熔焊原理-7.1 Ferrite-to-Austenite Transformation in Austenitic Stainless Steel Welds

7.1 Ferrite-to-Austenite Transformation in Austenitic Stainless Steel Welds7.1.1 Primary Solidification ModesThe welds of austenitic stainless steels normally have an austenite (fcc) matrix with varying amounts of δ-ferrite (bcc). A proper amount of δ-ferrite in austenitic stainless steel welds is essential-too much δ-ferrite (≥10 vol %) tends to reduce the ductility, toughness, and corrosion resistance, while too little δ-ferrite (≤5 vol %) can result in solidification cracking.A.Phase DiagramFigure 7.1 The Fe-Cr-Ni ternary system: (a) liquidus surface; (b) solidus surface Figure 7.1 shows the ternary phase diagram of the Fe-Cr-Ni system. The heavy curved line in Figure 7.1a represents the trough on the liquidus surface, which is called the line of twofold saturation. The line declines from the binary Fe-Ni peritecticreaction temperature to the ternary eutectic point at 49Cr-43Ni-8Fe.Alloys with a composition on the Cr-rich (upper) side of this line have δ-ferrite as the primary solidification phase, that is, the first solid phase to form from the liquid. On the other hand, alloys with a composition on the Ni-rich (lower) side have austenite as the primary solidification phase. The heavy curved 1ines on the solidus surface in Figure 7.1b more or less follow the trend of the liquidus trough and converge at the ternary eutectic temperature.Figure 7.2 Schematics showing solidification and postsolidification transformation in Fe-Cr-Ni welds: (a) interdendritic ferrite; (b) vermicular ferrite; (c) lathy ferrite; (d) vertical section ofternary-phase diagram at approximately 70% Fe.The development of weld metal microstructure in austenitic stainless steels is explained in Figure 7.2. The weld metal ferrite can have three different types of morphology: interdendritic (Figure 7.2a), vermicular (Figure 7.2b), and lathy (Figure 7.2c). Figure 7.2d shows a schematic vertical (isoplethal) section of the ternary phase diagram in Figure 7.1, for instance, at 70 wt % Fe and above 1200°C. This has also been called a pseudo-binary phase diagram. The apex (point 1) of the three-phase eutectic triangle (L+γ+δ) corresponds to the intersection between the vertical section and the heavy curved line in Figure 7.1a.The two lower corners (points 2 and 3) of the triangle, on the other hand, correspond to the intersections between the vertical section and the two heavy curved lines in Figure 7.1b.B. Primary Austenite For an alloy on the Ni-rich (left-hand) side of the apex of the three-phase eutectic triangle, austenite (γ) is the primary solidification phase. Thelight dendrites shown in Figure 7.2a are austenite, while the dark particles between the primary dendrite arms are the δ-ferrite that forms when the three-phase triangle is reached during the terminal stage of solidification. These are called the interdendritic ferrite. For dendrites with long secondary arms, interdendritic ferrite particles can also form between secondary dendrite arms.C. Primary Ferrite For an alloy on the Cr-rich (right-hand) side of the apex of the three-phase eutectic triangle, δ-ferrite is the primary solidification phase. The dark dendrites shown in Figure 7.2b are δ-ferrite. The core of the δ-ferrite dendrites, which forms at the beginning of solidification, is richer in Cr (point 4), while the outer portions, which form as temperature decreases, have lower chromium contents. Upon cooling into the (δ+γ) two-phase region, the outer portions of the dendrites having less Cr transform to austenite, thus leaving behind Cr-rich “skeletons” of δ-ferrite at the dendrite cores. This skeletal ferrite is called vermicular ferrite. In addition to vermicular ferrite, primary δ-ferrite dendrites can also transform to lathy or lacy ferrite upon cooling into the (δ+γ) two-phase region, as shown in Figure 7.2c.D. Weld Microstructure Figure 7.3a shows the solidification structure at the centerline of an autogenous gas-tungsten arc weld of a 310 stainless steel sheet, which contains approximately 25% Cr, 20% Ni, and 55% Fe by weight. The composition is on the Ni-rich (left) side of the apex of the three-phase eutectic triangle, as shown in Figure 7.4a, and solidification occurs as primary austenite. The microstructure consists of austenite dendrites (light etching; mixed-acids etchant) and interdendritic δ-ferrite (dark etching; mixed-acids etchant) between the primary and secondary dendrite arms, similar to those shown in Figure 7.2a.Figure 7.3b, on the other hand, shows the solidification structure at the centerline of an autogenous gas-tungsten arc weld of a 309 stainless steel sheet, which contains approximately 23 wt% Cr, 14 wt% Ni, and 63 wt % Fe. The composition lies just to the Cr-rich side of the apex of the three-phase eutectic triangle, as shown in Figure 7.4b, and solidifies as primary δ-ferrite. The microstructure consists of vermicular ferrite (dark etching; mixed-acids etchant) in an austenite matrix (light etching; mixed-acids etchant) similar to those shown in Figure 7.2b. In both welds columnar dendrites grow essentially perpendicular to the teardrop-shaped pool boundary as revealed by the columnar dendrites.Kou and Le quenched welds during welding in order to preserve the as-solidified microstructure, that is, the microstructure before post-solidification phase transformations. For stainless steels liquid-tin quenching is more effective than water quenching because steam and bubbles reduce heat transfer. With the help of quenching, the evolution of microstructure during welding can be better studied. Figure 7.5 shows the δ-ferrite dendrites (light etching; mixed-chloride etchant) near the weld pool of an autogenous gas-tungsten arc weld of 309 stainless steel, quenched in during welding with liquid tin before the δ→γtransformation changed it to vermicular ferrite like that shown in Figure 7.3b. Liquid-tin quenching was subsequently used by other investigators to study stainless steel welds.Figure 7.3 Solidification structure at the weld centerline:(a) 310 stainless steel; (b) 309 stainless steel. Magnification 190×.Figure 7.4 The Fe-Cr-Ni pseudo-binary phase diagrams:(a) at 55 wt % Fe; (b) at 63wt % Fe; (c) at 73 wt % Fe.Figure 7.5 Liquid-tin quenched solidification structure near the pool of an autogenous gas-tungsten arc weld of 309 stainless steel. Magnification 70×. Mixed-chloride etchant.7.1.2 Mechanisms of Ferrite FormationInoue et al. studied vermicular and lathy ferrite in autogenous GTAW of austenitic stainless steels of 70% Fe with three different Cr-Ni ratios. It was found that, as the Cr-Ni ratio increases, the ratio of lathy ferrite to total ferrite does not change significantly even though both increase. A schematic of the proposed formation mechanism of vermicular and lathy ferrite is shown in Figure 7.6. Austenite first grows epitaxially from the unmelted austenite grains at the fusion boundary, and δ-ferrite soon nucleates at the solidification front. The crystallographic orientation relationship between the δ-ferrite and the austenite determines the ferrite morphology after the postsolidification transformation. If the closed-packed planes of the δ-ferrite are parallel to those of the austenite, the δ→γ transformation occurs with a planar δ/γ interface, resulting in vermicular ferrite. However, if the so-called Kurdjumov–Sachs orientation relationships, namely, (-110)δ//(-111)γ and [-1-11]δ//[-1-10]γ, exist between the δ-ferrite and the austenite, the transformation occurs along the austenite habit plane into the δ-ferrite dendrites. The resultant ferrite morphology is lathy, as shown in Figure 7.7. For the lathy ferrite to continue to grow, the preferred growth direction <100> of both δ-ferrite and austenite must be aligned with the heat flow direction.Figure 7.6 Mechanism for the formation of vermicular and lathy ferrite.Figure 7.7 Lathy ferrite in an autogenous gas-tungsten arc weld of Fe-18.8Cr-11.2Ni.7.1.3 Prediction of Ferrite ContentSchaeffler first proposed the quantitative relationship between the composition and ferrite content of the weld metal. As shown by the constitution diagram in Figure 7.8, the chromium equivalent of a given alloy is determined from the concentrations of ferrite formers Cr, Mo, Si, and Cb, and the austenite equivalent is determined from the concentrations of austenite formers Ni, C, and Mn. DeLong refined Schaeffler’s diagram to include nitrogen, a strong austenite former, as shown in Figure 7.9. Also, the ferrite content is expressed in terms of the ferrite number, which is more reproducible than the ferrite percentage and can be determined nondestructively by magnetic means. Figure 7.10 shows that nitrogen, introduced into the weld metal by adding various amounts of N2to the Ar shielding gas, can reduce the weld ferrite content significantly. Cieslak et al., Okagawa et al., and Lundin et al. have reported similar results previously.Figure 7.8 Schaeffler diagram for predicting weld ferrite content and solidification mode.Figure 7.9 DeLong diagram for predicting weld ferrite content and solidification mode.Figure 7.10 Effect of nitrogen on ferrite content in gas-tungsten arc welds of duplex stainlesssteel.Figure 7.11 WRC-1992 diagram for predicting weld ferrite content and solidification mode.The WRC-1992 diagram of Kotecki and Siewert, shown in Figure 7.11, was from the Welding Research Council in 1992. It was modified from the WRC-1988 diagram of McCowan et al. by adding to the nickel equivalent the coefficient for copper and showing how the axes could be extended to make Schaeffler-like calculations for dissimilar metal joining.Kotecki added the martensite boundaries to the WRC-1992 diagram, as shown in Figure 7.12. More recent investigations of Kotecki have revealed that the boundaries hold up well with Mo and N variation but not as well with C variation. Balmforth and Lippold proposed the ferritic-martensitic constitution diagram shown in Figure 7.13. Vitek et al. developed a model FNN-1999 using artificial neural networks to improve ferrite number prediction, as shown in Figure 7.14. Twelve alloying elements besides Fe were considered: C, Cr, Ni, Mo, N, Mn, Si, Cu, Ti, Cb, V, and Co. The model is not in a simple pictorial form, such as the WRC-1992 diagram, because it allows nonlinear effects and element interactions.Figure 7.12 WRC-1992 diagram with martensite boundaries for 1, 4, and 10% Mn.Figure 7.13 Ferritic-martensitic stainless steel constitution diagram containing a boundary for austenite formation and with iso-ferrite lines in volume percent of ferrite.Figure 7.14 Experimentally measured ferrite number (FN) versus predicted FN:(a) FNN-1999; (b) WRC-1992.7.1.4 Effect of Cooling RateA. Changes in Solidification Mode The prediction of the weld metal ferrite content based on the aforementioned constitution diagrams can be inaccurate when the cooling rate is high, especially in laser and electron beam welding. Katayama and Matsunawa, David et al., and Brooks and Thompson have compared microstructures that form in slow-cooling-rate arc welds with those that form in high-cooling-rate, high-energy-beam welds. Their studies show two interesting trends. For low Cr-Ni ratio alloys the ferrite content decreases with increasing cooling rate, and for high Cr-Ni ratio alloys the ferrite content increases with increasing cooling rate. Elmer et al.(33) pointed out that in general low Cr-Ni ratio alloys solidify with austenite as the primary phase, and their ferrite content decreases with increasing cooling rate because solute redistribution during solidification is reduced at high cooling rates. On the other hand, high Cr-Ni ratio alloys solidify with ferrite as the primary phase, and their ferrite content increases with increasing cooling rate because the d Æ g transformation has less time to occur at high cooling rates.Elmer et al. studied a series of Fe-Ni-Cr alloys with 59% Fe and the Cr-Ni ratio ranging from 1.15 to 2.18, as shown in Figure 7.15. The apex of the three-phase triangle is at about Fe-25Cr-16Ni. Figure 7.16 summarizes the microstructural morphologies of small welds made by scanning an electron beam over a wide range of travel speeds and hence cooling rates. At low travel speeds such as 0.1-1mm/s, the cooling rates are low and the alloys with a low Cr-Ni ratio (especially alloys 1 and 2) solidify as primary austenite. The solidification mode is either single-phase austenite (A), that is, no ferrite between austenite dendrites or cells (cellular-dendritic A), or primary austenite with second-phase ferrite (AF), that is, only a small amount of ferrite between austenite dendrites (interdendritic F). The alloys with a high Cr-Ni ratio (especially alloys 5-7), on the other hand, solidify as primary ferrite.The solidification mode is primary ferrite with second-phase austenite (FA), that is, vermicular ferrite, lacy ferrite, small blocks of austenite in a ferrite matrix (blocky A), or Widmanstatten austenite platelets originating from ferrite grain boundaries (Widmanstatten A).Figure 7.15 Vertical section of Fe-Ni-Cr phase diagram at 59% Fe showing seven alloys withCr-Ni ratio ranging from 1.15 to 2.18.Figure 7.16 Electron beam travel speed (cooling rate) versus composition map of microstructural morphologies of the seven alloys in Figure 7.15 (A and F denote austenite and ferrite, respectively). The solid lines indicate the regions of the four primary solidification modes, while the dashed lines represent the different morphologies resulting from postsolidification transformation from ferrite to austenite.At very high welding speeds such as 2000mm/s, however, the cooling rates are high and the alloys solidify in only the single-phase austenite mode (A) or the single-phase ferrite mode (F). An example of the former is the alloy 3 (about Fe-24.75Cr-16.25Ni) shown in Figure 7.17a. At the travel speed of 25mm/s (2×103 °C/s cooling rate) the substrate solidifies as primary austenite in the AF mode, with austenite cells and intercellular ferrite. At the much higher travel speed of 2000mm/s (1.5×106°C/s cooling rate) the weld at the top solidifies as primary austenite in the A mode, with much smaller austenite cells and no intercellular ferrite (cellular A). An example of the latter is alloy 6 (about Fe-27.5Cr-13.5Ni) shown in Figure 7.17b. At 25mm/s the substrate solidifies as primary ferrite in the FA mode, with blocky austenite in a ferrite matrix. At 2000mm/s the weld at the top solidifies asprimary ferrite in the F mode, with ferrite cells alone and no austenite (cellular F).Figure 7.16 also demonstrates that under high cooling rates an alloy that solidifies as primary ferrite at low cooling rates can change to primary austenite solidification. For instance, alloy 4 (about Fe-25.5Cr-15.5Ni) can solidify as primary ferrite at low cooling rates (vermicular F) but solidifies as primary austenite at higher cooling rates (intercellular F or cellular A).Another interesting point seen in the same figure is that at high cooling rates alloy 5 can solidify in the fully ferritic mode and undergoes a massive (diffusionless) transformation after solidification to austenite (massive A). Under very high cooling rates there is no time for diffusion to occur.Figure 7.17 Microstructure of the low-cooling-rate substrate (2 × 103°C/s) and thehigh-cooling-rate electron beam weld at the top: (a) alloy 3 in Figure 7.15; (b) alloy 6.B. Dendrite Tip Undercooling Vitek et al. attributed the change solidification mode, from primary ferrite to primary austenite, at high cooling rates to dendrite tip undercooling. Brooks and Thompson explained this under-cooling effect based on Figure 7.18. Alloy C0solidifies in the primary ferrite mode at low cooling rates. Under rapid cooling in laser or electron beam welding, however, the melt can undercool below the extended austenite liquidus (C Lg), and it becomes thermodynamically possible for the melt to solidify as primary austenite. The closer C0 is to the apex of the three-phase triangle, the easier sufficient undercooling can occur to switch the solidification mode from primary ferrite to primary austenite.Figure 7.18 Vertical section of Fe-Cr-Ni phase diagram showing change in solidification fromferrite to austenite due to dendrite tip undercooling.Kou and Le made autogenous gas–tungsten arc welds in 309 stainless steel, which has a composition close to the apex of the three-phase triangle, as shown in Figure 7.4b.At 2mm/s (5ipm) welding speed, primary ferrite was observed across the entire weld (similar to that shown in Figure 7.3b). At a higher welding speed of 5mm/s (12ipm), however, primary austenite was observed along the centerline, as shown in Figure 7.19. Electron probe micro-analysis (EPMA) revealed no apparent segregation of either Cr or Ni near the weld centerline to cause the change in the primary solidification phase. From Equation (6.3) the growth rate R = Vcosα,where α is the angle between the welding direction and the normal to the pool boundary. Because of the teardrop shape of the weld pool during welding (Figure 2.15), α drops to zero and R increases abruptly at the weld centerline. As such, the cooling rate (GR) increases abruptly at the weld centerline, as pointed out subsequently by Lippold.Elmer et al. calculated the dendrite tip undercooling for the alloys in Figure 7.16 under various electron beam travel speeds. An undercooling of 45.8°C was calculated at the travel speed of 175mm/s, which is sufficient to depress the dendrite tip temperature below the solidus temperature (Figure 7.15). This helps explain why alloy 4 can change from primary ferrite solidification at low travel speeds to primary austenite solidification at much higher travel speeds.Figure 7.19 Weld centerline austenite in an autogenous gas-tungsten arc weldof 309 stainless steel solidified as primary ferrite.7.1.5 Ferrite Dissolution upon ReheatingLundin and Chou observed ferrite dissolution in multiple-pass or repair austenitic stainless steel welds. This region exists in the weld metal of a previous deposited weld bead, adjacent to but not contiguous with the fusion zone of the deposited bead under consideration. Both the ferrite number and ductility are lowered in this region, making it susceptible to fissuring under strain. This is because of the dissolu tion of δ-ferrite in the region of the weld metal that is reheated to below the γ-solvus temperature. Chen and Chou reported, in Figure 7.20, a significant ferrite loss in a 316 stainless steel weld subjected to three postweld thermal cycles with a 1250°C peak temperature, which is just below the γ + δ two-phase region of about 1280-1425°C.Figure 7.20 Effect of thermal cycles on ferrite content in 316 stainless steel weld: (a) as welded;(b) subjected to thermal cycle of 1250°C peak temperature three times after welding.。