Heat Treatment and Secondary Phase Formation in FeCrNi Medium Entropy Alloys

(1)

INOM

EXAMENSARBETE MATERIALDESIGN, AVANCERAD NIVÅ, 30 HP

STOCKHOLM SVERIGE 2021 ,

Heat Treatment and Secondary Phase Formation in FeCrNi

Medium Entropy Alloys

AMANDA CARSBRING

KTH

(2)

Abstract

The topics of high entropy alloys (HEA) and medium entropy alloys (MEA) have been heavily researched in recent years. A HEA usually consists of five or more base elements, and a MEA would have three or four base elements. These types of alloys are multi-principal element alloys (MPEA) that have been thought to have interesting properties due to their high configurational entropy, which was thought to be the reason for stabilized simple solid solution phase in the HEA. The high entropy effect contributing to stable single phase in these alloys has been discussed and has not been found to be a predicament to which MPEA that will present as single phase at lowered temperatures. Still, some of the HEA and MEA investigated have interesting properties such as high ductility and good thermal properties, as is the case for the commonly researched CoCrFeMnNi HEA and the CoCrNi MEA which are both solid solution FCC phase at lower temperatures.

This master thesis aims to investigate one of the less commonly researched MEA: equimolar FeCrNi.

This alloy has been studied previously, and it was found there might be a possibility of precipitation hardening the alloy. To further study this alloy system, three FeCrNi alloys in the close-to equimolar range were produced and underwent a series of aging heat treatments to study the amount of precipitated secondary phase with composition changes and different aging temperatures. The objective is to evaluate and interpret the data found in the different CALPHAD databases used in Thermo-Calc and FactSage software and make comparisons to the experimental results. This to discuss the possibilities of hardening this alloy through aging treatment.

The alloys selected and produced are 33Fe33Cr33Ni, 40Fe30Cr30Ni and 45Fe30Cr25Ni, all in mol%.

Through experimental investigation using x-ray diffraction (XRD) analysis, it is found that Cr-rich BCC phase is formed in all alloys after most of the aging treatments performed. The volume fraction of BCC was quantified through the reference intensity ratio (RIR) method. From quantification, the largest volume fraction BCC is found in the equimolar 33Fe33Cr33Ni alloy, and the lowest fraction BCC is shown in the 40Fe30Cr30Ni alloy. The increased volume fraction of BCC coincides with an elevated hardness in all three alloys. It is also found that out of the three equilibrium phase

calculations used in this project, the ThermoCalc steel database TCHEA4 seems to give results that are in closest agreement with the experimental results. For future studies in this subject, the

recommendation is to further study the mechanical properties of the FeCrNi MEA and assess possibilities for application.

(3)

Sammanfattning

Högentropilegeringar och mediumentropilegeringar har studerats närmre de senaste åren på grund av deras intressanta mekaniska egenskaper. En högentropilegering består vanligtvis av fem eller flera baselement, och en mediumentropilegering har tre eller fyra baselement. Detta skiljer dem från konventionella legeringar som i regel har ett, maximalt två, baselement. Dessa typer av multi- baslegeringar har ansetts ha intressanta egenskaper på grund av deras höga konfigurationsentropi, vilken tros vara orsaken till stabilisering av fast lösningsfas i legeringarna. Högentropieffekten som bidrar till stabil enfas i dessa legeringar har diskuterats och har emellertid inte visat sig vara tillförlitligt för att förutsäga vilka multi-baslegeringar som blir fast lösningsfas vid ett lägre

temperaturintervall. Ändå har några av de undersökta legeringarna intressanta egenskaper som hög duktilitet och goda termiska egenskaper, vilket är fallet för högentropilegeringen CoCrFeMnNi och mediumentropilegeringen CoCrNi som båda är enkel FCC-fas vid lägre temperaturer.

Detta examensarbete syftar till att undersöka en av de mindre omnämnda

mediumentropilegeringarna: ekvimolär FeCrNi. Denna legering har studerats tidigare, och det visade sig att det kan finnas en möjlighet att utskiljningshärda legeringen. För att ytterligare studera detta legeringssystem producerades tre FeCrNi-legeringar i ett sammansättningsintervall nära ekvimolär sammansättning, som sedan genomgick en serie åldringsbehandlingar för att studera mängden utskild sekundär fas beroende på sammansättning och åldringstemperatur. Målet är att utvärdera och tolka data som finns i de olika CALPHAD-databaserna som används i beräkningsverktygen Thermo-Calc och FactSage, och därefter göra jämförelser med experimentresultaten. Detta för att diskutera möjligheterna att härda dessa legeringar med partikelutskiljning.

De legeringar som valts och tillverkats är 33Fe33Cr33Ni, 40Fe30Cr30Ni och 45Fe30Cr25Ni, alla angivna i molprocent. Genom experimentella undersökningar med röntgendiffraktionsanalys hittades att BCC-fas med höga halter Cr bildats i alla legeringar efter majoriteten av åldringsbehandlingarna som utförts. Volymfraktionen av BCC kvantifierades genom beräkningar från

referensintensitetsförhållande (RIR). Från kvantifiering fås den största volymfraktionen BCC i den ekvimolära 33Fe33Cr33Ni-legeringen, och den lägsta fraktionen BCC i 40Fe30Cr30Ni-legeringen.

Högre volymfraktion BCC sammanfaller med en förhöjd hårdhet i alla tre legeringarna. Det visar sig också att utav de tre jämviktsfasberäkningarna som används i detta projekt så är det ThermoCalc- ståldatabasen TCFE10 som gett resultat som överensstämmer med experimentresultaten. För framtida undersökningar inom detta område rekommenderas att studera de mekaniska egenskaperna hos FeCrNi och bedöma möjligheterna för tillämpning av materialet.

(4)

1 Introduction

Alloying is the very core of metallurgy. This because most metals are not very useful in their pure form and much more useful after alloying elements are added. Traditionally, alloys have mainly been designed through choosing a base element with most of the desired material properties, and then altering the properties through addition of smaller amounts of alloying elements. However, a vast number of multicomponent alloys were investigated in the works by Franz Karl Achard which he published as a book of alloy data in 1788 [1]. Achard showed a first attempt at studying

multicomponent alloys, but got little attention for this publication, which was brought up again almost 200 years later when mentioned in 1963 by Professor Cyril Stanley Smith. However, the concept of alloying with multiple base elements was not taken into serious consideration until the independent discoveries of Cantor and Yeh in 2004, where an equimolar five-component alloy showed interesting properties: CoCrFeMnNi [2, 3]. This led to the relatively new concept in materials design which is commonly referred to as the concept of high entropy alloys (HEA), because of the suggestion of high configurational entropy playing a big part in creating the single phase structure in these alloys. The discovery of equimolar CoCrFeMnNi presenting in the form of simple FCC structure was unexpected, and sparked interest in investigating the general properties of equimolar multicomponent alloys and what possibilities they might have for application. This has created a relatively unexplored area in materials design and engineering which is currently in need of investigation.

Further research in the field of HEA led to finding a highly interesting three-component alloy, namely the equimolar CoCrNi, which showed good thermal properties and has been observed to have simple FCC structure at 600°C [4-6]. The equimolar or close-to equimolar three- or four-component alloys are commonly referred to as medium entropy alloys (MEA). The CoCrNi showed better material properties than that of CoCrFeMnNi. However, an issue with the CoCrNi MEA is the hardness, which should ideally be higher for better applicability. Additionally, the price of the alloy is quite steep, and with Co-production being problematic in terms of impact on environment and human health it would be beneficial to find alternatives [7]. Because of this, there have been studies investigating a cheaper alternative: the FeCrNi MEA. The FeCrNi MEA is not single phase at 600°C, but there are hopes that the alloy could be further engineered to be a more cost-effective alternative to CoCrNi with sufficiently similar properties. Additionally, the possibility of achieving precipitation hardening through formation of secondary phase in the alloy has been studied and was deemed interesting from a materials design perspective [8].

1.1 Scope and aim

Because of previous studies on precipitation hardening FeCrNi MEA through formation of Cr-rich BCC it was deemed interesting to further study this alloy system for this master thesis project. The aim is to investigate the phases present in the FeCrNi system in the close-to equimolar composition range. This to find the effects of reducing the amount of Cr and Ni to further economize the alloy. The ultimate objective is to investigate the applicability of thermodynamic database models from FactSage and Thermo-Calc, which are both CALPHAD based software, as well as the Thermo-Calc PRISMA module for modeling precipitation of secondary phase. This through comparison with experimental results.

The scope of this project is to first perform an extensive literature study and carry out a series of thermodynamic calculations in ThermoCalc and FactSage to select alloy compositions and

temperatures for heat treatment. Thereafter the further plan is to produce the three FeCrNi alloys in the near-equimolar composition range, ensure material quality before further experiments (with ICP- OES), and then to conduct a series of heat treatments at different times and temperatures to be able to study precipitation behavior in the alloy at a large extent. The samples produced are tested for

hardness with hardness vickers, analyzed for phases present with x-ray diffraction, and EDS analysis is performed to some of the samples to validate composition of secondary phases. The results are then discussed to come down to the final conclusions from this project.

(6)

2 Background

There are many types of steel, such as martensitic, ferritic, duplex and austenitic steel to name a few of the most common ones [9]. Austenitic steels are used because of properties such as ductility,

toughness and resistance to corrosion, and they cannot achieve hardening from heat treatment [10].

They are typically used in a wide range of applications spanning from architectural structures to equipment for food processing and chemical industries. For precipitation hardening stainless steel alloys the composition could also be comprised out of other alloying elements such as Al, Cu, Nb, and commonly N. In these types of alloys, a moderate corrosion resistance is usually achieved, as well as high strength. The strength would usually be achieved after heat treatment in the lower temperature range of 500-800°C, where precipitates are formed.

In this project the alloy system in focus is stainless steel in the near equimolar composition range, and also semi-austenitic precipitation hardened steel. However, in the equimolar composition range it is important to consider that the material is no longer a Fe-based alloy and should more correctly be viewed as an MPEA. Keeping this in mind, knowledge on stainless FeCrNi steels with high Cr- and Ni- content can still provide useful information for observing the FeCrNi MEA. Stainless steel contains a significant amount of Cr (11 wt.% Cr or more), which is the alloying element that hinders oxidation of stainless steel in water [10]. Cr has this effect through creating a passive layer on the steel surface which creates a barrier of stable Cr oxide. This is the main reason to alloy Fe with Cr. Other effects of alloying Fe with Cr are increased heat resistance, and increased hardenability during heat treatment.

Austenitic stainless steels are FeCrNi alloys with an austenitic FCC structure, and have a Cr content of 16-28% and Ni content of 3.5-32%. Also, some Mo can be added.

Elevated levels of Cr in the alloy creates the possibility of precipitating Cr rich BCC particles. This has been studied in equimolar FeCrNi and has been shown to contribute with precipitation hardening in the alloy [8]. Precipitation of Cr BCC particles is generally viewed as disadvantageous because it removes Cr from the matrix phase, decreasing the anti-corrosive effects from Cr-addition. However, if it is possible to achieve higher hardness in FeCrNi in this manner without losing too much of the anti- corrosive properties it could be of high interest.

The addition of Ni is important because Ni stabilizes the FCC austenitic phase, compared to Cr which destabilizes FCC in favor of BCC [10]. Stainless steels alloyed with Ni are usually easier to weld and have good formability, due to the austenitic structure. Addition of Ni also removes the magnetic properties in Fe alloys through the stabilization of FCC. Additionally, it makes the alloy ductile at lowered temperatures, even at cryogenic temperatures, while suitable in high temperature applications as well.

2.1 HEA

Steel alloys are alloys that consist mainly of Fe, as it is the base metal in the system, with additions of smaller amounts of alloying elements. Most traditional metallic alloys are built up in this manner and are based on one or two base elements with smaller amounts of other elements added to add to the existing properties of the base material [4, 8, 11]. However, in the early 2000s, interest spurred for another approach to alloy production, when some five-component alloys were shown to unexpectedly form a single-phase solid solution [2, 3]. This type of alloy is referred to as HEA or MPEA, but because HEA is more commonly used as a term in literature that is the term used in this report. HEA are alloys that do not have one single base metal and can sometimes be referred to as baseless alloys [11]. They consist of larger amounts of five or more base elements, usually close to the equimolar composition but that is not a requirement. Alloys with three or four base elements are referred to as MEA, because of the lower configurational entropy compared to HEA. By definition there is typically a concentration of 5-35 mol% for each of the base elements in HEA, but the earlier studies are focused around alloys

(7)

consisting of five elements in equimolar quantities. There has been discussion on the effects from the high entropy contribution to the material properties in these types of alloys, since the high

configurational entropy was thought to be the main reason to some of these alloys being simple solid solution phase at lower temperatures. However, this contribution is only dominant in single phase solid solution [12]. Reports on HEA include their many beneficial properties, such as low-temperature ductility and fracture toughness, and resistance to wear and corrosion [8, 11]. Additionally, the high entropy super alloys (HESA) have a similar microstructure compared to Ni-base super alloys and are observed having superior strength to commercial super alloys, especially in higher temperatures [13].

Because of this, HESA have been considered as an alternative to Ni-base alloys in high-temperature applications.

Some core effects affecting the HEAs are the high-entropy effect, the effect of sluggish diffusion, severe lattice distortion, and the cocktail effect [11, 14]. These four phenomena have been discussed as the reason behind the specific set of properties found commonly in HEA. The high-entropy effect relates to the high configurational entropy in HEA and is the reason for the name HEA specifically. High entropy is related to a low Gibbs free energy in the system which impedes the formation of intermetallic phases and makes it easier to form solid solution phases. However, this is not synonymous to all HEA being random solid solutions, since there are other factors that contribute as well. The high entropy effect is lower at lower temperatures, which makes it possible to precipitate intermetallic compounds if annealed in the right temperature range.

The effect of sluggish diffusion refers to the fact that diffusion is slower in HEA compared to traditional alloys [11]. Sluggish diffusion in HEA has been demonstrated but is yet to be fully

understood. One conclusion that has been drawn is that diffusion seems to get slower with an increase in number of base elements in the alloy, but the results are ambiguous as to why this is the case. Some explain this phenomenon as being caused by the variety in adjacent atoms, while others look more towards the crystallographic structure of HEAs to find an explanation. Additionally, there is conflicting research that suggest no decrease in diffusion of Ni with increasing number of elements in

conventional HEA systems. As for severe lattice distortion, it has been confirmed in a few studies of conventional HEA. This leads to lattice strain in the material, which then can lead to an increase in hardness and strength. Studies also state that the distortion is not remarkably greater in HEA compared to regular solid solutions in the same alloying system, but it is slightly greater in HEAs.

What makes severe lattice distortion in HEA more difficult to model with traditional solid solution strengthening models is that there is no way to distinguish between solvent and solute elements.

The cocktail effect plays an important part as well, meaning the different properties of the elements put into the alloy will affect the properties of the alloy similarly [11]. For example, the addition of low- density elements like Al will give the HEA a lowered density, and by adding refractory elements like Nb, the material can become more resistant to higher temperatures. This is if the material stays in simple solid solution, and effects of combining certain elements should also be taken into

consideration.

2.2 MEA

For a material to be considered HEA, it should consist of five or more base elements, but conventional solid solution alloys usually only include one, sometimes two, base elements [11]. This leaves a category in between, namely the alloys consisting of three to four base elements [6]. These alloys are referred to commonly as MEA and have been shown to have interesting properties as well. The most commonly studied three component MEA is the equimolar CoCrNi, which is a random solid solution at 600°C [4, 6]. This alloy has many of the wanted characteristics from, or sometimes better than, the conventional HEA of equimolar CoCrFeMnNi, for example in measures of hardness and fracture toughness [6, 15, 16]. CoCrNi was shown to have the most desirable properties out of the other

(8)

equimolar combinations of two, three or four of the constituent elements of the CoCrFeMnNi alloy system. However, among other alloys studied, there were still a few other combinations that showed a single-phase solid solution structure after both casting and homogenization.

2.3 FeCrNi

One alloy combination which is simple FCC phase as cast and after homogenization is FeCrNi. This alloy was further investigated in the works by Liang et al [8]. The idea behind their work was to find a way to make the FCC structure MEA stronger without losing too much of the ductility in the material.

Although the alloy is purely FCC as cast and after homogenization, it does have more equilibrium phases around 600°C. Still, the idea behind selecting FeCrNi as a material in this study was to replace Co from the CoCrNi alloy with the cheaper Fe. Additionally, Co-production has damaging effects on both the environment and on the health of people working in production as well as surrounding areas, which makes it a less desirable component in alloy production. Hence, it would be of interest to replace it with a cheaper element such as Fe.

According to previous studies FeCrNi shows potential of forming BCC precipitates which has been hoped to make it possible to achieve precipitation hardening in the material in previous studies, although this is usually disadvantageous due to the risk of Cr-depletion in the matrix phase which can lead to stress corrosion cracking [17]. For studying hardness related to this formation of secondary phase, Liang et al. created an ultrafine grained (UFG) FCC MEA through the powder sintering method.

Results from their measurements show that BCC phase is indeed precipitated, and as predicted the BCC was rich in Cr. Conclusions were drawn, stating that the UFG FCC MEA with Cr-rich BCC precipitates showed 26% elongation and a tensile strength of 826 MPa which was deemed a good combination of strength and ductility and hence, a promising material for structural applications. It was also concluded that the dominant strengthening mechanisms in this material was grain boundary strengthening and precipitation strengthening.

(9)

3 Method

This part of the report accounts for the methods behind the several experiments conducted for this project. Composition selection, calculational models, sample preparation, experimental equipment and methodology are explained.

3.1 Choice of compositions and heat treatment

Three alloy compositions were chosen for experimental investigation, all of which being FeCrNi MEA.

The criteria required were that they should be single phase after homogenization and have a secondary equilibrium phase at lower temperatures for the possibility of precipitation of particles. Preferably, this secondary phase should be BCC, and not sigma phase, to compare to previous studies. The aim is to investigate the difference in alloy microstructure and phase constitution with decreasing the amount of Ni and Cr in the material as to lower the price of the alloy.

The alloys chosen are the following (presented in mol%); 40Fe30Cr30Ni, 45Fe30Cr25Ni and the equimolar composition 33Fe33Cr33Ni. These three were chosen because they all show the possibility of producing single-phase FCC structure at elevated temperature and they are equimolar or close to equimolar, fulfilling criteria for qualifying as MEA. Further they all present a temperature region for possible precipitation of secondary phase, which should be BCC or sigma phase (figure 1). 33Fe, 40Fe, and 45Fe are abbreviations of 33Fe33Cr33Ni, 40Fe30Cr30Ni and 45Fe30Cr25Ni respectively.

Figure 1: ThermoCalc HEA database equilibrium calculation for three different compositions. Equilibrium phases are BCC, Sigma and FCC (not shown in graph).

(10)

The conditions for heat treatment were chosen from the equilibrium calculations as well. The heat treatment cycle was selected to investigate formation of secondary phase in the three alloys.

Calculations were performed in Thermo-Calc and FactSage. Because all the alloys should be single phase FCC at temperatures above 1000°C, 1100°C was chosen as the annealing temperature. Then, a wide range of temperatures were chosen for aging treatment. Two different aging times were also chosen: 15 minutes and 1 hour. For the one-hour treatment, eight temperatures from 650°C to 1000°C with a 50°C step were selected. To be able to see the initial precipitation of secondary phase at the lower end of the temperature range, the 15-minute treatment was done at the five temperatures from 650°C to 850°C (also with a 50°C step). For clarity, all samples for aging treatment are listed in table 1, constituting in total 39 samples for aging. These temperatures were selected because they are within the multiphase equilibrium range and they should give a high enough mobility for the possibility of secondary phase formation.

Table 1: List of samples for aging treatment.

3.2 Thermodynamic calculations

Two different thermodynamic software are used in this project: Thermo-Calc and FactSage. Both software have databases built on CALPHAD models and can be used to find phase diagrams and equilibrium phases for a great number of alloy systems [18, 19]. The Thermo-Calc databases are thermodynamic databases TCFE10 and TCHEA4 which are used separately, and mobility databases MOBFE5 and MOBHEA2 which are used together with their corresponding thermodynamic database respectively. The FactSage databases used are thermodynamic databases FSstel and FactPS which are used simultaneously to calculate phase equilibria.

The first batch of calculations were performed to predict the experimental outcome. These were done with default settings mostly, but also with variations in the number of nucleation sites for the PRISMA precipitation models. When experimental results had been studied, the models were changed to fit the result, to find the settings needed to make the model show the precipitation behavior. Parameters adjusted were precipitate interfacial energy, mobility enhancement factor, and dislocation density in the matrix. The mobility enhancement factor has a default value of 1, but up to 3 is usually realistic although this number can increase greatly with deformation induced vacancies [20]. Different values were tested for this parameter. The precipitate interfacial energy was set initially to default calculated value from matrix settings, and then changed to 0.1, which is the lowest realistic value for this

parameter [21]. The dislocation density was changed between the default 5E12 m^-2 and a realistic maximum for highly deformed metals which is 1E16 m^-2 [22].

3.3 Alloy preparation

The first step of alloy preparation is to assemble the constituent raw material for the alloy. Mol% was recalculated to wt% to prepare the materials measured in grams. It was decided to create batches of 600g per alloy. The calculated and measured weight of the raw materials is shown in table 2. The raw materials are pure bulk metals Fe, Cr and Ni.

[˚C] 650 700 750 800 850 900 950 1000

33Fe 15min+1h 15min+1h 15min+1h 15min+1h 15min+1h 1h 1h 1h 40Fe 15min+1h 15min+1h 15min+1h 15min+1h 15min+1h 1h 1h 1h 45Fe 15min+1h 15min+1h 15min+1h 15min+1h 15min+1h 1h 1h 1h

(11)

Table 2: Input material weight for alloy preparation

Composition (Mol%) 33Fe33Cr33Ni 40Fe30Cr30Ni 45Fe30Cr25Ni Theoretical Mass Fe (g) 201,19 241,28 272,14

Theoretical Mass Cr (g) 187,32 168,49 168,92 Theoretical Mass Ni (g) 211,49 190,23 158,93 Theoretical Mass total (g) 600,00 600,00 599,99

Measured Mass Fe (g) 201,25 241,50 272,15

Measured Mass Cr (g) 187,33 168,43 168,93

Measured Mass Ni (g) 211,43 190,22 158,93

Measured Mass Total (g) 600,01 600,15 600,01

The three alloys were produced through induction melting in an alumina crucible, using an Eltek induction heater (shown in figure 2), creating approximately 600g of melt. The alloys were

isothermally processed at the temperature 1500°C, and ingots were produced through suction casting into glass tube molds. After casting the ingots were quenched in saltwater.

Figure 2: Schematic of the induction furnace used for alloy production. [Hanyang University]

(12)

3.4 Sample preparation

The ingots were machined and proceeded to be homogenized though annealing in Ar-atmosphere at 1100°C for 24 hours. The homogenized ingots underwent cold rolling with 80% reduction before aging treatment. The as-cast, homogenized, cold rolled, and aged ingots were cut to approximately 5x5 mm, and polished by hand using sandpaper of increasing grit size (600,800,1000,2000) to get a smooth surface. Samples were mounted on a sample holder using wax, and the wax was removed using acetone after polishing. Samples for EDS analysis and hardness Vickers testing were electropolished after manual polishing.

3.5 ICP-OES method

The samples were analyzed through ICP-OES to study the composition of the alloys. ICP-OES is short for Inductively Coupled Plasma Optical Emission Spectroscopy, and is a common method for

determining composition and studying trace elements in many types of samples [23]. The two main components in ICP-OES are the plasma torch and the spectrometer. In the plasma torch, the sample is turned into aerosol spray which is then turned into plasma. The atoms in the sample emit light with element-specific wavelengths, that are picked up in the spectrometer and thus the sample constituent elements are found.

For the chemical analysis of the FeCrNi MEA, sample pieces weighing between 0.2 and 0.3 grams were prepared after machining of the as cast alloys. A total of 6 metal pieces (2 of each alloy composition) were cut and placed in individual tubes for solution treatment. The solvent used is a mixture of 5ml 60%nitric acid, 10ml hydrochloric acid and 10ml distilled water per sample. When the samples were dissolved, the solution was weighed and filtered before measuring small amounts of solution for further dilution before the samples were analyzed.

The equipment was calibrated trough measuring a total of nine samples consisting of 1ppm, 10ppm and 20ppm respectively made using solutions of pure Fe, Cr and Ni. After successful calibration, the six samples of FeCrNi solution were analyzed. New samples were prepared in the same manner, but with different dilution, to analyze Al impurities. This was done to find if the impurity amounts were in an acceptable range. It is assumed to find a small amount of impurities in all samples because the bulk material contains trace amounts of Al.

3.6 XRD analysis

XRD analysis was prepared to investigate the presence and amount of secondary phase for the different heat treatment conditions. XRD is a common method of identifying species of crystalline materials and can be used for quantifying phases in a multiphase crystalline sample [24, 25]. In XRD, the sample is subjected to x-ray beams which are scattered by the sample. The diffracted x-ray beam is picked up by a detector, and the result is shown as peaks in the XRD diagram. The peak patterns in XRD are specific for each phase, and the peak intensities imply relative quantity of a phase in a multiphase sample.

Samples for XRD were prepared from the three different alloy compositions at the following process steps: as-cast, post homogenization treatment, post cold rolling, and post annealing at different times and temperatures. With the results from XRD analysis it is possible to analyze the changes in

microstructure through the different materials processing steps and see the difference in outcome from heat treatment at different temperatures. The samples were sent to the XRD-lab where XRD was performed at 2-theta 20-100 degrees. The X-ray diffractometer used is a D/Max-2500/PC, made by

(13)

Rigaku, Japan. Monochromatized CuKα radiation was used as the X-ray source (λ = 1.5418 Å) at 40 kV and 100 mA.

3.7 RIR quantification method

By using the results from XRD analysis it is possible to evaluate the quantity of the different phases in the material. One such quantification method is the Reference Intensity Ratio (RIR) method [26]. The RIR method uses the integrated intensity of the peaks in the XRD graph and combines this with lattice parameters to calculate the relative quantities of the known phases in the system. This is done by utilization of the intensity equation shown below, where 𝐼_𝑖𝛼 is the intensity from diffraction line i and phase α, 𝑋_𝛼 is the weight fraction of phase α, 𝜌_𝛼 is the density of α, and (𝜇 𝜌⁄ )_𝑚 is the mass absorption coefficient in the mixture. 𝐾_𝑖𝛼 is a specific constant for the diffraction line i, the phase α and the specific conditions of the experimental setup such as the incident beam intensity and the diffractometer radius.

𝐼_𝑖𝛼= 𝐾𝑖𝛼𝑋𝛼

𝜌_𝛼(𝜇 𝜌⁄ )_𝑚

This method is typically used on powder materials because they are assumed to be distributed entirely at random, and hence no dominant lattice plane can skew the results. This makes it possible to look at the intensity values and compare them to reference values, and hence receive the RIR quantity through comparison. However, this method is less reliable for non-powdered samples, and because of that the method has been tweaked to be able to give an estimate quantification even for experiments on solid metal pieces. The method used in this project is provided by normalizing the intensity calculations over all the intensity peaks in the XRD graph. This leads to most of the 𝐾𝑖𝛼 factor being removed due to normalization, since the experimental setup is the same in all the XRD measurements. Remaining are the multiplicity value for i, phase α unit cell volume, the structure factor for α and i, and the Lorenz polarization for the incident angle. Through input of these data, the normalized RIR quantified value is found. The statistical error of the RIR method and other quantitative analysis methods based on intensity is around 5% if done correctly.

3.8 Hardness test

Hardness testing through the Hardness Vickers method was performed on selected samples to see if the amount of secondary phase in the material was corresponding to the quantified amount of secondary phase. Hardness Vickers is a method for hardness testing, where an indentation is created in the material under a specific load [27]. The indentation is a square pyramid shape for hardness Vickers measurements, and the value is found through measuring the two diagonals of the indentation and taking the mean value of the two. The load force is accounted for as well to get the value for hardness measurement according to the following equation, where F is the load force and d is the mean diagonal length:

𝐻𝑉 = 0.1891 ∗ 𝐹 𝑑²

The load for the hardness test on the FeCrNi MEA samples was 500gf. All samples heat treated at 800- 950°C for a duration of 1 hour were tested for hardness. Samples were prepared by electropolishing the surface to remove residual stress in the material. All samples, in total 12 pieces, were tested a total of 15 times to achieve a good average value of the hardness.

(14)

4 Results

This part of the report aims to present the experimental results in their entirety. The results shown here are discussed later in this report.

4.1 Thermo-Calc and FactSage

The thermodynamic data in this report is taken from calculations made primarily using ThermoCalc thermodynamic software, but comparisons are also made with FactSage software to discuss which equilibrium calculation corresponds best with findings from experimental data. In ThermoCalc software, two different thermodynamic databases were used: steel database TCFE10 and HEA database TCHEA4. When studying the different phase equilibrium diagrams, ThermoCalc HEA database and FactSage steel database seemed to correspond quite well to each other, while the ThermoCalc steel database showed higher equilibrium fraction of sigma phase over a wider

temperature range than the other two databases showed. The data is shown in figures 1,3,4. Note the difference in unit in the FactSage calculations, due to calculations being based on input experimental sample weight instead of mol%.

Figure 3: ThermoCalc steel database equilibrium calculation for three different compositions. Equilibrium phases are BCC, Sigma and FCC (not shown in graph).

(15)

Figure 4: FactSage steel database equilibrium calculation for three different compositions. Equilibrium phases are BCC, Sigma and FCC (not shown in graph).

Examples of ternary phase diagrams were produced from TCFE10 database in Thermo-Calc as seen in table 3. Table 3-figure a is an isothermal section at 800°C for the FeCrNi ternary system, and it shows the equilibrium phases for the three selected alloys at 800°C, marked out with red crosses in the phase diagrams. In this image the 45Fe30Cr25Ni and 40Fe30Cr30Ni alloys are in the FCC + sigma 2-phase region, while 33Fe33Cr33Ni alloy is in the 2-phase region for FCC + BCC, although very close to the 3- phase region. The anticipated BCC would be high in Cr according to the phase diagrams.

Comparing the figure a phase diagram in table 3 to the figure e phase diagram, none of the alloys are in the 2-phase FCC + sigma region at 500°C. 33Fe33Cr33Ni and 40Fe30Cr30Ni alloys are in the 2- phase FCC + BCC region, while 45Fe30Cr25Ni is in the 3-phase region. According to the two diagrams there could be a possibility to form both sigma and Cr-rich BCC from supersaturated FCC in all three alloys over the temperature range.

(16)

Table 3: Ternary phase diagram examples for comparing TCFE10 and TCHEA4 databases

TCFE10 TCHEA4

800°C

a b

650°C

c d

500°C

e f

There is also the possibility of metastable precipitation of sigma or BCC in all alloys, exemplified in table 4, which shows isothermal sections at 650°C made with TCFE10 and TCHEA4 in Thermo-Calc.

The upper row displays the metastable phase diagram when disabling BCC phase in the calculation.

Likewise, the lower of the rows in the table shows the metastable phase diagram when disabling sigma phase. Throughout the figures the four figures in table 4 all three alloys are in the 2-phase region for precipitation of BCC or sigma phase respectively.

FCC + BCC FCC + BCC + sigma FCC + sigma

FCC + BCC FCC + BCC + sigma FCC + sigma FCC + BCC

FCC + BCC + sigma FCC + sigma

(17)

Table 4: Metastable phase diagrams from TCFE10 and TCHEA4

TCFE10 650C TCHEA4 650C

No BCC

No Sigma

The PRISMA precipitation module in Thermo-Calc was used to try and analyze the precipitation behavior in the three alloys. The results from six such computations are shown in figures 5-10, displayed here as TTT-diagrams. The many precipitation simulations performed for this project have been condensed to the eight most important TTT-diagram calculations, with simulation specifications stated in table 5. The simulations are labeled in the format [XY-Z] where X represents the precipitate phase with B for BCC precipitation or S for sigma phase precipitation. Y represents the selected nucleation sites for the simulation, which is G for grain boundary sites and D for dislocation sites. Z is specific for the combination of the parameter values for dislocation density, interfacial energy, and the mobility enhancement factor.

FCC + BCC FCC + sigma

(18)

Table 5: Specifications of parameter values for TTT-diagram calculations Simulation

Label

Precipitate phase

Nucleation site

Dislocation density

Interfacial energy

Mobility factor BG-1 * BCC Grain boundary 5E12 (default) (default) 1 (default)

BG-2 BCC Grain boundary 1E16 0.1 3

BD-2 BCC Dislocations 1E16 0.1 3

BD-3 BCC Dislocations 1E16 (default) 1 (default)

BD-4 BCC Dislocations 5E12 (default) 0.1 3

BD-5 BCC Dislocations 1E16 0.1 5

SG-1 Sigma Grain boundary 5E12 (default) (default) 1 (default)

SG-2 * Sigma Grain boundary 1E16 0.1 3

SD-2 * Sigma Dislocations 1E16 0.1 3

Simulations BG-1, SG-2 and SD-2 are denoted with a star (*) in table 5 to emphasize that there was no result displayed for these simulations. The other TTT-diagram simulations are found in figures 5-10.

Figure 5: TTT-diagram from simulation BG-2

(19)

Figure 6: TTT-diagram from simulation BD-2

(20)

Figure 9: TTT-diagram from simulation SG-1

(21)

4.2 Results from ICP-OES

Sample composition was chemically analyzed through ICP-OES. After successful calibration the six samples of FeCrNi solution were analyzed producing the results shown in the table below. Samples 33- 1 and 33-2 refer to the two samples from the mol% 33Fe33Cr33Ni alloy, samples 40-1 and 40-2 refer to the mol% 40Fe30Cr30Ni alloy, and samples 45-1 and 45-2 refer to the 45Fe30Cr25Ni alloy. In table 6 the raw data from ICP analysis is presented.

Table 6: Mean value results from ICP-OES analysis.

Sample name Cr (mg/l) Fe (mg/l) Ni (mg/l) Total (mg/l)

33-1 10.174 11.176 11.156 32.506

33-2 10.491 11.536 11.5 33.527

40-1 9.378 13.85 10.26 33.488

40-2 7.976 11.682 8.706 28.364

45-1 11.324 18.725 10.314 40.363

45-2 9.872 16.41 8.985 35.267

As seen in table 7, the normalized mole fraction of Cr in all three alloys corresponds with the measured fraction of Cr before melting. The normalized fraction of Fe in all alloys all deviate with +0.01 to the measured fraction of Fe before melting, while the normalized fraction of Ni consistently deviates with - 0.01 from the expected value.

(22)

Table 7: Conversion from mmol/l to normalized mole fraction.

Sample name Cr Fe Ni Total

33-1 0.33 0.34 0.32 1.00

33-2 0.33 0.34 0.32 1.00

40-1 0.30 0.41 0.29 1.00

40-2 0.30 0.41 0.29 1.00

45-1 0.30 0.46 0.24 1.00

45-2 0.30 0.46 0.24 1.00

After the main composition analysis new samples were prepared, with much higher metal

concentration to evaluate for Al impurities in the material. This was done because the bulk material used in alloy production contains trace amounts of Al. Results are found in table 8.

Table 8: Trace amount of Al in alloys, showed in ppm

Sample 33-3 33-4 40-3 40-4 45-3 45-4

ppm 250.4 243.8 307.6 297.1 359.6 370.4

Because the bulk Fe used in alloy production has the least Al impurity, the measured values were plotted together with the calculated Al content to see if these correspond to each other. This is shown is figure 11. The black dots represent the experimental values, and the blue dots represent the calculated values from the known Al content in each bulk material.

(23)

Figure 11: Al impurity content plotted against Fe content

From the figure it is visible that the measured Al content is consistently higher than the calculated value. The amount is also seemingly increasing with an increased amount of Fe which is contradictory to the known impurity content in bulk materials.

4.3 Results from XRD

Results from XRD analysis are received as a diffraction pattern. The results from measurements on samples before aging treatment show typical FCC pattern with varying intensities from the different crystallographic planes. This is shown in figure 12. No sign of secondary phase existence is shown in these samples. Important to point out are the high intensity peaks for the 40Fe and 45Fe homogenized samples, which have been cut out of the image for visual reasons.

45-3 45-4

40-3 40-4

33-333-4

150 200 250 300 350 400

30 32 34 36 38 40 42 44 46 48 50

Al content [ppm]

mol% Fe in sample

(24)

Figure 12: XRD pattern from pre-aging analyzed samples.

After 15 minutes heat treatment at the five chosen temperatures (650-850°C) all of the samples deviate from the pure FCC diffraction pattern according to figure 13-15. This implies formation of secondary phase. Deviations resemble the BCC diffraction pattern and can be seen in all samples if studied closely. Secondary phase is marked with red rectangles in figure 13-18.

Figure 13: XRD patterns for the mol% 33Fe33Cr33Ni alloy samples after 15 minutes heat treatment at five different temperatures.

30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 2 Theta

45Fe homogenized 45Fe cold rolled 45Fe as cast 40Fe homogenized 40Fe cold rolled 40Fe as cast 33Fe homogenized 33Fe cold rolled 33Fe as cast

30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

2 Theta

850C 800C 750C 700C 650C FCC: (111)

BCC: (200) (220)

(311) (110) (200) (211) (222) FCC: (111) (200) (220)

(311) (222)

(25)

After one hour of heat treatment at the extended temperature range (650-1000°C) existence of

secondary phase is more visible than after 15 minutes heat treatment, although not in all samples. This is shown in figure 16-18. There is consistently no existence of secondary phase shown in any of the samples treated at 1000°C. Furthermore, for the 40Fe sample there is no existence of secondary phase implied for 900°C and 950°C either.

30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

2 Theta

850C 800C 750C 700C 650C FCC: (111)

BCC: (200) (220)

(311) (110) (200) (211) (222)

30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

2 Theta

850C 800C 750C 700C 650C FCC: (111)

BCC: (200) (220)

(311) (110) (200) (211) (222)

(26)

Figure 16: XRD patterns for the mol% 33Fe33Cr33Ni alloy samples after one hour of heat treatment at eight different temperatures.

30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

2 Theta

1000C 950C 900C 850C 800C 750C 700C 650C FCC: (111)

BCC: (200) (220)

(311) (110) (200) (211) (222)

30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

2 Theta

1000C 950C 900C 850C 800C 750C 700C 650C FCC: (111) (220)

BCC: (200)

(311) (110) (200) (211) (222)

(27)

The patterns shown in the previous figure 13-18 show the existence of secondary phase. In the next paragraph an attempt is made to quantify these results to measure the amount of said secondary phase.

4.4 Quantified RIR values

After quantification with the RIR method, the numbers for the following graphs (figure 19-20) were obtained. Because the diffraction pattern resembled that of BCC structure calculations were performed with BCC data.

Figure 19: RIR quantified data from XRD measurements on samples after one hour of heat treatment at varying temperatures.

30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

2 Theta

1000C 950C 900C 850C 800C 750C 700C 650C FCC: (111)

BCC: (200) (220)

(311) (110) (200) (211) (222)

0 0,01 0,02 0,03 0,04 0,05 0,06 0,07 0,08 0,09

650 700 750 800 850 900 950

Volume fraction BCC

Temperature [C]

33Fe 40Fe 45Fe

(28)

Figure 20: RIR quantified data from XRD measurements on samples after 15 minutes of heat treatment at varying temperatures.

Quantified amounts of secondary phase are consistently the highest in the mol% 33Fe33Cr33Ni samples, except for the condition 750°C for 15 minutes where the mol% 45Fe30Cr25Ni sample showed a slightly higher quantified amount. The lowest amount of secondary phase is similarly consistently found in the mol% 40Fe30Cr30Ni samples except for the condition 700°C for 15 minutes where the mol% 45Fe30Cr25Ni sample shows a slightly lower quantified amount. The secondary phase seemingly never exceeds a volume fraction of 0.09.

4.5 Results from Hardness-Vickers test

Below, in figure 21, the average measured hardness is presented. The average is taken over the 15 measurements for each sample. As seen in the graph in figure 21 the hardness decreases in each of the alloys with an increase in temperature. The mol% 33Fe33Cr33Ni alloy consistently shows the highest hardness for each heat treatment temperature studied. Similarly, the mol% 40Fe30Cr30Ni alloy consistently shows the lowest hardness among the three alloys.

Figure 21: Average hardness measurement in 12 selected samples after 1 hour aging treatment.

0 0,01 0,02 0,03 0,04 0,05 0,06 0,07 0,08 0,09

650 700 750 800 850

Volume fraction BCC

Temperature [C]

33Fe 40Fe 45Fe

100 130 160 190 220 250 280

800 850 900 950

Hardness Vickers

Temperature [C]

33Fe 40Fe 45Fe

(29)

4.6 Results from EDS analysis

Microstructural images of the three alloys are seen in figure 22-23. It is visible in all four of these images that particles are dispersed in the matrix phase as predicted from XRD analysis. Results from chemical analysis of the particles conclude higher amount Cr and lower in Ni than matrix phase, however with varying degrees. Both the lighter and darker particles show concentrations higher in Cr than the matrix phase, shown in tables to right of figure 23.

Figure 22: a) 33Fe33Cr33Ni, b) 40Fe30Cr30Ni, c) 45Fe30Cr25Ni, microstructure after 1h aging at 850C

Figure 23: mol% 33Fe33Cr33Ni (1hour, 800°C sample) EDS analysis points and data tables per measuring point.

a. b. c.

Element Wt% At%

Cr 53.47 55.77

Fe 26.19 25.44

Ni 20.34 18.79

Cr 46.27 48.62

Fe 28.59 27.98

Ni 25.14 23.40

Matrix point 1

Cr 31.23 33.36

Fe 32.67 32.49

Ni 36.10 34.15

Matrix point 2

Cr 29.38 31.45

Fe 33.07 32.96

Ni 37.54 35.59

(30)

5 Discussion

The model alloy system chosen for this project is different to the often-studied CoCrNi MEA in the sense that equimolar FeCrNi has higher possibility of forming secondary phase during isothermal treatment. It was suggested by Liang et al. that this creates a possibility of achieving precipitation hardening, which in combination with a UFG matrix can create sufficient hardness in the material without decreasing ductility too much [8]. The CoCrNi equimolar alloy has good fracture toughness, and good hardness as well, but the high price of Co could make the UFG FeCrNi MEA a better candidate for some applications. It is also beneficial to remove the Co from the alloy to decrease environmental impact and health concerns due to Co in production. If the FeCrNi alloys can be precipitation hardened while keeping most of the ductility it could be an interesting structural material. Experimental findings show formation of secondary phase for a wide range of aging

temperatures in all three of the FeCrNi alloys studied in this project. This should be studied further to find out if FeCrNi can be successfully strengthened by particle precipitation.

5.1 Alloy quality assurance

Analysis of Cr content shows the expected values with mole fractions measuring 0.3-0.33. Fe content is consistently 0.01 higher than expected, ranging through mole fractions 0.34-0.46 instead of the expected 0.33-0.45. This at the expense of Ni content, which is consistently 0.01 lower, with the mole fraction range 0.24-0.32. Because both samples from each alloy respectively show the same

composition, the numbers are deemed trustworthy. Also, because the analysis shows only slight deviation from the goal composition this allows for furthering the analysis of the samples produced.

The results for impurity Al analysis, also through ICP-OES, show that the ppm level of Al in the samples are all around 300 ppm which is expected. There is consistently more Al impurities in the samples with higher content Fe which leads to the thought that the bulk Fe material could have higher Al impurity content than the other bulk material. As can be seen in the chart in figure 24 below, the Al impurity content seems to increase almost proportionally to the Fe content of the sample. This leads to the assumption that a great amount of Al impurity comes from the bulk metal Fe used for producing the alloys. However, it is known that the bulk Fe has the least amount of Al impurity out of the three bulk materials. Bulk Fe should contain close to no Al at all, while there should be 670ppm Al in bulk Cr and 30ppm Al in bulk Ni. Because of this information, the Al impurity analysis results are less

intuitive. However, it is decided that the results are reliable due to the consistent result for each alloy composition.

Figure 24: Al impurity [ppm] relation to Fe content [mol%]

45Fe1 45Fe2

40Fe2 40Fe3

33Fe1 33Fe2

200 220 240 260 280 300 320 340 360 380

30 32 34 36 38 40 42 44 46 48 50

Al content [ppm]

mol% Fe in sample

(31)

A first consideration is the fact that the alloys were manufactured using an alumina crucible. This could possibly lead to Al from the crucible material ending up in the final alloy. Still, because the Al level is low enough in all samples, the alloys were found to have good enough quality for further use in the project.

5.2 Secondary phase analysis

The initial questions posed during this project have been if secondary phase will form and, in that case, if it would be presented as sigma or BCC phase. The equilibrium phase diagrams from Thermo-Calc and FactSage imply possibility of forming both phases at different temperatures in the range, and mobility calculations suggest rapid sigma formation in all alloys. The results from XRD confirm that the alloys are all single-phase FCC as cast, as homogenized and after cold rolling. The samples annealed at 1000°C are also fully FCC across all three alloys. The lower annealing temperatures, 650- 950°C, all yielded formation of secondary phase in alloys 45Fe and 33Fe, while secondary phase was only formed after aging at 650-850°C in the 40Fe alloy. However, from studying results, specifically from XRD analysis, it is found that the secondary phase formed in the three alloys seems to be BCC, and sigma phase cannot be confirmed. Comparing this to equilibrium phases found from using Thermo-Calc and FactSage these results seem unlikely and there could be undetectable amounts of sigma that might be found using something other than XRD. The size of the particles imply that the particles were formed during heat treatment, and not in any other step in the process.

This result suggests that a metastable BCC reaction takes place during heat treatment, because BCC is found in all samples where secondary phase could be observed. It is interesting because the initiation of this project was motivated with the intention to compare the experimental findings to the findings of Liang et al. [8] where a powder sintered ingot showed precipitation of BCC. However, it was initially hypothesized in this new project that because the manufacturing method is different, there might be a difference in outcome. Also, the mobility calculations in Thermo-Calc PRISMA module predicted faster formation of sigma phase. However, sigma was not found in the experiments, which means that the metastable BCC reaction has been favored in this case or that sigma was not detected with the method used (XRD). If sigma was not formed, this could be due to the high dislocation density in the material after cold rolling, even though attempts were made to model based on this property as well.

With a high number of dislocations, it could be difficult to model the precipitate formation.

Nevertheless, this way of modeling this precipitation behavior was less successful in terms of experimental results matching the model. Adjustments to the calculations are probably needed and might yield a more satisfactory result.

When the RIR quantified volume fraction of secondary BCC phase was studied alongside the other experimental results, it seems that the volume fraction of BCC particles correlates to the results from hardness measurement. A higher volume fraction coincides with a higher hardness in all samples.

When this is viewed together with the information that BCC particles are indeed, as hypothesized, rich in Cr and low in Ni, shown from EDS results, it seems that there could be precipitation hardening effects in the alloys from the Cr rich particles produced. As seen in figure 25 the curves for quantified volume fraction and for hardness Vickers measurements have very similar appearance and shape.

(32)

Figure 25: Comparison between RIR calculated volume fraction BCC and Hardness Vickers for the three alloys.

As seen in the database values from ThermoCalc software, shown in figure 26, the relative equilibrium BCC content between the three alloys is similar to the relative RIR quantified volume fraction BCC.

This implies it is expected for the 40Fe alloy to show the least amount of BCC secondary phase and the 33Fe alloy to show the highest amount of BCC secondary phase. This might seem counterintuitive, but this is due to the FCC stabilizing Ni and the BCC stabilizing Cr. The 33Fe alloy has the highest Cr content and the highest Ni content, while the 45Fe has the lowest Ni content and, together with the 40Fe alloy, the lowest Cr content. Hence, the high Cr content destabilizes the FCC phase in the 33Fe alloy, and the low Ni content destabilizes the FCC phase in the 45Fe alloy. The combination of the lowered Cr content and slightly higher Ni content in 40Fe seems to be the reason that less secondary phase is produced in this alloy.

Figure 26: Values from ThermoCalc software steel database TCFE10 showing metastable equilibrium volume fraction BCC in each alloy with regards to temperature.

5.3 Comparing calculated models

One of the aims for this project has been to investigate the usefulness of the different databases used in FactSage and Thermo-Calc software. Models were created with three different software database packages, first as a way to try and predict the experimental outcome and then, after studying experimental results, to fit the model to these results.

0 0,02 0,04 0,06 0,08

800 850 900 950

Volume fraction BCC

Temperature [C]

33Fe 40Fe 45Fe

130 170 210 250 290

800 850 900 950

Hardness Vickers

Temperature [C]

33Fe 40Fe 45Fe

0 0,05 0,1 0,15 0,2 0,25

500 600 700 800 900 1000

Volune fraction BCC

Temperature [C]

33Fe 40Fe 45Fe

(33)

5.3.1 Equilibrium calculations

Three phase equilibrium diagrams are presented in this report displayed in figures 1, 3 and 4. All three of the diagrams contain the same equilibrium phases: FCC, BCC and sigma phase. The diagram from FactSage steel databases (figure 4) is quite similar to the diagram from Thermo-Calc HEA database (figure 1) in the amount and species of secondary phase predicted for the temperatures in the interval.

The biggest discrepancy between the two diagrams is the prediction of smaller amounts of BCC phase at elevated temperatures for all three of the alloys studied in the FactSage diagram. The Thermo-Calc HEA database shows only BCC as a possible secondary phase for the equimolar 33Fe alloy, while it shows equilibrium for sigma formation in the other two of the alloys. Similarly, the FactSage diagram shows both BCC and sigma in equilibrium with FCC in the 40Fe and 45Fe alloys, but only BCC as secondary phase for the 33Fe alloy. However, in the Thermo-Calc HEA calculation, there is no BCC in equilibrium for the temperatures above the sigma region in any of the alloys, which is the case in FactSage diagram.

The diagrams in figure 1 and figure 4 are hence quite similar, except for smaller discrepancies.

However, the third phase equilibrium diagram (from Thermo-Calc Steel database TCFE10) shown in figure 3 has greater differences from the other two calculations and shows sigma and BCC phase in equilibrium with FCC phase for all three alloys, and not only in the 40Fe and 45Fe alloys. Additionally, the temperature ranges for sigma equilibrium are wider and the equilibrium fraction of sigma phase is larger over the entire range of sigma equilibrium temperatures.

When the three phase equilibrium diagrams are compared to the results from XRD phase analysis it is quite clear that none of the three models are perfect for prediction of secondary phase precipitation, because sigma is not found in any of the samples, at least not in detectable amounts. However, it can be inferred that the equilibrium diagram that corresponds the best to the experimental results is the one produced with TCHEA4 database, because it shows a greater equilibrium of BCC formation, as well as no secondary phase at 1000°C. It is evident that equilibrium calculations alone are insufficient for predictions of secondary phase formation.

With the Thermo-Calc databases, TCHEA4 and TCFE10, ternary phase diagrams were calculated in addition to the phase equilibrium diagrams in figures 1, 3 and 4. These phase diagrams are displayed in tables 3 and 4, whereas in table 3 all equilibrium phases are displayed, while table 4 shows the metastable phase diagrams for FCC in equilibrium with sigma phase and BCC phase separately. From studying these phase diagrams together with experimental data, it is inferred that metastable

precipitation of BCC is what could have occurred in the studied material samples. According to the phase diagrams, the BCC phase should be higher in Cr-content than the matrix FCC phase, which is in accordance with the results from EDS analysis. It is important to point out that the sigma phase should also be higher in Cr than the FCC phase. However, because no sigma could be detected in XRD testing, it is unlikely that particles of the size shown in microstructural images could be sigma phase, because the amount should in that case be high enough to detect in XRD peaks.

5.3.2 Precipitation module calculations

The precipitation behavior in the three alloys was simulated in Thermo-Calc PRISMA module, with the database packages for HEA and steel alloys. However, the two database packages gave highly similar results, and it was deemed unnecessary to display results from both packages, hence why only the TCFE10/MOBFE5 results are displayed in this project report. All PRISMA calculations discussed in this section are disclosed in table 5, where the input parameter values are also stated. Prisma simulations will be referred to by their labels stated in table 5.

(34)

The first simulation, BG-1, is performed for precipitation of BCC phase at grain boundary sites, for parameter values attempting to simulate precipitation in previously annealed material where it would be assumed that grain boundaries are the most dominant sites for nucleation of secondary phase. BCC B2 was selected because of the equilibrium calculations showing ordered phase, but results were the same for both ordered and disordered BCC when performing PRISMA calculations. Default values are used for interfacial energy, dislocation density and the mobility enhancement factor. In this

calculation, no initial precipitation of BCC was found for any of the alloys for the 30-hour heat

treatment simulation. The second simulation, also for precipitation of BCC on grain boundary sites, is called BG-2 (figure 5) and is an attempt to model grain boundary precipitation directly on cold rolled material. Input parameters are set to an estimated highest realistic value for dislocation density and mobility enhancement factor, and the lowest possible interfacial energy is used. The mobility

enhancement could be higher, but this will be discussed for future work in this area. The BG-2 results show initial precipitation of BCC within one hour of heat treatment for some temperatures in the 33Fe equimolar alloy, but precipitation occurs much later in the other two alloys. However, using grain boundaries as nucleation sites might be a disadvantageous way of modeling heavily deformed material, which is why the same calculation was performed using dislocation sites for nucleation instead in BD-2 (figure 6). From simulation BD-2 it can be inferred that dislocations are more advantageous for nucleation because the model shows rather early initial precipitation of BCC in the 33Fe alloy, and although 40Fe and 45Fe alloys are not showing initial precipitation within the one-hour mark, they still occur much earlier in BD-2 than in BG-2.

The same simulations were run for formation of sigma secondary phase, in simulations SG-1 (figure 9), SG-2 and SD-2. From SG-1, which had the same input parameters as the BG-1 simulation, results showed very speedy initial formation of sigma phase for all three alloys. However, the SG-2 and SD-2 simulations, which correspond to the BG-2 and BD-2, yielded no result for 30 hours heat treatment.

This was interpreted as there being a higher chance of getting more BCC phase and no sigma phase if aging heat treatment was performed on highly deformed material. Hence, with these six initial simulations the decision was made to perform the aging treatment cycles directly on cold rolled material, and by doing so disregarding the conventional annealing usually performed before aging treatment. This was done because sigma phase is usually highly disadvantageous for material quality due to it being brittle, and because BCC was the precipitate phase studied in earlier works.

Because experimental results could only confirm BCC precipitation, the BD-2 model was assumed to be the most correctly corresponding to these values, and further simulations were performed to find if the model could be made to fit better to the quantified XRD data. BD-3,4 and 5 (figure 7, 8 and 10 respectively) are simulations preformed after analysis of experimental results and are all modified versions of the BD-2 simulation. BD-3 was performed with default dislocation density, while keeping the elevated mobility factor and the lowered interfacial energy. The result showed almost no

precipitation at all, except for some precipitation of BCC in the 45Fe alloy after several hours. This confirmed the necessity of elevating the dislocation density for the model to work. BD-4 simulation was performed to find if the model could work with only the elevated dislocation density, without increasing the mobility enhancement factor and lowering the interfacial energy. This result showed some possibility of BCC formation in the 33Fe alloy, but still quite far from the BD-2 model. Hence, after these two further simulations it was seemingly important to keep the changed parameter values as they were in BD-2 to keep the model somewhat in accordance with experimental values. However, after further studying of literature and quantified XRD values, it was found that the one parameter that could possibly be further altered to make the model fit better is the mobility enhancement factor.

It had been assumed before experimental investigations that the mobility enhancement factor could not realistically be higher than 3, but from results it was inferred that perhaps a higher value should be used in the case for heavily deformed material. To see if the model could fit better with elevated mobility enhancement factor, the BD-2 model was altered once more to create simulation BD-5, where the mobility enhancement factor was set to 5 instead of 3. With this change in parameter value the result showed faster initial precipitation of BCC in all alloys. This was interpreted as being an improvement to the model in terms of fit to the experimental values for BCC fraction. When even

Heat Treatment and Secondary Phase Formation in FeCrNi Medium Entropy Alloys

INOM

EXAMENSARBETE MATERIALDESIGN, AVANCERAD NIVÅ, 30 HP

STOCKHOLM SVERIGE 2021 ,

Heat Treatment and Secondary Phase Formation in FeCrNi

Medium Entropy Alloys

AMANDA CARSBRING

KTH

Abstract

Sammanfattning

Table of Contents

1 Introduction

2 Background

3 Method

Composition (Mol%) 33Fe33Cr33Ni 40Fe30Cr30Ni 45Fe30Cr25Ni Theoretical Mass Fe (g) 201,19 241,28 272,14

Theoretical Mass Cr (g) 187,32 168,49 168,92 Theoretical Mass Ni (g) 211,49 190,23 158,93 Theoretical Mass total (g) 600,00 600,00 599,99

Measured Mass Fe (g) 201,25 241,50 272,15

Measured Mass Cr (g) 187,33 168,43 168,93

Measured Mass Ni (g) 211,43 190,22 158,93

Measured Mass Total (g) 600,01 600,15 600,01

4 Results

a. b. c.

5 Discussion