The Influence of Molybdenum on the Microstructure of Stainless Steel ...
IIW-1749-06 IX-2202-06 IX-H-647-07
The Influence of Molybdenum on the Microstructure of Stainless Steel Welds Timothy Anderson Lehigh University Advisor: Prof. J.N. DuPont
Abstract The excellent corrosion resistance of superaustenitic stainless steel (SASS) alloys has been shown to be a direct consequence of high concentrations of Mo. The presence of Mo can have a significant effect on the microstructural development of welds in these alloys. In this research, the microstructural development of welds in the Fe-Ni-Cr-Mo system was analyzed over a wide variety of Cr/Ni ratios and Mo contents. The system was first simulated by construction of multi-component phase diagrams using the CALPHAD technique. Data gleaned from vertical sections of these diagrams were inserted in the compositional range of a liquidus projection to produce diagrams that can be used as a guide to understand the influence of composition on microstructural development. A large number of experimental alloys were then prepared via arc-button melting for comparison with the diagrams. Each button was characterized using various microscopy techniques. The expected δ-ferrite and γ-austenite phases were accompanied by martensite at low Cr/Ni ratios and by σ-phase at high Mo contents. The results were used to construct a map of expected phase transformation sequence and resultant microstructures as a function of composition. Electron microprobe measurements of selected alloys were performed to compare the distribution of Mo solute between different solidification modes. Severe Mo microsegregation was observed in alloys that solidified directly as austenite; this was attributed to the low diffusivity of Mo in austenite. The level of Mo microsegregation could be significantly reduced in alloys that
solidified as δ-ferrite first and subsequently transformed to austenite by a solid state transformation. Magnetic ferrite measurements were also used to produce quantitative relationships between alloy composition and ferrite content, and the results were then plotted on the WRC-1992 diagram. The results of this work provide a working guideline for future base metal and filler metal development of these alloys. Background SASS alloys with high Mo concentrations are used in a wide variety of applications that require good toughness and corrosion resistance. Unfortunately, the corrosion resistance of welds is never as good as the corresponding wrought base metal [1]. The microstructural development and resultant corrosion resistance of austenitic alloys in this compositional regime has been shown to be strongly controlled by the microsegregation of Mo to the intercellular regions [2]. The ensuing depletion of Mo in the cell core results in preferential corrosive attack [3]. Alloys that solidify as primary austenite are also susceptible to solidification cracking caused by the accumulation of tramp elements (such as P and S) and concomitant formation of low melting point liquid films in the intercellular regions [4].
Previous studies have also shown that the
microsegregation of Mo will proceed over a wide range of welding conditions and compositions [2,5,6]. Although high Mo nickel base filler metals can be used to help mitigate this problem, the microsegregation problem still persists, and careful control of welding variables must be exercised for close control over weld metal composition. In contrast, it is well known that stainless steels that solidify as ferrite exhibit significant backdiffusion and good resistance to solidification cracking due to high solubility of tramp elements [4]. Thus, identifying high Mo SASS alloys that solidify first as ferrite and subsequently transform to austenite via a solid state reaction may help eliminate problems associated with solidification cracking and microsegregation-induced corrosion. A wide range of microstructures are possible for near-eutectic Mo-bearing stainless steels, owing to a variety of solidification modes and a possible solid-state phase transformations that occur from the ferrite to the austenite phase. Multi-component phase diagrams provide an efficient way for understanding microstructure formation and serve as a useful tool for alloy development. The presentation of multi-component phase
diagrams has commonly been accomplished through the use of two different displays: the liquidus projection and the pseudo-binary vertical isopleth. The liquidus projection is capable of displaying the primary solidification mode and terminal solidification reactions over a wide range of compositions, but is unable to describe subsequent solidstate phase transformations that may further influence the final microstructure. The vertical isopleth, while being able to describe post-solidification behavior, is hampered by a slim range of concurrent alloy compositions. For the quaternary Fe-Cr-Ni-Mo system being studied in this research project, an attempt was made to form a composite type diagram that would communicate the microstructural development behavior of a greater number of alloys concurrently.
Efforts were concentrated to combine the
strengths of both diagrams in order to bound the regions of different solidification mode. Thus, the objectives of this research were to design a diagram for phase prediction in the Fe-Ni-Cr-Mo system and to use those diagrams to locate compositions that can experience a solid-state transformation to austenite after primary ferrite solidification. A matrix of alloys were constructed to validate the diagrams and to compare the concentration profiles of Mo solute in austenite derived by either solidification or solidstate transformation.
Experimental Procedure A large matrix of vertical isopleths was constructed of the Fe-Cr-Ni-Mo system using the CALPHAD software Thermo-Calc [7] in conjunction with the Iron Alloy Database [8]. Only the ferrite and austenite phases were included in the calculations, and a range of temperatures was used that would span from the fully liquid state to any solidstate phase-transformations that might transpire. Vertical isopleths with constant Mo additions of 0, 2, 4, 6, 8, and 10 wt% were constructed for several Ni+Cr contents. The software package allows specific portions of a generated diagram to be enlarged; this feature was used to acquire the specific temperature and composition of four special points of interest: a. the peak of the eutectic triangle b. the γ-side vertex of the eutectic triangle c. the δ-side vertex of the eutectic triangle
d. the maximum solubility of Cr in the austenite phase Figure 1 illustrates the locations of these points of interest on a sample isopleth. Further isopleths were assembled in order to ascertain these same locations at the terminating edges of the phase diagram, where either no Cr or Ni is present. In addition, liquidus projections were calculated at each Mo content in order to validate the locations of the eutectic as well as observe the impingement of any other solidifying phases. The completed phase stability diagrams were then created by plotting the four points of interest on the liquidus projection of corresponding Mo content. Figure 2 shows a typical example for a 6 wt% Mo alloy. The temperature axis is normal to the plane of the diagram. Equivalent points were then connected by curves in order to highlight several significant features of the overall phase diagram. The martensite boundary from the Schaeffler Diagram [9] was also incorporated onto the diagrams. The use of these diagrams will be discussed in more detail in the next section. Experimental study and verification of the new diagrams was sought through laboratory experiment. 96 alloys across both sides of the eutectic line in each of the six modeled Mo contents were selected for construction. The alloys were produced from virgin elements using the arc button melting (ABM) method. Carefully weighed portions of each constituent were placed under vacuum, which was then backfilled with inert argon to serve as a shielding gas. A manually-controlled TIG torch running at 300A and 10V and a water-cooled Cu hearth were used to produce 50g buttons. Metallographic preparation and subsequent microstructural analysis were used to reveal and identify the particular solidification and transformation mechanisms experienced by each alloy. The relative locations and shape of the ferrite and austenite phases were used as indications of the solidification mode and morphological type in a manner previously described by Elmer [10]. A magnetic ferrite detector was used to determine the specific weight percentage of ferrite within each alloy. Precision was maintained through the use of calibrated standards of known ferrite content. These data were used to relate composition to the ferrite content. The related data of solidification mode and measured ferrite content was then plotted on the respective phase stability diagram. Electron Probe Micro-Analysis (EPMA) analysis was conducted on select alloys for each of the observed solidification modes in order to
measure the distribution of each element within the microstructure. The analysis was conducted using the corrections outlined by Heinrich, et al [11].
Results & Discussion
Phase Stability Diagrams An example of the multi-component phase stability diagrams is shown in Figure 2. The band across the center describes the three vertices of the eutectic triangle. The dashed line that intersects this band represents the maximum solubility of Cr in austenite, which curves below the eutectic triangle. A reduction in Fe content (i.e., increasing Ni and Cr) can be seen to expand the width of the eutectic triangle. The set of diagrams given in Figure 3 shows the effect of increasing the Mo content, as well as a set of data points denoting the compositions of the alloys prepared for experimental validation. Successively higher additions of Mo cause the eutectic triangle not only to widen, but also to veer its location towards higher Ni contents. The latter effect is a result of the ferrite-stabilizing effect of Mo comparable to that of Cr. In addition, the continued addition of Mo widens the space between the eutectic and the boundary representing the maximum solubility of Cr in austenite. Within this range, an alloy that solidifies as primary ferrite can cool into a region of the phase diagram in which only single-phase austenite is stable and thus exhibits the thermodynamic potential for a ferrite to austenite solid state transformation. The phase stability diagrams can be used to predict the solidification mode of a wide range of Fe-Cr-Ni-Mo alloys that simulate the compositions of SASS alloys. Alloys above and to the left of the eutectic line should exhibit primary austenite solidification, while compositions below and to the right of this line will exhibit primary ferrite solidification.
Alloy compositions that solidify as primary ferrite and exhibit the
thermodynamic potential to experience a solid-state transformation to the austenite phase during cooling can also be identified in the diagrams. Compositions above and to the left of the maximum solubility of Cr in austenite should experience significantly more solidstate transformation into austenite, as the composition may now enter the single-phase
austenite region of the phase diagram. Compositions below and to the right may also experience solid state transformations by entering the δ + γ two-phase regions. Because δ is still thermodynamically stable in these alloys, the transformation does not proceed as efficiently.
However, since the diagrams take into account only thermodynamic
considerations, they are not proficient at predicting the type or amount of transformation that will occur. This factor can only be found through an analysis of the kinetics of the system, which takes into account the cooling conditions present. Therefore, the diagrams represent a necessary, but not sufficient, condition for attaining a fully-austenitic alloy derived by ferritic solidification. The phase stability diagrams calculated in this research were experimentally validated. The compositions of alloy buttons were plotted by their nominal Cr and Ni contents on the phase stability diagram of the corresponding Mo content.
The
summarized results for all alloy compositions are plotted in Figure 3. The corresponding weight percentages of ferrite phase are included with each data point. The calculated eutectic lines were fairly accurate in separating alloys between the austenitic and ferritic primary solidification modes, especially in the lower Mo diagrams. A third of the samples showed evidence of the martensite phase. Microstructural analysis and determination of the solidification mode of these 32 alloys was not conducted, in order to maintain focus on alloys composed predominately of austenite. Observations of a segment of these alloys did show evidence of martensite being created from each of the possible solidification modes. Of the 64 alloys that contained no martensite, only eight were found to disagree with the solidification modes predicted by the phase stability diagrams.
In nearly all cases, their nominal compositions were very close to the
calculated eutectic lines.
Phase Transformation Sequences and Microstructural Evolution A total of eighteen distinct transformation sequences and resultant microstructures were observed in the alloys of this study and are summarized in Table I. The numerous microstructures are a result of four possible solidification modes (A, AF, FA, F) and the possibility of three solid-state phase transformations (δ → γ, δ → γ + σ and γ → martensite). The proceeding sections will deal with each of the solidification modes
individually, with discussion included in each detailing the requirements and characteristics of the solid-state transformations that occur after the completion of solidification.
The micrographs displayed in Figure 4a and b show examples of the austenitic (A) and austenitic-ferritic (AF) structures, respectively.
All structures
containing primary austenite solidified on the Ni-rich side of the eutectic line. The intercellular regions of the A mode structure were etched away because of disparities in composition between the cores and the outer edges of the cells. The compositions of alloys belonging to the AF solidification mode were close enough to the eutectic line to drive the solidification path through the eutectic triangle of the phase diagram, at which point the ferrite phase was stabilized by the excess Cr and Mo that partitioned to the remaining liquid. The quantity of ferrite produced by solidification was reduced by a solid-state transformation of the type δ → γ. The transformation proceeded from the austenite/ferrite phase boundary in the intercellular regions, so no nucleation was required. The small amounts of ferrite remaining after this transformation, as shown by the weight percentage of ferrite in each alloy in Figure 3, made direct observation of it in the microstructure difficult. Only those alloys that had a detectable quantity of ferrite via magnetic measurement were thus designated to the AF mode. The number of structures possible from the primary solidification of austenite was magnified by the initiation of two further solid-state transformations, the eutectoid-type δ → γ + σ transformations and the shear-type γ → martensite transformation. It was shown that the composition of the alloy button was a controlling factor in the presence of each type. All alloys containing 20wt% Ni + Cr and alloys with 0-2wt% Mo and 25wt% Ni + Cr showed evidence of the martensite phase initially through ferrite measurements. Alloys that were expected to have low ferrite contents were shown to exhibit a strong magnetic signature due to the presence of martensite. Laths of martensite were then revealed via metallography and observed directly on a portion of these samples. Figure 4c shows several laths of martensite crossing multiple cells of austenite. As shown in Figure 3, nearly all compositions that provided evidence of martensite were encompassed
by the line conceived by Schaeffler to bound martensite-containing compositions. The only martensitic alloys to escape this boundary contained high (8-10wt%) additions of Mo, a compositional range that was likely not included in the creation of the Schaeffler diagram. Martensite was shown to form not only in alloys consisting solely of austenite, but also those compositions with intercellular ferrite. Since the only martensitic samples containing 25wt% Ni+Cr were limited to 2wt% Mo, it may be theorized that Mo solute will inhibit the shear transformation to martensite. It should be noted though that the addition of Mo is increasing the Cr-equivalency of the alloy to a level above that of martensitic compositions, as shown in the Schaeffler diagram. Higher Mo contents were shown to be sufficient to cause the nucleation and growth of σ-phase, a higher-order intermetallic that is known for deleterious effects on corrosion resistance and mechanical properties [12]. The σ-phase, shown in Figure 4d, was identified via Backscattered Electron Kikuchi Pattern (BEKP) analysis. Recent research [13] has demonstrated that the σ-phase in these alloys forms by a eutectoid-type δ → γ + σ solid-state transformation, the details of which will be discussed in a separate article. Based on the location of the σ-phase field in the phase diagram (see Figure 5), it is believed that the eutectoid transformation interrupts the regular δ → γ solid-state transformation during cooling. A secondary condition for the presence of σ-phase, beyond a sufficiently high Mo content, was revealed:
the proximity of the alloy composition to the eutectic line.
Thermodynamic calculations of the system show that the phase fields in which σ-phase is stable reach the highest temperatures at compositions closest to the eutectic.
This
characteristic supplies those particular alloys with near-eutectic compositions the necessary time and energy to nucleate and grow the σ-phase. A schematic detailing this behavior is given in Figure 5.
Ferritic-Austenitic Solidification Modes At least one alloy from each Mo content group was found to belong to the ferriticaustenitic (FA) solidification mode. This mode was differentiated from the AF modes by comparing the size, shape, distribution, and coloration of the ferrite phase, which were useful in discerning the locations of the phases relative to the initial solidification
structure. Termination of solidification occurred at the eutectic line, where the soluterich liquid in the intercellular region then solidified as austenite, producing an effective nucleus from which the ensuing solid-state transformation could grow. The same δ → γ solid-state transformation that occurred in the AF alloys also took place in the FA mode alloys. Although both transformations proceed from the austenite/ferrite phase boundary in the intercellular, the process will now consume the solidified ferritic cell, whereas before the transformation consumed the intercellular ferrite of the AF mode. Because of this difference, there is far more of the parent phase to expend, and the residual ferrite is located at what was formerly the cell core. The skeletal “ribs” of the residual ferrite, as seen in Figure 4e possess many ~90o angles, likely a result of the easy growth direction. Both the martensite transformation and the eutectoid transformation to σ-phase were shown to occur in alloys with the FA solidification mode. The same compositional requirements for each transformation discussed in the preceding section were shown to still be a controlling factor. Alloys with 20wt% Ni+Cr experienced a transformation to martensite. The δ → γ transformation must have occurred first for there to be a parent phase to generate the martensitic structure.
The location of the eutectoid γ +
σ constituent at the cell cores, as seen in Figure 4f, proved that ferrite was indeed the parent phase. This solidification mode did however require a higher Mo content for the eutectoid transformation. While 4wt% Mo was shown to be sufficient for AF structures to nucleate the σ-phase, FA structures required a concentration of 8wt% Mo. This can be taken as a comparison of the ability to distribute solute between the microsegregation that occurs during solidification and the solute partitioning that occurs during the δ → γ transformation. Microsegregation was able to drive the intercellular regions to higher concentrations than partitioning could cause in the residual ferrite, even if the nominal concentration of Mo was equivalent in each case.
Ferritic Solidification Modes Alloys on the Cr-rich side of the phase diagram further away from the eutectic line experienced the ferritic (F) solidification mode. The liquid was not enriched to the eutectic; therefore, ferritic solidification occurred throughout the entire structure. A
representative microstructure of this mode is shown in Figure 4f. Viewing of the cellular solidification structure was not possible because of the elevated diffusivity of the δ-ferrite phase. The open nature of the bcc crystal structure allowed for solute diffusion from the intercellular regions down the concentration gradients towards the cell core. Due to the homogenization of the structure via this elevated diffusion, etching techniques were not capable of highlighting solidification cells. This featureless solidification structure was seen in alloy compositions furthest from the eutectic line, where only regions nearest to grain boundaries experienced the solid-state δ → γ transformation. Alloy buttons that exhibited this phase transformation did not contain austenite at the termination of solidification, therefore no pre-existing phases were available from which the transformation could proceed. All alloys of this type, however, entered a region of the phase diagram where the austenite phase became stable. The instability of the solidified ferrite phase caused the spontaneous nucleation and growth of austenite by way of two morphologies.
The first of these involved the nucleation of austenitic
allotriomorphs along the ferritic grain boundaries, where free energy is highest. The easy growth of the product phase along grain boundaries allowed nuclei to encompass entire grains with austenite. This is a result of rapid growth along incoherent interfaces along the grain boundary, as compared to slow growth into the grain where an orientation relationship, likely the Kurdjumov-Sachs[14] type, often existed. The growth of these allotriomorphs was eventually slowed by the drop in temperature, at which time the solidstate transformation proceeded into the ferrite grains via the Widmanstatten mechanism. This type of austenitic product appears as needles or platelets that extend from the allotriomorphic product inward to the grain. There is a degree of shear inherent to this transformation product, similar to that observed in bainitic structures. Alloys closest to the eutectic line experienced the greatest degree of solid-state transformation. All alloys of the F mode, however, contained greater amounts of residual ferrite, as they most often fell into the δ + γ region of the phase diagram, where ferrite was still a stable phase. Since alloys of the F mode still contained some of the austenite phase, the martensite transformation was still found to take place in these alloys. The measured quantity of a magnetic phase was much greater in martensitic F mode alloys than found in other F mode alloys, indicating the presence of martensite in the microstructure. The
discovery of σ + γ eutectoid constituent between the austenitic Widmanstatten platelets, as shown in Figure 4h, was one of many pieces of evidence that proved the incidence of this solid-state transformation, (rather than the formation of σ-phase by a terminal solidification reaction, which is occasionally observed[2]). The eutectoid constituent is only found in inter-platelet regions, the boundaries of which were defined only after solidification.
The solute partitioning that arose in the σ-containing alloys was once
again responsible for the elevated levels of Mo necessary for nucleation and growth of the σ−phase in this type of morphology.
EPMA An example of the difference in solute profiles between solidified and transformed austenite for the 10wt% Mo system are shown in Figure 6 and Figure 7. The Mo profile shows a definite increase in concentration in the distance between a cell core and the neighboring intercellular region produced by AF solidification. This was caused by the solidifying cell rejecting Mo into the liquid. The solute accumulates in the intercellular region, since it is the last to solidify. A similar process occurs in the solidification of ferrite, with solute being rejected to the liquid. Owing to the improved diffusivity of solute in the open bcc structure of ferrite, the excess Mo in the intercellular region is able to diffuse down the concentration gradient towards the cell core. The transformed austenite regions in Figure 7 show a more uniform concentration of Mo. Since the central axis of the austenite regions represents the former intercellular regions of the solidified ferrite cells, it is established that the initial concentration gradient of Mo was significantly reduced via backdiffusion, prior to the γ → δ transformation. The significant backdiffusion of solute in the ferrite phase is also verified in Figure 8, which shows the microstructure produced by F mode solidification. The central region of this solute profile shows the uniform distribution of all measured elements across a region of untransformed ferrite that encompasses several cell widths. Despite this success, the subsequent solid-state transformation from ferrite to austenite did cause a difference in Mo concentration between the austenite and residual ferrite. The partitioning of Mo during the transformation is apparent by the high levels of
Mo seen in the ferrite/σ-phase dual structure of Figure 7. As shown by peaks in the solute profile, the elements Cr and Mo, both stabilizers of the ferrite-phase, were rejected into the residual ferrite during the transformation process. The austenite matrix resulting from the FA solidification mode contains slightly less than nominal, yet essentially uniform, concentrations of these solute atoms. Conversely, austenite-stabilizing Ni was drawn into the newly created austenite, resulting in a uniform profile of Ni within the austenite that was greater than the nominal composition. The residual ferrite was thus depleted of Ni. A summary of the results of solute partitioning for a single austenitic transformation product can be examined using the grain boundary allotriomorph contained in Figure 8. While the concentrations of solute are relatively uniform within the allotriomorphic transformation product, there exist distinct differences in concentration between it and the neighboring ferrite grain. The amount of Mo and Cr are lower in the allotriomorph, while the amount of Ni has increased as a result of transformation. The ferrite nearest to the allotriomorph even shows raised intensities of Mo and Cr, the result of being partitioned to the parent ferrite. The general upheaval of concentration brought about by solid-state transformations can also be seen on the right hand side of the micrograph. The production of Widmanstatten austenite platelets has caused the uniform distribution of solute in the parent ferrite to be disrupted. Although the platelets are too small to be individually measured, the partitioning of solute is apparent in the measured chemical profile. Solute partitioning resulting from solid-state transformation does not, however, cause a chemical profile equivalent to that produced by microsegregation such as during AF solidification. Figure 9 compares the profile of Mo solute between a single solidified austenite cell brought about by AF solidification with a region of transformed austenite derived by the FA solidification mode.
While neither austenite region reaches the
nominal concentration of ~6wt% Mo, the concentration in the FA austenite is certainly closer to 6wt% across its entire width. The advancing transformation front partitioned Mo to the remaining ferrite regions, where the excess Mo led to the formation of σ-phase, preventing the austenite regions from reaching the nominal level of 6wt% Mo.
Effects of Composition on Microstructure and Ferrite Content As would be expected, the weight percentage of ferrite measured in the alloys increased as a function of Creq/Ni (Creq = wt% Cr + wt% Mo). The data is plotted in Figure 10 with the alloys differentiated according to the solidification mode of the alloy. The data indicates that each solidification mode belongs to a specific regime of Creq/Ni ratio.
The boundary between austenitic and ferritic mode alloys that is found at
1.5Creq/Ni is in good agreement with previous findings[4]. Another significant boundary can be seen between FA and F mode alloys at ~1.7Creq/Ni. The plot also shows that the FA mode was far more efficient at reducing the amount of the primary solidified ferrite, as the detected ferrite content in F mode alloys ranged between 3 and 55wt%. Meanwhile, the ferrite content of FA alloys never surpassed 10wt%. A polynomial trendline has been included in this figure that relates the ferrite content to the Creq/Ni ratio.
Only those alloys with a detectable amount of ferrite were included in this
calculation. Figure 11 displays another plot which compares the experimental findings in this research with a model that relates the ferrite content to alloy composition. The model set forth by Seferian[15] indicates that the weight percent of ferrite can be predicted by the composition according to the relationship: Wt% Ferrite = 3[Creq – 0.93Nieq -6.7] Each data point in Figure 11 denotes an alloy composition. The nominal Cr and Ni equivalencies of each alloy button were inputted to attain a predicted weight percent of ferrite. By separating the data into σ-free and σ-containing alloys, the effects of the δ Æ σ + γ eutectoid transformation are highlighted.
The measured ferrite content in σ-
containing alloys was always less than that predicted by the model. However, the ferrite content of alloys lacking in σ-phase shows good agreement with the Seferian model up to 20wt% ferrite. Alloys beyond this limit were observed to contain large degrees of untransformed ferrite such that Widmanstatten austenite was found only near grain boundaries. The solidification mode and presence of sigma phase are mapped out as a function of Mo concentration and Creq/Nieq ratio in Figure 12. As shown in Figure 10, a
switch in solidification mode from AF to FA occurs at a Creq/Nieq ratio of ~ 1.5. Also, the switch from FA to F mode is found at a Creq/Nieq ratio of ~ 1.7. Where as the previous figure plotted all alloys by one compositional parameter (Creq/Ni), Figure 12 shows that the previously observed boundaries do not change as the Mo content is altered. These findings seem to indicate that the long-standing assumption that Mo possesses a ferritestabilizing strength equal to that of Cr is valid for the high Mo alloys. Boundaries are also included that separate σ-containing samples from σ-free samples. Figure 12 thus represents a map by which microstructures may be predicted based on composition for a Mo-bearing stainless steel. Of particular interest are the near-eutectic alloys with 4-6wt% Mo.
Alloys that solidified as primary austenite showed substantial amounts of
interdendritic σ-phase. On the other side of the eutectic, FA mode alloys showed no evidence of σ-phase, save for a few isolated particles in the 6Mo-16Cr-14Ni sample. The lack of σ-phase in these alloys implies that the Mo content was more uniformly distributed throughout the microstructure as opposed to residing in the interdendritic regions forming σ-phase.
Analysis of Data Using Accepted Stainless Steel Diagrams The most widely used constitution diagram in current research is the WRC-1992 diagram. The advantage of this diagram is its inclusion of boundaries which demarcate the various solidification modes available in the range of industrial stainless steels. Although previously included in the Schaeffler and DeLong diagrams, the WRC-1992 diagram does not include the martensite phase. The martensite boundary devised by Schaeffler in his diagram was used in this research to border the samples expected to contain the martensite phase. Recent research efforts have sought to locate a similar line within the WRC-1992 diagram. Kotecki has proposed a boundary [16] shown in Figure 13, in which the set of 96 alloys are also shown, plotted by their nominal Cr and Ni equivalencies. In his work, Kotecki identified the presence of martensite in his samples through the observation of brittle fracture in flexural bend tests. Martensite was revealed in this research instead by magnetic measurements and the use of metallography to directly observe the martensite laths. 32 of the experimental alloys indicated the presence of the martensite phase. These were used to corroborate the boundary proposed by
Kotecki. Twelve of the martensitic samples from this study are actually located above this boundary. A majority of these samples contain high amounts of Mo additions. Clearly, the effects of this element on the martensitic transformation need to be taken into account if a martensite boundary is to be an official inclusion on the WRC-1992 diagram for high Mo stainless steels. The WRC-1992 diagram was generally derived from low Mo alloys. The list of original samples [17] used shows that only four samples contained greater than 5wt% Mo, the highest reaching 6.85wt% Mo. Plotting the samples according to solidification mode on the WRC-1992 diagram (as in Figure 13) shows generally good agreement between the predictions of the diagram and the experimentally observed solidification mode.
Conclusions 1) The superimposition of data collected from vertical isopleths of multi-component phase diagrams onto a liquidus projection can be done to create a phase stability diagram. 2) A phase stability diagram of the Fe-Cr-Ni-Mo system can be used to predict the primary solidification mode of a particular alloy composition as well as the solidstate transformations that it will experience simultaneously over a wider range of compositions possible in a pseudo-binary phase diagram. 3) Eighteen distinct microstructural development sequences were observed and mapped in the near-eutectic Fe-rich corner of the Fe-Cr-Ni-Mo system. Four solidification modes and three solid-state phase transformations accounted for the variety in microstructure.
The physical characteristics and compositional
requirements of each phase were described in detail. 4) Additions of Mo up to 10wt% did not significantly affect the shift from the austenitic to ferritic PSM occurring at 1.5 Creq/Ni. 5) The shift from FA to F solidification mode occurs at 1.7 Creq/Ni, and is also insensitive to Mo and Cr content.
6) The FA solidification mode produced a uniform distribution of Mo solute in the transformed austenite phase that was greater in concentration than observed in the cores of solidified austenite cells of the AF mode. 7) Stainless steel alloys containing high additions of Mo undergo a martensite transformation in alloys not previously predicted by the Schaeffler diagram or Kotecki’s proposed additions to the WRC-1992 diagram. A new boundary line has been proposed to encompass these alloys, many of which contained high Mo contents.
Acknowledgments The author would like to thank his advisors, Profs. John DuPont and Arnold Marder, for their assistance. The technical and physical contributions of Dr. Matthew Perricone were also invaluable during all phases of this research.
Reference List 1. A. Garner, The Effect of Autogenous Welding on Chloride Pitting Corrosion in Austenitic
Stainless Steels. Corrosion 35, 108-114 (1979). 2. S. Banovic, J. DuPont, and A. Marder, Dilution and microsegregation in dissimilar metal welds between super austenitic stainless steel and nickel base alloys. Science and Technology of Welding and Joining (UK) 7, 374-383 (2002). 3. J. DuPont, L. Friedersdorf, A. Marder, and S. Banovic. Weldability and Corrosion Performance of Welds in AL-6XN Superaustenitic Stainless Steel. Lehigh University ATLSS Report No. 01-03. 2001. 4. J. Brooks and A. Thompson, Microstructural Development and Solidification Cracking Susceptibility of Austenitic Stainless Steel Welds. International Materials Reviews 36, 1644 (1991). 5. S. Banovic, J. DuPont, and A. Marder, Dilution Control in Gas Tungsten Arc Welds Involving Superaustentic Stainless Steels and Nickel Based Alloys. Metallurgical and
Material Transactions 32B, 1171-1176 (2001). 6. S. Banovic, J. DuPont, and A. Marder, Microstructural Evolution and Weldability of Dissimilar Welds between a Super Austenitic Stainless Steel and Nickel-Based Alloys. Welding Journal 82, 125-135 (2003). 7. B. Sundman. Thermo-Calc. S-100 44[[N]]. 2001. Stockholm, Sweden, Department of Materials Science and Engineering, KTH. 8. N. Saunders. Fe-Data Thermodynamic Database. [[3.0]]. 2001. The Surrey Research Park, Guildford, UK, Thermotech, Ltd. 9. A. Schaeffler, Constitution Diagram for Stainless Steel Weld Metal. Metals Progress 56, 680-680B (1949). 10. J. Elmer, S. Allen, and T. Eagar, Microstructural Development During Solidification of Stainless Steel Alloys. Metallurgical Transactions A 20A, 2117-2131 (1989). 11. K. HEINRICH, R. L. MYKLEBUST, H. YAKOWITZ, and S. D. RASBERRY. SIMPLE CORRECTION PROCEDURE FOR QUANTITATIVE ELECTRON-PROBE MICROANALYSIS. NBS Technical Note 719, 1-31. 1972. 12. T. Koseki and T. Ogawa, An Investigation of Weld Solidification in Cr-Ni-Fe-Mo Alloys. Welding International 6, 516-522 (1992). 13. Dr. Matthew J. Perricone. Effect of Composition, Cooling Rate, and Solidification Velocity on the Microstructural Development of Mo-bearing Stainless Steels. Thesis, Lehigh University. 14. G. Kurdjumov and G. Sachs, Z. Physics 64, 325-343 (1930). 15. D. Seferian, Metallurgie de la Soudure (1959). 16. D. Kotecki, A Martensite Boundary on the WRC-1992 Diagram. Welding Journal 78, 180192 (2000). 17. C. McCowan, T. Siewart, and D. Olson. Stainless Steel Weld Metal: Prediction of Ferrite Content. Bulletin 342. 1989. Welding Research Council. Welding Research Council Bulletin.
Table I. The eighteen microstructural development sequences found to result in Fe-Ni-Cr-Mo alloys from the arc-melt condition. Solidification Solidification Mode
Solid-State
Solidification Primary Sequence Transformation
A
L Æ L + γp Æ γp
−−
AF
L ÆL + γp ÆL + γp + (δ/γ)e Æγp + (δ/γ)e
δe Æ γtf
FA
L ÆL + δp ÆL + δp + (δ/γ)e Æδp + (δ/γ)e
δp, δeÆ γtf
F
L Æ L + δp Æ δp
δp Æ γgb + γWid
Secondary Transformations
Final Microstructure
Final Phases
−− γp Æ Μ −− −− γÆΜ γÆΜ δe Æ (γ/σ)eutectoid δe Æ (γ/σ)eutectoid both both −− γÆΜ δp Æ (γ/σ)eutectoid both
γp γp + M γp + γe + γtf γp + γe + δe + γtf γp + γe + γtf + M γp + γe + δe + γtf + M γp + γe + γtf + (γ/σ)eutectoid γp + γe + δe + γtf + (γ/σ)eutectoid γp + γe + γtf + (γ/σ)eutectoid + M γp + γe + δe + γtf + (γ/σ)eutectoid + M δp + γe + γtf δp + γe + γtf + M δp + γe + γtf + (γ/σ)eutectoid δp + γe + γtf + (γ/σ)eutectoid + M
−− γÆΜ δp Æ (γ/σ)eutectoid both
δp + γgb + γWid δp + γgb + γWid + Μ δp + γgb + γWid + (γ/σ) eutectoid δp + γgb + γWid + (γ/σ) eutectoid + M
γ γ+Μ γ γ+δ γ+Μ γ+δ+Μ γ+σ γ+δ+σ γ+σ+Μ γ+δ+σ+Μ γ+δ γ+δ+Μ γ+δ+σ γ+δ+σ+Μ γ+δ γ+δ+Μ γ+δ+σ γ+δ+σ+Μ
L = liquid; γ = austenite phase; δ = ferrite phase; σ = sigma phase; M = martensite; p = primary solidification product; γ’ = phase of reduced concentration; e = eutectic solidification product; tf = product of long-range diffusional solid-state phase transformation; eutectoid = product of eutectoid phase transformation; gb = grain-boundary allotriomorph; Wid = Widmanstatten platelets.
Figure 1. Vertical isopleth of the Fe-Cr-Ni-Mo system. The shaded zone demarcates compositions that will experience primary ferrite solidification in addition to entering the single-phase austenite region of the phase diagram. The points labeled represent: a) the eutectic line; b) γ-vertex of the eutectic triangle; c) δ-vertex of the eutectic triangle; d) the maximum solubility of chromium in the austenite phase.
Figure 2. Phase stability diagram of the Fe-Cr-Ni-6Mo alloy system. The shaded region denotes the predicted zone of compositions that will experience FA solidification.
Figure 3. Phase stability diagram for all Mo contents. Each data point represents the nominal alloy compositions. The shape and color of each data point represents the solidification mode, while the adjacent values represent its wt% ferrite.
Figure 4. Representative microstructures of the following morphological designations: a) A mode e) FA mode
b) AF mode
c) A w/ martensite
f) FA w/ σ−phase
g) F mode
d) AF w/ σ−phase h) F w/ σ−phase
Figure 5. Vertical section of Fe-Cr-Ni-Mo phase diagram at 6wt% Mo and 64wt% Fe, with σ-phase included in calculations. The compositions of 6Mo-12Cr-18Ni (a) and 6Mo-14Cr-16Ni (b) are plotted. Alloy b, which contains σ-phase, enters the σ-phase field at a higher temperature, allowing more time and heat to initiate the reaction. Alloy a, which does not contain σ-phase in its microstructure, enters at a lower temperature. This behavior was notable in alloys containing intermediate concentrations of Mo (4-8wt%)
Figure 6. EPMA linescan data for solidified austenite cells with interdendritic ferrite and sigma phases produced by AF solidification mode.
Figure 7. EPMA linescan data of transformed austenite and precipitated sigma in the residual ferrite, produced by FA solidification mode.
Figure 8. EPMA linescan of austenitic allotriomorph and Widmanstatten platelets bordering a ferrite grain.
Weight % Mo
8
6
4
FA
AF
FA Nominal
AF Nominal
2 0
5
10
15
20
25
Distance (microns)
Figure 9. Comparison of Mo solute profile generated by the AF and FA solidification modes. The nominal Mo concentrations of the particular alloys are included for comparison.
60
50
Weight % Ferrite
A
AF
FA
F
40
30
2
y = 33.622x - 86.546x + 56.041 2 R = 0.8791
20
10
0 0
0.5
1
1.5
2
2.5
(%Cr+%Mo)/(%Ni)
Figure 10. Measured weight percentage of ferrite phase as a function of Creq/Ni ratio.
3
Seferian Ferrite Predictions (wt%)
35
30
25
20
15
10
Sigma-free
5
Sigma-containing
0 0
10
20
30
40
50
60
Experimental Ferrite Measurements (wt%) Figure 11. Comparison of experimental ferrite measurements on FA and F mode alloys with Seferian’s model for prediction of ferrite content as a function of composition. The δ Æ σ transformation expends ferrite, causing these alloys to deviate from the Seferian model significantly.
12
Mo content (wt%)
10
8
σ
A
AF
FA
6
4
F 2
0 0
0.5
1
1.5
2
2.5
3
(%Cr + %Mo) / (%Ni) Figure 12. Microstructural map of alloy solidification mode based on Creq/Ni ratio and Mo content. Boundaries are included to separate regions by solidification mode and σ-content
Ni equivalent = %Ni + 35(%C) + 20(%N) + 0.25(%Cu)
WRC-1992 - All Compositions 22 20 18
A
20
16
AF
40 60
FA
14
80
F
12
100
10
MARTENSITE 8
proposed
6
Kotecki 4 10
12
14
16
18
20
22
24
26
28
30
32
Cr equivalent = %Cr + %Mo + 0.7(%Nb)
Figure 13. WRC-1992 Diagram with all alloy button compositions superimposed. The solidification mode of each alloy is represented by shape and color of the data point, while its location is representative of its composition. The martensite boundary proposed by Kotecki16 is also included.
The Influence of Molybdenum on the Microstructure of Stainless Steel Welds Timothy Anderson Lehigh University Advisor: Prof. J.N. DuPont
Abstract The excellent corrosion resistance of superaustenitic stainless steel (SASS) alloys has been shown to be a direct consequence of high concentrations of Mo. The presence of Mo can have a significant effect on the microstructural development of welds in these alloys. In this research, the microstructural development of welds in the Fe-Ni-Cr-Mo system was analyzed over a wide variety of Cr/Ni ratios and Mo contents. The system was first simulated by construction of multi-component phase diagrams using the CALPHAD technique. Data gleaned from vertical sections of these diagrams were inserted in the compositional range of a liquidus projection to produce diagrams that can be used as a guide to understand the influence of composition on microstructural development. A large number of experimental alloys were then prepared via arc-button melting for comparison with the diagrams. Each button was characterized using various microscopy techniques. The expected δ-ferrite and γ-austenite phases were accompanied by martensite at low Cr/Ni ratios and by σ-phase at high Mo contents. The results were used to construct a map of expected phase transformation sequence and resultant microstructures as a function of composition. Electron microprobe measurements of selected alloys were performed to compare the distribution of Mo solute between different solidification modes. Severe Mo microsegregation was observed in alloys that solidified directly as austenite; this was attributed to the low diffusivity of Mo in austenite. The level of Mo microsegregation could be significantly reduced in alloys that
solidified as δ-ferrite first and subsequently transformed to austenite by a solid state transformation. Magnetic ferrite measurements were also used to produce quantitative relationships between alloy composition and ferrite content, and the results were then plotted on the WRC-1992 diagram. The results of this work provide a working guideline for future base metal and filler metal development of these alloys. Background SASS alloys with high Mo concentrations are used in a wide variety of applications that require good toughness and corrosion resistance. Unfortunately, the corrosion resistance of welds is never as good as the corresponding wrought base metal [1]. The microstructural development and resultant corrosion resistance of austenitic alloys in this compositional regime has been shown to be strongly controlled by the microsegregation of Mo to the intercellular regions [2]. The ensuing depletion of Mo in the cell core results in preferential corrosive attack [3]. Alloys that solidify as primary austenite are also susceptible to solidification cracking caused by the accumulation of tramp elements (such as P and S) and concomitant formation of low melting point liquid films in the intercellular regions [4].
Previous studies have also shown that the
microsegregation of Mo will proceed over a wide range of welding conditions and compositions [2,5,6]. Although high Mo nickel base filler metals can be used to help mitigate this problem, the microsegregation problem still persists, and careful control of welding variables must be exercised for close control over weld metal composition. In contrast, it is well known that stainless steels that solidify as ferrite exhibit significant backdiffusion and good resistance to solidification cracking due to high solubility of tramp elements [4]. Thus, identifying high Mo SASS alloys that solidify first as ferrite and subsequently transform to austenite via a solid state reaction may help eliminate problems associated with solidification cracking and microsegregation-induced corrosion. A wide range of microstructures are possible for near-eutectic Mo-bearing stainless steels, owing to a variety of solidification modes and a possible solid-state phase transformations that occur from the ferrite to the austenite phase. Multi-component phase diagrams provide an efficient way for understanding microstructure formation and serve as a useful tool for alloy development. The presentation of multi-component phase
diagrams has commonly been accomplished through the use of two different displays: the liquidus projection and the pseudo-binary vertical isopleth. The liquidus projection is capable of displaying the primary solidification mode and terminal solidification reactions over a wide range of compositions, but is unable to describe subsequent solidstate phase transformations that may further influence the final microstructure. The vertical isopleth, while being able to describe post-solidification behavior, is hampered by a slim range of concurrent alloy compositions. For the quaternary Fe-Cr-Ni-Mo system being studied in this research project, an attempt was made to form a composite type diagram that would communicate the microstructural development behavior of a greater number of alloys concurrently.
Efforts were concentrated to combine the
strengths of both diagrams in order to bound the regions of different solidification mode. Thus, the objectives of this research were to design a diagram for phase prediction in the Fe-Ni-Cr-Mo system and to use those diagrams to locate compositions that can experience a solid-state transformation to austenite after primary ferrite solidification. A matrix of alloys were constructed to validate the diagrams and to compare the concentration profiles of Mo solute in austenite derived by either solidification or solidstate transformation.
Experimental Procedure A large matrix of vertical isopleths was constructed of the Fe-Cr-Ni-Mo system using the CALPHAD software Thermo-Calc [7] in conjunction with the Iron Alloy Database [8]. Only the ferrite and austenite phases were included in the calculations, and a range of temperatures was used that would span from the fully liquid state to any solidstate phase-transformations that might transpire. Vertical isopleths with constant Mo additions of 0, 2, 4, 6, 8, and 10 wt% were constructed for several Ni+Cr contents. The software package allows specific portions of a generated diagram to be enlarged; this feature was used to acquire the specific temperature and composition of four special points of interest: a. the peak of the eutectic triangle b. the γ-side vertex of the eutectic triangle c. the δ-side vertex of the eutectic triangle
d. the maximum solubility of Cr in the austenite phase Figure 1 illustrates the locations of these points of interest on a sample isopleth. Further isopleths were assembled in order to ascertain these same locations at the terminating edges of the phase diagram, where either no Cr or Ni is present. In addition, liquidus projections were calculated at each Mo content in order to validate the locations of the eutectic as well as observe the impingement of any other solidifying phases. The completed phase stability diagrams were then created by plotting the four points of interest on the liquidus projection of corresponding Mo content. Figure 2 shows a typical example for a 6 wt% Mo alloy. The temperature axis is normal to the plane of the diagram. Equivalent points were then connected by curves in order to highlight several significant features of the overall phase diagram. The martensite boundary from the Schaeffler Diagram [9] was also incorporated onto the diagrams. The use of these diagrams will be discussed in more detail in the next section. Experimental study and verification of the new diagrams was sought through laboratory experiment. 96 alloys across both sides of the eutectic line in each of the six modeled Mo contents were selected for construction. The alloys were produced from virgin elements using the arc button melting (ABM) method. Carefully weighed portions of each constituent were placed under vacuum, which was then backfilled with inert argon to serve as a shielding gas. A manually-controlled TIG torch running at 300A and 10V and a water-cooled Cu hearth were used to produce 50g buttons. Metallographic preparation and subsequent microstructural analysis were used to reveal and identify the particular solidification and transformation mechanisms experienced by each alloy. The relative locations and shape of the ferrite and austenite phases were used as indications of the solidification mode and morphological type in a manner previously described by Elmer [10]. A magnetic ferrite detector was used to determine the specific weight percentage of ferrite within each alloy. Precision was maintained through the use of calibrated standards of known ferrite content. These data were used to relate composition to the ferrite content. The related data of solidification mode and measured ferrite content was then plotted on the respective phase stability diagram. Electron Probe Micro-Analysis (EPMA) analysis was conducted on select alloys for each of the observed solidification modes in order to
measure the distribution of each element within the microstructure. The analysis was conducted using the corrections outlined by Heinrich, et al [11].
Results & Discussion
Phase Stability Diagrams An example of the multi-component phase stability diagrams is shown in Figure 2. The band across the center describes the three vertices of the eutectic triangle. The dashed line that intersects this band represents the maximum solubility of Cr in austenite, which curves below the eutectic triangle. A reduction in Fe content (i.e., increasing Ni and Cr) can be seen to expand the width of the eutectic triangle. The set of diagrams given in Figure 3 shows the effect of increasing the Mo content, as well as a set of data points denoting the compositions of the alloys prepared for experimental validation. Successively higher additions of Mo cause the eutectic triangle not only to widen, but also to veer its location towards higher Ni contents. The latter effect is a result of the ferrite-stabilizing effect of Mo comparable to that of Cr. In addition, the continued addition of Mo widens the space between the eutectic and the boundary representing the maximum solubility of Cr in austenite. Within this range, an alloy that solidifies as primary ferrite can cool into a region of the phase diagram in which only single-phase austenite is stable and thus exhibits the thermodynamic potential for a ferrite to austenite solid state transformation. The phase stability diagrams can be used to predict the solidification mode of a wide range of Fe-Cr-Ni-Mo alloys that simulate the compositions of SASS alloys. Alloys above and to the left of the eutectic line should exhibit primary austenite solidification, while compositions below and to the right of this line will exhibit primary ferrite solidification.
Alloy compositions that solidify as primary ferrite and exhibit the
thermodynamic potential to experience a solid-state transformation to the austenite phase during cooling can also be identified in the diagrams. Compositions above and to the left of the maximum solubility of Cr in austenite should experience significantly more solidstate transformation into austenite, as the composition may now enter the single-phase
austenite region of the phase diagram. Compositions below and to the right may also experience solid state transformations by entering the δ + γ two-phase regions. Because δ is still thermodynamically stable in these alloys, the transformation does not proceed as efficiently.
However, since the diagrams take into account only thermodynamic
considerations, they are not proficient at predicting the type or amount of transformation that will occur. This factor can only be found through an analysis of the kinetics of the system, which takes into account the cooling conditions present. Therefore, the diagrams represent a necessary, but not sufficient, condition for attaining a fully-austenitic alloy derived by ferritic solidification. The phase stability diagrams calculated in this research were experimentally validated. The compositions of alloy buttons were plotted by their nominal Cr and Ni contents on the phase stability diagram of the corresponding Mo content.
The
summarized results for all alloy compositions are plotted in Figure 3. The corresponding weight percentages of ferrite phase are included with each data point. The calculated eutectic lines were fairly accurate in separating alloys between the austenitic and ferritic primary solidification modes, especially in the lower Mo diagrams. A third of the samples showed evidence of the martensite phase. Microstructural analysis and determination of the solidification mode of these 32 alloys was not conducted, in order to maintain focus on alloys composed predominately of austenite. Observations of a segment of these alloys did show evidence of martensite being created from each of the possible solidification modes. Of the 64 alloys that contained no martensite, only eight were found to disagree with the solidification modes predicted by the phase stability diagrams.
In nearly all cases, their nominal compositions were very close to the
calculated eutectic lines.
Phase Transformation Sequences and Microstructural Evolution A total of eighteen distinct transformation sequences and resultant microstructures were observed in the alloys of this study and are summarized in Table I. The numerous microstructures are a result of four possible solidification modes (A, AF, FA, F) and the possibility of three solid-state phase transformations (δ → γ, δ → γ + σ and γ → martensite). The proceeding sections will deal with each of the solidification modes
individually, with discussion included in each detailing the requirements and characteristics of the solid-state transformations that occur after the completion of solidification.
The micrographs displayed in Figure 4a and b show examples of the austenitic (A) and austenitic-ferritic (AF) structures, respectively.
All structures
containing primary austenite solidified on the Ni-rich side of the eutectic line. The intercellular regions of the A mode structure were etched away because of disparities in composition between the cores and the outer edges of the cells. The compositions of alloys belonging to the AF solidification mode were close enough to the eutectic line to drive the solidification path through the eutectic triangle of the phase diagram, at which point the ferrite phase was stabilized by the excess Cr and Mo that partitioned to the remaining liquid. The quantity of ferrite produced by solidification was reduced by a solid-state transformation of the type δ → γ. The transformation proceeded from the austenite/ferrite phase boundary in the intercellular regions, so no nucleation was required. The small amounts of ferrite remaining after this transformation, as shown by the weight percentage of ferrite in each alloy in Figure 3, made direct observation of it in the microstructure difficult. Only those alloys that had a detectable quantity of ferrite via magnetic measurement were thus designated to the AF mode. The number of structures possible from the primary solidification of austenite was magnified by the initiation of two further solid-state transformations, the eutectoid-type δ → γ + σ transformations and the shear-type γ → martensite transformation. It was shown that the composition of the alloy button was a controlling factor in the presence of each type. All alloys containing 20wt% Ni + Cr and alloys with 0-2wt% Mo and 25wt% Ni + Cr showed evidence of the martensite phase initially through ferrite measurements. Alloys that were expected to have low ferrite contents were shown to exhibit a strong magnetic signature due to the presence of martensite. Laths of martensite were then revealed via metallography and observed directly on a portion of these samples. Figure 4c shows several laths of martensite crossing multiple cells of austenite. As shown in Figure 3, nearly all compositions that provided evidence of martensite were encompassed
by the line conceived by Schaeffler to bound martensite-containing compositions. The only martensitic alloys to escape this boundary contained high (8-10wt%) additions of Mo, a compositional range that was likely not included in the creation of the Schaeffler diagram. Martensite was shown to form not only in alloys consisting solely of austenite, but also those compositions with intercellular ferrite. Since the only martensitic samples containing 25wt% Ni+Cr were limited to 2wt% Mo, it may be theorized that Mo solute will inhibit the shear transformation to martensite. It should be noted though that the addition of Mo is increasing the Cr-equivalency of the alloy to a level above that of martensitic compositions, as shown in the Schaeffler diagram. Higher Mo contents were shown to be sufficient to cause the nucleation and growth of σ-phase, a higher-order intermetallic that is known for deleterious effects on corrosion resistance and mechanical properties [12]. The σ-phase, shown in Figure 4d, was identified via Backscattered Electron Kikuchi Pattern (BEKP) analysis. Recent research [13] has demonstrated that the σ-phase in these alloys forms by a eutectoid-type δ → γ + σ solid-state transformation, the details of which will be discussed in a separate article. Based on the location of the σ-phase field in the phase diagram (see Figure 5), it is believed that the eutectoid transformation interrupts the regular δ → γ solid-state transformation during cooling. A secondary condition for the presence of σ-phase, beyond a sufficiently high Mo content, was revealed:
the proximity of the alloy composition to the eutectic line.
Thermodynamic calculations of the system show that the phase fields in which σ-phase is stable reach the highest temperatures at compositions closest to the eutectic.
This
characteristic supplies those particular alloys with near-eutectic compositions the necessary time and energy to nucleate and grow the σ-phase. A schematic detailing this behavior is given in Figure 5.
Ferritic-Austenitic Solidification Modes At least one alloy from each Mo content group was found to belong to the ferriticaustenitic (FA) solidification mode. This mode was differentiated from the AF modes by comparing the size, shape, distribution, and coloration of the ferrite phase, which were useful in discerning the locations of the phases relative to the initial solidification
structure. Termination of solidification occurred at the eutectic line, where the soluterich liquid in the intercellular region then solidified as austenite, producing an effective nucleus from which the ensuing solid-state transformation could grow. The same δ → γ solid-state transformation that occurred in the AF alloys also took place in the FA mode alloys. Although both transformations proceed from the austenite/ferrite phase boundary in the intercellular, the process will now consume the solidified ferritic cell, whereas before the transformation consumed the intercellular ferrite of the AF mode. Because of this difference, there is far more of the parent phase to expend, and the residual ferrite is located at what was formerly the cell core. The skeletal “ribs” of the residual ferrite, as seen in Figure 4e possess many ~90o angles, likely a result of the easy growth direction. Both the martensite transformation and the eutectoid transformation to σ-phase were shown to occur in alloys with the FA solidification mode. The same compositional requirements for each transformation discussed in the preceding section were shown to still be a controlling factor. Alloys with 20wt% Ni+Cr experienced a transformation to martensite. The δ → γ transformation must have occurred first for there to be a parent phase to generate the martensitic structure.
The location of the eutectoid γ +
σ constituent at the cell cores, as seen in Figure 4f, proved that ferrite was indeed the parent phase. This solidification mode did however require a higher Mo content for the eutectoid transformation. While 4wt% Mo was shown to be sufficient for AF structures to nucleate the σ-phase, FA structures required a concentration of 8wt% Mo. This can be taken as a comparison of the ability to distribute solute between the microsegregation that occurs during solidification and the solute partitioning that occurs during the δ → γ transformation. Microsegregation was able to drive the intercellular regions to higher concentrations than partitioning could cause in the residual ferrite, even if the nominal concentration of Mo was equivalent in each case.
Ferritic Solidification Modes Alloys on the Cr-rich side of the phase diagram further away from the eutectic line experienced the ferritic (F) solidification mode. The liquid was not enriched to the eutectic; therefore, ferritic solidification occurred throughout the entire structure. A
representative microstructure of this mode is shown in Figure 4f. Viewing of the cellular solidification structure was not possible because of the elevated diffusivity of the δ-ferrite phase. The open nature of the bcc crystal structure allowed for solute diffusion from the intercellular regions down the concentration gradients towards the cell core. Due to the homogenization of the structure via this elevated diffusion, etching techniques were not capable of highlighting solidification cells. This featureless solidification structure was seen in alloy compositions furthest from the eutectic line, where only regions nearest to grain boundaries experienced the solid-state δ → γ transformation. Alloy buttons that exhibited this phase transformation did not contain austenite at the termination of solidification, therefore no pre-existing phases were available from which the transformation could proceed. All alloys of this type, however, entered a region of the phase diagram where the austenite phase became stable. The instability of the solidified ferrite phase caused the spontaneous nucleation and growth of austenite by way of two morphologies.
The first of these involved the nucleation of austenitic
allotriomorphs along the ferritic grain boundaries, where free energy is highest. The easy growth of the product phase along grain boundaries allowed nuclei to encompass entire grains with austenite. This is a result of rapid growth along incoherent interfaces along the grain boundary, as compared to slow growth into the grain where an orientation relationship, likely the Kurdjumov-Sachs[14] type, often existed. The growth of these allotriomorphs was eventually slowed by the drop in temperature, at which time the solidstate transformation proceeded into the ferrite grains via the Widmanstatten mechanism. This type of austenitic product appears as needles or platelets that extend from the allotriomorphic product inward to the grain. There is a degree of shear inherent to this transformation product, similar to that observed in bainitic structures. Alloys closest to the eutectic line experienced the greatest degree of solid-state transformation. All alloys of the F mode, however, contained greater amounts of residual ferrite, as they most often fell into the δ + γ region of the phase diagram, where ferrite was still a stable phase. Since alloys of the F mode still contained some of the austenite phase, the martensite transformation was still found to take place in these alloys. The measured quantity of a magnetic phase was much greater in martensitic F mode alloys than found in other F mode alloys, indicating the presence of martensite in the microstructure. The
discovery of σ + γ eutectoid constituent between the austenitic Widmanstatten platelets, as shown in Figure 4h, was one of many pieces of evidence that proved the incidence of this solid-state transformation, (rather than the formation of σ-phase by a terminal solidification reaction, which is occasionally observed[2]). The eutectoid constituent is only found in inter-platelet regions, the boundaries of which were defined only after solidification.
The solute partitioning that arose in the σ-containing alloys was once
again responsible for the elevated levels of Mo necessary for nucleation and growth of the σ−phase in this type of morphology.
EPMA An example of the difference in solute profiles between solidified and transformed austenite for the 10wt% Mo system are shown in Figure 6 and Figure 7. The Mo profile shows a definite increase in concentration in the distance between a cell core and the neighboring intercellular region produced by AF solidification. This was caused by the solidifying cell rejecting Mo into the liquid. The solute accumulates in the intercellular region, since it is the last to solidify. A similar process occurs in the solidification of ferrite, with solute being rejected to the liquid. Owing to the improved diffusivity of solute in the open bcc structure of ferrite, the excess Mo in the intercellular region is able to diffuse down the concentration gradient towards the cell core. The transformed austenite regions in Figure 7 show a more uniform concentration of Mo. Since the central axis of the austenite regions represents the former intercellular regions of the solidified ferrite cells, it is established that the initial concentration gradient of Mo was significantly reduced via backdiffusion, prior to the γ → δ transformation. The significant backdiffusion of solute in the ferrite phase is also verified in Figure 8, which shows the microstructure produced by F mode solidification. The central region of this solute profile shows the uniform distribution of all measured elements across a region of untransformed ferrite that encompasses several cell widths. Despite this success, the subsequent solid-state transformation from ferrite to austenite did cause a difference in Mo concentration between the austenite and residual ferrite. The partitioning of Mo during the transformation is apparent by the high levels of
Mo seen in the ferrite/σ-phase dual structure of Figure 7. As shown by peaks in the solute profile, the elements Cr and Mo, both stabilizers of the ferrite-phase, were rejected into the residual ferrite during the transformation process. The austenite matrix resulting from the FA solidification mode contains slightly less than nominal, yet essentially uniform, concentrations of these solute atoms. Conversely, austenite-stabilizing Ni was drawn into the newly created austenite, resulting in a uniform profile of Ni within the austenite that was greater than the nominal composition. The residual ferrite was thus depleted of Ni. A summary of the results of solute partitioning for a single austenitic transformation product can be examined using the grain boundary allotriomorph contained in Figure 8. While the concentrations of solute are relatively uniform within the allotriomorphic transformation product, there exist distinct differences in concentration between it and the neighboring ferrite grain. The amount of Mo and Cr are lower in the allotriomorph, while the amount of Ni has increased as a result of transformation. The ferrite nearest to the allotriomorph even shows raised intensities of Mo and Cr, the result of being partitioned to the parent ferrite. The general upheaval of concentration brought about by solid-state transformations can also be seen on the right hand side of the micrograph. The production of Widmanstatten austenite platelets has caused the uniform distribution of solute in the parent ferrite to be disrupted. Although the platelets are too small to be individually measured, the partitioning of solute is apparent in the measured chemical profile. Solute partitioning resulting from solid-state transformation does not, however, cause a chemical profile equivalent to that produced by microsegregation such as during AF solidification. Figure 9 compares the profile of Mo solute between a single solidified austenite cell brought about by AF solidification with a region of transformed austenite derived by the FA solidification mode.
While neither austenite region reaches the
nominal concentration of ~6wt% Mo, the concentration in the FA austenite is certainly closer to 6wt% across its entire width. The advancing transformation front partitioned Mo to the remaining ferrite regions, where the excess Mo led to the formation of σ-phase, preventing the austenite regions from reaching the nominal level of 6wt% Mo.
Effects of Composition on Microstructure and Ferrite Content As would be expected, the weight percentage of ferrite measured in the alloys increased as a function of Creq/Ni (Creq = wt% Cr + wt% Mo). The data is plotted in Figure 10 with the alloys differentiated according to the solidification mode of the alloy. The data indicates that each solidification mode belongs to a specific regime of Creq/Ni ratio.
The boundary between austenitic and ferritic mode alloys that is found at
1.5Creq/Ni is in good agreement with previous findings[4]. Another significant boundary can be seen between FA and F mode alloys at ~1.7Creq/Ni. The plot also shows that the FA mode was far more efficient at reducing the amount of the primary solidified ferrite, as the detected ferrite content in F mode alloys ranged between 3 and 55wt%. Meanwhile, the ferrite content of FA alloys never surpassed 10wt%. A polynomial trendline has been included in this figure that relates the ferrite content to the Creq/Ni ratio.
Only those alloys with a detectable amount of ferrite were included in this
calculation. Figure 11 displays another plot which compares the experimental findings in this research with a model that relates the ferrite content to alloy composition. The model set forth by Seferian[15] indicates that the weight percent of ferrite can be predicted by the composition according to the relationship: Wt% Ferrite = 3[Creq – 0.93Nieq -6.7] Each data point in Figure 11 denotes an alloy composition. The nominal Cr and Ni equivalencies of each alloy button were inputted to attain a predicted weight percent of ferrite. By separating the data into σ-free and σ-containing alloys, the effects of the δ Æ σ + γ eutectoid transformation are highlighted.
The measured ferrite content in σ-
containing alloys was always less than that predicted by the model. However, the ferrite content of alloys lacking in σ-phase shows good agreement with the Seferian model up to 20wt% ferrite. Alloys beyond this limit were observed to contain large degrees of untransformed ferrite such that Widmanstatten austenite was found only near grain boundaries. The solidification mode and presence of sigma phase are mapped out as a function of Mo concentration and Creq/Nieq ratio in Figure 12. As shown in Figure 10, a
switch in solidification mode from AF to FA occurs at a Creq/Nieq ratio of ~ 1.5. Also, the switch from FA to F mode is found at a Creq/Nieq ratio of ~ 1.7. Where as the previous figure plotted all alloys by one compositional parameter (Creq/Ni), Figure 12 shows that the previously observed boundaries do not change as the Mo content is altered. These findings seem to indicate that the long-standing assumption that Mo possesses a ferritestabilizing strength equal to that of Cr is valid for the high Mo alloys. Boundaries are also included that separate σ-containing samples from σ-free samples. Figure 12 thus represents a map by which microstructures may be predicted based on composition for a Mo-bearing stainless steel. Of particular interest are the near-eutectic alloys with 4-6wt% Mo.
Alloys that solidified as primary austenite showed substantial amounts of
interdendritic σ-phase. On the other side of the eutectic, FA mode alloys showed no evidence of σ-phase, save for a few isolated particles in the 6Mo-16Cr-14Ni sample. The lack of σ-phase in these alloys implies that the Mo content was more uniformly distributed throughout the microstructure as opposed to residing in the interdendritic regions forming σ-phase.
Analysis of Data Using Accepted Stainless Steel Diagrams The most widely used constitution diagram in current research is the WRC-1992 diagram. The advantage of this diagram is its inclusion of boundaries which demarcate the various solidification modes available in the range of industrial stainless steels. Although previously included in the Schaeffler and DeLong diagrams, the WRC-1992 diagram does not include the martensite phase. The martensite boundary devised by Schaeffler in his diagram was used in this research to border the samples expected to contain the martensite phase. Recent research efforts have sought to locate a similar line within the WRC-1992 diagram. Kotecki has proposed a boundary [16] shown in Figure 13, in which the set of 96 alloys are also shown, plotted by their nominal Cr and Ni equivalencies. In his work, Kotecki identified the presence of martensite in his samples through the observation of brittle fracture in flexural bend tests. Martensite was revealed in this research instead by magnetic measurements and the use of metallography to directly observe the martensite laths. 32 of the experimental alloys indicated the presence of the martensite phase. These were used to corroborate the boundary proposed by
Kotecki. Twelve of the martensitic samples from this study are actually located above this boundary. A majority of these samples contain high amounts of Mo additions. Clearly, the effects of this element on the martensitic transformation need to be taken into account if a martensite boundary is to be an official inclusion on the WRC-1992 diagram for high Mo stainless steels. The WRC-1992 diagram was generally derived from low Mo alloys. The list of original samples [17] used shows that only four samples contained greater than 5wt% Mo, the highest reaching 6.85wt% Mo. Plotting the samples according to solidification mode on the WRC-1992 diagram (as in Figure 13) shows generally good agreement between the predictions of the diagram and the experimentally observed solidification mode.
Conclusions 1) The superimposition of data collected from vertical isopleths of multi-component phase diagrams onto a liquidus projection can be done to create a phase stability diagram. 2) A phase stability diagram of the Fe-Cr-Ni-Mo system can be used to predict the primary solidification mode of a particular alloy composition as well as the solidstate transformations that it will experience simultaneously over a wider range of compositions possible in a pseudo-binary phase diagram. 3) Eighteen distinct microstructural development sequences were observed and mapped in the near-eutectic Fe-rich corner of the Fe-Cr-Ni-Mo system. Four solidification modes and three solid-state phase transformations accounted for the variety in microstructure.
The physical characteristics and compositional
requirements of each phase were described in detail. 4) Additions of Mo up to 10wt% did not significantly affect the shift from the austenitic to ferritic PSM occurring at 1.5 Creq/Ni. 5) The shift from FA to F solidification mode occurs at 1.7 Creq/Ni, and is also insensitive to Mo and Cr content.
6) The FA solidification mode produced a uniform distribution of Mo solute in the transformed austenite phase that was greater in concentration than observed in the cores of solidified austenite cells of the AF mode. 7) Stainless steel alloys containing high additions of Mo undergo a martensite transformation in alloys not previously predicted by the Schaeffler diagram or Kotecki’s proposed additions to the WRC-1992 diagram. A new boundary line has been proposed to encompass these alloys, many of which contained high Mo contents.
Acknowledgments The author would like to thank his advisors, Profs. John DuPont and Arnold Marder, for their assistance. The technical and physical contributions of Dr. Matthew Perricone were also invaluable during all phases of this research.
Reference List 1. A. Garner, The Effect of Autogenous Welding on Chloride Pitting Corrosion in Austenitic
Stainless Steels. Corrosion 35, 108-114 (1979). 2. S. Banovic, J. DuPont, and A. Marder, Dilution and microsegregation in dissimilar metal welds between super austenitic stainless steel and nickel base alloys. Science and Technology of Welding and Joining (UK) 7, 374-383 (2002). 3. J. DuPont, L. Friedersdorf, A. Marder, and S. Banovic. Weldability and Corrosion Performance of Welds in AL-6XN Superaustenitic Stainless Steel. Lehigh University ATLSS Report No. 01-03. 2001. 4. J. Brooks and A. Thompson, Microstructural Development and Solidification Cracking Susceptibility of Austenitic Stainless Steel Welds. International Materials Reviews 36, 1644 (1991). 5. S. Banovic, J. DuPont, and A. Marder, Dilution Control in Gas Tungsten Arc Welds Involving Superaustentic Stainless Steels and Nickel Based Alloys. Metallurgical and
Material Transactions 32B, 1171-1176 (2001). 6. S. Banovic, J. DuPont, and A. Marder, Microstructural Evolution and Weldability of Dissimilar Welds between a Super Austenitic Stainless Steel and Nickel-Based Alloys. Welding Journal 82, 125-135 (2003). 7. B. Sundman. Thermo-Calc. S-100 44[[N]]. 2001. Stockholm, Sweden, Department of Materials Science and Engineering, KTH. 8. N. Saunders. Fe-Data Thermodynamic Database. [[3.0]]. 2001. The Surrey Research Park, Guildford, UK, Thermotech, Ltd. 9. A. Schaeffler, Constitution Diagram for Stainless Steel Weld Metal. Metals Progress 56, 680-680B (1949). 10. J. Elmer, S. Allen, and T. Eagar, Microstructural Development During Solidification of Stainless Steel Alloys. Metallurgical Transactions A 20A, 2117-2131 (1989). 11. K. HEINRICH, R. L. MYKLEBUST, H. YAKOWITZ, and S. D. RASBERRY. SIMPLE CORRECTION PROCEDURE FOR QUANTITATIVE ELECTRON-PROBE MICROANALYSIS. NBS Technical Note 719, 1-31. 1972. 12. T. Koseki and T. Ogawa, An Investigation of Weld Solidification in Cr-Ni-Fe-Mo Alloys. Welding International 6, 516-522 (1992). 13. Dr. Matthew J. Perricone. Effect of Composition, Cooling Rate, and Solidification Velocity on the Microstructural Development of Mo-bearing Stainless Steels. Thesis, Lehigh University. 14. G. Kurdjumov and G. Sachs, Z. Physics 64, 325-343 (1930). 15. D. Seferian, Metallurgie de la Soudure (1959). 16. D. Kotecki, A Martensite Boundary on the WRC-1992 Diagram. Welding Journal 78, 180192 (2000). 17. C. McCowan, T. Siewart, and D. Olson. Stainless Steel Weld Metal: Prediction of Ferrite Content. Bulletin 342. 1989. Welding Research Council. Welding Research Council Bulletin.
Table I. The eighteen microstructural development sequences found to result in Fe-Ni-Cr-Mo alloys from the arc-melt condition. Solidification Solidification Mode
Solid-State
Solidification Primary Sequence Transformation
A
L Æ L + γp Æ γp
−−
AF
L ÆL + γp ÆL + γp + (δ/γ)e Æγp + (δ/γ)e
δe Æ γtf
FA
L ÆL + δp ÆL + δp + (δ/γ)e Æδp + (δ/γ)e
δp, δeÆ γtf
F
L Æ L + δp Æ δp
δp Æ γgb + γWid
Secondary Transformations
Final Microstructure
Final Phases
−− γp Æ Μ −− −− γÆΜ γÆΜ δe Æ (γ/σ)eutectoid δe Æ (γ/σ)eutectoid both both −− γÆΜ δp Æ (γ/σ)eutectoid both
γp γp + M γp + γe + γtf γp + γe + δe + γtf γp + γe + γtf + M γp + γe + δe + γtf + M γp + γe + γtf + (γ/σ)eutectoid γp + γe + δe + γtf + (γ/σ)eutectoid γp + γe + γtf + (γ/σ)eutectoid + M γp + γe + δe + γtf + (γ/σ)eutectoid + M δp + γe + γtf δp + γe + γtf + M δp + γe + γtf + (γ/σ)eutectoid δp + γe + γtf + (γ/σ)eutectoid + M
−− γÆΜ δp Æ (γ/σ)eutectoid both
δp + γgb + γWid δp + γgb + γWid + Μ δp + γgb + γWid + (γ/σ) eutectoid δp + γgb + γWid + (γ/σ) eutectoid + M
γ γ+Μ γ γ+δ γ+Μ γ+δ+Μ γ+σ γ+δ+σ γ+σ+Μ γ+δ+σ+Μ γ+δ γ+δ+Μ γ+δ+σ γ+δ+σ+Μ γ+δ γ+δ+Μ γ+δ+σ γ+δ+σ+Μ
L = liquid; γ = austenite phase; δ = ferrite phase; σ = sigma phase; M = martensite; p = primary solidification product; γ’ = phase of reduced concentration; e = eutectic solidification product; tf = product of long-range diffusional solid-state phase transformation; eutectoid = product of eutectoid phase transformation; gb = grain-boundary allotriomorph; Wid = Widmanstatten platelets.
Figure 1. Vertical isopleth of the Fe-Cr-Ni-Mo system. The shaded zone demarcates compositions that will experience primary ferrite solidification in addition to entering the single-phase austenite region of the phase diagram. The points labeled represent: a) the eutectic line; b) γ-vertex of the eutectic triangle; c) δ-vertex of the eutectic triangle; d) the maximum solubility of chromium in the austenite phase.
Figure 2. Phase stability diagram of the Fe-Cr-Ni-6Mo alloy system. The shaded region denotes the predicted zone of compositions that will experience FA solidification.
Figure 3. Phase stability diagram for all Mo contents. Each data point represents the nominal alloy compositions. The shape and color of each data point represents the solidification mode, while the adjacent values represent its wt% ferrite.
Figure 4. Representative microstructures of the following morphological designations: a) A mode e) FA mode
b) AF mode
c) A w/ martensite
f) FA w/ σ−phase
g) F mode
d) AF w/ σ−phase h) F w/ σ−phase
Figure 5. Vertical section of Fe-Cr-Ni-Mo phase diagram at 6wt% Mo and 64wt% Fe, with σ-phase included in calculations. The compositions of 6Mo-12Cr-18Ni (a) and 6Mo-14Cr-16Ni (b) are plotted. Alloy b, which contains σ-phase, enters the σ-phase field at a higher temperature, allowing more time and heat to initiate the reaction. Alloy a, which does not contain σ-phase in its microstructure, enters at a lower temperature. This behavior was notable in alloys containing intermediate concentrations of Mo (4-8wt%)
Figure 6. EPMA linescan data for solidified austenite cells with interdendritic ferrite and sigma phases produced by AF solidification mode.
Figure 7. EPMA linescan data of transformed austenite and precipitated sigma in the residual ferrite, produced by FA solidification mode.
Figure 8. EPMA linescan of austenitic allotriomorph and Widmanstatten platelets bordering a ferrite grain.
Weight % Mo
8
6
4
FA
AF
FA Nominal
AF Nominal
2 0
5
10
15
20
25
Distance (microns)
Figure 9. Comparison of Mo solute profile generated by the AF and FA solidification modes. The nominal Mo concentrations of the particular alloys are included for comparison.
60
50
Weight % Ferrite
A
AF
FA
F
40
30
2
y = 33.622x - 86.546x + 56.041 2 R = 0.8791
20
10
0 0
0.5
1
1.5
2
2.5
(%Cr+%Mo)/(%Ni)
Figure 10. Measured weight percentage of ferrite phase as a function of Creq/Ni ratio.
3
Seferian Ferrite Predictions (wt%)
35
30
25
20
15
10
Sigma-free
5
Sigma-containing
0 0
10
20
30
40
50
60
Experimental Ferrite Measurements (wt%) Figure 11. Comparison of experimental ferrite measurements on FA and F mode alloys with Seferian’s model for prediction of ferrite content as a function of composition. The δ Æ σ transformation expends ferrite, causing these alloys to deviate from the Seferian model significantly.
12
Mo content (wt%)
10
8
σ
A
AF
FA
6
4
F 2
0 0
0.5
1
1.5
2
2.5
3
(%Cr + %Mo) / (%Ni) Figure 12. Microstructural map of alloy solidification mode based on Creq/Ni ratio and Mo content. Boundaries are included to separate regions by solidification mode and σ-content
Ni equivalent = %Ni + 35(%C) + 20(%N) + 0.25(%Cu)
WRC-1992 - All Compositions 22 20 18
A
20
16
AF
40 60
FA
14
80
F
12
100
10
MARTENSITE 8
proposed
6
Kotecki 4 10
12
14
16
18
20
22
24
26
28
30
32
Cr equivalent = %Cr + %Mo + 0.7(%Nb)
Figure 13. WRC-1992 Diagram with all alloy button compositions superimposed. The solidification mode of each alloy is represented by shape and color of the data point, while its location is representative of its composition. The martensite boundary proposed by Kotecki16 is also included.
