The Applied Processes, Inc., (AP) heat treatment plant, in Oshkosh, Wisconsin, works primarily in the austempering of ductile iron and wrought steel. By this process, the iron or steel is heated to above the A3 temperature and held to produce a fully austenitized microstructure and then quenched in an oil or salt bath. AP's facilities are equipped solely with salt bath quenches. Once the temperature is between the A1, the temperature above which austenite forms, and the martensite start (Ms), temperature below which martensite forms, the steel is held isothermally until the austenite transforms into a bainitic microstructure (Figure 1). This process increases the ductility of the iron or steel, as well as produces a more uniform hardness, a greater wear resistance, and higher impact and fatigue strengths for the hardness of the material. To date, austempering has been used almost exclusively to treat ductile iron and wrought steel, not just at AP, but also in the field of heat treatment as a whole.
Figure 1. A schematic of a typical austempering cycle, courtesy of Hayrynen [19]
A. Problem
In 2009, one of AP's customers approached them about applying the austempering treatment to a cast steel. To date, the customer has used a quench and temper process for their high hardness applications, but they believe that austempering could produce a premium product that will justify the additional cost through longer part life. The team at AP determined that the best way to achieve the customer's specifications, discussed below, is to produce a microstructure containing solely lower bainite, a form of bainite that develops during low temperature austempering, though small amounts of retained austenite are
Table 1. The chemical composition of the cast steel alloy from which the digger teeth first treated by AP were made.
acceptable, so long as the austenite is stable and will not transform into martensite during use. Therefore, AP applied a variety of heat treatments to a cast steel digger tooth commonly produced by the customer (Table 1) in an attempt to obtain the desired microstructure. The teeth were expected to austemper well, but the simulation used to draw this conclusion was developed for wrought steel rather than cast. The closest microstructure that AP was able to produce contained large amounts of martensite, an undesirable microstructure (Figure 2). It was determined by AP that full lower bainite could not be achieved in the customer's alloy without an additional heat treatment prior to austempering. However, the facilities at AP are not suited to the addition of such a step, and it was, therefore, determined that, rather than adjusting the treatment to produce bainite in the given alloy, a new alloy should be developed which will produce bainite when treated within AP's capabilities (Table 2).
B. Objectives
In order to meet the customer's specifications, AP has commissioned the research team to develop a cast steel alloy which can be austempered to produce a microstructure containing only lower bainite, without prior heat treatment. Per the customer's request, the final hardness of the heat-treated steel must be in the mid-fifties Rockwell-C and must have a nominal impact toughness of 25J at -40ï‚°C/F. AP has requested that the steel be through hardenable to a section thickness of at least four inches, in other words to have a hardenability factor of at least four inches, and that the alloy be designed so that the treatment takes a relatively short amount of time, specifically no more than three to eight hours. In order for this to be achieved, alloying elements must be selected to reduce solute segregation within the casting, to minimize the martensite start temperature so that the formation of this microstructure is more easily avoided, to place the bainite start temperature within the temperature range which is achievable by AP's facilities (Table 2), and to discourage the development of upper bainite. Though full lower bainite is the target of the project, AP has determined that retained austenite is acceptable, so long as it is stable and will not transform into martensite.
Table 2. The temperature ranges achievable in each stage of AP's austemper treatment equipment.
II. Literature
A. Bainite Research
1. Background of Bainite Research
Bainite was first recognized in the late 1920's by Davenport and Bain, who identified an "acicular, dark etching aggregate" unlike pearlite or martensite while studying isothermal transformations of austenite at temperatures above the martensite start temperature. [1] The name bainite was coined in 1934 by the research team at the United States Steel Corporation Laboratory in New Jersey. [2] In 1936, Vilella et al. proposed that bainite is formed by the abrupt development of flat plates of ferrite supersaturated with carbon, which would then diffuse into the surrounding microstructure, at a rate determined by temperature, to form carbide particles. [3] Three years later, Bain confirmed this postulation, and presented the finding that bainite is "tougher for the same hardness than tempered martensite." [4]
Competition between the ferrite and cementite known to comprise the bainite aggregates was proposed as the explanation for the variations seen in bainite, with the ferrite in control at high temperatures and the cementite in control at lower temperatures, by Greninger and Troiano in 1940, [5,8] and confirmed by Smith and Mehl in 1942. [6,8] In 1944, Klier and Lyman presented the incomplete-reaction phenomenon, by which the complete transformation of austenite to bainite was not observed at any temperature studied. Instead, pearlite was found to form in the residual austenite after a long delay for well-alloyed steels, and after a short delay for carbon steels. They went on to propose that bainite forms in the wake of the austenite segregating into carbon-rich and carbon-lean regions. [7,8] This, however, requires uphill diffusion. [8]
In 1946, Zener attempted to produce a thermodynamic model of the phase transformations of steel, based on the supposition that bainite growth is diffusionless, and that carbon super-saturation is relieved after the ferrite formation. He believed that, unlike martensite, there is no strain energy associated with bainite growth, but did not present an explanation for this difference. Zener also went on to propose that the limited size to which he observed the initial plates of bainite to grow was the result of cementite formation retarding further growth. [9,8]
In 1947, Hultgren confirmed that the cementite that forms during the bainite transformation is distinct from that which forms during the pearlite transformation with magnetic, chemical, and X-ray data. He went on to combine this idea and the idea that bainite is initiated by the nucleation of ferrite [10,8] which was proposed in 1939 by Mehl [11,8], to postulate that upper bainite, the bainite which forms at high temperatures, begins with the nucleation of ferrite of paraequilibrium carbon content, which enriches the residual austenite with carbon, and is a kind of reconstructive transformation. He also admits that lower bainite, the bainite which forms at low temperatures, may form with a supersaturation of carbon, but he does not discuss the mechanisms for the occurance of paraequilibrium conditions in bainite or for the lower bainite formation. [10,8] The idea that some microconstituents are formed by displacive transformation of the iron lattice, while others are formed by reconstructive transformation, was later explored by Buerger in 1951 [12,8] and then by Christian in 1965. [13,8]
In 1952, Ko and Cottrell demonstrated that bainite grows relatively slowly with the use of hot-stage light microscopy. They also showed that bainite formation alters the shape of the transformed region, the change being qualitatively characterized as an invariant-plane strain, and that bainite cannot grow past the prior austenite grain boundries or twins, a distinction from pearlite. [14,8]. Bainite was further distinguished from pearlite in 1962, when Hillert demonstrated that pearlite is formed by the cooperative growth of two intertwining crystals, one of ferrite and the other of cementite, while bainite forms as many plates of ferrite, which are followed by many particles of cementite. [15,16,8]
Based upon all of the various distinctions between bainite and pearlite, and bainite and martensite, Aaronson developed a microstructural definition for bainite in 1969, where bainite is considered simply a nonlamellar aggregate of ferrite and cementite which form consecutively, as opposed to cooperatively. Aaronson's model put the upper limit to bainite formation at the A1 without regard to the bainite start temperature. The distinct upper limit for bainite formation was explained as the result of different alloying elements, which affect the growth kenetics. [17,8] [1]
2. The Microstructural definition of bainite
In 2002, Aaronson summarized the microstructural definition of bainite as a "microstructural continuum" of pearlite, the structural variations from which result entirely from differences in the relative nucleation rates, and stated that bainite is a "non-lamellar, two-phase product resulting from the competitive mode of eutectoid decomposition," thereby dismissing the morphology of the ferrite as being relevant to the definition of bainite. [18] He went on to give descriptions of the overall reaction kenetics definition of bainite, and the surface-relief effect definition, but, as AP has used metallography to determine the bainite content of all samples produced and studied during the initial work on cast steel austempering, the research team will be using the microstructural definition to determine if a fully bainitic structure has been achieved.
B. Past work in cast steel austempering
1. Background of Austempering
The process of austempering, as discussed above, was first developed and implemented by Davenport and Bain for their research concerning the transformation of austenite below the critical temperature but above the martensite start temperature, the same research that lead to the discovery of bainite. [1] As austempering is the means of producing, all of the research discussed above was indirectly concerned with austempering, but the amount of work concerned specifically with the heat treatment process itself is considerably more limited. As Bain and Davenport's initial work was concerned primarily with steel, this was the first time the austempering process was applied to that material. Austempering was first applied to cast iron by Flinn in the early 1940's in a study of the affects of the treatment on gray cast iron, but would also be applied to ductile gray iron after its development in 1948, by the British Cast Iron Research Association (BCIRA) and the International Nickel Company (INCO). [19]
Even though the technology for austempered ductile iron existed by the beginning of the 1950s, it was not until the 1970s that it was implemented on a commercial scale, such as is produced at AP. In 1990, ASTM published the first standard specifications for Austempered Ductile Iron (ADI) in the United States. [19] The most up-to-date version of this standard was published in 2006, and includes the current grade definition for ADI. [20] ADI has been the subject of an increasing amount of study in recent years, but as the major body of this is inapplicable to the austempering of cast steel, it will not be further explored here.
Table 3. The compositions of the alloys found in the literature that are known to produce lower bainite when austempered as cast. Those alloys that best fit the criteria discussed in the objectives are highlighted. [21, 22, 23]
2. Alloys developed by various research teams
Though, as previously stated, austempering has been used primarily for the treatment of ductile iron and wrought steel, past work has been done in the field of austempered cast steel. Alloys which were found to contain carbides in lower bainite were developed by Austin and Schwartz in 1952, 1955, Matas and Hehemann in 1961, Deliry in 1965, Oblak and Hehemann in 1967, Lai in 1975, Huang and Thomas in 1977, Dorazil and Svejcar in 1979, Sandvik in 1982, Dubensky and Rundman in 1985, Miikhkinen and Edmonds in 1987, and by Honeycombe and
Bhadeshia in 1995. [21] [2] 2005 saw Mawella's patent on Bainitic steel, no. 6884306 B [22], and 2006, Voigt and Bendaly's work in austempering high silicon steels. [23] The compositions of these alloys are summarized in Table 3.
C. Effects of Alloying Elements on Bainite
1. Manganese
Manganese additions are best known to improve hardenability inexpensively, particularly in austenite. It is also known to harden ferrite, and form carbides more readily than iron. It has little effect on tempering when used in moderate quantities. [24]
2. Silicon
Silicon is a strong deoxidizer and is known to strengthen low-alloy steels. It is also known to increase hardness in ferrite and sustain hardness during tempering. Silicon has a negative tendency towards carbide formation. [24]
3. Phosphorus
Phosphorus improves hardness in ferrite and hardenability in austenite. It is also known to strengthen low-carbon steels and improve machinability in free-cutting steels. [24]
4. Nickel
Nickel strengthens unquenched and annealed steels and has a negative tendency towards carbide formation. It toughens ferrite and improves the hardenability of austenite, but tends to promote retained austenite. [24]
5. Chromium
Chromium hardens ferrite and improves the hardenability of austenite. It also has a greater tendency than Mn to form carbides. Cr also improves strength at high temperatures and improves wear resistance, but is very expensive. [24]
6. Molybdenum
Molybdenum additions delay the pearlite transformation and suppress the martensite start temperature with little effect on the bainite transformation isothermally. The effect on the pearlite transformation is the most pronounced of the alloying elements, followed closely by manganese, but the effect on the martensite transformation is exceeded by manganese. [25]
D. Microstructure Simulation
1. Bainite Start
In 1956, Steven and Hayes proposed a simple equation by which the upper limit of the isothermal transformation temperature range, also known as the bainite start temperature can be calculated. This equation was based upon the quantitative impact of various alloying elements on the bainite start temperature and is written:
BS(%C)=830-270(%C)-90(%Mn)-37(%Ni)-70(%Cr)-83(%Mo) [26,27]
However, this equation only applies to hardenable low-alloy steels that contain between 0.1 and 0.55% C. For low-carbon steels, containing between 0.15 and 0.29% C, Bodnar et al developed a second equation in 1989, which is written:
BS(%C)=844-597(%C)-63(%Mn)-16(%Ni)-78(%Cr) [26,28]
Based on the alloy compositions given in Table 3, the research team has used the prior equation almost exclusively to provide the bainite start temperatures given in Table 4
Table 4. The A3 temperature as calculated using ASM's equation, the bainite start temperater as calculated using Steven and Hayes' equation, the martensite start temperature as calculated with Andrew's product equation, and the critical diameter. The final selection is highlighted. [21, 22, 23]
2. A1 , A3 Temperature
The A1 and A3 temperatures denote the start and finish of the austenite transformation in iron and steel. According to Aaronson's 1969 microstructural definition of bainite, the upper limit of the bainite transformation is the A1 temperature without regard to the bainite start temperature. However, this is not mentioned in Aaronson's 2002 microstructural definition.
According to ASM International, the A1 and A3 temperatures can be calculated using the composition of the steel and the following equations:
A1(°C) = 723 - 20.7(% Mn)- 16.9(%Ni) + 29.1(%Si) - 16.9(%Cr) [29]
A3(°C) = 910-203(√%C)-15.2(% Ni)+44.7(% Si)+104(% V)+31.5(% Mo) [29]
These equations are valid for steels with carbon contents over 0.8%, however the A3 was calculated in Table 4 to determine the approximate austenization temperature for each chemistry.
3. Martensite Start
A similar equation has been developed for the temperature below which martensite will form. The first were proposed by Payson and Savage [26,30] and by Carapella [26,31] in 1944. In 1946, equations were proposed by Rowland et al., [26,32] Grange et al., [26,33] and by Nehrenberg. [26,34] The first Celsius based equation was published by Steven and Haynes, along with their equation for the bainite start temperature, in 1956. [26,27] In 1965, Andrews published both linear and product equations. [26,35] The last of these, Andrews' product equation, was used to calculate the Ms temperatures shown in Table 4.
4. Critical diameter
The critical diameter, also called the hardenability factor, is a quantitative means of estimating the depth to which an alloy can be through hardened. It is found as the product of the base diameter of the base metal and the multiplying factors of each of the alloying elements used, each of which will have some affect on the hardenability of the alloy. Both the base diameter and the multiplying factors are
found through the use of graphs, like the one shown in Figure 3, based upon the chemical composition of the alloy and the austenitic grain size. [36] The results of this simulation for the alloys listed are summarized in Table 4.
Figure 3. A graph used to estimate the multiplying factor of a given element based on the weight percent of that element in the alloy. [35]
5. Scheil equation
The Scheil equation, which is written
cS=kco(1-fS)k-1
where co is the initial solute concentration, cS is the solute concentration in the solid, cL is the solute concentration in the liquid, fS is the weight fraction of the material to have solidified, and k=cS/cL, was developed as a means of calculating the extent of solute segregation in a casting. Though it requires several simplifying assumptions, including but not limited to negligible undercooling, complete diffusion through the liquid phase, and little to no diffusion through the solid phase, it is still considered an accurate model of solute enrichment throughout the solidification problem. [26] Software programs, such as Factsage, can be used to run Scheil analysis of complex alloys, ie. containing more than two components, to find the composition of the last solid to form during the solidification process, which can be analyzed by the same process as the average composition to determine if all specifications will be met by the whole of the microstructure.
Table 5. Applying the Scheil equation to the two alloys from the literature that most nearly met AP's criteria, the composition ranges for which are shown on the left, gave the expected compositions of the last fifteen percent of alloys to solidify when cast, shown on the right. The starting chemistries can be found in Table 3.
6. Alloying elements that discourage upper bainite
In their 1995 book, Steels: Microstructure and Properties, Honeycombe and Bhadeshia proposed that a high carbon content, in excess of 0.6 wt%, discourages the formation of upper bainite, in favor of lower bainite. This recommends those
alloys which contain higher carbon contents. [8] The results of applying this simulation to the two most promising alloys are given in Table 5.
III. Experimental Procedure
A. Base Alloy Selection
It has been recommended by AP that the research begin to develop a steel alloy that will form a fully lower bainitic microstructure when austempered by beginning with an alloy that is known to form lower bainite, such as those discussed above, and to adjust the composition of that alloy to improve the desired properties. As several such alloys exist, the research team began the selection process by simulating the product of each cast material with Steven and Hayes' bainite start temperature equation, ASM International's A1 and A3 equations, Andrew's martensite start temperature equation, the Scheil equation, and the critical diameter (Table 4). Specifically, the team looked for an alloy that would discourage the formation of upper bainite, minimize solute segregation, produce a critical diameter of at least four inches, minimize the martensite start temperature, and place the bainite start temperature in the 350-740F temperature range that is readily achievable by AP's equipment. Based on the carbon content of the alloys, the hardenability factor, the bainite start temperatures and the martensite start
Table 6. The calculated bainite start temperature, martensite start temperature, and hardenability factor for the two most promising alloys and the last 15% of each to solidify, based on the compositions calculated with the Scheil equation.
temperatures, two alloys were selected for further analysis (Table 5). The Scheil equation was applied to these two alloys to determine the compositions of the last 15% of each alloy to solidify. These compositions were then simulated as all of the base alloys had been, based upon carbon content, start temperatures, and critical diameters, and as both were found to be within acceptable ranges (Table 6). Therefore, Sandvik's alloy was selected based on the ease of production, as it can be prepared by simply adding small amount of alloying elements to 4140 steel, whereas the Oblak and Hehemann alloy would require more difficult and less accurate processes to recreate.
B. Sample Production
1. The Pattern and Molds
A pre-existing keel block (Figure 4) will be modified to produce a cube approximately 4 and 1/16 inches on each side. This will produce a casting that weighs approximately 25 pounds. Based on industrial practice, a large riser will be
used to ensure complete filling and minimum porosity in each casting, resulting in a final casting weight of approximately fifty pounds. From this pattern, no-bake sand molds will be produced from F-60 silica sand, each of which will be marked with a distinct number or letter on the keel block so that each casting will be easily distinguishable after shipping.
Figure 4. The keel block pattern that will be used to produce the molds once the shortest dimension has been extended to about 4 and 1/16 inches.
2. Heat Preparation and Ladle Additions
Each heat will consist of two hundred pounds of steel, poured into four castings. For the first pour, this base material will consist of 4140 steel, carbon, ferro-silicon, and ferro-manganese additions. Each of the four castings from this heat will have a different composition, achieved through the use of additions in the ladle, so that the effects of the alloying elements can be calculated. The first heat will therefore include castings with low manganese and low silicon, high manganese and low silicon, low manganese and high silicon, and both high manganese and high silicon. The samples will be allowed to mold cool for one hour before shakeout and a water quench, per industry practice.
3. Heat treatment
The finished samples will be shipped to AP for heat treatment. Though the exact treatment will be selected by AP employees and is at this time unknown to the research team, the temperature ranges that the samples will be treated at, during both the austenization treatment and the austemper are outlined in Table 2. The turn around time for the treatment, the time between AP receiving the samples and sending them back, is expected to be less than one week.
C. Sample Testing
Once an alloy has been selected and samples have been cast and heat treated, each sample will be metallographically analyzed to determine the final percent bainite in the microstructure. Rockwell-C hardness and impact toughness testing will also be run to determine if the samples meet AP's specifications.
1. Metallography
The metallographic analysis of the casting samples will be conducted according to AP's standard technique. The polished samples will be etched with 2% nital and then treated with 10% sodium metabisulfite, which will tint the bainite present in the microstructure blue, making it readily distinguishable from any martensite or retained austenite present in the microstructure. Image analysis will then be used to determine the percent lower banite contained in the microstructure. As stated in the objectives, the target is full lower bainite, though some stable retained austenite is acceptable.
2. Rockwell-C hardness
The current standard method of testing the Rockwell hardness of metallic materials is described in ASTM E18-08b. [37] Recall that the target hardness is mid-fifties Rockwell-C.
3. Impact Toughness
The impact toughness of a metallic material is determined by the impact testing of notched bars. The current, standardized method for this test is described in ASTM E23-07ae1. [38] Recall that the target is a nominal impact toughness of 25J at -40ï‚°C/F.
D. Sample Modification
The results of the tests described above will be used to determine which of the four compositions most nearly meets AP's request, and which properties will need to be further modified. This information in conjunction with known recovery values will be used to select alloying elements to be added to the second heat. Like the first heat, the second will consist of four castings of varying compositions, produced with ladle additions.
IV. Conclusion
The literature has provided a variety of steel alloys known to produce bainite when austempered without prior heat treatment or processing. Based upon the results of the simulations previously discussed and summarized in Table 4, the two chemistries that most nearly meet the project's specifications were selected. Of these, the alloy that can be most easily produced in the available facilities, the Sandvik alloy, was chosen as the project's starting point. Likewise, silicon and manganese have been selected as the alloying elements to be varied in the first heat to see if the microstructures of the austemepered samples can be improved. Base and alloying metals will soon be acquired and the first heat poured during the semester break. This heat will produce four castings, each of a different composition, by adding alloying elements in the ladle just before pouring. This serves the dual purpose of testing the team's addition and pouring technique and of
Figure 5. The work which has been done and which remains to be done for the austempering cast steel project for AP.
allowing the team to determine the recovery rate of the alloys chosen for adjustment. The castings will then be sent to AP for austempering.
When the samples are returned hardness and impact toughness tests, as well as metallography will be performed. The results of these tests will be used to determine what alterations, if any, must be made to the alloy to improve the results of the second pour, which is to be preformed next semester. The results of the first heat can also be used to verify the accuracy of the simulations before they are used to estimate the necessary alloy adjustments. An outline of what work has been done and what remains to be done is shown in Figure 5.
