ly develop this curve and more accurately identify the peak in cracking susceptibility. The effect of adding 2 and/or 4 wt-% molybdenum to alloys with 2, 4, or 6 wt-% Nb at the same pin length is shown in Fig. 4. The pin length, 1.375 in., selected for comparison of the Mo effect was one in which all three niobium compositions exhibited cracking prior to molybdenum addition. At the 2 and 4 wt-% Nb levels, there appears to be little effect from the Mo addition, and it may actually increase susceptibility in the 2% Nb alloy. A significant effect of the Mo addition can be seen in the 6Nb samples. The 6Nb4Mo composition results included pins that did not crack compared to 6Nb0Mo pins, which exhibited significant cracking. Thus, there appears to be a clear beneficial effect of Mo additions at the higher Nb level. The large error among circumferential cracking values in pins of the same composition makes it difficult to make definitive comparisons where small trends may appear, as for the 2Nb and 4Nb compositions. The data suggest that the 4Mo samples experienced less cracking regardless of niobium composition, but the error in the data will require more testing before definitive conclusions can be made. Characterization As shown in Fig. 5, there is a clear increase in the fraction eutectic as the Nb content is increased in the Alloy 690 base alloy. In the base alloy microstructure, virtually no eutectic phase is detectable, and only a slight increase is observed when 2 wt-% Nb is added, as evidenced by the dark particles at the solidification subgrain boundaries. At 4 wt-% Nb, considerable eutectic constituent is evident, and at 8 wt-% Nb, the eutectic constituent is almost continuous. Evidence of crack backfilling was detected in the 4, 6, and 8 wt-% niobium pins as well as in the Nb + Mo alloys. Optical and SEM micrographs that demonstrate the backfilling effect are shown in Figs. 6–8. With the exception of the 2 wt-% Nb alloys, all micrographs shown are from pin lengths at the cracking threshold, where no surface cracking was observed. In the 2 wt-% Nb alloy, evidence of very minor backfilling was only observed in pins containing cracks and is at such a low level that the contribution to crack healing is minimal. The backfilled cracks in the 8Nb0Mo pins appeared to contain fewer voids (cracks) and are more continuous than in the 4Nb0Mo sample. The backfilled cracks in the 8Nb0Mo sample run almost continuously through the sample, indicating an extensive crack healing effect in this alloy. In the 4Nb0Mo alloy, many open cracks are evident, indicating that insufficient liquid of eutectic composition is available to heal the cracks that form. Backfilled cracks were difficult to find in the lower niobium, 2 wt-% compositions, although some were observed in the 2Nb4Mo alloy, as shown in Fig. 7. While the beginnings of a small network of backfilled cracks appear, many voids remain, and the pin exhibited surface cracking. The addition of 4 wt-% Nb to this alloy increases the fraction eutectic from 1.4 to 7.5 vol-% (Table 2), leading to an increase in cracking susceptibility, as shown previously in Fig. 4. Crack healing via a backfilling mechanism is clearly dependent on the Nb content. While some crack healing may occur in the lower 2Nb and even 4Nb compositions, the level of backfilling is not sufficient to have a beneficial effect. In fact, the effect may actually be detrimental in that the solidification temperature range of the 2Nb and 4Nb alloys has increased, but the amount of eutectic formed is not enough to adequately fill voids and cracks to prevent cracking. Again, it is important to point out that crack healing is a phenomenon that may not always increase resistance to solidification cracking during welding. High levels of restraint may overwhelm any crack healing effects, even if the fraction eutectic is high. This is shown in Fig. 2, where at the longest pin lengths significant cracking is observed in the 6Nb0Mo and 8Nb0Mo alloys. The SEM micrographs in Fig. 8 better illustrate the difference in crack backfilling observed based on differences in Nb. In all cases, the backfilling occurs along solidification grain boundaries where preexisting cracks were present. In the 4Nb0Mo sample, the width of the backfilled liquid is relatively WELDING RESEARCH thin with several “breaks” along the boundary. As Nb increases to 6 and 8 wt-%, the degree of backfilling increases and the backfilled boundaries are much wider. While the backfilling network in the 8Nb0Mo sample is quite extensive, some small voids are still present internally, even though no cracking was observed on the surface. Based on prior research, cracks initiate at the surface and propagate toward the center and pins that are at or below the threshold do not contain internal cracks (Refs. 52, 53). Thus, for the 8Nb0Mo material, the crack that formed at the surface was nearly completely healed by backfilling. This backfilling results in the substantial drop in cracking susceptibility in the 6 and 8 wt-% Nb alloys relative to the 4 wt-% alloy, as shown in Fig. 2. Based on the data in Table 2, maximum cracking at 4 wt-% Nb occurs when there is ~7.5 vol-% eutectic present. By increasing the fraction eutectic above 15 vol-%, a significant reduction in cracking occurs, as shown in Fig. 2. SEM/EDS spot analyses were performed to determine the composition in the backfilled cracks for 4Nb0Mo, 6Nb0Mo, 8Nb0Mo, 2Nb2Mo, 2Nb4Mo, 6Nb2Mo, and 6Nb4Mo compositions. SEM micrographs and the location of the spot analysis for the 8Nb0Mo and 6Nb4Mo alloys are shown in Figs. 9 and 10, respectively. The “nominal” composition refers to the composition of the Alloy 690 base alloy used in this study prior to niobium and molybdenum additions, and the “adjusted” composition is the calculated composition of the new alloys taking the additions of Nb and Mo into account. The partially backfilled crack (spots 1–3) is enriched in Nb to an average of 24.34 wt-% while the matrix (fcc gamma phase) is depleted to an average of 3.56 wt-% Nb. These matrix values represent locations away from the solidification boundary where no eutectic constituent is present. Thermo- Calc™ predicted an average niobium concentration of 23 wt-% for the eutectic reaction that occurs in this alloy, so the EDS data is in good agreement with the computed value. It is important to point out that the measured values are average values. The EDS spot collected data from JULY 2016 / WELDING JOURNAL 235-s
Welding Journal | July 2016
To see the actual publication please follow the link above