Effect of Al content on nanoprecipitates, austenite grain growth and toughness in coarse-grained heat-affected zones of Al–Ti–Ca deoxidized shipbuilding steels

1.   Introduction

Welding efficiency has been improved by high-heat input welding (HHIW) technologies, which can attain a heat input over 400 kJ·cm⁻¹. The applications of HHIW technologies include shipbuilding, construction, machinery, and other fields [1–2]. However, with increased heat input, the peak temperature of coarse-grained heat-affected zones (CGHAZs) rises, and the exposure to high temperatures is prolonged, resulting in brittle microstructures, coarse grains, and deteriorated CGHAZ toughness [3–5]. Oxide metallurgy with Ca addition can be potentially used to improve the impact toughness of CGHAZ through promoting the formation of intragranular acicular ferrite (IAF) and refinement of prior austenite grains (PAGs) [6–10], which has been confirmed by JFE Steel [11] and Kobe Steel [12]. We previously showed that the number density (ND) of complex oxysulfides increases with the increase in the Ca content, which promotes IAF nucleation and the precipitation of nanosized TiN particles and considerably improves toughness in the CGHAZ [6,8,13].

The size and distribution of austenite contribute to the toughness of CGHAZ, with large PAGs promoting the formation of brittle microstructures and inhomogeneous microstructure distribution, both of which negatively affect toughness [14–15]. During HHIW, PAG growth suffers from the inhibition caused by the dragging force of the alloying elements segregated at grain boundaries (solute drag) and the pinning force of fine particles, with the latter exhibiting a notably stronger effect [16–18]. These processes, such as particle dissolution and reprecipitation related to PAG growth, cause substantial alterations in the chemistry, size distribution, and volume fraction of precipitates during the welding process [19–20]. Especially under HHIW, the CGHAZ experiences prolonged exposure to elevated temperatures, which exacerbates the dissolution or coarsening of pinning particles.

Given the high thermal stability of fine TiN precipitates at high temperatures, dispersed TiN precipitates are used to pin grain boundaries and control the size of PAG in the CGHAZ of thick steel plates under HHIW [21–22]. The addition of microalloying elements, such as Nb, V, B, and Mo, considerably affects PAG growth, microstructural evolution, and impact toughness of the CGHAZ of alloyed steels as a result of the influence of microalloying elements on the precipitation and thermal stability of TiN. Wu et al. [23] reported that the increase in the Mo content of steels not only increases the ND but also reduces the size of precipitates, which favors the improvement of impact toughness. Our previous work also reported that with the increase in Mo content, the concentration of Mo in Ti(C,N) particles and the thermal stability of TiN precipitates in high-Mo steel increase [24]. However, the addition of Nb results in coarse TiN precipitates in the CGHAZ of Mg/Ca-deoxidized shipbuilding steels and decreases the thermal stability of TiN due to the interaction between Ti and Nb atoms [1,25]. Pan et al. [26] demonstrated that the decrease in the Al content caused the refinement of precipitates in the CGHAZ of thick steel plates, which enabled the faster dissolution of the precipitates and growth of PAGs at a high rate during welding.

The very strong chemical attraction of Al to oxygen allows its frequent use as a deoxidizer in the steel deoxidation process. In the CGHAZ of Ca-treated low-alloy steels with a low Al content, effective inclusions and IAFs exhibit a higher likelihood of formation than steels with a high Al content [27–28]. Our previous work indicated that during HHIW, the microstructures of the CGHAZ of Ca-treated thick steel plates changed from IAF to bainitic ferrite with the increase in the Al content, and this condition resulted in a decrease in impact toughness [29].

Studies rarely focused on the evolution of nanoprecipitates in the CGHAZ of Al–Ti–Ca-deoxidized shipbuilding steels with various Al contents. At a low Al content, Ti and Ca are likely to participate in the deoxidation process, which affects not only the development of inclusions and microstructures but also the nucleation of nanoprecipitates. In this work, experiments and theoretical calculations are performed to investigate the influence of Al content on the nanoprecipitates, PAG growth, and impact toughness in CGHAZ. Therefore, the mechanism of toughness improvement in Al–Ti–Ca-deoxidized shipbuilding steels with various Al contents is clarified.

2.   Experimental

2.1   Steel preparation

The studied steels contained various amounts of Al and were prepared via a 50-kg vacuum induction melting furnace. Argon gas was blown into the furnace during the melting process. The melt was added with appropriate amounts of deoxidizers and alloying elements to achieve the desired steel composition. The melt was then cast into ingots (120 mm × 180 mm × 240 mm). Table 1 lists the chemical compositions of the studied steels analyzed through spark emission spectrometry (ARL 4460, Thermo Fisher, USA) and inductively coupled plasma atomic emission spectrometry (ICP-AES). The experimental steels were designated 60Al and 160Al based on their Al contents. The thermomechanical control process (TMCP) was performed to hot roll ingots at a thickness of 50 mm. The roughing and finishing rolling temperatures were above 930 and 770°C, respectively, with reduction ratios over 30%. Eventual cooling down of the steel plates was achieved from 750 to 400°C at 10°C·s⁻¹.

Table  1.  Chemical compositions of shipbuilding steels investigated in the present study wt%

Steel	C	Si	Mn	P	S	Nb	Ni	Cu	Al	Ti	Ca	O	N	Cr
60Al	0.075	0.21	1.6	0.0070	0.0050	0.015	0.36	0.10	0.0060	0.0090	0.0017	0.0028	0.0033	0.013
160Al	0.080	0.24	1.6	0.0070	0.0050	0.017	0.35	0.10	0.016	0.0085	0.0022	0.0024	0.0040	0.013

 | Show Table

DownLoad: CSV

2.2   Simulated HHIW experiment

The welding simulation involved obtaining 11 mm × 11 mm × 71 mm specimens obtained at 1/4 of the thickness and 1/4 of the width of the as-rolled steel plates (Fig. 1(a)). HHIW thermal cycle of the specimens was performed at 400 kJ·cm⁻¹ using a Glebble-3800 simulator (Dynamic Systems Inc., Poestenkill, NY, USA). The thermal cycle, represented by the red line in Fig. 1(b), was calculated using the Rykalin 3D model embedded in the welding software.

Fig. 1.  (a) Heating and cooling curves of CGHAZ simulation and in situ observation, (b) sample used in the simulated HHIW experiment, and (c) sample used in the in situ observation experiment.

DownLoad: Full-Size Img PowerPoint

2.3   High-temperature laser-scanning confocal microscopy (HTLSCM) experiment

In situ observations of the growth behavior of PAGs were conducted via HTLSCM. After machining the samples from the TMCP plates, they were mechanically polished into metallographic specimens. The samples measured 7.5 mm in diameter and 3 mm in height (Fig. 1(c)). During simulation, the samples were heated to 1400°C, held at this peak temperature (T_p1) for 3 s, and then cooled sequentially to 800 and 500°C at the cooling rates of 3.41 and 0.78°C·s⁻¹, respectively (blue line in Fig. 1(b)). A cooling time of 385 s was observed from 800 to 500°C (t_8/5). Afterward, the standard V-notch specimens with dimensions of 10 mm × 10 mm × 55 mm were prepared for impact testing. The ASTM E23 standard was used to test the specimens at −40°C. Each impact toughness value was calculated from three test findings.

2.4   Characterization of precipitates and austenite grains

Optical microscopy (DM2700M, Leica Microsystems) was performed for the observation of austenite grains in the simulated CGHAZ and HTLSCM samples. The metallographic specimens were polished and etched in a 4vol% nital solution for approximately 15 s. The average size of the PAGs was determined through measurement of more than 300 grains across more than 10 photographs at 50× magnification.

Replica studies and twin-jet electropolishing experiments were performed to analyze the morphology and size of nanoscale precipitates [1,25]. The precipitates on carbon replicas were investigated through transmission electron microscopy (TEM, JEM-2010F, Japan) at an accelerating voltage of 120 kV. The chemistry and crystallographic characteristics of the precipitates were analyzed via energy-dispersive X-ray spectroscopy (EDS, Oxford, UK) and the selected area diffraction pattern (SADP). To study the size distribution and ND, we measured more than 1000 precipitates per unit area of each sample using Image Pro Plus. Each sample underwent a replica study and precipitate analysis three times.

Nonaqueous electrolytic experiments were performed to determine the soluble Nb and Al contents in the matrix. Electrolytic experiments were conducted in a solution containing 2 mL triethanolamine + 1 g tetramethyl-ammonium chloride-methanol + 98 mL methanol with a current density of 10 mA·cm⁻² at approximately 0°C. After its filtration using a polycarbonate filter membrane, the electrolyte was subjected to microwave digestion and analyzed via inductively coupled plasma (ICP) atomic emission spectroscopy (Agilent 5800 ICP-OES, USA). The nonaqueous electrolysis and ICP experiments were repeated twice for each sample, and the findings were averaged.

3.   Results

3.1   Mechanical properties

Table 2 provides a summary of the mechanical properties of the base metals (BMs) and CGHAZs. The simulated CGHAZ of 160Al steel had a Charpy impact energy of 134 J, which is higher than that of 60Al steel (54 J); this finding indicates that the addition of an appropriate amount of Al favored improvement of the CGHAZ impact toughness. Supplementary Material contains further details on the fracture morphology analysis results.

Table  2.  Mechanical properties of BM and CGHAZ of the two steels

Steel	BM							CGHAZ
	Transverse yield strength / MPa	Longitudinal yield strength / MPa	Transverse tensile strength / MPa	Longitudinal tensile strength / MPa	Transverse toughness / J	Longitudinal toughness / J		Toughness / J
60Al	478	464	595	575	160	226		54 ± 25
160Al	540	544	643	645	151	239		134 ± 13

 | Show Table

DownLoad: CSV

3.2   Evolution of the austenite grain size with Al content

Fig. 2 shows the PAG growth behavior at various temperatures during the thermal cycle. Fig. 2(a), (b), (d), and (e) reveals that a notable acceleration in PAG growth occurred during heating from 1300 to 1400°C and during cooling. Abnormal-growth PAGs were observed on the 60Al steel, whereas the PAG size was more uniform in 160Al steel (Fig. 2(c) and (f)). The abnormally grown PAG contained more bainite than normal grains. The relative difference (RD) described by Eq. (1) [30] can be used to determine abnormally large grains.

Fig. 2.  SEM images of PAGs in (a) 60Al and (d) 160Al steels after heating to 1300°C, (b) 60Al and (e) 160Al steels after cooling to 900°C, and (c) 60Al and (f) 160Al steels after cooling to room temperature. (g) Average PAG size of both steels at various temperatures; (h) size distribution of PAGs after cooling to room temperature.

DownLoad: Full-Size Img PowerPoint

where ${\text{GS}}_{\text{abnormal}}$ and ${\text{GS}}_{\text{normal}}$ denote the sizes of abnormally large and normal grains, respectively. If $\text{RD}$ exceeded 0.9, the grains were considered to have abnormal growth [30]. In the 60Al steel, the largest grain was approximately 1 mm in diameter, which is almost twice the size of the largest normal grains. The abnormal grain growth in the 60Al steel substantially reduced the toughness of CGHAZ, and such a result can be attributed to the decreased pinning effect of the precipitates on grain boundaries.

Measurement and statistical analysis of the equivalent diameters of PAGs at various temperatures of the 60Al and 160Al steels were performed (Fig. 2(g)). Both steels exhibited slow-growing PAGs until a temperature of 1300°C. When the temperature was increased to 1100°C, PAGs in the 60Al steel attained an average size of 17.13 µm, which is slightly larger than that in the 160Al steel (16.37 µm). As the temperature increased to 1300°C, the average sizes of PAGs in the 60Al and 160Al steels were 47.18 and 44.82 µm, respectively, and the average sizes of PAGs in both steels were coarsened by about 1.75 times. When the temperature dropped below 1100°C, no evident changes in the PAG size occurred. The 60Al and 160Al steels achieved average PAG sizes of 189.73 and 174.70 µm, respectively, after cooling to room temperature. PAGs in both steels presented similar size distributions at room temperature when PAGs were smaller than 450 μm (Fig. 2(h)).

3.3   Evolution of precipitates

3.3.1   Typical precipitates in BM

Based on the morphologies and compositions of precipitates in the BMs, the precipitates were categorized into ellipsoidal, mixed morphology, and cubic. Fig. 3 illustrates the morphologies and SADP and EDS analysis results on the three types of precipitates. All three were found in the BMs of both steels. Type I precipitate has a cubic structure, Type II precipitate comprises a cubic precipitate as a core with caps on its faces (Fig. 3(b)), and Type III precipitate possesses an ellipsoidal shape that is finer than those of Types I and II precipitates and contains a higher Nb atomic ratio (Fig. 3(c)). The EDS analysis revealed that all three particle types consisted of (Ti,Nb)(C,N). However, the proportions of Ti and Nb varied within (Ti,Nb)(C,N) precipitates of various morphologies. The Type I precipitates attained a lattice parameter of 4.24 Å, which is close to that of TiN (lattice parameter = 4.239 Å) [31]. In the case of Type II precipitate, lattice parameters of 4.26 and 4.38 Å were observed for the core and cap, respectively (Fig. 3(b)). The lattice parameter for the cap was closer to that of NbC (lattice parameter = 4.47 Å) and larger than that of the core; this result indicates a higher Nb content in the cap, in agreement with the results by Moon et al. [31]. The precipitates exhibited an ellipsoidal shape (Type III) when Nb(C,N) precipitated or distributed at the periphery of Ti(C,N), with a lattice parameter equal to 4.58 Å. With the increase in the Nb content of the precipitates, the lattice parameters of the precipitates increase correspondingly.

Fig. 3.  Bright-field TEM micrographs patterns of three types of precipitates in the carbon extraction replicas of the BM before the thermal cycle, associated with SADPs and EDS analyses: (a) cubic (Ti,Nb)(C,N) precipitate, Type I precipitate; (b) cubic (Ti,Nb)(C,N) core with Nb-rich cap, Type II precipitate; (c) ellipsoidal Nb-rich precipitate, Type III precipitate.

DownLoad: Full-Size Img PowerPoint

3.3.2   Typical precipitates in CGHAZ

The CGHAZ of the two steels mostly contained Types I and II precipitates. However, the CGHAZ of the 60Al steel exhibited many irregular particles exceeding 80 nm. Several irregular particles were selected for the composition and structural analysis (Fig. 4). Fig. 4(a)–(c) displays the bright-field micrographs and high-angle annular dark field (HADDF) micrographs of irregular particles showing the growth of square particles from the substrates. Fig. 4(d)–(h) displays the SADP pattern, darkfield images, and element mappings for the irregular particles. The results presented in element mapping indicate that the particles comprised alumina inclusion (white line in Fig. 4(g)), Cu sulfide (Fig. 4(f)), and Ti–Nb carbonitride (red line in Fig. 4(g)). SADP analysis revealed the cubic structure of the different parts of composite particles. Cu sulfide had a lattice parameter of 5.62 Å, which suggests that Cu sulfide is Cu₂S (lattice parameter a = 5.55 Å) [32]. Ti–Nb carbonitride had a lattice parameter of 4.32 Å, which is close to that of TiN. The alumina inclusions detected in the CGHAZ of the 60Al steel presented a lattice parameter of 7.87 Å. These inclusions were assumed to include γ-Al₂O₃ (lattice parameter = 7.94 Å) due to their cubic structure [33]. The diffraction patterns showed the cube–cube orientation relationship between Cu₂S and (Ti,Nb)(C,N) and the cube-on-edge relationship observed between γ-Al₂O₃ and Cu₂S/(Ti,Nb)(C,N).

Fig. 4.  TEM analysis of irregular precipitates in the CGHAZ of the 60Al steel after the HHHIW thermal cycle: (a, b) bright-field micrographs and (c) HADDF micrograph of Al₂O₃-based precipitates; (d) bright-field micrographs with the high-resolution TEM micrograph of the precipitate. (e) Corresponding diffraction of incident beam direction parallel to [001]${}_{{\bf {Cu}}_{\boldsymbol{2}} {\bf {S}}}$//[001]_(Ti,Nb)(C,N)//[114]_γ${}_{-{\bf{Al}}_2{{\bf{O}}_{\bf{3}}}}$, (f, g) dark-field images obtained from the precipitate, and (h) element mappings of the precipitate.

DownLoad: Full-Size Img PowerPoint

3.3.3   Precipitates distribution in the BM and CGHAZ

Fig. 5 revealed the cluster and dispersed distributions in the BM and CGHAZ. The characteristic of cluster distribution is that the precipitates belongs to a group or is in contact with each other (Fig. 5(a)). Notably, most of the precipitates that formed clusters were Type III precipitates. This pattern was commonly observed in the BM of the 60Al steel but was uncommon in the 160Al steel. The precipitates were predominantly dispersed Types I and II precipitates in the CGHAZ of both steels.

Fig. 5.  TEM images of the replica showing precipitates in the (a, b) BM and (c, d) CGHAZ after the HHIW thermal cycle: (a, c) 60Al and (b, d) 160Al steel.

DownLoad: Full-Size Img PowerPoint

Fig. 6 displays the TEM micrographs of precipitates in the CGHAZ of the 60Al steel. The coarsened Type I and Nb-rich (Type III) precipitates aggregated within the grain and showed no pinning effect on PAG boundaries (Fig. 6(a)). Fig. 6(b) demonstrates the distribution and elemental mappings of Al₂O₃-based particles in the CGHAZ, with most of the coarsened particles contained within the grains. For the three particles distributed at the grain boundaries, P1 exhibited no pinning effect, and P2 demonstrated a relatively evident pinning effect on grain boundaries, followed by P3. This observation aligns with the findings, implying a strong pinning effect on small precipitates [34].

Fig. 6.  TEM images of thin foils revealed precipitates in the CGHAZ of the 160Al steel: (a) coarsened Types I and III precipitates that aggregated within the grains; (b) dispersed composite precipitates along the boundaries and elemental mappings.

DownLoad: Full-Size Img PowerPoint

Table 3 lists the changes in average size, ND, and proportion of precipitates observed from the BM to the CGHAZ of the two steels. The precipitates had an average size of 16.29 nm in the BM of the 60Al steel and 13.48 nm in 160Al steel. After being subjected to HHIW, both steels experience a substantial increase in the precipitate size, with an average of 58.91 nm for 60Al steel and 44.64 nm for 160Al steel. In the BM of both steels, the precipitates attained NDs of 11.37 × 10⁵ and 13.88 × 10⁵ mm⁻², respectively. In the CGHAZ, the NDs decreased substantially to 6.12 × 10⁵ mm⁻² in the 60Al steel and to 9.28 × 10⁵ mm⁻² in the 160Al steel, respectively. In addition, Type III precipitates accounted for 38.12% of the total in the BM of the 60Al steel and 6.39% of that in the 160Al steel. Therefore, the BM of the 160Al steel were mostly Types I and II precipitates, with far fewer Type III precipitates than those in the 60Al steel.

Table  3.  Average size, ND and proportion of various precipitates in the BMs and CGHAZs of both steels

Steels	BM				CGHAZ
	Average size / nm	ND / (105 mm−2)	Proportion of Type III precipitates / %		Average size / nm	ND / (105 mm−2)	Proportion of Al2O3- based precipitates / %
60Al	16.29	11.37	38.12		58.91	6.12	29.78
160Al	13.48	13.88	6.39		44.64	9.28	0.00

 | Show Table

DownLoad: CSV

Fig. 7 illustrates the distribution of precipitate diameters for both steels. The BM of both steels showed unimodal and similar distributions of precipitates over the entire diameter range (Fig. 7(a)). After welding, a bimodal precipitate diameter distribution was observed in the CGHAZ of the 60Al steel. Peak 2 (Fig. 7(b)) in the CGHAZ of the 60Al steel was caused by the Al₂O₃-based precipitates, which accounted for 30.93% of the total in the CGHAZ (Table 3). Excluding the Al₂O₃-based precipitates, pure carbonitrides in the CGHAZ of the 60Al steel had an average size of 46.21 nm, similar to that of the 160Al steel. However, the ND of precipitates in the CGHAZ of 60Al steel (6.12 × 10⁵ mm⁻²) decreased more substantially than that of the 160Al steel (9.28 × 10⁵ mm⁻²) after welding.

Fig. 7.  Distribution of precipitate diameters observed in the (a) BM and (b) CGHAZ of 60Al and 160Al steels.

DownLoad: Full-Size Img PowerPoint

Table 4 lists the elemental concentrations measured by ICP in the matrix and precipitates. The Nb content in the precipitates exhibited a decrease from 0.010wt% in the 60Al steel to 0.0054wt% in the 160Al steel, which suggests that the increase in Al content inhibited the precipitation of Nb-rich precipitates in the steel, which is the same trend as the results displayed in Fig. 7. The inclusions of both steels contained very close amounts of Al.

Table  4.  Element concent in the matrix and precipitates wt%

Steel	Nb			Al
	Matrix	Precipitates		Matrix	Oxides
60Al	0.0046	0.010		0.0053	0.00068
160Al	0.011	0.0054		0.015	0.00084

 | Show Table

DownLoad: CSV

4.   Discussion

4.1   Thermodynamic analysis

Factsage 8.1 software was used to perform equilibrium calculations for the inclusions and carbonitride precipitates (Fig. 8). When the temperature decreased from 1500 to 800°C, both steels revealed the formation of various inclusions, such as perovskite (Ca₂Ti₂O_(5,6)), Al₂O₃, CaO·6Al₂O₃, sulfides (MnS and CaS), and carbonitride precipitates (Fig. 8(a)). In the 60Al steel, the perovskite (green solid line) transformed to carbonitrides (blue solid line) and Al₂O₃ (cyan solid line) at 1388°C, and the CaO·6Al₂O₃ achieved the stable phase from 1080 to 1237°C. With the increase in the Al content, the predicted phase changed in the following manner: (1) The temperature at which perovskite (green dashed line) transformed into carbonitrides (blue dashed line) and calcium aluminates (red dashed line) increased to 1438°C. (2) The Al₂O₃ became the stable phase at temperatures lower than 1264°C, whereas CaO·6Al₂O₃ was the stable phase at temperatures above this threshold.

Fig. 8.  Thermodynamic calculation results: (a) phase transformation in two steels during solidification; (b) changes in the specific compositions of carbonitrides (blue carbonitrides in (a)) with temperature.

DownLoad: Full-Size Img PowerPoint

Fig. 8(b) shows that the TiN particles were the main phase of carbonitrides during the decrease in the temperature from the peak temperature to 1000°C. This finding suggests that the TiN precipitates exerted a pinning effect at high CGHAZ temperatures, and the Nb-rich particles exhibited poor high-temperature stability and thus cannot pin PAG boundaries. As the TiN content in the 60Al steel was lower than that in the 160Al steel at temperatures above 1100°C, the precipitates in 60Al exhibited a weaker pinning effect on grain boundaries compared with those in the 160Al steel. This finding indicates a faster growth rate of normal PAGs in the 60Al steel than in the 160Al steel. Based on the formation sequence shown in Fig. 8(b), the precipitate cores and caps observed in Fig. 3(b) corresponded to the TiN and NbC phases, respectively. This sequence is consistent with the larger size of Ti-rich precipitates and the smaller size of Nb-rich precipitates.

4.2   Coarsening of precipitates after the thermal cycle

4.2.1   Coarsening of pure carbonitride precipitates

According to the theory of Lifshitz, Slyozov, and Wagner, the coarsening of Ti–nitride precipitates can be calculated using Eq. (2) [35–36]:

where ${r}$ and ${{r}}_{\text{0}}$ denote the average and initial radius of the precipitates, respectively, $\gamma$ represents the interphase surface energy between the matrix and precipitate, ${{V}}_{\text{m}}$ represents the molar volume of precipitates (11.53 cm³·mol⁻¹ for TiN precipitates), ${{C}}_{\text{m}}$ indicates the solute concentration in the matrix, and ${{D}}_{\text{m}}$ refers to the diffusivity of solute atoms (cm²·s⁻¹), t (s) and T (K) represent the time and absolute temperature, respectively. As calculated using Eq. (2), the degree of precipitate coarsening was primarily influenced by ${{V}}_{\text{m}}$ and ${{C}}_{\text{m}}$ under given t and T conditions. As the precipitates in both steels were (Ti,Nb)(C,N), both steels were assumed to have the same ${{V}}_{\text{m}}$ value. As the Nb-rich particles failed to pin the PAG boundaries at 1100°C, only the coarsening of Ti-rich particles was considered during calculations. The red lines in Fig. 8(b) indicate the higher ${{C}}_{\text{m}}$ of the 60Al steel than the 160Al steel. This result indicates the slightly elevated coarsening rate of carbonitride precipitates in 60Al steel. Therefore, the (Ti,Nb)(C,N) observed in the CGHAZ of 60Al steel were slightly larger than those of the 160Al steel.

4.2.2   Interfacial reactions between inclusions and matrix for the formation of nano Al₂O₃ particles

According to experimental results, in the CGHAZ of 60Al steel, γ-Al₂O₃ served as a nucleation site for (Ti,Nb)(C,N) and Cu₂S (Fig. 4). From a thermodynamic perspective, if the γ-Al₂O₃ particles act as nucleation sites, γ-Al₂O₃ should form prior to the precipitation of (Ti,Nb)(C,N) and Cu₂S. Fig. 8(a) reveals that during the cooling process of HHIW, Al₂O₃ formed in the 60Al and 160Al steels at temperatures around 1080 and 1264°C, respectively. However, a large number of (Ti,Nb)(C,N) had already precipitated (Fig. 8(b)). Therefore, the Al₂O₃ formed during the cooling process failed to satisfy the thermodynamic requirement. During the heating process, CaO·6Al₂O₃ in the 60Al steel transformed into Al₂O₃ at 1237°C (Fig. 8(a)), and this phenomenon was not observed in the 160Al steel. In the subsequent cooling process of the thermal cycle, TiN particles precipitated again with a decrease in the temperature. Therefore, in the 60Al steel, the Al₂O₃ particles that formed during the heating process met the thermodynamic requirements as nucleation sites for TiN.

Based on the above analysis and the results in Fig. 8(a), during the heating process, the Al₂O₃ (cyan solid line) first transformed into CaO·6Al₂O₃ in the 60Al steel, followed by a retransformation back into Al₂O₃. Therefore, Ca reacted with Al₂O₃ in the oxides, which resulted in the increased CaO content in the oxide and Al diffusion into the matrix (Part I in Fig. 9). Subsequently, with the increase in the temperature, Al replaced Ca in the oxides again and formed Al₂O₃. Although transformations of oxide compositions in solid steel during heating or cooling processes have been previously studied [37–38], little attention has been given to the formation of nanoscale Al₂O₃ particles. Al-deoxidation experiments revealed the nano Al₂O₃ particles formed by O in oxides and dissolved Al in the matrix [39]. In the present paper, the Al₂O₃ nanoparticles were formed by the Al isolated from the oxides and O in the matrix. The transformation of Al₂O₃ into CaO·6Al₂O₃ can be described using Eq. (3), and that of CaO·6Al₂O₃ to Al₂O₃ using Eqs. (4) and (5). In Eq. (5), the elemental Al in the alloy was mainly supplied by the Al₂O₃ in the oxide according to Eq. (3). The Supplementary Information further discusses the formation process of nanoscale alumina.

Fig. 9.  Schematic of transformation of precipitates and PAGs in the CGHAZ of the 60Al and 160Al steels.

DownLoad: Full-Size Img PowerPoint

4.2.3   Precipitation of Cu₂S based on structure mismatch analysis

In the present experiment, the detection of Cu–sulfide in the precipitates extracted from CGHAZ samples of 60Al steel by carbon replica deviated from the predictions obtained through thermodynamic equilibrium calculations. This phenomenon can be explained by the classical nucleation theory. Phase transition free energy per unit volume (${\Delta }{{G}}_{\text{v}}$), which can be expressed using Eq. (6), is one of the primary factors controlling the nucleation energy (${\Delta }{{G}}^{\text{*}}$):

where ${{T}}_{\text{e}}$ denotes the equilibrium temperature; ${\Delta }{T}$ and ${Q}$ are the undercooling below the ${{T}}_{\text{e}}$ and free energy of the solution, respectively. For the solubility product of MX in Fe, ${Q}$ can be evaluated using Eq. (7) and parameter A in Eq. (8) [40]:

where R is the universal gas constant, 8.314 J·mol⁻¹·K⁻¹. The solubilities of MnS and Cu₂S in γ-Fe were determined as follows [40–41]:

Based on the present steel compositions, the calculated ${{Q}}_{\text{MnS}}$ was 83235 J/mol for [%Mn] = 1.55, and ${{Q}}_{{\text{Cu}}_{\text{2}}\text{S}}$ was 286900 J/mol [42]. The calculated equilibrium precipitation temperature (${{T}}_{\text{e}}$) for MnS and Cu₂S reached 1434 and 1257°C, respectively. Fig. 10 shows the variation in the calculated ${\Delta }{{G}}_{\text{v}}$ with the temperature for MnS and Cu₂S in γ-Fe. At temperatures below 1200°C, the ${\Delta }{{G}}_{\text{v}}$ for Cu₂S was larger than that for MnS, which means that Cu₂S exerted a greater driving force for nucleation.

Fig. 10.  Driving force for the nucleation of MnS and Cu₂S at various temperatures.

DownLoad: Full-Size Img PowerPoint

The mismatch theory can accurately reflect the capacity for heterogeneous nucleation. The interfacial energy ($\gamma$), which determines substrate capability for promoting nucleation, was influenced by the lattice mismatch between the substrate and the nucleating parameter (δ) proposed by Bramfitt [43] was used to calculate the mismatch using Eq. (11):

where ${\text{(}{hkl}\text{)}}_{\text{s}}$ refers to a low-index lattice plane of the substrate (nucleating agent), ${\text{(}{hkl}\text{)}}_{\text{n}}$ indicates a low-index lattice plane of the nucleating phase, ${\text{[}{uvw}\text{]}}_{\text{s}}$ means a low-index direction in the ${\text{(}{hkl}\text{)}}_{\text{s}}$ plane, ${\text{[}{uvw}\text{]}}_{\text{n}}$ indicates the low-index direction in the ${\text{(}{hkl}\text{)}}_{\text{n}}$ plane, ${{d}}_{{\text{[}{uvw}\text{]}}_{\text{s}}}^{{i}}$ represents the atomic spacing along ${{[uvw]}}_{\text{s}}$, ${{d}}_{{\text{[}{uvw}\text{]}}_{\text{n}}}^{{i}}$ means the atomic spacing along ${\text{[}{uvw}\text{]}}_{\text{n}}$, and θ is the angle between the ${\text{[}{uvw}\text{]}}_{\text{s}}$ and ${\text{[}{uvw}\text{]}}_{\text{n}}$. The substrate can play a very effective role in heterogeneous nuclei when $\delta$ is less than 6.0%; $\delta$ between 6% and 12% indicates a moderate degree of effectiveness, and $\delta$ above 12% is invalid [43].

According to Eq. (11), the mismatch values between MnS and γ-Al₂O₃ for various interfaces were calculated (detailed results are provided in the Supplementary Information). The mismatch for the (110)_MnS//${\text{(110)}}_{\text{γ-}{\text{Al}}_{\text{2}}{\text{O}}_{\text{3}}}$ interface reached 7.449%, which is smaller than those for other planes. This result suggests that the γ-Al₂O₃ can potentially induce the heterogeneous nucleation of MnS. The lattice parameters of TiN were used for the mismatch calculation between (Ti,Nb)(C,N) and γ-Al₂O₃. The SADP results indicate that ${\text{(1}\overline{\text{1}}\text{0)}}_{\text{γ-}{\text{Al}}_{\text{2}}{\text{O}}_{\text{3}}}$ was a low-index plane that matched with the ${\text{(1}\overline{\text{1}}\text{0)}}_{{\text{Cu}}_{\text{2}}\text{S}}$ and (1$\overline{\text{1}}$0)_(Ti,Nb)(C,N) plane. Therefore, the ${\text{(1}\overline{\text{1}}\text{0)}}_{\text{γ-}{\text{Al}}_{\text{2}}{\text{O}}_{\text{3}}}$//(1$\overline{\text{1}}$0)_(Ti,Nb)(C,N)//${\text{(1}\overline{\text{1}}\text{0)}}_{{\text{Cu}}_{\text{2}}\text{S}}$ interface was applied in $\delta$ calculations. Fig. 11 displays the atomic configurations of the ${\text{(1}\overline{\text{1}}\text{0)}}_{\text{γ-}{\text{Al}}_{\text{2}}{\text{O}}_{\text{3}}}$, (1$\overline{\text{1}}$0)_(Ti,Nb)(C,N) and ${\text{(1}\overline{\text{1}}\text{0)}}_{{\text{Cu}}_{\text{2}}\text{S}}$ planes. The atoms used for $\delta$ calculation between γ-Al₂O₃ and (Ti,Nb)(C,N) or Cu₂S are indicated with red and blue dashed lines in these configurations, respectively. The calculated $\delta$ between γ-Al₂O₃ and (Ti,Nb)(C,N) on the ${\text{(1}\overline{\text{1}}\text{0)}}_{\text{γ-}{\text{Al}}_{\text{2}}{\text{O}}_{\text{3}}}$//(1$\overline{\text{1}}$0)_(Ti,Nb)(C,N) interface was 6.409% (<12%), which suggests that γ-Al₂O₃ can provide effective nucleation conditions for the precipitation of (Ti,Nb)(C,N). The calculated $\delta$ between γ-Al₂O₃ and Cu₂S on the ${\text{(1}\overline{\text{1}}\text{0)}}_{\text{γ-}{\text{Al}}_{\text{2}}{\text{O}}_{\text{3}}}$//${\text{(1}\overline{\text{1}}\text{0)}}_{{\text{Cu}}_{\text{2}}\text{S}}$ interface was 1.088% (<6%), which implies the higher likelihood of heterogeneous nucleation between γ-Al₂O₃ and Cu₂S than that between γ-Al₂O₃ and MnS. The (Ti,Nb)(C,N) can also act as a core for the heterogeneous nucleation of Cu₂S. Therefore, a lower interfacial energy was observed between γ-Al₂O₃ and Cu₂S than between γ-Al₂O₃ and MnS.

Fig. 11.  Atomic configurations of ${{\boldsymbol{(1}}\overline{{\boldsymbol{1}}}}{\boldsymbol{0}})_{\text{γ-}{\bf Al}_{\boldsymbol{2}}{\bf {O}}_{\boldsymbol{3}}}$, ${({\boldsymbol{1}}\overline{\boldsymbol{1}}}{\boldsymbol{0}})_{{\bf {Cu}}_{\boldsymbol{2}}\bf {S}}$, and ${({\boldsymbol{1}}\overline{\boldsymbol{1}}}{\boldsymbol{0}})$_(Ti,Nb)(C,N) planes.

DownLoad: Full-Size Img PowerPoint

Based on the above ${\Delta }{{G}}_{\text{v}}$ and $\delta$ results, the heterogeneous nucleation activation energy for Cu₂S on γ-Al₂O₃ was lower than that for MnS, which implies that Cu₂S was more easily nucleated than MnS at temperatures below 1200°C. Moreover, the good lattice matching relationship promoted the precipitation of (Ti,Nb)(C,N) and Cu₂S on γ-Al₂O₃. Thus, the larger-sized complex precipitates (γ-Al₂O₃–(Ti,Nb)(C,N)–Cu₂S) formed at high temperatures. Considering that TiN is the main component of carbon nitride, complex precipitates were expressed as γ-Al₂O₃–TiN–Cu₂S. Reduction in the amount of independently precipitated (Ti,Nb)(C,N) and the formation of large-sized γ-Al₂O₃–TiN–Cu₂S drastically reduced the pinning force of the precipitates, which led to abnormal grain growth.

4.3   Effects of precipitates on the PAG size and CGHAZ toughness

Fig. 9 shows the transformation of precipitates and PAGs in CGHAZ after the HHIW for the 60Al and 160Al steels. In the CGHAZ, controlling the growth of PAG was crucial, as grain refinement played a crucial role in the enhancement of steel toughness. Experimental evidence indicates that the growth behavior of PAGs was influenced by the presence of Nb- and Ti-rich precipitates. Specifically, during welding, more Nb-rich precipitates in the 60Al steel dissolved at temperatures below 1100°C compared with that in the 160Al steel, which promoted normal PAG growth (Figs. 2 and 8(b)). The normal growth of PAGs was affected by the dissolution and reprecipitation of TiN precipitates during the heating process from 1100 to 1400°C and the subsequent cooling process. When the temperature dropped from the peak temperature of 1400°C, abnormally large grains appeared in the CGHAZ of the 60Al steel but not in the 160Al steel, and this finding was attributed to the formation of γ-Al₂O₃ nanoparticles in the former. These γ-Al₂O₃ particles served as nucleation sites for TiN and Cu₂S, which facilitated the heterogeneous nucleation of TIN and Cu₂S and reduced the amount of independent TiN precipitates. When the complex precipitates γ-Al₂O₃–TiN–Cu₂S were located near the grain boundaries, the driving force for grain growth exceeded the pinning force exerted by precipitates on such areas, which led to abnormal grain growth. Therefore, in the 60Al steel, large PAG grains and abnormal-growth PAGs contributed to the heterogeneity of grain size distribution. The enlargement of grains increased the hardening capacity, which rendered the material nonuniform and potentially reduced its toughness due to the presence of an inhomogeneous microstructure [44]. Cracks in the 60Al steel were more prone to propagate in the grains, which resulted in a decreased impact toughness from 134 to 54 J at −40°C with a decrease in the Al content from 0.016wt% to 0.006wt% (Fig. 9).

5.   Conclusions

The present study examined the characteristics of precipitates and their influence on austenite grain growth and impact toughness in the CGHAZ of Al–Ti–Ca deoxidized shipbuilding steel plates (containing 0.006wt% and 0.016wt% Al) subjected to HHIW at 400 kJ·cm⁻¹. The major conclusions are as follows:

(1) Three types of precipitates formed in the BM of two steels: Type I: cubic (Ti,Nb)(C,N); Type II: precipitate with cubic (Ti,Nb)(C,N) core and Nb-rich cap; Type III: ellipsoidal Nb-rich precipitate. With the increase in the Al content, the ND of Type I precipitates in the matrix increased, which contributed to the suppression of austenite grain growth.

(2) In the low-Al steel (0.006wt% Al), the good lattice mismatch between (Ti,Nb)(C,N) and γ-Al₂O₃ promoted the heterogeneous nucleation of TiN on the γ-Al₂O₃ surface during the welding thermal cycle. As a result, the ND of independently precipitated (Ti,Nb)(C,N) was reduced. This condition drastically lowered the pinning force of the precipitates on austinite grain boundaries.

(3) The low nucleation energy of Cu₂S and excellent lattice mismatch (1.088%) between Cu₂S and γ-Al₂O₃ caused the precipitation of Cu₂S on γ-Al₂O₃ and the formation of large-sized γ-Al₂O₃–TiN–Cu₂S.

(4) As the Al content was decreased, the reduction in the amount of independently precipitated (Ti,Nb)(C,N) and the formation of large-sized γ-Al₂O₃–TiN–Cu₂S triggered the increase in austenite grain size and formation of nonuniform microstructures. Ultimately, the low-temperature impact toughness of the low Al steel was substantially reduced.