2015 Volume 55 Issue 11 Pages 2435-2442
Using permeation tests, determination of hydrogen trapping and microhardness measurements, the effect of the microstructure on the hydrogen diffusivity was analyzed in the welding zone of two high-strength experimental microalloyed steels called M1 steel, with a martensite and bainite microstructure and M2 steel with a quasi-polygonal ferrite (QPF), acicular ferrite (AF) and martensite-austenite (M/A) microstructure. Determination of the diffusivity was performed using electrochemical permeation tests, and hydrogen traps were determined with the silver decoration technique.
One-pass welds without filler material were simulated with the Gas Tungsteng Arc Welding (GTAW) process, and samples of the welding zone were collected: base material (BM), heat affected zone (HAZ) and fusion zone (FZ). A Devanathan and Stachurski type electrochemical cell was constructed to conduct permeation tests. From the microstructural analysis, the permeation testing and silver decoration, it was observed that the hydrogen diffusivity decreases with the increase in traps, promotion of the formation of the M/A microconstituent and AF and the reduction of martensite and bainite microconstituents. The lower diffusivity of all zones of both steels is presented by the BM of M2 steel, which is associated with a QPF, AF and M/A microstructure and low microhardness. The highest relative amount of traps are in the coarse grained heat affected zone (CGHAZ) of both steels, however because these are reversible traps, these subzones could be the most susceptible to hydrogen induced cracking (HIC).
Hydrogen has a detrimental effect on the mechanical properties and mechanical behavior of steels.1,2,3,4) It is known and accepted that increasing the strength of steels also increases the susceptibility to hydrogen induced cracking (HIC).5,6,7) In particular, it is known that the microstructure and hardness of the heat affected zone (HAZ) of steel welds have a great influence on HIC. For example, the martensite formation is related to this phenomenon,8,9,10) so to avoid or mitigate HIC, materials are selected and/or the welding process is designed to inhibit the formation of this microconstituent.
In recent years, great interest has been taken in the effect of hydrogen on the properties and mechanical behavior of microalloyed steels intended for the oil industry.11,12) This is because there may be presence of H2S during the exploitation, processing and/or transport of hydrocarbons; thus through corrosion reactions, hydrogen ions can be produced, which can adsorb to form atomic hydrogen, absorb and diffuse to a high chemical activity zone, creating HIC problems. Additionally, it is expected that in the coming decades hydrogen will be the fuel to replace hydrocarbons as a primary energy source.13) To do this, will require materials resistant to the harmful effect of hydrogen; for example, hydrogen will need to be transported from its place of production to the distribution terminals, for which the currently installed or to be installed infrastructure of pipelines will have to be used. Therefore, all knowledge generated in this regard will serve in the near future.
The problem with hydrogen is its low solubility in the lattice, its high power to be adsorbed, absorbed and diffused and its ability to localize at internal sites such as voids, precipitates, inclusions, grain boundaries and regions with high stresses;5,14,15) as well as its ability to react with certain elements to form hydrides. The driving force for hydrogen diffusivity is a difference of activities presented as a concentration gradient, stress gradient, electrical gradient and temperature gradient.7,16,17) Grain boundaries can increase hydrogen diffusion providing fast diffusion paths18) or they can reduce the mobility of hydrogen sites acting as reversible hydrogen trapping sites at nodes and connecting points.19) As a result of these traps, solubility is inversely proportional to the effective diffusion coefficient20) so that a material with a greater number of traps could be less susceptible to HIC,21,22) thus proper selection of the traps can control the hydrogen content and distribution.23) In short, hydrogen diffusion in steel is influenced by two factors: a) a difference of chemical activities and b) hydrogen traps (grain boundaries, solute atoms, voids, inclusions, precipitates, phases and dislocations).12,15,24)
Although residual stress and yield stress are the important parameters to produce HIC, for example, it has been shown that with increased yield stress, increases the accumulation of hydrogen in the area of crack tip corresponding to the zone where is located the maximum triaxial tensile stress;5,6) all mechanisms that attempt to explain the damage caused by hydrogen, agree that for this phenomenon to exist requires that the hydrogen diffuses towards a high activity zone. Thus, to study the susceptibility of a steel or a microstructure to HIC, hydrogen diffusivity should be determined. Therefore, hydrogen permeation tests were carried out in this paper in the fusion zone (FZ), HAZ and base material (BM), to determine how hydrogen diffusivity is affected based on the microstructure developed by welding thermal cycles in two high-strength experimental microalloyed steels. Additionally, the relative amount of hydrogen traps was determined with the silver decoration technique and related to the diffusivity and microstructure of each welding zone.
The nominal chemical composition of the two microalloyed steels under study (named M1 steel and M2 steel) are given in Table 1, where according to the Ito-Bessyo formula the carbon equivalent is respectively: 0.142 and 0.146. Basically the two steels have the same chemical composition but a different microstructure because they received a different thermomechanical treatment. The yield stress of M1 steel is 971 MPa and for M2 steel is 789 MPa.
The BMs are in the form of 11 mm thickness plates of which two plates were cut with dimensions of 100 mm long by 100 mm wide, retaining the original thickness. The surfaces of the plates were roughed up to 600 grit sandpaper to remove oxide formed by the thermomechanical process. Prior to welding, the plates were cleaned and degreased with alcohol to prevent contamination with organic and inorganic materials and to prevent the introduction of hydrogen due to the presence of moisture. One-pass welds were simulated in these plates with the autogenous Gas Tungsteng Arc Welding (GTAW) process. To obtain a high quality welding, the welding torch was adapted to a plasma cutting device, with which welding speed was achieved as well as distance between the electrode and the plate were constants. During the welding process, 200 A of direct current with negative electrode (DCEN) were applied. The plates welding were performed in the direction of the thermomechanical lamination process. Due to the carbon equivalent of these steels, no preheating or postweld heat treatment of the plates were performed, and they were allowed to cool in still air to room temperature.
Once the welding process is completed, a sample was collected to reveal the microstructure of the welding zone as follows: for each weld plate a sample was collected in a remote zone away from the start and end of the weld area through cross sections to the direction of the welding with an industrial hacksaw at low speed to avoid any microstructural change. These samples were ground up to 600 grid sandpaper, and polished with alumina of 1, 0.3 and 0.05 μm of particle size. Finally revealing the microstructure of the welding zone was performed with the following sequence of reagents: Nital 2, Picral 4 and LePera.
In the welding zone, microhardness measurements were made with the model HMV-2 Shimadzu Microhardness device, which were made with a force of 980.5 mN with a time of 15 s.
To determine the relative amount of hydrogen traps of the different microstructures developed in the welding zone, the silver decoration technique was used. To evaluate that the silver is actually deposited on sites where hydrogen is present, the decoration test was performed for three different samples:
Case 1: samples that were not charged with hydrogen.
Case 2: samples that were charged with hydrogen.
Case 3: samples that were charged with hydrogen and subjected to a low temperature heat treatment (baking) to remove trapped hydrogen.
Samples were collected in the perpendicular direction of the weld (Fig. 1), which were roughened up to 600 grit sandpaper and then polished with alumina of 1, 0.3 and 0.05 μm of particle size.
Scheme of sampling zones for permeation and silver decoration tests.
An electrochemical cell was available for the hydrogen charging, where the samples were electrolytically charged in a 0.5 M H2SO4 solution with 0.2 g of As2O3 per liter of solution. The As2O3 was used to prevent the recombination of adsorbed hydrogen to form molecular hydrogen. This charging was performed for 180 minutes by applying a current of 40 mA/cm2.
For the silver decoration, two solutions were prepared, one of AgNO3 0.86 M and another of KCN 1.72 M. According to Shomer and Dieker25) and Yao and Cahoon,26) one milliliter of KCN solution was added drop by drop to one milliliter of AgNO3 solution stirring to dissolve the precipitate that formed. With this the compound K[Ag(CN)2] is formed. Subsequently, this solution was diluted with water at a ratio of 1:50.
The decoration process consisted in immersing the samples to be decorated in the K[Ag(CN)2] dissolution for a period of two hours. After the treatment, the samples were rinsed in absolute ethyl alcohol. Finally, to determine the distribution of the silver particles deposited in the welding zone, images of the different areas were obtained by using the scanning electron microscope (SEM).
From the welding zone, block samples were cut and both sides were roughened up to 600 grit sandpaper, and then a macroetch with nital 6 was performed to reveal the macrostructure and be able to distinguish between the BM, the HAZ and the FZ. Once this was done, new cuts were made to collect samples from these three zones (Fig. 1). Lastly the final finish was done by roughening both faces of the sample with a 600 grit sandpaper to a final sample thickness of 1 mm.
A Devanathan and Stachurski type cell was constructed for permeation tests. This type of cell has two half cells, one where the sample is charged with atomic hydrogen and the other where the oxidation of the atomic hydrogen takes place. A 0.5 M H2SO4 solution with 0.2 g of As2O3 per liter of solution was used in the charging cell and a solution of 0.1 M NaOH was used in the oxidation cell. Both solutions were rebuffed with nitrogen for a period of 20 minutes prior to the start of the permeation tests. Figure 2 shows the permeation cell and the sample arrangement referred as working electrode (WE), which, due to the design of the cell, presents an exhibition area of 0.785 cm2. The sample was placed between the two half cells so that one side of the sample would undergo the process of reduction of hydrogen ions and the other side the atomic hydrogen oxidation process. The charging half cell was connected to a power source to generate a current equivalent to 40 mA/cm2 and the oxidation half cell was connected to a potentiostat to maintain a constant 300 mV potential with respect to the Ag/AgCl electrode. Graphite electrodes were used in both half cells as auxiliary electrodes and the solutions were maintained with a constant flow of nitrogen during the course of testing.
Permeation cell. RE: reference electrode; AE: auxiliary electrode; WE: working electrode.
Figures 3 and 4 present the microstructures of the welding zones and HAZ subzones for M1 steel and M2 steel respectively. The BM of M1 steel has light blue martensite and brown bainite; while in the M2 steel, quasi-polygonal ferrite (QPF) and acicular ferrite (AF) are observed, both blue and the martensite-austenite microconstituent (M/A) white; in addition, precipitates are observed in both steels. The percentages of microconstituents in the BMs (Table 2) were determined by quantitative metallography.
Optical micrographs of the welding zone of M1 steel. a) BM, b) ICHAZ, c) RCHAZ, d) CGHAZ and e) FZ.
Optical micrographs of the welding zone of M2 steel. a) BM, b) ICHAZ, c) RCHAZ, d) CGHAZ and e) FZ.
Figures 3(b) and 4(b) show the microstructure developed in the intercritical heat affected zone (ICHAZ). Because the peak temperatures that develop in this subzone (between Ac1 and Ac3), part of the original microstructure did not transform to austenite during the heating cycle, as seen in the case of M2 steel, where the austenite transformed to QPF and promoted the forming of the M/A microconstituent and the agglomeration and growth of precipitates. For M1 steel, part of the martensite and bainite were transformed to form austenite in the heating cycle in which, during the cooling cycle transformed to QPF, additionally to the formation of the M/A microconstituent. In the recrystallized heat affected zone (RCHAZ) (Figs. 3(c) and 4(c)), peak temperatures reached are above the critical temperature Ac3, but lower than 1573 K, so that the original microstructures of the BMs were fully austenized during the heating cycle. In the final microstructure no martensite formation was observed, which can be attributed to the austenite grain not growing excessively, which led to the existence of many nucleation sites for the formation of new phases during cooling. Additionally, it was observed that the formation of polygonal ferrite (PF) for both steels and the decrease of the M/A microconstituent in M2 steel were promoted.
According to the dissolution temperature of carbonitrides (1273 K) obtained from Eq. (1),27) niobium carbonitrides presented in the ICHAZ and part of the RCHAZ, which in addition to contributing to mechanical resistance because of the precipitation mechanism, controlled grain size during welding.
In the coarse grained heat affected zone (CGHAZ) M1 steel presents a martensite microstructure and M2 steel a combination of AF and M/A microconstituent (Figs. 3(d) and 4(d) respectively). Due to the high temperatures experienced in the CGHAZ (between 1573 K and 1726 K), dissolution phenomena of most of the precipitates (except titanium nitrides) occur which promotes grain growth. As a result, there are fewer nucleation sites for the new phases which were formed during cooling, causing the kinetic transformation of diffusional phases to decrease, making it easier to form martensite and/or bainite. Finally, in the FZ of M1 steel presents a bainite microstructure and M2 steel an AF microstructure (Figs. 3(e) and 4(e) respectively). In both the CGHAZ and FZ of the M2 steel, the formation of M/A microconstituent is observed, where this microconstituent delimits the AF boundaries.
To corroborate that the white areas in Figs. 3 and 4 are actually the M/A microconstituent, a microanalysis of these areas was performed as shown in Fig. 5, where it is observed that the particle is composed of iron.
M/A microconstituent in the ICHAZ of M2 steel. a) MO; b) SEM; c) diffractogram.
Figure 6 shows the relative amount of M/A microconstituent for each welding zone and HAZ subzone of both steels. In the BM of M1 steel there is no presence of the M/A microconstituent; on the other hand, the highest percentage of M/A microconstituent is observed in the ICHAZ. By comparison, M2 steel presents a much higher amount of M/A microconstituent in all welding zones and HAZ subzones. The highest amount of M/A microconstituent (2.96% and 2.02% respectively) and the highest aspect ratio, i.e., the highest length over width ratio of the particle occurs in the FZ and CGHAZ. The highest amount of M/A microconstituent that occurs in this zone and subzone can be explained by the formation of AF by a shear mechanism, where carbon enrichment occurs at the boundary of this phase. This made small M/A particles to be produced and distributed evenly. On the other hand, the irregular forming of larger M/A in the ICHAZ compared to the other zones and subzones is observed, that is fewer but larger particles. It is observed that the morphology and amount of this microconstituent is a function of the heat cycle experienced in the weld and the original microstructure.
Relative amount of M/A as function of the welding zone.
Figure 7 presents the microhardness in each welding zone and HAZ subzones, which is related to the microstructures present. In M2 steel an increase in microhardness is observed from the BM with QPF, AF and M/A microstructure to the CGHAZ with AF and M/A microstructure. A decrease in the FZ is observed later, as this zone, being a solidification structure, presents columnar grains with a thicker AF microstructure compared to the CGHAZ. M1 steel has a softening in the HAZ, particularly in the subcritical heat affected zone (SCHAZ), ICHAZ and RCHAZ to subsequently increase in the CGHAZ and decrease again in the FZ. This softening is usually present in steels containing martensite.28) The softening in the SCHAZ is because this subzone experienced a thermal cycle similar to tempering; in the ICHAZ, part of the microstructure that did not transform to austenite, was tempered as in the case of the SCHAZ, and that which transformed to austenite in the heating cycle, formed ferrite in the cooling cycle, so this subzone is even softer than the SCHAZ; and in the RCHAZ, all the original microstructure transformed to austenite in the heating cycle and formed ferrite in the cooling cycle, so that there is no trace of the original microstructure, with this subzone being the softest of all. In the CGHAZ microhardness is higher than that of the BM due to the formation of martensite, and in the FZ microhardness decreased due to the formation of bainite. Increasing the hardness of the welding zone, is related to the increase in yield stress and the generation of residual stresses by thermal gradients and phase transformations (high chemical activity zone), and these in turn are related to increased susceptibility to HIC,6,15,29,30) so it would expect that the CGHAZ to be the subzone more susceptible of both steels, particularly the subzone of the M1 steel.
Microhardness of the welding zone of the microalloyed steels.
During the preparation of the solution for silver decoration, the most likely reactions that present are:25)
With the subsequent dilution of the compound K[Ag(CN)2], the dissociation reaction is promoted:
When this solution comes into contact with the surface containing trapped hydrogen the reduction of silver and oxidation of hydrogen is produced simultaneously.
Figure 8 presents the results of silver decoration of the BM of M2 steel. The image of Fig. 8(a) corresponds to silver decoration from a sample that underwent charging with hydrogen and silver decoration, where the presence of silver precipitates can be observed which are corroborated by the microanalysis of the Fig. 8(b). The image of Fig. 8(c) corresponds to silver decoration of a sample that was not charged with hydrogen, for this reason no silver precipitation is observed. Finally, in the image of Fig. 8(d), no silver precipitation is observed, this is because although this sample underwent the same hydrogen charging and silver decoration procedure, between these two procedures a baking heat treatment was performed to remove trapped hydrogen. From these results, it can be concluded that an oxide-reduction reaction occurs between the silver ions and trapped hydrogen, making it possible to determine the hydrogen traps as suggested by Eqs. (5) and (6), and these traps are reversible traps.
Silver decoration of BM of M2 steel. a) sample charged with hydrogen; b) diffractogram of the particles; c) sample without hydrogen charged treatment; d) sample charged with hydrogen and with baking heat treatment.
Figure 9 presents the relative amount of hydrogen traps present in the welding zone. The more hydrogen traps there are, it is expected that the welding zone present a lower hydrogen diffusivity because the traps are filled first.15) From this figure and Fig. 7, it can be observed that the hydrogen tends to concentrate on areas of higher hardness. In both steels, in the CGHAZ the higher relative amount of traps is presented, but because these are reversible, they will function as a hydrogen source, which can diffuse to a high chemical activity zone as in the area of the crack tip where high stresses are developed.31) It is observed that in each welding zone, M1 steel presents the least amount of hydrogen traps compared to M2 steel, except for RCHAZ where M2 steel has the lower amount.
Relative amount of traps in the welding zone.
Regarding permeation tests, current density, which is related to the hydrogen flux, was measured experimentally (J(t)) by Eq. (7):
Hydrogen flux at steady state (Jss) and the effective diffusion coefficient (Deff) were determined by the Eqs.:
Regarding M1 steel, an increase is observed in the effective diffusion coefficient in the following order: FZ<HAZ<BM. The FZ with bainite microstructure presents lower diffusivity compared to the BM with martensite and bainite microstructure, because in the FZ there is a higher number of traps (Fig. 9), besides it can also be due to the FZ having a larger grain size and thus smaller grain boundary area so fewer diffusion paths are presented.18,32) In the case of M2 steel, an increase of the hydrogen effective diffusion coefficient is observed in the following order: BM<HAZ<FZ.
With respect to BMs, M2 steel has a lower diffusivity than M1 steel, which can be attributed that the M2 steel has higher number of traps (Fig. 9), i.e. it must first fill the hydrogen traps before reaching stable state. This corresponds to the fact that martensite is more susceptible to HIC.9,10)
It is known that M/A microconstituent is a hydrogen trap, thus it promotes decreased diffusivity,33,34,35) although it is also known that this microconstituent is related to a decrease in toughness.36) In Fig. 10 is observed that a higher number of M/A microconstituent in the BM and HAZ, is related to decreased diffusivity. However in the M2 steel, the FZ with the highest number of M/A microconstituent has a higher diffusivity than the BM.
Relative amount of M/A and effective diffusion coefficient of the welding zone.
Figure 11 shows the microhardness of the welding zone and correlates with the effective diffusion coefficients obtained from the permeation tests. For M2 steel, it is observed that an increase in microhardness is accompanied by an increase in the effective diffusion coefficient for the case of BM and HAZ; however, in the case of the FZ, where the microhardness decreases, the effective diffusion coefficient increases even further; this is because this zone presents fewer traps (Fig. 9). In the case of M1 steel, a similar behavior is observed, that is, in the FZ the diffusivity decreases because the microhardness decreases and in the HAZ the diffusivity decreases because a softening is presented in the RCHAZ, ICHAZ and SCHAZ.
Microhardness and effective diffusion coefficient of the welding zone.
In Fig. 12 the relationship between the relative amount of traps and the effective diffusion coefficient is observed. It is expected that with a higher amount of traps, diffusivity will be lower as hydrogen is trapped and can not diffuse, i.e., traps increase the solubility of hydrogen and decrease diffusivity.12,37) For the case of BMs, it is observed that M2 steel shows the highest amount of hydrogen traps, which corresponds with its lower diffusivity and lower microhardness compared to M1 steel. However, the opposite occurs in the FZ, i.e., the FZ of M1 steel with the highest number of traps has the highest diffusivity compared to the FZ of M2 steel. If this result is analyzed only observing M1 steel, it is observed that the BM has fewer traps with a high diffusivity, and the FZ has more traps with low diffusivity, which corresponds to the expectations: more traps lead to lower diffusivity. The same applies to M2 steel, where it is seen that the BM has more traps and lower diffusivity compared to the FZ. The reason that the diffusivity in the FZ of M1 steel is higher than that the FZ of M2 steel is because the former (bainite microstructure) has greater microhardness, that is related to higher yield stress and residual stress i.e. higher chemical activity, and the latter (AF microstructure) has a lower microhardness, that is related to lower yield stress and residual stress i.e. lower chemical activity, where it is known that AF decreases the diffusivity;38) and the reason why the FZ of M1 steel has a lower diffusivity than in the BM with both having a similar microhardness, is because in the FZ there are more hydrogen traps than in the BM (Fig. 12). Then, for HIC there must be a combined effect of diffusivity and trapping, because a low diffusivity by increasing traps is no guarantee of lower susceptibility to HIC, since these can act as sources of hydrogen.
Relative amount of traps and effective diffusion coefficient of the welding zone.
The effect of the microstructure developed in the welding zone on the hydrogen diffusivity of two high strength experimental microalloyed steels was determined. For this, hydrogen permeation experiments, silver decoration and microhardness measurements were performed.
It was observed that the combination of increasing traps, promoting the formation of M/A microconstituent and AF and decreasing martensite and bainite microconstituents (decrease in microhardness), are related to a decrease in hydrogen diffusivity.
In the case of the M2 steel weld, the BM with QPF, AF and M/A microstructure has the lowest diffusivity, and the FZ with of AF and M/A microstructure the highest, i.e., diffusivity increases in the order: BM<HAZ<FZ. In the case of the M1 steel weld, the BM with martensite and bainite microstructure has the highest diffusivity, and the FZ with bainite microstructure and highest number of traps has the lowest diffusivity, i.e., diffusivity increases in the order: FZ<HAZ<BM. All zones in the M2 steel weld had a lower diffusivity compared to the M1 steel zones.
The CGHAZ of both steels presented the highest microhardness and the highest relative amount of traps, however because these are reversible traps, these subzones could be the most susceptible to HIC.
The authors are grateful to CONACYT (grant CB-178777 and 178511) for the financial support and for the scholarship (No 174555) to E. L.-M. Also authors are grateful to UNAM PAPIIT grant IN118714 for the final support.