Influence of the Heat Input on the Dendritic Solidification Structure and the Mechanical Properties of 2.25Cr-1Mo-0.25V Submerged-Arc Weld Metal

The investigated 2.25Cr-1Mo-0.25V multilayer all-weld metal specimens were produced with single-wire SAW using two 24 mm thick 2.25Cr-1Mo-0.25V steel plates and a 2.25-1Mo backing strip. A root gap of 22 mm and a bevel angle of 0° were chosen. Table 1 shows the welding parameters in detail.

The heat input H for submerged-arc welding where the thermal efficiency factor equals approximately 1 is defined as the product of welding current I and voltage U divided by the welding speed v according to Eq 1 (Ref 12).

Within this study, the heat input was varied only by changing the welding speed. Table 2 gives an overview of the details of the investigated weld metals, labeled with A to D. The table shows that weld metal D with the highest heat input of 32 kJ cm⁻¹ has the same amount of layers as weld metal C with a lower heat input, but exhibits a higher weld reinforcement. In addition, the calculated t_8/5-times are shown for the last weld beads, which are defined as the time for cooling the weld metal from 800 to 500 °C. The calculation was conducted by means of the voestalpine welding calculator according to DIN EN 1011-2 (Ref 25, 26).

Figure 1 shows the calculated t_8/5-times depending on the applied welding speed and the heat input. A lower welding speed during SAW causes a bigger melt pool and therefore increases the t_8/5-time and thus prolongs the time for solidification.

The mean values of the chemical compositions of the four weld metals A, B, C and D are given in Table 3. Furthermore, the values of the Bruscato (X) and the Watanabe factor (J), which were calculated according to Eq 2 and 3, are provided (Ref 27).

The value of the J Factor is below the limit of J ≤ 100 and the value of the X Factor is below the limit of X ≤ 15, which according to API Recommended Practice 934-A (Ref 27) minimizes a low alloyed CrMo(V) steels and weld metals risk to temper embrittlement.

Beside the elements shown in Table 3, the weld metals also contain P and S in small amounts (≤ 100 ppm).

To contrast the primary and secondary dendrites of the welds in their as-welded state, the all-weld metal specimens were cut normal to the welding direction (cross section) and ground with SiC paper. After grinding, the samples were polished with 3 µm and 1 µm diamond suspension and treated with Béchet-Beaujard etchant [100 ml cold saturated filtered picric acid, 1 g natriumdodecyl-benzolsulfonat, 1-2 drops hydrochloric acid (Ref 28-31)]. Further information about the application of Béchet-Beaujard etchant on creep-resistant 2.25Cr-1Mo-0.25V weld metal can be found elsewhere (Ref 32).

To analyze the mean primary (λ_prim) and mean secondary dendrite spacings (λ_sec), light optical microscopy (LOM) analysis was conducted on the etched cross sections. For robust statistics, 5 LOM pictures within the last weld bead were taken for each all-weld metal specimen, in order to analyze a sufficient number of primary and secondary dendrites. For this purpose, a LOM Zeiss Axio Imager M1m was used. Figure 2 shows a schematic drawing of the weld metals cross section. The LOM pictures were taken in the last weld bead of the multilayer weld metals in order to exclude influences resulting from reheating from further weld beads and to ensure a sharply contrasted dendritic solidification structure, see green colored and blue outlined area in Fig. 2. The green arrows exemplarily show the direction of the dendritic solidification. The dendrites are oriented approximately parallel to the surface of the all-weld metal cross section.

For the measurement of λ_prim and λ_sec, the LOM pictures were taken with a magnification factor of 50 and 100, respectively. Figure 3(a) shows a schematic drawing of the definition of λ_prim and λ_sec. It has to be emphasized that λ_prim corresponds to the dendrite trunk spacing and λ_sec corresponds to the primary dendrite arm spacing (Ref 33). As the dendrites are oriented approximately parallel to the etched cross section, the evaluation of λ_prim and λ_sec was conducted with the linear intercept method using Olympus Stream Motion image processing software. Here, a line is placed normal to a number of n primary or secondary dendrites, which reaches from the middle of the first dendrite to the middle of the n^th dendrite. The mean dendrite spacing is then calculated by dividing the length of this measurement line by the number of crossed dendrites n minus 1 (Ref 34). The graphic in Fig. 3(b) schematically shows how λ_sec is determined from the LOM images using the linear intercept method. For each specimen and etchant, more than 200 primary and secondary dendrites were analyzed to determine the mean dendrite spacings λ_prim and λ_sec.

Macro-images of the Nital (3% HNO₃ in ethanol) etched weld metals were taken in order to investigate the influence of the heat input on the size of the weld beads.

For the determination of the prior austenite grain size, the all-weld specimens were heat treated at 700 °C for 10 h, ground, polished and etched with Béchet-Beaujard etchant. Ten LOM dark field images were taken in the HAZ between the last and the subjacent weld bead (magnification of 500) and in the last weld metal (magnification of 200) of each all-weld metal specimen. The grain size analysis was conducted with an Olympus Stream Motion image processing software using the linear intercept method. Therefore, a pattern of ten horizontal measurement lines was chosen and the points of intersection with prior austenite grain boundaries were set manually for each LOM image.

A scanning electron microscope (SEM) equipped with a FEI Versa 3D Dual DualBeam and an electron backscatter diffraction (EBSD) system from EDAX Hikari XP was used to investigate the microstructure within the last weld bead, see Fig. 2.

Additionally, S and P distribution mappings of weld metal D were performed with a Cameca SX 100 micro-probe with five spectrometers. In order to investigate a cross section with a sufficient number of primary dendrite trunks, an area of 256 times 256 µm² was analyzed using a step size of 2.5 µm and 80 ms dwell time. The acceleration voltage was 15 kV. The all-weld metal specimen was in the as-welded state without any heat treatment.

For EBSD experiments, the specimens were ground and polished with diamond suspension (3 and 1 µm) and etched with Nital before the final polishing step using Struers OP-S colloidal Si suspension. The specimen tilt was 70°, the scanning mode was analytical mode, the working distance was 17 mm, the acceleration voltage was set to 30 kV, the step size was 0.2 µm, and 6 × 6 binning was used. To determine the bainitic grain size, an area of 160 × 160 µm² was mapped to ensure the analysis of a sufficient number of grains. The evaluation of the EBSD scans was conducted with the software OIM data Analysis 8 (EDAX inc.) and Origin Pro 2017G.

One sample for tensile testing according to DIN EN ISO 5178:2019 (Ref 35) and five Charpy V-notch impact testing samples according to DIN EN ISO 148-1:2009 (Ref 36) were machined out of the four welded joints. The tensile specimens were taken from the middle of the weld seam with the longitudinal axis in welding direction (all-weld metal specimens), and the Charpy V-notch specimens were machined out normal to the welding direction (cross-weld specimens). According to DIN EN ISO 9016:2012 (Ref 37), the position of the V-notch was VWT 0/7.0 which means that the notch was located in the middle of the weld metals cross section where the left and right weld beads overlap. Before mechanical testing, the weld metals did undergo a dehydrogenation treatment (soaking at 350 °C for 4 h). To relief the stresses from the welding operation, the weld metals for the tensile test were post-weld heat treated at 705 °C for 8 h before they were machined and tested at room temperature. The Charpy V-notch specimens from the same heat-treated weld metals were tested at − 30 °C.

Additionally, Vickers hardness measurements were conducted on the all-weld metal specimens after a PWHT of 8 and 32 h at 705 °C, respectively. Ten HV10 macro-hardness indents were made on a line in vertical and horizontal direction on the weld metal cross sections.

According to ASME Boiler and Pressure Vessel Code Section VIII Division 2:2017 (Ref 38), stress rupture tests were conducted at 540 °C and 210 MPa. All stress rupture tests were performed after a PWHT at 705 °C and an annealing time of 32 h.

Figure 14 shows the results of the HV10 macro-hardness measurements in horizontal and vertical direction of the all-weld metal specimens produced with different heat inputs of 20, 24, 28 and 32 kJ cm⁻¹. The specimens were heat treated at 705 °C for 8 and 32 h, respectively. Although the hardness values show some deviations depending on the position of the indent on the weld metals cross section, there appears to be no connection between the macro-hardness and the heat input during SAW. Nonetheless, it can be seen that the PWHT-time has a major impact on the macro-hardness of the weld metal. After a short PWHT-time of 8 h, the hardness of the all-weld samples is significantly higher than after the longer PWHT-time of 32 h.

Tensile tests at room temperature and Charpy V-notch impact tests at − 30 °C were conducted for each of the four weld metal specimens. Figure 15(a) reveals that there is no significant change of the 0.2%-yield strength (YS) and the ultimate tensile strength (UTS) with increase in heat input during welding. No correlation between elongation at fracture and reduction of area with the heat input was found either. Figure 15(b) shows that even though there is no influence of the heat input on the ductility of the weld metal, the heat input strongly influences the weld metal’s toughness. After a PWHT of 8 h at 705 °C, a strong decrease in impact energy with increase in heat input from 20 to 28 kJ cm⁻¹ was observed. Concerning the range from 28 to 32 kJ cm⁻¹, the impact toughness remains nearly constant.

Figure 16 shows the normalized stress rupture time of the all-weld metal specimens A to D versus the heat input during SAW. The normalized stress rupture time equals the stress rupture time of the respective all-weld metal specimen divided by the minimum stress rupture time of the whole data set. With increase in heat input, the stress rupture time increases linearly between weld metal A (20 kJ cm⁻¹ heat input) and weld metal C (28 kJ cm⁻¹ heat input). A further increase in heat input from 28 to 32 kJ cm⁻¹ (weld metal D) did not lead to a further enhancement of the stress rupture time. The behavior of stress rupture time and Charpy impact energy is reverse; with higher heat input during SAW the stress rupture time increases and the Charpy impact energy decreases simultaneously. Between 28 and 32 kJ cm⁻¹ heat input, both the impact energy and the stress rupture time stay nearly constant.

The LOM investigation of the Béchet-Beaujard etched all-weld metals fabricated with different heat inputs revealed that a higher heat input during SAW evokes a coarser dendritic solidification structure. Even though the change in heat input was only moderate, it affected both mean dendrite spacings λ_prim and λ_sec.

The increase in dendrite spacing with higher heat input during welding is a consequence of the lowered cooling rate which is defined as the product of temperature gradient G and growth rate R, see green arrow in schematic drawing in Fig. 17 (Ref 14, 15, 41).

This decrease in cooling rate (G * R) with increase in heat input leads to a prolonged solidification time and thus more time for diffusion. Furthermore, a longer solidification time enables the reduction of surface energy via the growth of larger dendrite arms at the expense of smaller dendrite arms which exhibit more surface area per unit volume (Ref 14, 15). The relatively high standard deviation of the determined mean dendrite arm spacings λ_prim and λ_sec is a result of the inhomogeneous distribution of cooling rates within the weld bead. This is due to the fact that the temperature gradient is higher at the fusion line than at the centerline of the weld bead and that vice versa the growth rate is much higher at the centerline compared to the fusion line of the weld bead. Therefore, the cooling rate is highest at the weld beads centerline and lowest at the fusion line and changes gradually between these lines. As a consequence, the solidification structure is the finest in welding direction and gets enlarged with increase in distance from the weld metals centerline, leading to a variety of dendrite spacings within a single weld bead caused by local changes in the cooling rate (Ref 14).

Further on, the different heat inputs during SAW did not lead to a significant alteration of the ratio G/R within the last weld bead, as no change in solidification mode from columnar to equiaxed dendritic was visible via LOM investigation.

An increase in heat input during welding is known to possibly influence the local concentration of impurity elements such as S, P or B, which tend to segregate to the grain boundary during solidification. There, they can form compounds with low melting points which widen the solidification temperature range and lead to an increased mushy zone which is prone to solidification cracking (Ref 14, 15, 41, 42). The lower the content of these residual elements is and the more homogeneous they are distributed within the grain or sub-grain area; the less harmful they are and the lower is the risk of solidification cracking (Ref 15). The analysis of the 2.25Cr-1Mo-0.25V all-weld metal specimen produced with the highest heat input of 32 kJ cm⁻¹ revealed that despite the broad solidification structure (increased dendrite spacing) still no significant variations of the S and P content were recognizable across the dendritic solidification structure by means of micro-probe analysis. Therefore, it is assumed that increasing the heat input up to 32 kJ cm⁻¹ by reducing the welding speed does not significantly increase the weld metals risk for solidification cracking.

The tensile testing of the four all-weld metal samples revealed that despite the broader and thicker weld beads generated via welding with higher heat input, no influence on the strength and ductility was found.

In contrast, there was a strong correlation between the heat input during SAW and the weld metal’s toughness. A lower heat input, realized by welding with higher welding speed, improves the weld metal’s toughness. This effect can be attributed to the bead sequence consisting of a higher number of weld beads compared to a weld metal produced with a higher heat input. Each weld bead consists of a fine upper bainitic microstructure with large prior austenite grains which are elongated in the direction of the solidification (columnar grained zone). Between the individual weld beads of the multilayer weld, metal fine grained HAZs (equiaxed grained zone) have formed due to reheating by further weld beads throughout the welding process. A finer bainitic microstructure is beneficial for the weld metals toughness, as more interfaces provide more positions for stress and strain redistribution. Furthermore, the grain size analysis showed that the mean bainitic and prior austenite grain size increases with higher heat input during SAW for both investigated zones, the last weld bead and the HAZ.

While the average bainitic and prior austenite grain size was found to correlate with the heat input during SAW, the variation of heat input did not affect the microstructure constituents of the weld metals. Despite the broad range of t_8/5-times obtained by varying the heat input, the microstructure in the as-welded state was fully upper bainitic for the four investigated all-weld metal specimens. So the decrease in weld metal impact toughness with higher heat input cannot be attributed to the negative influence of a certain microstructure constituent which forms at lower cooling rates.

It is therefore concluded that solely the finer microstructure is responsible for the higher toughness when welding is performed with lower heat input. Weld metals produced with lower heat input consist of a higher number of weld layers with less thick weld beads. Hence, the microstructure changes more often between coarse grained weld beads and fine grained HAZs between these weld beads. As a result, the Charpy V-notch specimens machined out of the weld metals produced with lower heat input contain a higher amount of fine grained HAZ per unit volume compared to the ones welded with higher heat input, see Fig. 18.

The higher proportion of fine grained zone and the finer bead sequence, which causes a more frequent change between coarse and fine grained microstructure, is beneficial for the weld metals toughness. Wenkai et al. (Ref 17) found a similar behavior during Charpy impact testing of narrow-gap metal active gas welded AISI 326LN ASS plates. They experienced that cross-weld specimens with the V-notch positioned in the cross section (normal to welding direction, many weld beads) absorbed more energy during Charpy testing than specimens with the V-notch on top of their surface (in welding direction, single weld bead). This anisotropy in the cryogenic impact performances was found for all investigated weld metals joined with different heat inputs and was explained by the anisotropic dendritic microstructure. However, for the submerged-arc weld metal investigated within this study, the effect of decreasing toughness with increase in heat input is not visible for weld metal D (32 kJ cm⁻¹). This is assumed to be linked to its bead sequence which is similar to weld metal C (28 kJ cm⁻¹). Despite the higher heat input, weld metal D exhibits the same amount of weld layers as weld metal C, causing no significant difference of the amount of fine grained zones between these two specimens. The lacking difference in impact toughness of weld metal C and D leads to the assumption that the weld metal’s bead sequence, and therefore, the change between coarse and fine grained zones is more decisive for the weld metals toughness than solely the bainitic grain size.

Nonetheless, when comparing Charpy impact energies, further influencing factors which are responsible for the relatively large scattering of the values should be discussed. In multilayer weld metal, the exact position of the V-notch shows great influence on the impact energy during Charpy V-notch testing, see schematic drawing in Fig. 19. To eliminate the influence of a finer bead sequence, only Charpy V-notch specimens with the same heat input (same amount of weld layers) are compared in Fig. 19. Ideally, the V-notch should be positioned in the middle of the weld metal where the individual weld beads cross each other, see green mark in Fig. 19. Then, the notch lies in a zone, which was reheated several times and therefore exhibits a high amount of fine grained HAZs. Such a frequent change between fine and coarse grained regions has been proven to be beneficial for toughness. In comparison, if the notch lies right to the ideal position (red mark in Fig. 19), where the left and right weld beads do not cross, the amount of fine grained HAZ in the tested area is lower and there is a less frequent change between coarse grained weld beads and HAZs. Equally, when the notch lies left to the ideal position (blue mark in Fig. 19), the testing area also does not lie in the region of crossing weld beads anymore, and therefore, the amount of fine grained HAZ again is lower, which has also a decreasing effect on the impact toughness. Depending on the exact position of the notch on the cross section of the weld seam, and whether it lies in the ideal position (Fig. 19—green notch), higher or lower impact energy values are determined by means of Charpy notch testing. This leads to quite large scattering of the impact energy values as the region where the left and right weld beads cross each other (highest amount of fine grained zone) might slightly shift over the length of the whole weld seam.

The influence of the heat input on the stress rupture time is contrary to its influence on the weld metal’s toughness. An increasing heat input during SAW of 2.25Cr-1Mo-0.25V alloy leads to an enhanced stress rupture time which indicates that a higher heat input is beneficial for the weldments creep resistance. Whereby similar to the toughness, the dependence of the stress rupture time on the heat input is strongly pronounced until a heat input of 28 kJ cm⁻¹ (weld metal C) is reached and less pronounced for the highest heat input of 32 kJ cm⁻¹ (weld metal D). This curve characteristic can also be traced back to the bead sequence of the weld metal specimens as the number of weld layers decreases from weld metal A (20 kJ cm⁻¹) to weld metal C (28 kJ cm⁻¹) and stays constant between weld metal C to weld metal D (32 kJ cm⁻¹). This entails an approximately same amount of coarse grained zone which is beneficial for the materials behavior at elevated temperatures and therefore a similar stress rupture behavior was determined for weld metal C and D. Furthermore, the increase in bainitic grain size with increasing heat input in the last weld bead and the HAZ between the last and the subjacent weld bead leads to the assumption that the creep rupture strength increases with the size of the bainitic structure, which was also demonstrated on the size of martensite variants of differently heat-treated 9% Cr CrMoV-steels (Ref 43, 44).

Finally, it has to be emphasized that for creep-resistant 2.25Cr-1Mo-0.25V SAW weld metal, stress rupture behavior and toughness have proven to be contrary. Therefore, to obtain an optimal behavior at creep conditions a beneficial compromise between good creep strength and sufficient toughness has to be found by choosing the optimal welding parameters and heat treatment.