloride concentrations leading to higher selective dissolution rates.

• The selective dissolution occurs at the external surfaces as well as on the fracture surfaces.

• The SCC cracking of carbon steel in FGE initiates as transgranular cracking at the pearlite and as intergranular cracking in the ferrite phase during the SSRT testing in FGE with negligible chloride concentrations.

• The SCC cracking of carbon steel in FGE initiates at the ferrite phase as intergranular cracking changes to transgranular cracking in the pearlite during SSRT testing in FGE with negligible chloride concentrations.

• The propagation of intergranular SCC cracking of carbon steel in FGE stops at the pearlite nodules, and the pearlite is selectively dissolved, leaving empty spaces inside the SCC cracks. The SCC cracking mechanism does not change from intergranular cracking to transgranular cracking in the failed FGE pipeline samples.

As the propagation of the intergranular SCC cracks stops at the pearlite nodules and changes to localized corrosion at the pearlite nodule, it indicates that there is a microgalvanic coupling between the cementite and ferrite lamellae of the pearlite nodule. Due to the microgalvanic coupling, the cementite seems to become the cathode, while ferrite dissolves as the anode, leading to localized corrosion of ferrite phase of the pearlite. As a result, crevices are generated between the cementite lamellae left on the steel surface. The most important observation is that, as the cementite becomes the cathode, the intergranular SCC cracking mechanism stops.

It is known that chlorides can cause acidification of the localized anodic corrosion sites during pitting or, for example, inside the SCC cracks of austenitic stainless steels in chloride containing conditions. The influence of chlorides on corrosion of martensitic stainless steel in ethanol-containing gasoline was studied by Abel and Virtanen (2015). It was observed that pitting corrosion occurs with extremely low chloride concentrations, which did not lead to passivity breakdown in a purely aqueous solution. In this case, it can be considered that the result of localized corrosion at the pearlite due to microgalvanic coupling is that chlorides get concentrated at these locations. This leads to acidification and more severe localized corrosion, as were observed in the investigated samples. The FGE specifications limit the chloride concentrations so that, usually, the chloride concentration of the bulk FGE is below 10 wt.-ppm. In ethanol-gasoline blends with 85% ethanol, only 2 wt.-ppm of chlorides is enough to cause localized corrosion at the pearlite (Torkkeli et al., 2014). In similar, ethanol-gasoline blends with negligible chloride concentration the corrosion appeared to be more general, and similar localized corrosion was not observed. These observations show that, with carbon steels, the chlorides concentrate at the pearlite in a similar way as they concentrate inside the pits in pitting corrosion of stainless steels. Although our observations of how chlorides affect the corrosion of carbon steel in FGE are related to localized corrosion at the pearlite nodules or inside the transgranular SCC crack, it is important to note that chlorides can also increase the dissolution rate of iron oxides or lead to oxide film breakdown even in neutral solutions.

The propagation of the intergranular SCC cracks stops at the pearlite nodules even with negligible chloride concentrations below the analysis limit. Chlorides increase this localized corrosion effect at the pearlite phase significantly. This is most likely why only 2 wt.-ppm of chlorides in the FGE can prevent the intergranular SCC from occurring (Torkkeli et al., 2014). Transgranular SCC can initiate at the pearlite even with negligible chloride concentrations during SSRT testing. Also, the intergranular SCC changes its fracture mode to transgranular cracking at the pearlite during SSRT testing. With the field samples, the intergranular SCC does not change the fracture mode at the pearlite, but localized corrosion occurs instead. The difference in these two conditions is the stress/strain level, which is much higher during the SSRT testing. The most important observation here is that even though the intergranular SCC mechanism stops at the pearlite with lower stress levels based on the field samples, it can continue as transgranular SCC with higher stress/strain levels of the SSRT specimens.

These observations give some information of the differences between the intergranular and transgranular SCC. As the propagation of the intergranular SCC cracks stops at the pearlite nodules and changes to localized corrosion at the pearlite nodule, it indicates that there is a microgalvanic coupling between the cementite and ferrite lamellae of the pearlite nodule. Due to microgalvanic coupling, the cementite becomes the cathode, which affects the potential at the intergranular crack tip enough to stop the crack propagation. This also explains why the intergranular SCC cannot initiate in conditions containing chlorides. This is because the chlorides catalyze the effect of the microgalvanic coupling by increasing the localized corrosion and conductivity of the solution, and the cementite lamellae on the steel surface become stronger cathodes. As the conductivity of the solution increases, the galvanic coupling between the cementite and ferrite affects a wider area around the pearlite bands, preventing the initiation and propagation of the intergranular SCC.

In most cases, the failures in the field service have been due to intergranular SCC. All the failed field samples studied by us have shown only intergranular SCC. Small amounts of chlorides are able to prevent the intergranular SCC while promoting transgranular SCC, which indicates that the transgranular SCC mechanism is more likely in more conductive conditions, especially containing chlorides. With negligible amounts of chlorides, the intergranular SCC stops at the pearlite but is able to continue as transgranular SCC at a high stress level. The transgranular SCC mechanism does not initiate at the stress levels of 80% of yield strength of the steel, while for the intergranular SCC mechanism, it is possible. This indicates that the transgranular SCC mechanism requires higher stress levels than does the intergranular SCC mechanism. In field conditions, the pipelines are usually designed with high safety factors, and due to this reason, stress levels in the field service are mainly weld residual stresses and can be relatively low, especially after PWHT. In laboratory SSRT testing, the test specimen is strained until fracture, so the stress levels are much higher than what are expected in the field service, which promotes the transgranular SCC mechanism. In commercial FGE, the chloride levels are usually low, as the chloride concentration is limited by the specifications. In the laboratory testing, in addition to leaking reference electrodes, simulated FGEs are often used, with higher concentrations of chlorides and other impurities to simulate the most impure FGE. Therefore, in the laboratory studies, the conductivity of the simulated FGE can be higher than what is expected in the field service, which promotes the transgranular SCC mechanism. It is therefore important to separate these two mechanisms as they seem to behave differently.

It is unclear whether something forming at the pearlite, for example, in the cathodic reaction, can change the SCC mechanism from intergranular to transgranular cracking. An attempt to measure the compounds formed at the cementite during acidic corrosion of steel in aqueous conditions was made by Staicopolus (1963) by using mass spectrometric techniques, and the results were as follows: H₂, 87.8%; CH₄, 4.5%; CO, 3.2%; C₃H₈ and C₄H₈ (each), <1.0%; C₂H₂, C₂H₆, C₄H₁₀, and H₂O (each) <0.5%; and traces of CO₂. This shows that quite high amounts of carbon monoxide can form at the cementite. The corresponding cathodic reactions were considered to be as follows:

2Fe 3 C + H 2 O → 6Fe + -CH 2 - + CO Fe 3 C + -CH 2 - + H 2 O → 3Fe + CH 4 + CO

Hypothesis made by Newman (2008) related the transgranular SCC mechanism to CO-CO₂ SCC and the intergranular SCC to a mechanism caused by the incomplete electrochemical oxidation products of ethanol. It is well known that CO-CO₂ can cause transgranular SCC of carbon steel (Kirkham, 1996). Depending on the CO-CO₂ concentrations and pHe conditions, the SCC mechanism could also be a form of carbonate cracking. The similarities between the mechanisms observed here and the SCC mechanisms in carbonate solutions are clear. The SCC mechanisms in carbonate solutions can be divided into high pH intergranular SCC and near-neutral transgranular SCC (Sutcliffe et al., 1972; Parkins et al., 1994; Fang et al., 2003; Lu et al., 2010). Selective dissolution of pearlite nodules has been observed in carbonate solutions at specific potential range, which is also the potential range where the high pH intergranular SCC occurs (Sutcliffe et al., 1972). In these conditions, a more dense oxide layer containing magnetite and iron carbonate forms. It could also be the potential range where iron carbonate and magnetite have stability boundary, which is where intergranular SCC has been observed in DI-isopropanolamine (DIPA) solutions used in refinery amine systems containing sulfur (Aaltonen et al., 1988, 1989). The near-neutral transgranular SCC prefers to initiate at the pearlite bands (Fang et al., 2003). It was also found that the steels with ferrite-pearlite structures are more susceptible for both SCC mechanisms (Fang et al., 2003). The near-neutral transgranular SCC occurs often in dilute bicarbonate-carbonate groundwater containing small quantities of chloride in coating disbonding of pipelines (Liu & Mao, 1995). All the mechanistic findings related to the SCC mechanisms in carbonate solutions are very similar if not identical to our observations on the SCC mechanisms in FGE solutions. This supports the hypothesis made by Newman that the carbon steel SCC in FGE can be a form of CO-CO₂ or carbonate cracking (Newman, 2008).

3.1 Electrochemical measurements

Electrochemical measurements were made with rectangular (1.65 cm²) specimens of pure iron (99.99% pure iron, Goodfellow, Huntingdon, UK) and cylindrical specimens (2.00 cm²) of pure cementite (Cylindrical rod, CAS: 12011-67-5, American Elements, Los Angeles, CA, USA) to get a better understanding of the electrochemical differences between these two phases. All specimens were dry-polished to 2000 grit, degreased with acetone, and dried in blowing air just before every measurement. The cementite specimen was porous, containing pores with an average diameter of 9.1 μm and pore density of 8100 pores/cm². This was taken into account for current density calculations for cementite by assuming spherical pores (0.01-cm² pore surface area). Due to the pores, the current density values for cementite show a small random error. It should also be noted that the bottoms of these pores are not dry-polished. The electrochemical measurements were made in FGE with some added NaCl (CAS: 7647-14-5, Sigma-Aldrich, Germany) and with various hydrochloric acid (HCl) (Product number 258148, CAS: 7647-01-0, Sigma-Aldrich, Germany) concentrations. The chemical composition of the FGE used is given in Table 1.

All the potentiodynamic polarization curves were measured using potentiostat (PCI4, Gamry Instruments, Warminster, PA, USA) in a sealed three-electrode electrochemical testing cell. The reference electrode used was a silver-silver chloride-based reference electrode Ag/AgCl/1 m LiCl, EtOH. The purging gas used for unaerated measurements was dry nitrogen and for aerated measurements, dry air. The electrochemical measurements in pure FGE were not possible due to very low conductivity, which is why some chlorides were added to the FGE for the measurement. Due to the very low conductivity, IR compensation was needed for the measurements. For IR compensation, positive feedback and current interrupt methods were both evaluated. Positive feedback method could have been used with lower conductivity than that of current interrupt method, but it was not taken into account that the conductivity of the solution may change quite significantly during the long measurement. Due to this reason, the current interrupt method by Gamry Instruments was used for the IR compensation. During testing, it was noticed that at least 50 wt.-ppm of chlorides was required for the current interrupt method to work reliably down to the target potential of −1.2 V. Cathodic potentiodynamic polarization measurements were started quickly as soon as the potential reached steady slope after the specimens were placed to the electrochemical testing cell to minimize the oxidation of the surface before the measurement. Open circuit potential (OCP) was measured versus time and before starting the cathodic potentiodynamic polarization measurement. All tests were performed at ambient temperature.