1 Introduction

With the extensive exploitation of ocean resources, steel used in ocean engineering has been designed for high strength, toughness, and corrosion resistance.[1] In general, microalloying elements such as Nb, V, or Ti are widely added to produce steel equipped with high strength, resistance to brittle fracture, and good weldability.[2] Niobium can readily form NbC, which has high hardness, and a high melting point, and is very effective in restraining the growth austenite grains and inhibiting austenite recrystallization.[3] NbC precipitates also improve the strength and elongation of the weld metal and the resistance of hydrogen-induced delayed fracturing.[4,5,6]

The heat-affected zone (HAZ) of a weld joint in high-strength low-alloy (HSLA) steel is generally categorized into five distinct regions[7]: the base metal (BM), the tempered zone (TZ), the intercritical HAZ (ICHAZ), the fine-grained HAZ (FGHAZ), and the coarse-grained HAZ (CGHAZ). In regard to mechanical properties, the CGHAZ, which has coarse grains, and the ICHAZ, which has an uneven microstructure, have the lowest toughness among the weld joint regions. Moreover, the HAZ corrodes more easily than the BM,[8] and the passive current density produced from the HAZ is found to be higher than that of the BM in a NaHCO3/Na2CO3 solution.[9] Aside from the above, the differences in composition and microstructure between the HAZ and the BM can create local electrochemical potential variations, leading to galvanic corrosion.[10]

The effects of Nb microalloying on the mechanical properties and microstructure of steel have been widely studied. Previous reports revealed that Nb improves properties by refining the grains, and making the microstructure of base steel and weld joints more uniform, etc.[11,12,13,14,15] Meanwhile, a large amount of research on the electrochemical corrosion behavior in HAZ of weld joints were reported.[8,9] However, except for our relative work,[16,17] the roles of Nb in the corrosion behavior of HSLA steels, particularly for each part of the weld HAZ, were rarely reported.

In this work, microstructures of the HAZ of HSLA steels with and without Nb alloy contents were investigated. The effects of nanoscaled Nb precipitates on the mechanical properties, especially, the corrosion behavior of simulated weld HAZ of HSLA steel were studied based on the changes in microstructures of different HAZ caused by Nb. The results are beneficial to the development of Nb microalloying steel, and for the service and safety of HSLA steel weld joints used in the ocean environment.

2 Experimental

2.1 Materials

Two kinds of HSLA steels were used in this experiment. Table I gives the chemical compositions of the as-produced materials. Except for Nb, the concentrations of other elements are almost the same.

Table I Chemical Compositions of Base Material (Weight Percent)
Full size table

2.2 Preparation of Simulated HAZ Microstructures

The simulated HAZ microstructures were prepared according to theoretical calculations and observations of real HAZ microstructures. Although researchers have achieved great success in simulating weld HAZ using hot simulation tests, the size of the real HAZ or hot simulated HAZ limited the practicability to conduct some experiments such as immersion tests. The heat treatment process simulating HAZ microstructures was confirmed to be feasible by some researchers based on the estimated phase transformation temperatures.[18]

The A c1 and A c3 temperatures of the experimental steels were estimated according to the empirical equation,[19] shown below:

$$ A_{\text{c1}} (^\circ {\text{C}}) \, = { 723} - 10. 7\omega \left( {\text{Mn}} \right) - 3. 9\omega \left( {\text{Ni}} \right) + 2 9\omega \left( {\text{Si}} \right) + \,1 6. 7\omega \left( {\text{Cr}} \right) + 2 90\omega \left( {\text{As}} \right) + 6. 3 8\omega \left( {\text{W}} \right), $$
$$ A_{\text{c3}} (^\circ {\text{C}}) \, = { 910} - 3 20\omega \left( {\text{C}} \right)^{0. 5} - 1 4\omega \left( {\text{Ni}} \right) - 1 2\omega \left( {\text{Cu}} \right) - 10\omega \left( {\text{Mn}} \right) + 5\omega \left( {\text{Cr}} \right) + 1 4\omega \left( {\text{Mo}} \right) + 1 8\omega \left( {\text{Si}} \right), $$

where ω is the mass fraction of each element.

According to the equations above, A c1 and A c3 for the Nb-bearing steel were 710 °C and 873.7 °C, respectively, and A c1 and A c3 for the Nb-free steel were 710 °C and 874.9 °C, respectively.

Based on the phase transformation temperatures, the heat treatment temperature was chosen. The base materials were put in the furnace at 1300 °C, 950 °C, 780 °C, and 500 °C, maintained for 15 minutes, and subsequently cooled in water at 30 °C to reproduce the targeted simulated microstructures of the CGHAZ, FGHAZ, ICHAZ, and TZ, respectively.

2.3 Microstructure Observations and Tensile Tests

The microstructures of the specimens were observed using optical microscopy (OM, Reichert-Jung Polyvar MET). In addition, a stereo microscope (model VHX-100; hereafter SZM) and high-resolution transmission electron microscopy (Tecnai 30; hereafter HRTEM) were used to observe microstructures and nanosized NbC precipitates. The TEM specimens were prepared by slicing, grinding, and ion thinning. Tensile tests of the as-received HSLA steel and the simulated HAZ were conducted at room temperature using a computerized tensile testing system on an Instron-type testing machine.

2.4 Conventional Electrochemical Measurements and Short-Term Immersion Tests

Prior to the electrochemical experiments, samples were ground using 2000 grit SiC sand paper, rinsed with double-distilled water, and subsequently dried using ethanol. Open circuit potential measurements (OCP), electrochemical impedance spectroscopy (EIS), and potentiodynamic tests were performed using a VersaSTAT 3F, for which a conventional three-electrode electrochemical cell was used. For the counter electrode, a sheet of Pt was used. All potentials in the present study were measured against a saturated calomel electrode (SCE). EIS tests were applied using a potential amplitude of 10 mV on the OCP. The frequency range of 0.01 to 1,00,000 Hz was used for EIS experiments. The potentiodynamic polarization curves were measured in the potential region from −0.3 to 0.3 V (vs OCP) at a scanning rate of 0.5 mV/s.

All of the electrochemical experiments were conducted in simulated seawater solutions, and the main compositions are given in Table II. To accord with the pH value of Pacific Ocean water,[20] a sodium hydroxide solution was added to the simulated seawater to adjust the pH to 8.0.

Table II Chemical Compositions of the Test Solution (g/L)
Full size table

ICHAZ electrochemical samples were chosen for short-term immersion tests. After immersion for 24, 48, and 72 hours, EIS tests were performed on the samples. The morphology of corrosion products of the immersed samples was observed on a Stereo Microscope, and the morphology of cross sections of the corrosion products was observed using a scanning electron microscope (SEM, Quanta™ 250). After rust removed, the corrosion morphology was observed using an SEM (JCM-6000PLUS).

3 Results and Discussion

3.1 Characterization of Simulated HAZ Microstructures

3.1.1 Optical metallographic microstructure

Figure 1 presents the OM microstructures of the HSLA steel and the corresponding simulated HAZ microstructures after different heat treatments. As a general observation, the size of grains in the Nb-bearing steel is finer than the size of grains in the Nb-free steel. It is obvious that Nb addition effectively inhibits the growth of grain boundaries.

Fig. 1
figure 1

Microstructures of HSLA steel and the corresponding simulated HAZ: (a) Nb-bearing BM, (b) Nb-free BM, (c) Nb-bearing TZ, (d) Nb-free TZ, (e) Nb-bearing ICHAZ, (f) Nb-free ICHAZ, (g) Nb-bearing FGHAZ, (h) Nb-free FGHAZ, (i) Nb-bearing CGHAZ, (j) Nb-free CGHAZ

Full size image

Figures 1(a) and (b) show that the BM with fine microstructures consists of granular bainite (GB) and bainite-ferrite (BF); otherwise, several grains of ferrite are present in Figure 1(b). When the heating temperature was below A c1, the steel was tempered. As shown in Figures 1(c) and (d), the microstructures of the TZ did not change too much in comparison to those in the BM. When the steel was heated to a temperature between A c1 and A c3, part of microstructure austenitized. Therefore, the ICHAZ with uneven microstructures consists of polygonal ferrite (PF), BF, and bainite with a net-like distribution of M-A islands along the primary austenite grain boundaries, as shown in Figures 1(e) and (f). As shown in Figures 1(g) and (h), the FGHAZ mainly consists of fine granular bainite and a few grains of BF, and the FGHAZ microstructure is smaller, denser, and more homogeneous than that of the BM. As shown in Figures 1(i) and (j), the CGHAZ grains were coarsened severely and are composed of granular bainite and a few grains of lath bainite with coarse primary austenite grain boundaries, because of the high heating temperature. Moreover, the microstructures of the Nb-bearing samples were finer and more homogeneous than those of the Nb-free ones.

3.1.2 Nanosized precipitates

Figure 2 shows the distribution of nanosized precipitates in different zones. Figures 2(a) through (e) show fine and evenly distributed precipitates observed in Nb-bearing samples, while Figure 2(f) shows that there is no obvious precipitate in Nb-free samples. The electron diffraction pattern of precipitates is indexed as niobium carbide, which has a particle size that varies from 3 to 10 nm.

Fig. 2
figure 2

Precipitates in HSLA steel and the corresponding simulated HAZ: (a) Nb-bearing BM, (b) Nb-bearing TZ, (c) Nb-bearing ICHAZ, (d) Nb-bearing FGHAZ, (e) Nb-bearing CGHAZ, (f) Nb-free BM

Full size image

The number and size of precipitates were controlled by the temperature of the solute.[21,22] As for the BM, TZ, and ICHAZ, the size and distribution of the precipitates barely exhibited obvious differences. The reason is that the precipitates do not have enough diffusion driving force and time to grow on account of the low heating temperature and short holding time, such as < A c3 (780 °C, 500 °C) and 15 minutes in the experiment. For the FGHAZ, the heating temperature was above A c3 and below the solution temperature of niobium carbide (approximately 1200 °C).[23] At the temperature of 950 °C, a large number of nanosized NbC particles precipitated in the matrix resulting in a fine grain size. At a high heating temperature (> 1200 °C), the finer precipitates redissolved, which caused the amount of precipitates in the CGHAZ to decrease. Meanwhile, at the temperature, grains grew larger, companied by the weakening of pinning effect of nanoparticles, which caused the matrix to possess a large grain size.

3.2 Mechanical Properties of the Simulated HAZ

The yield strength, tensile strength, and elongation of simulated microstructures with and without Nb addition are presented in Figure 3. Overall, the Nb-free simulated HAZ had lower yield and tensile strength than those of the Nb-bearing HAZ. Of the five subdivided regions, the ICHAZ had the minimum strength, and the CGHAZ had the minimum elongation, although the former had the largest elongation, and the latter had the highest strength. To explain the test results, an analysis of the microstructure, solid solution, and nanosized precipitates is necessary.

Fig. 3
figure 3

Mechanical properties of the simulated HAZ: (a) Stress, (b) Elongation

Full size image

In the simulated HAZ, the mechanisms that influence strengthening include solid solution, grain refinement, precipitation, and phase transformations,[24] where solid solution formation and precipitation of C and Nb are reciprocal to strength.

For the Nb-bearing samples, both the yield strength and tensile strength obviously enhanced compared with the Nb-free ones. This can be attributed to the following mechanisms: grain refinement resulting from NbC pinning to grain boundaries, precipitates strengthened by NbC particles, promotion of bainite transformation, and alloy element solutions enhancing the stability of austenite with dissolved Nb. On the other hand, the elongation of the BM, TZ, and ICHAZ slightly decreased along with the addition of Nb and increasing strength. Otherwise, the elongation of Nb-bearing CGHAZ was higher than that of Nb-free CGHAZ. At the high temperature of 1300 °C, C and Nb fully dissolved in steel, and after quenching, parts of C and Nb precipitated as NbC particles from matrix. In this case, the solid solution concentration of C in Nb-bearing sample was lower than that in Nb-free sample, which was advantageous to ductility. As for the FGHAZ, the heating temperature and elongation of the Nb-free sample were moderate, causing the inconspicuous effect of decreased C concentration on the elongation enhancement of Nb-bearing sample.

With heat treatments at different temperatures, strength and elongation of the HAZ change according to the complex synergies of solid solutions, grain sizes, precipitates, and phases. In the ICHAZ for instance, the declined strength and enhanced elongation decisively resulted from the increased amount of ferrite. Meanwhile, for the Nb-bearing steel, the tensile strength of the ICHAZ was only very slightly lower than that of TZ, revealing the distinct effect of Nb microalloying on the mechanical property of the weakest weld HAZ. The existent Nb impeded the transformation from partly austenitized phase to ferrite, but promoted the transformation to bainite or bainite-ferrite with supersaturated C and high strength. Compared to the Nb-free steel, both tensile and yield strength of the ICHAZ for the Nb-bearing steel had the maximum increase in all of the five zones. In the FGHAZ and CGHAZ, the increased strength was mainly because of the increased amount of bainite and the alloy solution, while the decreased elongation of CGHAZ resulted from the strengthened microstructures and coarsened grains. The changes in the mechanical properties of the TZ may be caused by the decomposition of M-A islands at the tempering temperature. As mentioned above, solid solution concentration of C in sample played a role in the ductility.

3.3 Electrochemical Corrosion Testing in Simulated Seawater

3.3.1 Potentiodynamic polarization

Figure 4 shows potential dynamic polarization curves of samples in seawater. The figure includes curves for the BM and the simulated HAZ samples. The corrosion current density, i corr, and corrosion potential, E corr, were extracted from the potentiodynamic curves and presented in Figure 5. The results revealed that the BM samples and the corresponding simulated HAZ in seawater manifested as the anodic process in an active state and cathodic process were dominated by oxygen reduction reactions. However, for each of the simulated microstructures, both i corr and E corr exhibited obvious differences.

Fig. 4
figure 4

Potential dynamic polarization curves of HSLA steel and the corresponding simulated HAZ: (a) Nb-bearing steel, (b) Nb-free steel

Full size image
Fig. 5
figure 5

Date electrochemical data acquired from the linear polarization region in Fig. 4: (a) Corrosion potential, (b) Corrosion current density

Full size image

E corr and i corr of the samples were obtained from Figure 4 using the fitting software ZsimpWin, as shown in Figure 5. According to Figure 5, E corr of the BM and TZ in seawater, regardless of whether Nb was added, was more positive than that of the simulated HAZ, especially the ICHAZ and CGHAZ. The i corr values from low to high were BM, TZ, FGHAZ, ICHAZ, CGHAZ, which were consistent with E corr. The corrosion resistance of the BM was obviously better than for the HAZ, except for the TZ, which has similar value. ICHAZ, which had the second lowest corrosion resistance, was closely related to the double phase of ferrite and bainite, while the lowest corrosion resistance (CGHAZ) was mainly caused by the largest internal stress and number of dislocations in the supersaturated bainite.

However, Nb-bearing samples possessed higher corrosion resistance than the Nb-free samples. Compared with the effect of adding Nb on improved microstructure, aggravated corrosion to the anodic matrix by NbC precipitates as cathodic particles was more minor, and on the whole, the corrosion resistance of the Nb-bearing steel was obviously enhanced. The factors improving corrosion resistance include a more homogeneous microstructure, and lower supersaturation and internal stress resulting from Nb and NbC precipitation in the Nb-bearing steel.

Otherwise, galvanic corrosion is unavoidable in real welded joints owing to the galvanic interactions of the different regions in the joints.[25] According to prior electrochemical data for simulated HAZ, and considering galvanic corrosion, the CGHAZ and ICHAZ are the worst zones in the corrosion of real weld joints.

The potentiodynamic curves with two inflection points reveal a chain of cathodic reactions. Before the first inflection point, the cathodic reactions transformed from oxygen activation controls to oxygen diffusion controls. The cathodic reaction was as follows:

$$ {\text{O}}_{ 2} + {\text{ 2H}}_{ 2} {\text{O }} + {\text{ 4e }} = {\text{ 4OH}}^{ - } . $$
(1)

After the inflection point, the current density of the cathodic reaction was approached gradually to the current density limit of oxygen diffusion. As the cathodic potential turned negative and went below the hydrogen-charged potential, the following hydrogen evolution reaction occurred:

$$ 2 {\text{H}}^{ + } + {\text{ 2e }} = {\text{ H}}_{ 2} . $$
(2)

Based on the Nernst equation, the hydrogen-charging potential at 25 °C is equal to

$$ E_{\text{e}} \left( {{\text{H}}_{ 2} /{\text{ H}}^{ + } } \right) \, = \, - 0.0 5 9 1 {\text{ pH }} = \, - 0. 7 20 \, V_{\text{SCE}} . $$
(3)

As the cathodic potential continued to be more negative, the hydrogen evolution reaction was enhanced gradually, and the cathodic polarization control process gradually transformed from an oxygen diffusion process to a hydrogen evolution reaction process. Thus, a second inflection point formed.

3.3.2 EIS

Figure 6 shows Nyquist plots for the EIS experiments performed on BM and heat-treated samples using OCP measurements in the simulated seawater. The size of the impedance arc radius, which revealed the corrosion resistance from large to small, was in the order: BM, TZ, FGHAZ, ICHAZ, CGHAZ. These results are consistent with those obtained with the E corr values and i corr values. ZSimpWin 3.50 was used to fit the EIS data, and the equivalent circuit is shown in Figure 7. In this equivalent circuit, R s and R t represent solution resistance and charge-transfer resistance, respectively, and Q dl represents the electrical double-layer capacitance of metal interfaces. In this environment, the R t revealed the corrosion resistance, that is, the corrosion resistance increased as the R t value became larger.

Fig. 6
figure 6

Nyquist plots of HSLA steel and the corresponding HAZ: (a) Nb-bearing steel, (b) Nb-free steel

Full size image
Fig. 7
figure 7

Equivalent circuit R(QR) of EIS

Full size image

The charge-transfer resistance of HSLA steel and the corresponding HAZ are presented in Figure 8. It is obvious that R t for the BM and TZ samples were much higher than those for the other HAZ samples, and the ICHAZ and CGHAZ were the lowest among the all samples. The results are in accordance with potentiodynamic test results.

Fig. 8
figure 8

Charge-transfer resistance of HSLA steel and the corresponding HAZ

Full size image

3.4 Short-Term Immersion of ICHAZ in Simulated Seawater

3.4.1 EIS for immersed ICHAZ

Figure 9 presents Nyquist plots of ICHAZ immersed in simulated seawater for different times, and the equivalent circuits and corresponding resistances are shown in Figures 10 and 11, respectively. In the equivalent circuits, R s, R t, and R f represent solution resistance, charge-transfer resistance, and corrosion product resistances, respectively. CPE1 and CPE2 represent the capacitance of corrosion products and electrical double-layer capacitance of metal interfaces, respectively.

Fig. 9
figure 9

Nyquist plots of ICHAZ immersed in simulated seawater for different times: (a) Nb-bearing steel, (b) Nb-free steel

Full size image
Fig. 10
figure 10

Equivalent circuit R(Q(R(QR))) of EIS

Full size image
Fig. 11
figure 11

Corrosion product resistance and charge-transfer resistance of ICHAZ at different times: (a) Corrosion product resistance, R f, (b) Charge-transfer resistance, R t

Full size image

It is obvious that the sizes of the impedance arc radius and R t and R f increased as the immersion time rose. According to the EIS results, the corrosion product plating on sample surfaces partly protected the metal matrix within the immersion terms. Compared with the Nb-free samples, the Nb-bearing samples showed better corrosion resistance. According to the Rf, which indirectly reflects the integrality and compactness of corrosion product films, the Nb-bearing samples with more compact corrosion product films markedly improved the resistance to ion migration.

3.4.2 Corrosion morphology

Figure 12 shows the morphology of corrosion products of ICHAZ immersed in simulated seawater. The picture shows that the corrosion product layers were composed of a black inner rust layer and a yellow outer rust layer on two samples of steel. But the yellow outer rust layer on the Nb-free sample was incompact, and easy to break and fall, which caused lesser yellow rust in Figure 12(b). It was evident that the yellow corrosion product films on the Nb-bearing sample surface were more compact and firm compared to the Nb-free sample. Figure 13 is a cross-sectional view of the corrosion products of ICHAZ immersed in simulated seawater. As shown in Figure 13, more serious pitting corrosion was found on the surface of the Nb-free sample. In addition, the corrosion product layers on the Nb-bearing surface were undoped and thicker in comparison to the Nb-free sample. The corrosion resistance and corrosion morphology of HSLA samples were dependent on the integrality and compactness of corrosion product films. Compact corrosion product films with higher resistance to ion migration effectively reduced the destruction caused by the chloride ion. As seen in the SEM photographs of the corrosion morphology of ICHAZ after removing the rust, shown in Figure 14, a certain number of pitting holes appeared on the surface of the samples immersed in simulated seawater for 72 hours. It was obvious that the pitting holes on the Nb-bearing samples were considerably fewer and smaller than those on the Nb-free samples. More compact corrosion product films on the Nb-bearing samples increased the resistance to ion migration, and thus inhibited the initiation and propagation of pitting holes caused by the chloride ion. The results of the immersion tests are fully consistent with the electrochemical tests.

Fig. 12
figure 12

Corrosion morphology (with corrosion products) of ICHAZ immersed in simulated seawater for 72 h. (a) Nb-bearing, (b) Nb-free

Full size image
Fig. 13
figure 13

Cross-sectional view of corrosion products of ICHAZ immersed in simulated seawater for 72 h. (a) Nb-bearing, (b) Nb-free

Full size image
Fig. 14
figure 14

Corrosion morphology (without corrosion products) of ICHAZ immersed in simulated seawater for 72 h. (a) Nb-bearing (b) Nb-free

Full size image

4 Conclusions

  1. (1)

    The simulated HAZ of Nb-bearing and Nb-free HSLA steel consisted of five subdivided zones, among which the ICHAZ and CGHAZ possessed the worst overall mechanical properties. The corrosion resistance in descending order was the BM, TZ, FGHAZ, ICHAZ, and CGHAZ.

  2. (2)

    Compared with the Nb-free steel, Nb obviously enhanced the strength and corrosion resistance of weld joints in the Nb-bearing steel because Nb microalloying resulted in nanosized precipitation, homogeneous microstructures, and other related sequences.

  3. (3)

    During immersion in seawater, compact corrosion product films with higher resistance to ion migration formed on the surface of the Nb-bearing steel, and the initiation and propagation of pitting holes were effectively inhibited.