Investigation of Microstructure and Mechanical Properties in Hot-work Tool Steels

Transcription

1 DEGREE PROJECT IN MATERIALS SCIENCE AND ENGINEERING, SECOND CYCLE, 30 CREDITS STOCKHOLM, SWEDEN 2017 Investigation of Microstructure and Mechanical Properties in Hot-work Tool Steels TOMAS REY KTH ROYAL INSTITUTE OF TECHNOLOGY SCHOOL OF INDUSTRIAL ENGINEERING AND MANAGEMENT

2 Abstract Hot-work tool steels make up an important group of steels that are able to perform with good strength and toughness properties at elevated temperatures and stresses. They are able to gain this behavior through their alloy composition and heat treatment, which relies on the precipitation of alloy carbides to counter the loss in strength as the tempered material becomes more ductile. As demand grows for materials that are suitable for even harsher applications and that show improved mechanical qualities, the steel industry must continuously investigate the development of new steel grades. Within this context, the present work focuses on examining the mechanical properties and microstructure of two hot-work tool steels, of which one is a representative steel grade (Steel A) and the second a higheralloyed variant (Steel B), at different tempering conditions. To complement the experimental work, precipitation simulations are used to monitor the progression of secondary carbide precipitation and to examine the predicted microstructural changes through varying the alloy composition. The study finds that Steel B does not actually have improved properties with respect to Steel A and suggests that the precipitation behavior of both steels is virtually identical. Despite this, the simulation work reveals that this behavior can change dramatically to favor more positive hardness contributions by increasing the alloy content of V. In short, with the project being part of an ongoing investigation, there remain several areas of analysis that need to be completed before offering a complete picture that can ultimately play a part in the development of a new hot-work tool steel grade. Keywords: hot-work tool steel, secondary carbide precipitation, precipitation hardening, tempering, carbide reactions, precipitation simulation, microstructure i

3 Acknowledgements I would like to express my deep gratitude towards SSAB Oxelösund for providing me with the funding and opportunity to carry out this project. Especially, I would like to give special thanks to Dr. Magnus Andersson, my supervisor from SSAB Oxelösund, for guiding me through the project and offering advice and support at every step of the way. Also, I would like to acknowledge the rest of SSAB Oxelösund s Product Development department as well as the Metallography department for providing helpful input regarding the direction of the project and for providing training in sample preparation and the use of the lab furnace. I would also like to thank Dr. Fredrik Lindberg and PhD candidate Erik Claesson from Swerea KIMAB for their great help in preparing the XRD and TEM samples as well as carrying out the analysis. Additionally, I am thankful towards Associate Prof. Peter Hedström for his role as my KTH supervisor and for reviewing this work. Last but not least, I owe a lot of gratitude to my family and friends, who have stood by with constant love and support, and especially to my boyfriend, whose input was not only supportwise but also very helpful in ensuring that my thesis was at the best shape it could be. Tomas Rey ii

4 Table of Contents Abstract... i Acknowledgements... ii 1. Introduction HOT-WORK TOOL STEELS MARTENSITIC MICROSTRUCTURE HEAT TREATMENT ALLOYING CONTRIBUTIONS Precipitation Hardening Solution Hardening Alloy Carbides Carbide Reactions Factors influencing hardening contribution TEMPERING MEASUREMENTS SIMULATION SOFTWARE AIM OF THE PROJECT Methodology EXPERIMENTAL WORK Materials Tempering Trials Sample preparation and hardness testing Impact toughness measurements TEM AND XRD SAMPLE PREPARATION AND ANALYSIS PRECIPITATION SIMULATIONS Results and Discussion STEEL COMPOSITIONS HARDNESS RESULTS IMPACT TOUGHNESS RESULTS MICROSCOPY AND XRD RESULTS COMPUTATIONAL RESULTS Conclusion Recommendations for Future Work References iii

5 1. Introduction 1.1 Hot-work Tool Steels Hot-work tool steels are a class of tool steels able to maintain optimal properties at elevated temperatures. Because of this, they have a wide range of applications including extrusion tools, die casting, and forging dies. What sets hot-work tool steels apart is their high strength and toughness, which are the resistance to plastic deformation and resistance to crack propagation, respectively [1]. These properties are crucial for these applications, as finished products are required to perform in harsh conditions with widely varying temperatures and loads, where they must be able to keep their form. One of the best-known hot-work tool steels is AISI H11, a 5 wt% Cr martensitic steel used primarily for the die casting of aluminum alloys [2]. However, one serious drawback with this grade, and others similar to it, is that it requires products to be heat-treated after machining to achieve the correct properties. The problem with this is that overall production takes a longer time, properties may vary considerably depending on the heat treatment conditions used by individual companies, and there is a risk of dimensional changes during the heat treatment due to microstructural transformations and temperature gradients [3]. In response to these issues, Swedish steel company SSAB designed and manufactured a new line of hot-work tool steels under the brand TOOLOX. Though designed for the same uses as conventional hot-work steels, TOOLOX grades take a very different approach to manufacturing. Primarily, these grades are already heat-treated upon delivery and have a much leaner composition that allows for faster manufacturing times while still having similar, if not improved, mechanical properties [3]. A pre-heat-treated product means that the heat treatment done by the manufacturers is standardized, resulting in more consistent properties, and that time is not wasted trying to correct dimensional changes as machining is now the final step in production. In total, the TOOLOX brand has three separate steel grades, TOOLOX 33, 40, and 44, which differ with respect to their peak hardness after heat treatment, with TOOLOX 44 having the highest hardness at 45 HRC ( HV) and lowest toughness (30 J) at ambient temperature [4]. Hardness has been found to have a linear correlation with both yield and tensile strength in a wide variety of steels [5], meaning that hardness can be used to compare changes in a steel grade s strength. Even with the development of these versatile new steel grades, there is still a need to produce tool steels with even higher strength and hot hardness to be used in more demanding applications. 1

6 1.2. Martensitic Microstructure To begin to examine the development of steels with improved properties, it is imperative to understand the components that make up the steel. Steels obtain their properties from their microstructure, which is in turn determined by the steel composition, heat treatments, and cold working it has undergone. All of the previously mentioned tool steel grades are characterized by the presence of a martensitic matrix. Martensite is the metastable bodycentered tetragonal (BCT) phase formed through diffusionless transformation from the rapid quenching of face-centered cubic (FCC) austenite. In a binary Fe-C system, austenite is stable at elevated temperatures, beginning at 727 C, and is characterized by its relatively high carbon solubility, which reaches a maximum of 2.14 wt% [6]. This solubility is due to austenite s FCC crystal structure, which has larger octahedral interstitial sites where carbon atoms can readily diffuse. Below 727 C, denoted as the A 1 temperature, austenite becomes thermodynamically unstable and undergoes a eutectoid reaction, transforming into a mixture of body-centered cubic (BCC) ferrite and stoichiometric intermetallic cementite (Fe 3 C) [7]. Unlike that of austenite, ferrite s crystal structure has much smaller interstitial sites, which translates to a carbon solubility of wt% [6]. And so, as austenite transforms into ferrite, carbon atoms are forced out of the matrix, providing a strong driving force for the precipitation of cementite. Figure 1 shows the binary iron-carbon phase diagram at low carbon content, denoting the stable regions of austenite, ferrite, and cementite at equilibrium. Figure 1. Fe-C binary phase diagram at low carbon wt% [6] However, if the cooling rate is fast enough, carbon atoms do not have enough time to diffuse and instead they become trapped in the interstitial sites as austenite undergoes phase transformation. This results in a distortion in the crystal lattice of what would normally be 2

7 BCC ferrite. This phase is now BCT martensite and due to the shear stresses created by the distorted lattice, the steel now contains an extremely high dislocation density in the order of m -2 [8]. A high dislocation density means that dislocations form a tangled network, which impedes their movement. Because of this, a larger shear stress is required to move these dislocations, and this contributes to a sharp increase in yield strength in the material, known as quench hardening, which is characteristic for martensite [6]. The equation for quench hardening is shown in Eq. 1, τ = Kε! Eq.1 where τ is change in yield strength, K is the strength coefficient, ϵ is the strain in the material, and n is a coefficient that denotes the plasticity of the material and goes from 0 to 1. And so, for a given material with set coefficients, providing a larger strain on the matrix lattice increases the yield strength. This phase transformation is achieved through quenching, which is a cooling process in which the austenitized steel is rapidly cooled by being immersed in a medium such as water or oil. In addition to the cooling rate, martensite formation is dependent on temperature, only being able to start forming below a set temperature known as the M s temperature. The M s temperature is highly dependent on austenite grain size and alloying elements. It is above ambient temperature for low and medium carbon steels and generally shows a decrease with increased alloying [9,10]. If the M s temperature is low enough, it is possible for residual austenite to be present in the steel as the phase transformation to form martensite is incomplete, which would negatively impact the uniformity of properties throughout the material. The martensitic structure can be divided into two morphologies, lath and plate, based on the carbon content of the steel [9]. Lath martensite occurs in carbon steels with carbon contents between 0-0.6wt%, beyond which plate martensite begins appearing. In both of these cases, the previous austenite grain boundaries are retained as a hard boundary inside which the martensitic microstructures form. These differences in microstructure constitute important differences in properties, as the dislocation density, and as a result strength, reaches a peak as the carbon content increases to 0.6wt% [8,11]. For this reason, lath martensite is favored for applications requiring high strength. However, the increase in strength in martensitic steels is countered by a decrease in toughness. This leads to a material that is more prone to fracture, which is detrimental to its use in higher temperatures and in the presence of large stresses. To address this, martensitic steel is heat-treated at an elevated temperature below the A 1 temperature to produce a more ductile product in a process known as tempering [9] Heat Treatment Tempering involves a wide array of overlapping processes and reactions that together have a great impact on the final microstructure and properties of a steel. For carbon steels, these processes include segregation, carbide precipitation, retained austenite decomposition, and 3

8 recovery and recrystallization of martensite [9]. Carbon segregation is triggered during tempering at temperatures below 200 C and involves the diffusion of carbon atoms to interstitial sites that are close to dislocations. These sites are lower in energy and create carbon-rich zones within the martensitic matrix [12]. Between C, epsilon carbide, a HCP iron carbide with chemical formula Fe 2.3 C, precipitates along the carbon-rich dislocations. These metastable carbides are then replaced starting at 200 C by Hägg carbides (Fe 5 C 2 ) and ultimately by cementite, which forms between C [9,13]. Cementite initially appears in needle-like shapes, which slowly spheroidize with increasing tempering to reduce surface energy. In carbon steels, the precipitation and coarsening of cementite is a significant cause for the loss of strength in tempered steels [9]. In addition, if any austenite was retained after quenching, once it reaches C, it will decompose into bainite, a fine mixture of cementite and ferrite that has lower strength than martensite [9,14]. Recovery and recrystallization are two processes in steels that can reduce internal stresses and defects, and typically occur above 400 C for recovery and 600 C, for recrystallization. During recovery, dislocations, which are stuck in a dense network, start to move around and annihilate dislocations that face the opposite direction. Subsequently, remaining dislocations rearrange themselves to form aligned structures [15]. In recrystallization, new grains without dislocations nucleate and grow, replacing the old deformed grains. These two processes result in an additional drop in strength and increase in ductility, resulting in a lower dislocation density. However, the rate at which the processes happen can be significantly slowed down by secondary phases such carbides, as these have precipitated on the dislocations themselves. A summary of some of the processes present during tempering as well as their impact on hardness can be seen in Figure 2. Figure 2. Hardness in different martensitic carbon steels after 1 hour of tempering at varying temperatures [9] 4

9 1.4. Alloying Contributions So far, tempering in martensitic steels has been considered for simple binary carbon steels. However, what characterize hot-work tool steels are the alloying elements added during production, which have a profound impact on the properties and microstructure of tempered steels. Common alloying elements include Si, Cr, Mo, Ti, Nb, V, Mn and Ni. These alloy elements may be categorized based on the effect that they have on the steel s microstructure and properties. One type of alloying elements includes those that are strong carbide formers, such as V, Ti, Nb, Mo, Mn and Cr, while the other type remain in solution, including Si and Ni [16]. Within the group of carbide-formers, there are even those, namely Ti, V, and Nb, that readily form carbides with an alloy content of less than 0.1 wt%, known as microalloying elements [17]. The main role of alloying elements in tempered tool steels is to compensate for the drop in strength and hardness that is seen in heat-treated carbon steels through a process called secondary hardening, and this is done primarily through two hardening mechanisms, precipitation and solution hardening. This increase in hardening may even be larger than that of the original peak hardness after quenching (Figure 3). Figure 3. Secondary hardness peak in tempered martensitic steel at varying Mo concentrations [9] 5

10 1.4.1 Precipitation Hardening Precipitation hardening is a hardening method that involves the precipitation of secondary phases. As a secondary phase precipitates out from the primary matrix phase, the moving dislocations in the matrix encounter an obstacle, as the new phase distorts the matrix lattice. Depending on the size and hardness of these particles, the dislocations either have the ability to cut the smaller particles (particle cutting) or form a loop around larger ones (Orowan looping) [6]. Bypassing the particles requires an increase in shear stress, adding to the total yield stress of the material. Figure 4. Precipitation hardening mechanisms: Cutting (top) and Orowan looping (bottom) However, particle cutting and the Orowan mechanism show a different relation between particle radius and yield stress, as shown in Eq. 2-3, respectively, τ =!"#!" τ =!"!!!! Eq. 2 Eq. 3 where r is particle radius, L is distance between particles, γ is shear stress, b is the Burgers vector, and G is the shear modulus. What this shows is that as the particle radius increases, the strength contribution from particle cutting increases while that of Orowan looping decrease. Since particle cutting is favored at lower particle sizes and looping at higher ones, there is a peak hardness at a critical particle radius where the two processes occur [6] Solution Hardening Solution hardening follows a similar mechanism to that of precipitation hardening. However, instead of involving a secondary phase, hardening is achieved by the alloying elements dissolved in the matrix. Since alloying elements, such as Ni, have a different size than those in the pure matrix, in this case Fe, the lattice is distorted to accommodate the impurity. This distortion inhibits the mobility of dislocations, which again increases the yield strength of the material [6]. Eq. 4 shows the parameters governing solution hardening, 6

11 τ = Gb cε!! Eq. 4 where c is solute concentration and ϵ is the strain the solute puts on the lattice. The yield stress will be higher if both the strain and the solute concentration are increased Alloy Carbides The addition of alloying elements radically changes the microstructural evolution of the tempering process described earlier. Instead of a steady decrease in strength due to spheroidization, recovery, and recrystallization at higher temperatures, starting at 500 C, alloy carbides begin to precipitate as finely dispersed secondary phases, gradually replacing the larger cementite particles [9]. The more common carbides that precipitate at this stage include MC, M 2 C, M 7 C 3, M 23 C 6, and M 6 C, where M denotes a transition metal atom. All of these carbides are thermodynamically more stable than cementite, but they are kinetically less favorable since these alloying elements diffuse substitutionally, compared to carbon s interstitial diffusion [18]. This is because atoms of substitutional alloying elements must diffuse into vacancies in the matrix, since they reside within the lattice sites, while carbon atoms can diffuse regardless of the vacancy concentration (Figure 5). And so, carbon atoms can move relatively undeterred and faster compared to alloying elements, which are constantly slowed down by the other atoms in the lattice. Figure 5. Diffusion mechanism for substitutional (top) and interstitial (bottom) atom [6] The precipitation of a particular alloy carbide is dependent on the alloy composition of the steel and on how the different alloying elements interact with one another [19]. Furthermore, studies have reported that alloy carbides undergo a series of carbide reactions, where a progression of carbides precipitates sequentially, ultimately reaching the most stable carbide at equilibrium [2,18-20]. Certain carbides are rich in a particular element and become more stable with an increasing alloy composition. For example, M 7 C 3 is typically Cr-rich, MC is V or Nb-rich, and M 2 C is Mo-rich. The remaining carbides show a much wider variation, with 7

12 M 23 C 6 s being able to contain large amounts of Mo in addition to Cr and M 6 C s accepting Fe despite favoring Mo [19,20] Carbide Reactions In a study done by Tamaki and Suziki, the carbide microstructural evolution for the Fe-Cr- Mo-C quaternary system was examined with a varying composition of Cr and Mo [19]. The trials consisted of quenched martensitic steels tempered between C and between hours. Carbon content was fixed at 0.13wt% while Cr and Mo contents ranged from 0-5 wt% and 0.5-1wt%, respectively. The study found two primary carbide reactions, one for when Cr content is higher and another when Mo is dominant. In the former case, the carbide reaction progression is and in the latter case M! C M! C! M!" C! Eq. 5 M! C M! C M!" C! Eq. 6 where M 3 C refers to a cementite phase (Fe 3 C) with alloying elements present. When significant amounts of both Cr and Mo were present, the two carbide sequences were reported. Additionally, in a study conducted by Bungardt et al., where the Fe-Mo-C tertiary system was investigated at 700 C, a further carbide sequence was reported at an increased Mo/C content ratio (21): M! C M! C M! C Eq. 7 Already with a quaternary system, the interactions among alloying elements are significant and result in a complicated web of carbide precipitation. The situation becomes even more complex as more elements are added to the steel, being representative of a tool steel s actual composition. Vyrostkova et al. conducted a comprehensive study looking into the carbide reactions and equilibria of four tool steel grades tempered from C for hours [20]. The steel grades contained the following compositions: Table 1. Steel grade compositions in Vyrostkova et al. study [20] Steel C Mo Cr Mn Si V Here, there is mixture of carbide-forming (Mo, Cr, Mn, V) and non-carbide-forming (Si) alloying elements. It is important to first note the behavior of Mn in tool steels. While 8

13 classified as a carbide-former, Mn does not form its own alloy carbide, instead dissolving readily into cementite [17]. Additionally, it also exists dissolved in the martensitic matrix, contributing to a significant hardness increase through solution hardening. This means that Mo, Cr, and V are the primary concerns in the study, as they are the ones involve in alloy carbide formation. All alloy compositions are kept roughly constant for all 4 grades except those of Mo, which rises for Steel 4, and V, which increases for Steels 1-3 and is then kept constant for Steel 4. Results from this study can be best expressed through a series of time vs. temperature diagrams showing the metastable regions of each carbide (Figure 6). Figure 6. Time vs. temperature curves of metastable carbide formation [20] In terms of general carbide trends seen in the steel samples, cementite (M 3 C) is present at a very wide range of conditions, disappearing only during tempering runs at both long times and high temperatures. M 23 C 6 and M 6 C in this study are classified by the authors as not following the classical carbide sequence, meaning that their nucleation does not follow predictable patterns, unlike M 3 C, M 2 C, M 7 C 3, and MC [20]. Focusing on the classical carbides, the first carbide reaction is the formation of M 7 C 3 from M 3 C, followed by further dissolution of M 3 C to form M 2 C in the steels where it is present. MC is shown to precipitate either from existing M 2 C or directly from M 3 C [22,23]. Even M 6 C is believed to behave this way, as it nucleates at M 2 C particles in Steels 2-4 and from M 3 C in Steel 3 [24,25]. Ultimately, the appearance of M 23 C 6 so early during tempering is not discussed, although it may be have been present before tempering. However, a further carbide reaction, involving the breakdown of M 23 C 6 to form M 7 C 3, is posited to show what happens to M 23 C 6 as it becomes thermodynamically unstable at higher temperatures [20]. 9

14 Changes in steel composition among the four grades appear to have little effect on M 3 C stability. However, the Mo content and the ratio of V/Mo are very significant for the precipitation and stability of most of the remaining carbides. M 23 C 6 is stabilized by both Mo and V, as well as by a low V/Mo ratio. Both M 6 C and M 2 C precipitation are slowed down by low Mo and V contents, with M 23 C 6 s presence s providing a source of Mo depletion. Additionally, a high V/Mo ratio and low Mo content boost MC precipitation, as this increases the V content in the matrix as well as slows down the precipitation of M 6 C and M 2 C, which can take in V [20]. Finally, being a Cr-rich carbide, M 7 C 3 precipitation experiences little change in this study Factors influencing hardening contribution Identifying the alloy carbides that precipitate during a given tempering condition as well as the carbide reactions and the role that different alloying elements play shows how the microstructural evolution can create a suitable environment for secondary hardening. However, it does not describe the degree to which a particular dispersed alloy carbide phase will contribute to precipitation hardening in tool steel. For this, other factors must be considered, such as the total carbide content in the microstructure, the shape and size of the carbide particles, and whether some carbides are more effective at hardening than others. The total carbide content in a tool steel refers to the mole fraction of all the carbides precipitated in a sample. Increasing this amount, either through increasing the carbon content or the alloy concentration, can still lead to an increase in strength even if the precipitation sequence is not affected [18]. This was seen in a study done by Gingell and Bhadeshia which looked at carbide precipitation in three tool steels. These steels differed primarily on Cr, Mo, and C concentrations. However, after microstructural analysis, it was noted that there was little difference in the carbide size and distribution between the three steels at similar tempering conditions, despite a clear increase in hardness for the steel with slightly higher C content and high Cr and Mo contents. And so, the hardness increase was attributed to the higher mole fraction of carbides, rather than any microstructural change, meaning simply that there were more carbides present. The shape and size of a carbide particle have a significant impact on hardness. As mentioned earlier, precipitation hardening has a critical radius that denotes where hardness contributions from the sum of particle cutting and the Orowan looping are highest. As different alloy carbides all have different preferred orientations and shapes after nucleation, as well as different rates for growth and coarsening [2,18], it can be difficult to determine what the size of these carbides should ideally be to optimize hardness. In Cr-Mo-V steels, it has been shown that the greatest hardening contribution is provided by Mo and V-rich M 2 C carbides while M 7 C 3 and M 3 C have a less significant effect [26]. In particular, Cr-rich carbides (M 7 C 3 and M 23 C 6 ) are less favorable for hardening because of 10

15 their high coarsening rates, forming relatively large, globular particles [2,9]. On the other hand, M 2 C precipitates as thin sheets of needles while MC appears as thin plates, making them relatively better for precipitation hardening [9,18]. This distinction between inherently coarser or finer carbides is due to the crystal structures of the carbides which can be divided into two categories. The first includes carbides with complex crystal structures (M 3 C, M 23 C 6, M 7 C 3, M 6 C) while the second contains those with a simple crystal lattice (M 2 C, MC) [17]. The most finely dispersed phases come from the second group and generally have lower heats of formation, such as VC and NbC, compared to those in the first group that tend to form coarser dispersions [27]. An example of this can bee seen in TEM imaging of a tempered martensitic matrix containing M 23 C 6 and MC (Figure 7) after two hours. Here, M 23 C 6 particles dominate with nm radii while MC particles have radii below 40 nm [2]. It is thus beneficial to try to design alloys and heat treatments that encourage the nucleation of these finer carbide while limiting the presence of coarser ones. MC M 23 C 6 Figure 7. TEM imaging of tempered martensitic matrix (2). One final important topic to address affecting the formation of carbides in hot-work tool steels is the role of Si. Si is widely reported to inhibit the nucleation and growth of cementite, preventing the softening of the steel by keeping carbon in the matrix solution [2]. This has great consequences for the precipitation of alloy carbides, since these are nucleated from existing cementite particles. Cementite inhibition is due to Si s very low solubility in 11

16 cementite, and this causes Si to be expelled from the carbide and accumulate at the carbidematrix interface [28]. The Si flux moving away from the carbide is opposite to that of carbon going towards it, and this reduces the carbon flux that forms cementite. However, alloy carbides will actually form faster due to the fact that the Si flux is parallel to the carbon flux during cementite dissolution and that there is more carbon available in the matrix to separately nucleate a dispersed phase [2]. The overall effects this has on the tool steel are that the carbide size is smaller and volume fraction is higher for steels with higher Si content, as there are more nucleation sites for the alloy carbides. Also, the secondary hardness peak moves towards lower temperatures as a response to the faster alloy carbide precipitation (Figure 8). Figure 8. Si effect on secondary hardening peak [2] As can be seen, the many parameters and processes make producing desired properties very challenging, and control in both the composition of the steel and heat treatment are crucial. For example, even if the steel is produced through clean methods and has the desired composition, the martensitic matrix itself may not have the correct composition. This happens when there are already carbides present in the steel from early steps in production, such as annealing or austenitization [2]. This is due to either the carbides being thermodynamically stable above the austenitization temperature or there not being enough time for the carbides to dissolve back into the austenite matrix. This means that the martensitic matrix composition at the as-quenched state has lower alloy composition, which reduces the amount of secondary carbide precipitation and must be taken into account Tempering Measurements During tempering, the two relevant parameters taken into account are temperature and time. However, the work of Hollomon and Jaffe led to the creation of a tempering parameter meant to incorporate the effects of both time and temperature [29]. The parameter s validity rests on the assumption that the degree of tempering of a steel sample, which can be measured as the change in hardness, at a high temperature and short time is equal to one at a lower temperature and longer time. The Hollomon-Jaffe parameter takes the following form: 12

17 P! = T! C + log t 10!! Eq.8 where T C is temperature in Kelvin, t is time in hours, and C is a constant based on the type of steel. The Hollomon-Jaffe was originally designed to monitor tempering in carbon steels and because of this it may not be suitable for some types of steels. For example, a similar formula looking at hardness decrease was needed for 5% Cr hot-work tool steels [30]. Still, it is appropriate for use in low-alloyed steels, where the parameter constant is assumed to be 20 by convention [31]. Despite its possible limitations, the Hollomon-Jaffe parameter is useful for describing tempering processes in a simple manner and for comparing the extent of different heat treatments. As mentioned earlier, hardness correlates linearly with yield stress. This becomes important when measuring the mechanical properties of steel samples, as microhardness testing offers non-destructive measurements as well as local changes in properties. These measurements affect only the surface and give a hardness profile across a sample, allowing one to see variations within the sample (32). Testing consists of a series of indentations done on the steel s smooth surface, each with a specific load and indentation time. On the other hand, tensile testing used to directly measure yield strength destroys the material and provides a single value Simulation Software Recent studies involving carbide precipitation have increasingly used the aid of computational models to examine the thermodynamics and kinetics of their respective steel grades [18,33]. These software programs, such as Thermo-Calc, are based on the CALPHAD method. The CALPHAD method relies on the collection of experimental and theoretical data about the thermodynamic properties and phase equilibria of a variety of systems ranging from pure elements to complex alloys. This data is described through their Gibbs free energy and inserted into a model that can then calculate the equilibrium conditions of new systems [34]. Additionally, Thermo-Calc has a precipitation module called TC-Prisma that takes into account thermodynamic data as well as atomic mobilities to simulate the nucleation, growth, and coarsening of particles in a given system. Within TC-Prisma, some of the most important adjustable parameters for precipitation modeling include the number of nucleation sites and the interfacial energy. The former dictates the number of individual sites where a new phase can nucleate, and this is closely related to where the precipitation occurs (i.e. grain boundary, grain edge, dislocation, etc.). For example, for a phase that primarily appears along dislocations in a crystalline matrix, TC- Prisma calculates the nucleation sites in the following way: N! = ρ!!!!!!!! Eq. 9 13

18 where ρ! is the matrix s dislocation density, N A is Avogadro s number, and V!! is the molar volume of the matrix [35]. Interfacial energy is crucial for determining the rates of nucleation, growth, dissolution, and coarsening during precipitation models. It refers to the free energy per area associated with the interface between the matrix and secondary phases that arises from changes in chemical bonding at the interface. This value can vary widely from around J/m 2 and TC-Prisma uses the extended Becker s model and data from its databases to calculate it [35]: σ! =!!!!!!!! E! Eq. 10 where n! is the number of atoms per unit area at the interface, z! is is the number of cross bonds per atom at the interface, z! is an atom s coordination number inside the crystal lattice, and E! is the energy of the system involving the matrix and secondary phase. These computational tools are useful for examining the microstructure of tool steels because they show the relative stability of phases both in and out of equilibrium. Furthermore, one can use such models in conjunction with experimental data, making it possible to fit the data to the model and then being able to vary parameters such as alloy composition or dislocation density without the need for additional physical samples Aim of the project The steel industry has come a long way in designing steel grades for an increasing amount of applications. Particularly, SSAB s creation of the TOOLOX brand has improved upon the previous class of hot-work tool steels, resulting in grades with enhanced properties, greater reliability, and lower manufacturing times. Nevertheless, it is still an important goal to build upon the brand s success and strive for products that have even better mechanical properties. The present work is aimed at identifying possible candidates for a new hot-work tool steel grade that has higher strength and hot hardness while not sacrificing toughness. To provide relevant comparison, the study will examine a contemporary and representative tool steel grade and a higher alloyed variant of it. The primary focus will be on the microstructural evolution of these tool steels during tempering and the role of alloy carbides in secondary hardening, including carbide types and particle size distributions. This structural data will then be linked to changes in properties, namely hardness and impact toughness, in order to explain the steels behavior. Precipitation simulations and thermodynamic calculations will also be used to map out a wide range of possible compositions and heat treatments which will help to determine what parameters a new hot-work tool steel grade would need to have. And 14

19 so, this project takes a multi-faceted approach involving experimental, analytical, and computational components. The investigation should be seen as an ongoing project that ultimately exceeds the scope and timeframe of the work described here. As such, this paper functions as a basis for future experimentation. Upon completion of the whole project, it is hoped that the findings will help SSAB in deciding what approaches to take as it seeks to develop a new steel grade. 2. Methodology 2.1. Experimental Work Materials As mentioned, two hot-work tool steels were used in the experimental portion of the investigation. The first, Steel A, is a quintessential lean hot-work tool steel grade and the second, Steel B, is a slightly higher-alloyed variant. Their chemical compositions, obtained through chemical analysis from SSAB s Testing House, are shown in Table 1. Table 1. Chemical Composition (wt%) of Tool Steel Grades C Mo Cr Mn Si V Ni Nb Ti N Steel A Steel B It is important to highlight the compositional differences between Steel A and Steel B. Steel B has a slightly higher C content, a 10% increase in Cr content, 25% in Mo content, and a doubling of Ni content. And so, it is hoped that these grades are compositionally different enough that one can attribute difference in behavior to their composition, yet still similar enough to show microstructural patterns common to the hot-working tool steel family. The two grades were produced in-house at SSAB Oxelösund and were received in the asquenched (Q) state. They were originally manufactured as large 70mm thick slabs, which were originally sectioned at the Testing house to several 235x250x70mm blocks. One block of each steel grade was further sectioned in order to obtain samples of suitable size for subsequent hardness analysis. 33 samples of each steel grade were requested, and these samples measured 20x20x70 mm, keeping the thickness of the original slab intact. To avoid issues such as carbon diffusion at the edges, the sectioning was done 10 mm from the edges (Figure 9). 15

20 235 mm c 20 mm 240 mm 10 mm Figure 9. Top view sketch of as-quenched steel slab and location of sectioned samples Tempering Trials A series of tempering trials were conducted for both steel grades. These are characterized by a steady heating ramp up to the desired tempering temperature, a holding time at that temperature, followed by rapid quenching with water. The first tempering trials varied both the tempering temperature, ranging from C, and the holding time, ranging from no holding time (0h) to 24 hours (24h). Table 2 shows the full overview of these tempering trials. Table 2. Tempering Trial Overview Temperature ( C) Holding Time 550 0h 0.5h 1h 4h 24h 575 0h 0.5h 1h 4h 24h 587 0h 1h 600 0h 0.5h 1h 4h 24h 612 0h 1h 625 0h 0.5h 1h 4h 24h A Carbolite-Gero CWF 12/65 laboratory furnace with a 301 PID furnace controller was used for the heat treatments. In addition to the temperature readings from the furnace itself, a thermocouple was fitted to each sample to monitor the temperature at the center of the sample. Every furnace trial included one sample from each grade, and the samples were placed approximately 2 mm from one another to ensure that they experienced the same heating. Since these heat treatments were meant to represent those that are done at an industrial scale, heat curves taken from the industrial tempering process for these steel types were used as a standard to establish heating rates and timescales. However, the furnace controller only allowed treatments consisting of one heating ramp and a set dwell time, 16

21 resulting in simplified heat curves that do not fully replicate the industrial curves at lower temperatures, as they show consistent undercooling towards the beginning of the treatment (Figure 10) Heat Treatment from Standard vs. Experimental Temperature ( C) Standard center Standard edge Exp Time (s) Figure 10. Comparison of Standard vs. Experimental Heat Treatment Curves at 600 C The point at which the standard heating curves reaches the tempering temperature, which was 70 minutes into the heat treatment, was designated as the zero point (0h), and all the tempering trials were designed to approach their target temperature at this time. Figure 11 shows the heating curves for all of the samples, taken from one of the thermocouples inside the samples. 17

22 0h 0.5h 1h 4h 24h Figure 11. Compilation of Heat Treatment curves Additionally, a second series of tempering trials was done for samples that were to be used for impact toughness testing, as the size of the hardness samples did not accommodate that used for this testing. For these tests, full-sized blocks of both steel grades were used for the heat treatment. The tempering conditions were at 600 C and 550 C at 0, 4, and 24h holding times. Rather than side-to-side, the size limitations of the lab furnace mean that one block was lying on top of the other, with the Steel B slab on top. Temperature was recorded using a thermocouple attached to the top surface of the Steel B slab. Due to the size and weight of the slabs, after tempering the samples were allowed to air-cool, resulting in a much slow cooling compared to the quenched hardness samples Sample preparation and hardness testing Each small tempered sample was ground and polished down to 1 µm diamond paste using automatic grinding and polishing equipment. Samples subject to longer tempering times had longer initial grinding steps, taking off approximately 1 mm off the surface instead of the usual 0.2 mm to avoid areas where carbon may have diffused out of the sample during tempering. Hardness measurements were subsequently done with a Qness Q30 microhardness tester down the length of the samples polished surface. Vickers microhardness testing was performed with a total of 70 indentations, using a 5 kg load with a loading time of 5 s (HV5/5) and a distance of 1 mm between indentations. The mean value of all the indentations in a sample was considered to be the hardness result. 18

23 Impact toughness measurements Impact toughness was measured using the Charpy V-notch test after the block samples were machined to the standard sample size of 10x10x55 mm [36]. Machining and the actual testing were done by the Testing House, where six samples were taken from each block, and the test was run at room temperature and at -20 C, using three testing samples for each temperature. The mean value of the impact test was considered the toughness result TEM and XRD sample preparation and analysis X-ray diffraction (XRD) and transmission electron microscopy (TEM) investigations were carried out in the Swerea KIMAB laboratories in Kista, Stockholm to examine the microstructure of the tempered samples. Six samples were used for this analysis, consisting of trials done at 600 C for both steels at 0, 1, and 24h holding times. The samples were cut into roughly 20 mm cubes and subsequently sectioned parallel to the polished surface. One of the pieces was used to make a carbon replica for TEM analysis and the other one to carry out an XRD investigation. These two methods would be used together to analyze the phases and components in the steels microstructure. These two halves were further ground and polished by hand up to 0.25 µm diamond paste. The carbon replica, which is essentially a thin graphite film that extracts carbide particles from the surface of the steel, was created by first etching the samples for 5 minutes in 2% Nital solution. A carbon layer was then coated onto the etched surfaces using a GATAN Model 682 Precision Etching Coating System (PECS) with a graphite target. The thickness of the layer was set at approximately 20 nm. After coating, the sample surface was scored into approximately 2x2 mm squares and the samples were each submerged in a 10% Nital solution until the carbon films detached from the surface. These films were rinsed in ethanol and distilled water before being mounted onto square mesh copper grids. A 200 kv JEOL JEM 2100F Transmission Electron Microscope was used to visualize the particles in the replica. Analysis consisted of identifying each type of carbide through Energy-dispersive X-ray Spectroscopy (EDS) detector and mapping out the particle size distribution for each sample. The XRD analysis was done as a complement to the TEM work to try to confirm the identity of the precipitated particles through their diffraction patterns. Additionally, limited light optical microscopy (LOM) was done at SSAB using unsectioned hardness samples Precipitation Simulations Computational simulations were carried out in Thermo-Calc-2016b using the TC-Prisma Precipitation module. Thermodynamic and mobility databases, CCTMA_Fe3C and CCTMOB, respectively, were supplied by KIMAB. This component of the project was split 19

24 into three parts. In the first part, the thermodynamic phase equilibrium was calculated for both Steels A and B between C to see the relative stability of carbides at equilibrium. Following this, precipitation simulations were run to mimic the experimental trials. The goal was both to compare how similar the models were to the experimental results as well as to help explain why the results look as they do. For this, the steel compositions used for these simulations were the composition of austenite at 930 C calculated at equilibrium for each grade. This was done to model the Q state, as the composition of austenite was assumed to be the same as that of the quenched martensite. Using these compositions for each steel grade, precipitation models were run isothermally for 24 hours at four different temperatures: 550 C, 575 C, 600 C, and 625 C. The matrix was assumed to be fully martensitic, denoted by choosing ferrite with a dislocation density of 5E-14, which is within the range of lowcarbon martensitic steels [8]. This ignores the presence of additional phases found in the austenization equilibrium, and so the growth of these particles is not considered. The possible carbides considered for precipitation were cementite, M 2 C, MC, and M 7 C 3. Molar volumes and interfacial energies were both calculated by TC-Prisma from the databases, and all precipitation was assumed to occur along dislocations. Finally, TC-Prisma assumes a spherical particle geometry. The third part of the investigation involved mapping out the carbide precipitation in a series of steel grades made by varying the C, Mo, and V content of Steel A. This was done to get a sense of how these different steel grades would alter the amounts and sizes of carbides and how they could lead to enhanced mechanical properties. These simulations were all done at 600 C using the previous austenite composition as the base martensitic matrix. Then, the matrix composition itself was directly changed to higher alloy contents. The V content was varied from wt%, the Mo content from wt%, and the C content from wt%. 3. Results and Discussion 3.1. Steel Compositions An overview of two steel grades compositions offer a good starting point for discussing their possible structures and behaviors. Containing a varied mix of carbide-forming and noncarbide forming elements, these steels are influenced by many factors that can determine carbide stability and mechanical properties. For example, the relatively high Si content in both grades is primarily meant to inhibit the growth of cementite, which will slow down the drop in hardness during tempering. Also, the C contents chosen for the steels (0.32 and 0.34 wt%) fall well within the approximate range of lath martensite and because both steels are relatively low-alloyed, their M s temperature is assumed to remain above ambient temperature. This means that lath martensite is expected to be the dominant feature in the microstructure 20

25 after quenching. Finally, the presence of several very strong carbide-forming microalloy such as Nb and V will help stabilize their respective carbides and increase precipitation. As highlighted previously, Steel B differs from Steel A by having increases in C, Cr, Mo, and Ni content. The purpose of this is to increase the amount and rate of carbide precipitation by providing more available C in the matrix and further stabilizing Cr and Mo-rich carbides. As Mo sees the largest increase in content, it is hoped that faster growing rates and a larger volume fraction for M 2 C will lead to a noticeable increase in peak hardness. Also projected to increase hardness is the doubling of Ni content through its role in solution hardening and increased hardenability Hardness Results As discussed previously, hardness measurements were taken along the entire thickness of each sample. Figure 12 shows an example of the Vickers hardness measurements across all the Steel A samples at 625 C Steel A 625 C Hardness Data Hardness (HV5) Distance (mm) Q 0h 0,5h 1h 4h 24h Figure 12. Hardness profile for Steel A samples at 625 C One thing that first stands out is the stability of the hardness measurements across the profile of the samples. Using the hardness data from Steel A s 625 C trials, it can be seen that there is no noticeable dip in hardness towards the center of the sample. Normally, a loss in hardness would be seen if the cooling rate at the center of the block during quenching was lowered enough to trigger the formation of bainite instead of martensite. And so, this suggests 21

26 that there was a large degree of temperature uniformity during manufacturing, and tempering as well, and that it is reasonable to assume a uniform martensitic matrix across these samples. The main hardness results are shown in three separate forms. The first looks at hardness over time (Figure 13), the second versus temperature (Figure 14), and the third versus the tempering parameter (Figure 15). For clarity, the data for each steel grade is shown in separate figures for the first two cases. As tempering temperatures and times are increased, the average hardness of both steel grade tends to drop. At constant temperature (Figure 13), this happens as cementite has more time to grow and coarsen while at constant time (Figure 14), the growth rate increases. The hardness that is lost relative to the least tempered samples varies greatly depending on the extent of the tempering conditions, with the change in hardness lying anywhere from 15 HV along samples at constant 550 C and those at constant 0h to exceeding 100 HV along samples with more extreme conditions. However, there are a few significant exceptions to this general trend. One example is Steel B s 575 C curve versus time, where the hardness at 1 hour is actually higher than that of the two previous timepoints. This is good evidence that there is some type of hardening contribution, such as an alloy carbide, which counters the overall softening of the steel at that particular tempering condition. Hardness (HV5) Steel B Hardness Data vs. Time 550 C 587 C 575 C 600 C 612 C 625 C Time (s) Hardness (HV5) Steel A Hardness Data vs. Time 550 C 575 C 587 C 600 C 612 C 625 C Time (s) Figure 13. Hardness measurements versus time for Steel B (top) and Steel A (bottom) 22

27 500 Steel B Hardness Data vs. Temperature Hardness (HV) Temperature ( C) 0h 0.5h 1h 4h 24h 500 Steel A Hardness Data vs. Temperature Hardness (HV) Temperature ( C) 0h 0.5h 1h 4h 24h Figure 14. Hardness measurements versus temperature for Steel B (top) and Steel A (bottom) The hardness curve versus the Hollomon-Jaffe parameter (Figure 15) allows one to track the changes in hardness as the degree of tempering increases, which makes it easier to pinpoint regions where secondary hardening may be present. It also allows one to directly compare the two steels in terms of hardness. The general trend sees the hardness staying largely stable between tempering values of 15.5 and 17.5 before increasingly dropping. Within the first range, the Steel B grade has a hardness that is approximately HV than Steel A. This is not an especially large difference, especially considering that the standard deviation of the measurements within each sample average around 9-10 HV. Still, Steel B is consistently harder until the samples reach the highest temperatures and longest times, at which point the rate of softening is faster in Steel B. 23

28 600 Hardness Data vs. Hollomon Parameter 550 Hardness (HV5) TX44 Steel A TS44 Steel B Hollomon Parameter Figure 15. Hardness measurements versus Hollomon-Jaffe parameter Taking a closer look at the stable region of the figure, there are two particular regions in both steels where secondary hardening appears probable: the areas around 17 and 17.5 tempering values. Here, there are several points that break the trend of decreasing hardness values, although the degree of hardening is limited to a maximum of 12 HV. While this difference is, again, not that significant, it is nonetheless interesting to see this pattern present in both steels. If secondary hardening were indeed present, this would indicate that there are areas where growing alloy carbides reach a peak hardening contribution at some critical size. Despite the possibility of secondary hardness in the steels, what is clear is that Steel B does not seem to experience a larger precipitation hardness contribution, even with its higher alloy content. In fact, other than an upward shift in hardness, the curves for the two steels look virtually the same except for the later values. And so, the difference in hardness can be attributed to solution hardening in the Steel B matrix brought upon by the much higher Ni content as well as the higher C content. As far as the higher softening rate towards the end of the curve, it is possible that at very long tempering condition the rate of carbide growth and coarsening in Steel B does exceed that of Steel A and leads to a negative hardening contribution due to coarse particles. Ultimately, one factor that is not taken into account is the relative amount of residual stresses and the degree of recovery in the steels. These factors would be able to account for differences in hardness as well as softening rates. 24

29 3.3. Impact Toughness Results The impact toughness results were divided based on the temperatures at which the Charpy V- notch tests were performed. Both of these, shown in Figure 16, are only shown across time since only two temperatures were examined. Impact toughness (Joule) Impact Toughness at Ambient Temperature 600C 600 C TX44 Steel A 600C 600 C TS44 Steel B 550C 550 C TX44 Steel A 550C 550 C TS44 Steel B Time (s) Impact toughness (Joule) Impact Toughness at -20 C 600C 600 C TX44 Steel A 600C 600 C TS44 Steel B 550C 550 C TX44 Steel A 550C 550 C TS44 Steel B Time (s) Figure 16. Impact toughness measurements versus time at ambient temperature (top) and -20 C (bottom) The toughness measurements feature a much more limited investigation compared to hardness. The main reason for this is that while a specific area of peak hardness was sought during the hardness investigate, the main purpose of the toughness trials was to see how well the impact toughness was maintained throughout the scope of the experimental trials for both grades. The first thing to note is that the toughness values in general are at least half of what 25

30 one would expect from this type of hot-work tool steel. The most likely reason for this is due to the way the toughness samples were cooled after heat treatment. Since the samples used were large and heavy blocks, it was not safe or easily feasible to quench the tempered samples. However, because the samples were air-cooled instead, they likely spent a significant amount of time within a temperature range that is associated with embrittlement. When held or slowly cooled between C, a tempered steel may undergo irreversible temper embrittlement, which is a process in which cementite grows and coarsens along the prior austenite grain boundaries and the lath boundaries, creating crack nuclei, weakening the boundaries, and resulting in a more brittle material [14]. Nonetheless, since all the samples from the same tempering temperatures followed virtually the same cooling curve, it will be assumed that the decrease in toughness is the same for all the measuring, thus allowing one to compare the results over time. Here, unlike with the hardness results, the two steels show opposite results as Steel B actually loses toughness while Steel A experiences an increase. Additionally, Steel A has a consistently higher toughness over time, going from a difference of 2 to 9 J for the 600 C samples tested at ambient temperature. Steel A s higher toughness is somewhat expected, as Steel B s higher hardness is due to impeding the movement of dislocations in the material, which increases internal stresses and can reduce toughness. The overall curve trend is also seen for the tests done at a low temperature, suggesting that Steel B s loss in toughness is not an outlier. One would expect a tool steel s toughness to increase as the degree of tempering increases, as the loss in residual stresses in the martensitic matrix would result in a more ductile material. In this case, Steel A appears to follow the expected result, while Steel B experiences quite severe embrittlement. To provide an overall view of the mechanical property changes in both steel grades and be able to put the toughness results in context, both hardness and toughness were plotted versus the tempering parameter (Figure 17). For simplicity, it was assumed that the parameter value was the same for hardness and toughness samples of the same heat treatment. 26

31 Hardness and Toughness Data vs. Hollomon 50 Parameter Hardness (HV5) Impact Toughness (Joule) TX44 Steel Hardness A Hardness TS44 Steel Hardness B Hardness TX44 Steel A Toughness TS44 Steel B Toughness Hollomon Parameter Figure 17. Hardness and Impact Toughness versus Hollomon-Jaffe parameter Looking at this combined figure, it may be possible to make some more sense of the data. Just like the hardness curves show an area of relatively stable hardness, the toughness results do appear to show little change in impact toughness at the lower tempering conditions. In fact, changes in toughness seem to coincide with the precipitous drop in hardness, in which Steel B was seen to drop at a faster rate. Given Steel B s higher Ni content, it is possible that another embrittlement process is at work here. At C, reversible temper embrittlement can occur, which is when trace impurities such as P segregate along the prior austenite grain boundaries and weaken the boundary. This process is significantly enhanced by a high content of Si and Ni, as these elements themselves build up along the grain boundary [14,37]. This could explain why the phenomenon is absent in Steel A but not Steel B and suggests the higher Ni content may not only be contributing to a hardness increase in early tempering but also a toughness decrease at higher tempering conditions. Nevertheless, it is crucial to reiterate the uncertainty in the toughness data given the limited amount of data points relative to those in the hardness data and given the uncertain extent to which irreversible temper embrittlement affects each toughness sample. For example, there are significant gaps in the toughness curve versus tempering parameter. This makes it difficult to know whether toughness does remain stable at the start and whether there is a peak toughness value for Steel B when the effects of embrittlement overcomes the effects of recovery. Also, the air-cooling of the samples poses some significant risks since there are many uncontrolled variables. First of all, the steel blocks were stacked one on top of the other in the furnace, meaning that the cooling curves might actually differ between the two grades. Moreover, it is unclear if irreversible temper embrittlement has a larger or smaller effect on 27

32 heavily tempered samples that are expected to already have a significant amount of carbide precipitation Microscopy and XRD Results The microscopy and XRD data has thus far yielded initial results regarding the extent of investigation. Prior to TEM and XRD work, a short preliminary examination was done with light optical microscopy (LOM) to get a general overview of the steels microstructure. As this was not a focal part of the project, only comparisons between Steel B s microstructure at the Q state and at 600 C 0h were considered in Figure 18. Figure 18. LOM micrographs of Steel B at Q state (top) and after 600 C tempering at 0h (bottom) 28

33 What is clear and most significant in both cases is that lath martensite dominates the microstructure. This helps bolster the claim that the matrix phase in all of these samples can be assumed to be fully martensitic. Additionally, one can just about see very small particles spread across the Q sample, which would suggest the presence of carbides left over from the manufacturing processes. In the tempered sample, this level of detail cannot be seen due to the microstructural changes associated with tempering, which in general highlights the fact that the scale of the carbides expected to precipitate in these steels is far lower than the resolution and magnification capabilities of LOM. For this reason, subsequent analysis was done with TEM. The present TEM analysis is part of an ongoing investigation regarding the visualization of secondary carbide precipitation in these types of tool steel. As such, a previous internal examination was conducted focusing on Steel B in the Q state and at a 620 C tempered state, which provides context to the present work [38]. Briefly highlighting the study and some findings, TEM analysis was done for replicas using an Al 2 O 3 film, the Q state shows the presence of Mo-rich M 6 C ( nm), MC (30-60 nm), and TiN (30-60 nm), while the tempered sample is characterized by a huge amount of large cementite particles ( nm). However, the report noted that the abundance of cementite made it impossible to see any nano-sized (2-5 nm) secondary alloy carbides that had been predicted by simulations. And so, one of the changes done in this present study is the use of a graphite film for the replicas. The result of this, it seems, is that carbides with an approximate 4 nm diameter can be seen in the tempered samples. At this time, the TEM results are limited to the identification of nano-sized MC/M 2 C alloy carbides in the carbon replicas. Already at 0h and 600 C in the Steel A grade, one can see the appearance of tiny Mo, Nb, and V-rich particles with an approximate diameter of 4 nm (Figure 19). In addition, it is important to note the presence of similarly shaped and sized particle rich in Fe and Cr. 29

34 Figure 19. TEM micrograph of Steel A-0h 600 C carbon replica. Spectra 52, 53, 54 denote V,Nb, and Mo-rich particles while Spectrum 54 is Fe and Cr-rich Unfortunately, due to resolutions and magnification limitations with the equipment, it is still challenging to properly see these small carbides and analyze their composition accurately. This issue is exacerbated by the fact that the graphite film distorts the C content analysed by EDS, forcing one to rely on the relative amounts of alloy elements alone to identify the carbides. Nonetheless, there is enough evidence to suggest that some of these particles are rich in V, Nb, and Mo while other similarly sized ones are Fe and Cr-rich. The size of these particles, as opposed to those found in the previous study, strongly suggest that they precipitated during tempering. XRD data for phase analysis was collected for all 6 samples. Overall, the samples had virtually the same diffraction pattern except for slight peak sharpening for the samples with longer tempering times. Figure 20 shows the diffraction pattern of the Steel B-0h sample along with phases that match up with the peaks. 30