IN
DEGREE PROJECT MATERIALS SCIENCE AND ENGINEERING,
SECOND CYCLE, 30 CREDITS ,
STOCKHOLM SWEDEN 2019
Understanding the effect of
temperature and time on Gamma
prime coarsening for Nickel-base
superalloy Haynes 282
KEVIN VATTAPPARA
KTH ROYAL INSTITUTE OF TECHNOLOGY
i
Abstract
Haynes 282 is a gamma prime (𝛾′)-strengthened nickel base superalloy developed in 2005, exhibiting a good combination of high temperature properties and fabricability. Microstructural features such as 𝛾′ and carbides play an important role in deriving the mechanical properties of the alloy during heat treatment. As Haynes 282 is a relatively new alloy with insufficient literature availability, the present thesis is aimed at studying the evolution of microstructure for different heat treatment times and temperature with a special focus on 𝛾′ phase precipitation kinetics with different initial conditions for the material. The study is divided into two sections with objectives which are focused on the different ends to the heat-treatment time scales.
The first objective of this study was to investigate γ' precipitation at short heat treatment times and develop Time-Temperature Precipitation (TTP) and Hardness (TTH) diagrams for Haynes 282 using a novel arc heat treatment. In this technique, a steady state temperature gradient, covering room temperature to liquidus, was created using stationary TIG arc on a disc mounted on a water-cooled chamber. Aged and solutionized samples were arc heat treated for 1.5 minutes, 30 minutes and 4 hours. The study was complemented with temperature modelling, thermodynamic calculations, and 𝛾′ precipitation simulation. A unique graded microstructure formed, consisting of dendritic region in fusion zone; dissolution area of all phases including MC carbides, grain boundary carbides, and 𝛾′; grain boundary carbide zone, 𝛾′ band; and base metal. 𝛾′ precipitate size increased with increasing time and temperature. 𝛾′ precipitation simulation model was developed, and it showed very good agreement with experimental results. Finally, the results were summarized in TTH and TTP diagrams.
The second objective in this work was to study understand the coarsening behaviour of 𝛾′ phase with an initial pre-heat-treated GKN heat treatment using furnace heat treatment. Isothermal heat treatments for temperatures from 800°C to 1120°C and times from 30 seconds to 96 hours were performed. Morphological changes in 𝛾′ phase, particle size distribution, grain sizes and hardness on these isothermal heat-treated states are presented in this work. Additionally, A TC PRISMA precipitation model was evaluated to predict 𝛾′sizes and compare it with the measurements. It was concluded that complex initial microstructure, containing bimodal distribution of 𝛾′ precipitates, caused deviations between predicted and measured values, while the model, in the previous objective, predicted the sizes in close approximation to the experimental values. Therefore, further understanding and development of precipitation kinetics with the software should be done to achieve closer results to the experiment.
ii
Sammanfattning
Haynes 282 är ett gamma prime (𝛾′) - förstärkt superlegering av nickelbas som utvecklades 2005 och uppvisar en god kombination av högtemperaturegenskaper och tygbarhet. Mikrostrukturella egenskaper såsom 𝛾′ och karbider spelar en viktig roll för att få de mekaniska egenskaperna hos legeringen under värmebehandling. Eftersom Haynes 282 är en relativt ny legering med otillräcklig litteraturtillgänglighet syftar den aktuella avhandlingen till att studera utvecklingen av mikrostruktur för olika värmebehandlingstider och temperatur med ett särskilt fokus på 𝛾′ fasutfällningskinetik med olika initiala förhållanden för materialet. Studien är uppdelad i två sektioner med mål som är inriktade på de olika ändarna på värmebehandlings tidsskalorna.
Det första syftet med denna studie var att undersöka 𝛾′nederbörd vid korta värmebehandlingstider och utveckla Time-Temperature Precipitation (TTP) och Hardness (TTH) diagram för Haynes 282 med användning av en ny bågvärmebehandling. I denna teknik skapades en jämn temperaturgradient, som täcker rumstemperatur till liquidus, med användning av stationär TIG-båge på en skiva monterad på en vattenkyld kammare. Åldriga och lösningsbara prover bågvärmebehandlades under 1,5 minuter, 30 minuter och 4 timmar. Studien kompletterades med temperaturmodellering, termodynamiska beräkningar och 𝛾′utfällningssimulering. En unik graderad mikrostruktur bildad, bestående av
dendritisk region i fusionszon; upplösningsområde för alla faser inklusive MC-karbider, korngränsande karbider och 𝛾′; korngränsen karbidzon, 𝛾′ band; och oädel metall. 𝛾′utfällningsstorlek ökade med
ökande tid och temperatur. 𝛾′utfällningssimuleringsmodell utvecklades, och den visade mycket bra överensstämmelse med experimentella resultat. Slutligen sammanfattades resultaten i TTH- och TTP-diagram.
Det andra syftet med detta arbete var att studera förstå det förgrovning beteendemönster hos 𝛾′ fasen
med en initial förvärmebehandlad GKN-värmebehandling med ugnsvärmebehandling. Isotermiska värmebehandlingar för temperaturer från 800 ° C till 1120 ° C och gånger från 30 sekunder till 96 timmar utfördes. Morfologiska förändringar i 𝛾′fasen, partikelstorleksfördelning, kornstorlekar och hårdhet på
iii
Abbreviations
TTT
Time-Temperature-Transformation
VOL%
Volume Percent
WQ
Water Quenched
AC
Air Cooled
HAZ
Heat Affected Zone
DOE
Design of Experiments
TIG
Tungsten Inert Gas
SEM
Scanning Electron Microscope
TC
Thermocouple
OM
Optical Microscope
TTH
Time-Temperature-Hardness
TTP
Time-Temperature-Precipitation
PSD
Particle Size Distribution
iv
List of Figures
v
sample of 1.5 minutes shows the start of the dissolution of the phases below the fusion zone. The samples
of 30 minutes and 4 hours show the area for dissolution of phases increasing with time. ... 24
Figure 22 Evaluation of γ́ band with increasing times for both initial conditions. The as received samples are etched for the γ′ band and the base metal while the solutionized samples are etched for only the γ′ precipitates... 24
Figure 23 (a) Image to depict the change in the microstructure in the different regions of the arc heat treated sample showing (b) MC carbide dissolution. (c) grain boundary carbide formation. (d) γ′formation ... 25
Figure 24 SEM micrograph showing fusion zone with dendritic structure and MC carbides formed due to segregation in the fusion zone (shown in the insert at higher magnification) along the boundary ... 26
Figure 25 Evolution of microstructure for γ́ precipitates within the γ′ band for the solutionized samples. ... 27
Figure 26 Grain boundary carbides and precipitates along a grain boundary. (a) High temperature grain boundary showing only grain boundary carbide with discrete morphology. (b) Grain boundary within the start of the γ' band showing a low fraction of γ' precipitates within the grain and a depleted region near the grain boundary. (c) & (d) Grain boundary at a lower temperature show a higher fraction of precipitates in the grain but still reduced fraction near the grain boundary. (e) & (f) show a similar trend for the grain boundaries at lower temperatures with higher fraction of precipitates in the grains and along the grain boundaries as well. ... 28
Figure 27 Micro-hardness map depicting the evaluation of hardness for the γ́ band for all the cases. The top hardness maps show arc heat treated as received condition, which have higher initial hardness than the solutionized samples. Arc heat treatment increased hardness within the γ′ band showed by different etching response in both sets of samples. ... 29
Figure 28 Equilibrium Diagram for Haynes 282 from JMatPro software. The different regions marked on the equilibrium diagram represent the different phases of interest for the study, region 1 for the precipitation of γ′, region 2 for grain boundary carbides precipitation and region 3 for the precipitation of MC carbides. ... 29
Figure 29 Simulated precipitate sizes at 760˚C, 820˚C, 950˚C and 980˚C. The values in the table shows the calculated sizes for 30 minutes and 4 hours. All the simulations were carried out taking solutionized initial condition... 30
Figure 30 Comparison between etched sample and temperature plot for arc-heat treatment. ... 31
Figure 31 Size comparison for γ' precipitates between the experimentally obtained values and the simulated values. The values for 820˚C and 760˚C are compared between the simulated (Thermo-Calc) and experimental values (S Haas et. al. and Joseph [4].) from literature. The values for 980˚C and 950˚C are compared between the simulated (Thermo-Calc) and experimental values (from this work). ... 32
Figure 32 Plot for TTH for Haynes 282 for the hardness values of 250 HV (Blue), 300 HV (Green), 350 HV (Orange). ... 33
Figure 33 Plot for TTP for Haynes 282 ... 34
Figure 34: Plot for precipitate sizes for various temperature values for heat-treated samples. ... 34
Figure 35 Flowchart describing the flow of work for Furnace heat treatments. ... 36
Figure 36 Schematic diagram for the workflow of simulation tools. ... 38
Figure 37 Simulation setup for Haynes 282 ... 39
Figure 38 Calibrated data vs experimental values. ... 40
Figure 39 (a) Optical image for the sheet (b) SEM image showing no γ′ precipitates [4]. ... 41
Figure 40 γ′Microstructure after solutionizing step of the GKN heat treatment. ... 41
Figure 41 Final γ′microstructure for the GKN heat treatment. ... 42
Figure 42 Grain Sizes for isothermal heat treatments. ... 42
Figure 43 Hardness for the different isothermal heat treatment conditions. ... 43
Figure 44 Hardness with respect to the coarsening time for the precipitates. ... 43
vi
Figure 46 Map for microstructural evaluation for all times and temperatures in the isothermal heat treatments. The temperatures marked 1050 - 1120˚C, depict the grain boundary carbides at higher magnification. The temperatures marked 800-1000˚C depict the γ′precpitates for the isothermal heat
treatments. ... 46
Figure 47 Grain boundary precipitation of γ'. (a) Shows the γ' precipitate free zone (PFZ) after the first step of GKN heat treatment. (b) Shows the γ' precipitation in the PFZ after the second step of GKN heat treatment. (c) Grain boundary showing a γ′ free zone at 1000˚C during the isothermal heat treatment after 96 hours. (d) Grain boundary showing a depleted γ′zone at 950˚C during the isothermal heat treatment after 96 hours. (e) Grain boundary showing a bimodal distribution of γ′ precipitates at 800˚C during the isothermal heat treatment after 96 hours. ... 47
Figure 48 (a) Equilibrium phase fraction for the desired phases with its respective (b) driving forces. 48 Figure 49 Comparison between the experimentally (solid lines) obtained and simulated (dashed lines) values. ... 48
Figure 50 GKN heat treatment thermal profile ... 49
Figure 51 Particle Size Distribution after GKN heat treatment ... 49
Figure 52 Comparison between experimental (solid lines) and simulated (dotted lines) with GKN PSD values ... 50
Figure 53 Parameters contributing to the prediction of the strength by the model ... 51
Figure 54 Experimental Particle Size distribution for the isothermal heat treatment temperatures. ... 51
Figure 55 Predicted strength values for different temperatures as compared to scaled yield strength values. ... 52
List of Tables
Table 1 Effects of alloying elements in Nickel alloys[15]. ... 6Table 2 Phases observed in Superalloys [5, 11]. ... 7
Table 3 Chemical Composition (wt%) for Haynes 282.[4, 7] ... 8
Table 4 Specifications for the various forms of products [9]. ... 8
Table 5 Different methods for calculating TTT diagram ... 13
Table 6 Designation of samples and arc heat treatment time. ... 19
Table 7 Metallographic preparation for arc-heated samples. ... 19
Table 8 Sizes with respect to the different times and temperatures for the solutionized samples. ... 27
Table 9 Isothermal times and temperatures for Furnace heat treatment trials ... 36
Table 10 Metallographic preparation for furnace heated samples. ... 37
Table 11 Test matrix for calibration of parameters ... 39
vii
Table of Contents
Abstract ... i
1
Introduction ... 1
1.1 Background... 2 1.2 Problem ... 3 1.3 Purpose ... 3 1.4 Goal ... 32
Literature StudyLiterature Study ... 4
2.1 Superalloys ... 4
2.1.1 Iron-Nickel-base Superalloys ... 5
2.1.2 Cobalt-base Superalloys ... 5
2.1.3 Nickel-base Superalloys ... 6
2.1.3.1 Phases in Ni-base Superalloys ... 7
2.2 Haynes 282 ... 8
2.2.1 Forms – Haynes 282 ... 8
2.2.2 Standard Heat Treatment (SHT) of Haynes 282 ... 9
2.2.2.1 Solutionizing treatment [4, 7] ... 9
2.2.2.2 Age-hardening treatment [4, 7] ... 10
2.3 Phase Transformations ... 11
2.3.1 Time-Temperature-Transformation (TTT) Diagram ... 11
2.3.2 Thermodynamic calculation background ... 14
2.4 Arc-heat Treatment ... 14
3
Physical simulation of short time aging time using Arc heat treatment ... 16
3.1 Experimental Work ... 16
3.1.1 Material ... 16
3.1.2 Experimental Setup ... 17
3.2 Metallography ... 19
3.3 Microstructural characterization ... 19
3.3.1 Microscopy (Stereo & OM) ... 19
3.3.2 Hardness Testing ... 20
3.3.3 Scanning electron microscopy (SEM) ... 20
3.4 Simulation setup ... 20
3.4.1 COMSOL Multi-physics® Modelling Software ... 20
3.4.2 Precipitation Calculations - Thermo-Calc PRISMA software ... 22
3.5 Results of first objective ... 22
3.5.1 Temperature Modelling ... 22
3.5.2 Initial conditions for the samples ... 23
3.5.3 Optical Microscopy ... 24
3.5.4 Scanning Electron Microscopy ... 26
3.5.5 Microhardness ... 28
3.5.6 Thermodynamic calculations ... 29
3.5.6.1 Equilibrium calculations ... 29
3.5.6.2 TC PRISMA Calculations ... 30
3.6 Discussions about the first objective ... 30
3.6.1 Microstructural map for arc heat treated sample ... 30
3.6.2 Region of dissolution of phases ... 31
viii
3.6.4 Time – Temperature Diagrams ... 33
3.6.4.1 Time – Temperature – Hardness (TTH) diagram ... 33
3.6.4.2 Time – Temperature – Precipitation (TTP) diagram ... 34
3.7 Conclusions for the first objective ... 35
4
Furnace heat treatment ... 36
4.1 Experimental Work ... 36
4.1.1 Material ... 36
4.1.2 Experimental Setup ... 36
4.2 Metallography ... 37
4.3 Microstructural Characterization... 37
4.3.1 Optical microscopy (OM) ... 37
4.3.2 Hardness ... 37
4.3.3 Scanning electron microscopy (SEM) ... 37
4.4 Thermodynamic, precipitation and strength simulations ... 37
4.4.1 Precipitation Calculations... 38
4.4.2 Strength Modelling ... 40
4.5 Results & Discussions about the second objective ... 40
4.5.1 Results from the Experimental work ... 40
4.5.1.1 Initial Microstructure ... 40 4.5.1.2 Grain Size ... 42 4.5.1.3 Hardness ... 42 4.5.1.4 𝛾′Precipitate Size ... 43 4.5.2 Simulation results... 47 4.5.2.1 Equilibrium Calculations ... 47 4.5.2.2 Precipitation calculations ... 48 4.5.2.3 Strength Modelling ... 50
4.6 Conclusions for the second objective ... 52
5
Final Conclusions ... 54
6
Future Work ... 55
7
Social and ethical aspects ... 56
8
Acknowledgement ... 57
9
References ... 58
1
1 Introduction
An engine is a complex arrangement of various components as shown in Figure 1, this assembly of parts makes the aero-engine one of the most researched areas in the aerospace industry. Casting of single pieces has been a traditional approach to produce some of the components in this industry. However, with improvements in manufacturing technology, more methods are available to produce parts with desirable shapes and properties using other techniques such as advanced welding processes and additive manufacturing.
Figure 1 Material selection for different engine parts [1].
2
1.1 Background
In the earlier decades of the 20th century, stainless steels were being used for high temperature applications. However, this class of materials were not able to keep with the temperature and corrosion demands imposed on them for the higher engine efficiency. This, therefore, lead that the designers have developed alloy compositions, which had superior high temperature capabilities. This application of “superior alloy compositions” has led to the development of a class of material, “Super-alloys” for high temperatures [5].
Superalloys is a “material developed for high temperature conditions and put into areas of application where a tendency of severe mechanical loading has been observed and there is a need for high surface integrity” [6]. This group of materials were developed to be applied at temperatures generally above 540⁰C. In the long history of development and usage of superalloys, many superalloys have been explored over the years but only a few have been used in the aerospace industry [3, 5].
For commercial production of the parts, as mentioned earlier, single piece castings are chosen but due to the high cost of manufacturing, and the limited number of suppliers, other solutions are being researched upon. The aerospace industry has been turning towards sectioning the large components and casting smaller parts; which has bigger market with larger number of suppliers. The component can later be joined by welding. This leads to producing low cost fabricated structures for the engine which are produced by joining various parts based on the criticality of the component in the engine [5].
3
Figure 2 Graph showing the higher stress-rupture strength for a larger range of temperatures for precipitation strengthened superalloys primarily nickel superalloys as compared to carbide and solution-strengthening found in cobalt and iron-nickel
alloys [4].
1.2 Problem
The quest for high temperature material led to the development of alloy Haynes 282 in 2005 by Haynes International. The alloy has attracted some interests in the aerospace as well as land-based power generation industries for applications in the high-temperature turbine components [4, 7-9]. Since 2005, microstructure, mechanical properties and thermo-mechanical response have been researched and reported for this alloy. However, the information about the precipitation of 𝛾′ phase in the heat-affected zone after welding of the alloy, the high temperature stability of carbides, validity of thermodynamic calculations, and precipitation and stability behavior of 𝛾′ is largely unstudied for Haynes 282.
1.3 Purpose
The main purpose is to understand the microstructural changes occurring in Haynes 282 with focus on 𝛾′ precipitation for different heat treatment times and temperatures.
1.4 Goal
4
2 Literature Study
This chapter briefly describes the literature study to understand superalloys, its classification and the alloy of interest Haynes 282. Figure 3 shows the flowchart to the different sections.
Figure 3 Flowchart for various sections of literature study.
2.1 Superalloys
5
Figure 4 Thermal profile for some of the extreme temperatures in an engine [10]. In high temperature part of the turbine (right hand), more high-performance alloys such Ni-based are used, while in lower temperatures more Al-based alloys are
used.
2.1.1 Iron-Nickel-base Superalloys
Iron-Nickel-base superalloys are classified as precipitation- and solid-solution strengthened alloys. These superalloys contain typically 25-60% Nickel and additions of 15-60% Iron [6]. Some of the alloys in this class are A-286, Inconel 718, and Hastelloy X [5, 6, 11].
This class of superalloys can be further divided into four types based on their strengthening mechanisms and compositions [6]:
• Strengthening of FCC structured γ’ phase, o Iron-rich
o Nickel-rich.
• Strengthening of BCT structured γ’’ phase. • Fe-Ni-Co system and FCC structured γ’ phase. • Strengthening of grain-boundary carbides.
Alloy 718 is one of the popular alloy in this class contributing to nearly 35% of the world’s superalloy production [12]. It is well-known for its high strength due to γ’’ strengthening phase and has good weldability especially for its post-weld cracking resistance [12-14].
2.1.2 Cobalt-base Superalloys
Cobalt-base superalloys are strengthened by carbides and solid-solution strengthening. It has second preference due to the absence of precipitate strengthening [1]. Typical compositions for this class of superalloy is 40-60% of Co, ̴ 20% Ni, 20-30% Cr, 5-10% W & 0.1 – 10% C. In recent years, Tantalum
6
has been used as successful replacement for Tungsten [1]. Additionally they have better weldability and thermal fatigue resistance than Ni-base superalloys [1, 6].
2.1.3 Nickel-base Superalloys
Usage of Nickel as an alloying element started in the mid-18th century but gathered importance in the 1900’s. Nickel is a reliable corrosion-resistant metal which can be easily alloyed with various elements such as chromium, copper, manganese, gold, iron, molybdenum, silicon, titanium, aluminum, niobium [15]. The contribution of these elements in the matrix to the overall properties for the materials are listed in Table 1.
Table 1 Effects of alloying elements in Nickel alloys[15].
Elements
Thermal properties
Other properties
Ni Stabilization of γ phase.
Helps in the formation of γ’ phase.
Good thermal stability and fabricability.
Cr Precipitation of M23C6.
Solid solution hardening.
In large amounts can decrease the precipitated
fraction of γ’ phase.
Mo Precipitation of M6C.
Solid solution hardening.
Al Helps in the precipitation of γ’ phase (Ni3 (Ti,Al)).
Suppress precipitation of η phase (Ni3Ti) Ti Helps in the precipitation of γ’ phase (Ni3
(Ti,Al)).
C Precipitation of M23C6, M6C and MC. Solid solution hardening.
Nb, Ta Helps in the precipitation of γ’ and γ’’ phase
in Fe-Ni superalloys.
Si Increases fluidity during
casting.
B, Zr Increases creep-rupture strength Suppress precipitation of η phase (Ni3Ti).
7
temperature applications. Another classification in this class is the oxide-dispersion-strengthened alloys which is strengthened by the precipitation of inert phase of Yttria in combination with 𝛾′ [1].
2.1.3.1 Phases in Ni-base Superalloys
In Ni-base superalloys, the matrix phase (𝛾 phase) has a FCC structure. As explained, 𝛾′ is the most important secondary phases in Ni-base superalloys, contributing to the precipitation strengthening for the superalloy class. It has a L12 structure.
In addition to 𝛾′, the MC carbides can be found in the microstructure, forming from melt due to the strong segregation of C and N. It has been reported that the MC carbide contribute to strengthening in superalloys [16]. After heat treatments and long hours of service completed, the MC carbides have been shown to decompose to other phases such as M23C6 at 760-980 °C and M6C at 815-980 °C [17, 18], respectively. The most common reactions are [6, 17]:
1. MC + 𝛾 → M23C6 + 𝛾′ 2. MC + 𝛾 → M6C + 𝛾′
The M23C6 and M6C could also transform from one another [6]. All phases observed have been listed in Table 2 below.
Table 2 Phases observed in Superalloys [5, 11].
Phase
Crystal
Structure
Comments
𝜸′ Ni3Al Ni3(Al, Ti) fcc (ordered L12)Principal strengthening phase in many nickel- and nickel-iron-base superalloys; shape varies from spherical to cubic; size varies with exposure time and temperature. Gamma prime is spherical in iron nickel-base and in some of the older base alloys. In the more recently developed
nickel-base alloys,γ' is generally cuboidal. MC
TiC
Cubic Composition is variable; appears as globular, irregularly shaped particles.
M23C6
Cr23C6
fcc It can precipitate as films, globules, platelets, lamellae, and cells; usually forms at grain boundaries.
M6C Fe3Mo3C, fcc M3B2 Ta3B2 V3B2 Nb3B2
8
2.2 Haynes 282
Haynes® 282® also represented as Haynes 282, a 𝛾′ strengthened wrought-processed alloy developed by Haynes International in the early 2000’s, as a superalloy for high-temperature applications in aero and land-based gas turbines with improvement in corrosion-resistance and creep strength due to the formation of M23C6 in block morphology at the grain boundaries [4, 7]. The alloy is also being looked at for producing structures in automotive and steam turbine application [7, 8]. In the alloy, it was seen that the equilibrium balance for 𝛾′ was carefully designed to give its excellent properties and easily fabricability. The content for 𝛾′ lead to having equivalent or better properties when compared to alloys with higher 𝛾′ content [7].
The composition for the alloy is as follows in Table 3.
Table 3 Chemical Composition (wt%) for Haynes 282.[4, 7]
Ni Cr Co Mo Al Ti C Mn Si Fe B
Bal. 19.44 10.22 9.42 1.41 2.15 0.067 0.06 0.07 0.92 0.004
Pike et al. (2008) [7] have reported that Haynes 282 has higher resistance to strain-age cracking compared to Waspaloy and Rene 41 thanks to the sluggish 𝛾′ precipitation. Kruger (2017) [8] has reported that the major contribution to strength loss of 𝛾′ precipitated alloys is coarsening and dissolution of 𝛾′ phase. Further works on various properties of the alloy, such as microstructure and hardness evolution, weldability, ductility, and creep have been widely performed for conventional and more modern processing techniques [4, 19-22].
2.2.1 Forms – Haynes 282
The different forms of availability of Haynes 282 is as shown in Table 4. For this study, Haynes 282 sheet and bar were used for the various experiments done during this project work.
Table 4 Specifications for the various forms of products [9].
Sheet, Plate & Strip AMS 5951
Billet, Rod & Bar B 637
AMS 5951
Seamless Pipe & Tube B 983
Forgings B 637
9
2.2.2 Standard Heat Treatment (SHT) of Haynes 282
The standard heat treatment (shown in Figure 5) for Haynes 282 comprises of solutionizing treatment and a two-step age hardening treatment. In the solutionizing treatment, the alloy is solutionized at a temperature range of 1120-1149 ˚C followed by water quenching [4, 9]. This step makes Haynes 282 readily formable.
In the age-hardening treatment, the 1st step is the carbide stabilization step at 1010˚C for 2 hours followed by air cooling (1010˚C/2hrs./AC). The 2nd step is the 𝛾′ precipitation step at 788 ˚C for 8 hours followed by air cooling (788˚C/8hrs./AC) [4, 7, 9, 15].
Figure 5 Summarized diagram for the different standard heat treatment steps [4].
2.2.2.1 Solutionizing treatment [4, 7]
After the solutionizing step [4, 7], all the phases are removed except the primary Titanium-rich MC carbides (marked with a white arrow in Figure 6) with a large, blocky morphology. This is because the solutionizing temperatures are above the solvus temperatures for M23C6 carbides, the grain boundaries are clean as shown in Figure 29.
10
Figure 6 Grain boundaries after the solution treatment, free of Carbides except for MC carbides (white arrow). In the annealed condition, a sheet material has been reported [7] to have an ASTM grain size of 4 to 4
1 ⁄ 2.
2.2.2.2 Age-hardening treatment [4, 7]
The 1st step of treatment (Carbide stabilization) consist of heat-treatment at 1010˚C/2hrs./AC where the formation of carbides take place at grain boundaries with a discrete morphology [4], as shown in Figure 7. This results in an improvement in ductility and creep-strength [4, 6].
Figure 7 Carbides (white arrow) at the grain-boundary with a discrete morphology.
11
Figure 8 𝛾′precipitate (yellow circle) distribution after the final step.
Joseph (2018) [4] has reported that the hardness due to the initial precipitation in the alloy depends on the type of cooling induced, a higher hardness was seen in the case for air-cooling as compared to water quenching the samples after the heat treatments [4]. In the case for isothermal heat treatments, the hardness was seen to be decreasing due to the reason that precipitate size were increasing and the number density was decreasing [4]. The coarsening kinetics for 𝛾′ precipitates depends on the size of 𝛾′ which are proportional to the cube root of the aging time [4]. It was also reported that the 𝛾′ solvus was estimated to be above 1010 ˚C but below 1024 ˚C [4].
2.3 Phase Transformations
Variations in the performance and properties of metallic materials can come through a variety of phase transformations, bringing about the alternation of the microstructure [23]. Precipitation is one of the most common phase transformations and occurs with nucleation and growth of the precipitate in the matrix phase.
Nucleation can be of two different forms; Homogenous and Heterogeneous, Homogenous nucleation occurs by the nucleation of the phase directly from a parent phase and requires high driving force. Heterogeneous nucleation, in contrast, happens by the formation of phase at structural deformities such as grain boundaries and dislocation and require much lower energy than homogenous nucleation [24].
2.3.1 Time-Temperature-Transformation (TTT) Diagram
The information received from any heat treatment (HT) is based on the knowledge about the effects of both time and temperature on the microstructure [25]. The progress of the isothermal transformations is shown by plotting the fraction of phase transformation time and temperature i.e. this translates to a TTT diagram. TTT diagram can be described as a plot of temperature versus time (on log scale) for an alloy with a definite composition, where the amount transformed of the phase is represented by a curve for the various temperatures. As shown in the Figure 9 (a), two curves show the 1% and 99% of phase transformation. Figure 9 (b) also shows the plot for the change in fraction of the formed phase for
12
different temperatures, indicating how a TTT diagram is extracted from the phase transformation data [23, 24].
Figure 9 (a) Percentage transformation (1% to 99%) for a phase versus time for different temperatures. (b) Increasing rate of change of fraction of a phase at different temperatures with respect to time as seen from the diagram on the top. [23] A TTT diagram could give a user the information about:
• Nature of the transformation • Type of transformation • Rate of transformations
• Approximate time and temperature of the phases to form for a definite composition of the alloy. • Qualitative information about the grain size scale of the product, in terms of precipitated phases
and about the possibility of grain coarsening [26].
The earliest TTT diagrams were introduced in 1930 based on the extensive study of steel phase transformation kinetics. These TTT diagrams were game-altering as they showed the complex reactions and processes in a more imaginable manner [27]. For Haynes 282, Fahrmann et al. [28] had created an experimental TTT diagram taking the annealed sheet Haynes 282 as the initial condition. The mill-annealed condition has shown the presence of fine cooling γ’, but this does not translate into higher
13
hardness. The γ’ is shown to have spherical in shape and have mono-modal distribution for isothermal exposures. In the case of secondary carbides, it was seen that the rate of precipitation of M23C6 carbides were two orders faster than M6C carbides [28]. The solvus temperatures of both secondary carbides were seen to be in the vicinity of 1121 °C [28].
To calculate and form a TTT diagram, it is necessary to identify microstructure/property of material and to able to measure the evolution of microstructure/properties to detect the phase transformations. Some of the methods of calculating TTT diagram are shown in Table 5 [29].
Table 5 Different methods for calculating TTT diagram
Category Techniques Description
Strain related Dilatometry
Measurement of dimensional changes due to thermal expansion to detect changes of density and coefficients of thermal expansion
caused by phase change. [30]
Resistivity Electrical Resistivity Measurement of differential electrical resistivity due to the variation in the solute concentration in the solid solution. [31]
Temperature related
Gleeble Thermo-mechanical simulator
Originally dedicated to thermomechanical testing, was selected to carry out the isothermal quenching resistivity measurements
because of its reliability in controlling temperature. [32] Calorimetry:
Differential Thermal Analysis (DTA)
Measurement of release or take-up of latent heat compared to a standard reference sample, associated with phase change. [31]
Furnace Heat treatment
Study of phases transformed across a range of temperature applied, on a single sample heat treated to various times. [25]
Arc Heat Treatment
Study of microstructure after the samples have been heat treated using TIG arc and a functionally graded microstructure formed at
different times. [28]
Based on the experimental data, semi-empirical approach such as the Kolmogorov, Johnson, Mehl and Avrami (KJMA) equation can also be used to predict the phase transformation at different temperatures and times. Avrami theory is based on nucleation and isothermal growth, where it is assumed that nucleation occurs at sites within the bulk of grain, which are progressively reduced in number. The KJMA equation can be expressed as:
Where 𝑉𝑓𝑎 is the volume fraction of phase a, n is a numerical exponent that can take values from 1 to 4
and 𝑧𝑎 is a temperature sensitive values which depends on the nucleation and growth rates [23, 33-35].
The different values for the numerical exponent “n”:
14
• 1D - 1 < n < 2 • 2D - 2 < n < 3 • 3D - 3 < n < 4
2.3.2 Thermodynamic calculation background
Phase diagrams have been an important part in understanding the interactions between composition, microstructure and processing for alloys [36]. They have calculated based on Gibbs energy calculations which have been developed since the early 1900’s [36]. The equilibrium state is calculated from thermodynamic functions (influenced by composition, temperature and pressure) [36]. Since 1970’s, when CALPHAD (CALculation of PHase Diagrams) has been continuously developed to assist in alloy design after it was first described in the book Computer Calculations of Phase Diagrams [36, 37]. The Calphad method is used in wide ranges of applications which involve the usage of Gibbs energy and its derivatives which is used to calculate the transformations and the properties in multicomponent materials. The method can be used to to evaluate the equilibrium phases fractions, precipitation and coarsening kinetics, diffusion of phases, phase transformations [38]. The second derivative of Gibbs energy equation i.e. chemical potential can be used for diffusion simulations while driving force for microstructural evolutions were based on Landau theory [36].
In theory, the precipitation process happens in three stages: nucleation, growth and coarsening. Thermo-Calc software has developed TC-PRISMA module which is being used to simulate the precipitation characteristics that could consider concurrent nucleation, growth, and coarsening and is modelled based on the LS theory and uses the KWN method for precipitation calculations [38, 39]. PRISMA accommodates the overlapping of different stages by calculating the evolution of the probability distribution of the particle number densities, termed as particle size distribution (PSD) [40, 41].
2.4 Arc-heat Treatment
Most conventional methods for producing TTT diagrams are time consuming. They involve a large number of samples and design of experiments (DOE) to cover most of the points for various temperatures and times in the TTT diagram. It is very difficult to cover all the temperatures and times using the conventional methods and apart from the large number of experiments, the calculations using the Avrami equation have also been used for extrapolating the transformations for the diagram [25, 42].
15
In this technique, a stationary tungsten inert gas (TIG) arc is applied on a water-cooled sample that is mounted on a holder. After turning on the arc, steady state melt pool forms on the disc thanks to the arc heating from the topside and water cooling on the bottom. This makes steady condition, as shown in Figure 10 with constant temperature at different locations. By variation of holding time, it is possible to get information about all temperature with only few samples. This methodology has been already used to draw TTT diagrams for phase transformations in super duplex stainless steels. The temperature readings were further used as an input in the simulation model to calculate the constant temperature distribution (as shown in Figure 11) in the heat-treated material [44, 45].
Figure 10 Schematic thermocouple data (representative marking) showing steady state achieved quickly for various points, which is maintained for the entire period of the heat treatment. The model on the top is for representative purposes, to depict
the sample with the arc on the top side forming a fusion zone
16
3 Physical simulation of short time aging time using Arc
heat treatment
As described in the previous chapter, arc heat treatment allows reaching the steady constant temperature quickly. The aim of this chapter, therefore, is to study 𝛾′ precipitation at short heat treatment times using this technique, create TTP and TTH diagrams, as well as evaluate 𝛾′ precipitation model using TC-PRISMA software. The outline for the work done for this part of the study is shown in Figure 12.
Figure 12 The methodology to physical simulation of short time aging using arc heat treatment. After AHT, the sample studied with different techniques to create TTP and TTH diagrams. Finally, precipitation model was evaluated using
TC-PRISMA
3.1 Experimental Work
3.1.1 Material
Haynes 282 bar samples of the chemical composition as listed in Table 3, with the dimension of 99 mm diameter x 6 mm thickness, shown in Figure 13, has been studied. The experiments were conducted on as-received (aged) sample with the initial hardness of 336 HV and solutionized (1120˚C for 90 minutes) sample with the initial hardness of 185 HV.
17
Figure 13 Sample for Arc-heat treatment with the size ∅ 99𝑚𝑚 𝑥 6 𝑚𝑚 𝑡ℎ𝑖𝑐𝑘.The lines on the sample represent positions
for the placement of the thermocouples.
3.1.2 Experimental Setup
The schematic diagram of the experimental setup for the heat treatment is as shown in Figure 14 [25]. As mentioned in the previous section, a stationary TIG was applied on a stationary disc water cooled from the backside to produce steady state melt pool and temperature gradient. The sample is bolted with the help of a holder creating a sealed chamber (formed by using a rubber gasket as seen in Figure 14). Water was circulated in this chamber for the symmetric heat removal from the sample during the heat treatment (the water flow pattern marked by arrows in Figure 14). The water flow rate was maintained at 1.4 l/min. The inlet water was maintained at 8.5˚C + 1 throughout the experiment. This helped to maintain a steady state condition in the sample.
18
Figure 15: Arc-heat treatment in progress with the different areas marked as seen in the schematic diagram. A Tungsten Inert Gas (TIG) torch was used to produce the arc with a programmable TIG COMMANDER AC/DC 400 power source. The electrode has a diameter of 2.4 mm and having a tip angle of 60°. The arc length was set to 3 mm. Figure 15 depicts the heat treatment in progress for one of the heat treatments.
Temperature measurement was done using 11 K-type thermocouples (TC) which were spot welded to the sample subjected to heat treatment 30 minutes as the calibration run. 9 of 11 TC were attached on the top side of the sample as shown in Figure 16, while the remaining 2 were attached at the bottom of the sample. The ones on the top side were spaced equally from the fusion zone.
Figure 16 Actual experimental setup with thermocouple attachment. The thermocouples are arranged in a circular format to capture the temperatures around the fusion zone.
TIG Torch
Arc
Aluminum Box
with the sample
19
The arc-heat treatment was conducted for various times, as given in the Table 6, for samples with 2 different initial states.
Table 6 Designation of samples and arc heat treatment time.
Time As received Solutionized
1.5 minutes 1a (Specimen marking) 1b
30 minutes 2a 2b
4 hours 3a 3b
3.2 Metallography
Heat treated samples were prepared for the microstructural analysis. The metallographic preparation steps for heat treated samples is as shown in Table 7:
Table 7 Metallographic preparation for arc-heated samples.
Sample preparation
step Equipment Consumables Parameters
Cutting Water-jet
Machining - -
Mounting Citopress 30 Polyfast Struers’
program
Grinding and
polishing Tegramin 30
MD-Piano discs:
80 grit, 220 grit, 360 grit, 500 grit
Struers’ program 9-micron diamond abrasives;
MD-Allegro disc
3-micron diamond abrasives; MD-Allegro
0.5/1 µm Alumina suspension; MDChem pad
Etching - Gamma prime etch
2% Oxalic acid
3V for 10 seconds
3.3 Microstructural characterization
3.3.1 Microscopy (Stereo & OM)
20
3.3.2 Hardness Testing
Hardness mapping with more than 400 indents on the cross section of each sample was performed using Qness 10 micro-hardness tester with a 200 grams load at SWERIM AB, Stockholm.
3.3.3 Scanning electron microscopy (SEM)
SEM analysis was done using LEO 1550 a high resolution FEGSEM with a working distance of 9 mm with the accelerating voltage of 10 kV. The images were taken at various locations in the sample to characterize the microstructure from the fusion zone to base metal. Image analysis was performed on SEM images for γ́ precipitate size prediction using ImageJ. Steps for image analysis by ImageJ were (a) Scaling the image (b) Thresholding the image (c) Measuring the area of the precipitates in the SEM.
3.4 Simulation setup
3.4.1 COMSOL Multi-physics® Modelling Software
COMSOL software was used to model the temperature distribution profile formed during the Arc-heat treatment. The model is based on steady state energy conservation with the dependent variable to be taken as temperature as:
Where ρ is the material density, C is the specific heat capacity, 𝜅 is the thermal conductivity, S is the source term for phase change and Q is the heat generation term. The source term and the heat generation terms are not taken into consideration for this calculation. This model has been adapted from an earlier work done by Kumara [46].
Model description: The setup for defining the geometry of the simulation domain is shown below in Figure 17, where different surfaces are marked and are used to describe the model and the simulation setup. As shown in Figure 13, the sample was asymmetric along the thickness (z-axis) with the TIG torch as the heat source was located at the center of the sample as seen in Figure 14 & Figure 15
.
Thus, using the axisymmetric axis, the geometry was converted from a 3D to a 2D geometry (shown in Figure 17) to reduce the computational time. Since the interest of creating this model was to predict the constant temperature distribution in the sample, only the fusion zone boundary was used as boundary condition [25, 46].𝜕(𝜌𝐶𝑇)
21
Figure 17 Domain for temperature modelling. The different terms marked on the image depict the different regions of interest in the sample. The fusion-zone boundary was marked as fusion zone, the thermocouple attached surface was marked as the
Top surface. “Axis” depicts the symmetric axis to model the bar shape of the sample.
For geometry definition of the model as seen in Figure 17, heat-treated Haynes 282 bar is modelled across its cross-section, with the dimensions being 49.5 mm x 6 mm thickness. The area marked as the "Fusion zone" (next to section a-a) is modelled based on the dimensions obtained from the measurements of the fusion zone using stereo microscopy. For the area marked by “Top surface - 1” (next to fusion zone), the TC were placed for temperature calibration which were taken as inputs for the model. The area marked “Top surface - 2” represents an area of low temperatures as being away from the fusion zone. Haynes 282-disc section which was in contact with the holder as seen in Figure 14 was marked as “side surface” and the base of the sample was marked as “bottom surface” respectively. A steady-state Dirichlet type boundary conditions for simulation were given at each of the marked surfaces:
• Top surface: The temperature readings from thermocouple measurements as against the distance from fusion zone, obtained during temperature calibrations.
• Side surface: Ambient conditions (Uniform room temperature).
• Bottom surface: The temperature readings from thermocouple measurements.
• Axis: 2D axis-symmetric boundary was used to represent the disc shape of the sample.
• The fusion zone boundary was assigned a constant temperature value to model the melt boundary using JMatPro software.
Material property data for density (gm/cm^3), thermal conductivity (W/ (m*K)) and specific heat (J/ (g*K)) for a range of temperatures were taken as material inputs, which were obtained from JMatPro version 6.2.1 [47]. A triangular mesh with fine element sizes was used as the meshing option for the model. For mapping of temperatures, a line segment a-a was defined as seen in Figure 17, values from which were imported, for further processing of results.
z y
Fusion Zone Top surface-1 Top surface-2
22
3.4.2 Precipitation Calculations - Thermo-Calc PRISMA software
The equilibrium phase fraction was calculated using JMatPro version 6.2.1 software using the actual chemical composition of the samples.
Precipitation calculations were carried out using TC PRISMA (2019a) [41] with the initial state of solutionized condition. The thermodynamic and mobility databases used for these calculations are TCNI9 and MOBNI4.1/5 respectively. The input for the simulations included the material composition, the desired temperatures and times and phases for defining the matrix and precipitate phase. Out of several parameters which were being used for the calculations, two parameters (interfacial energy & nucleation sites) were calibrated for the analysis in this work. A test matrix for the intended two parameters was created and the set of values for the parameters was selected based on the values obtained from the literature [4, 33, 48] and also examples in TC PRISMA user guide [41]. These parameters were scaled and optimized through calibrating them by predicting the radii values against experimental values for radii of 𝛾′ precipitates obtained from ref [4] for temperatures of 750˚C, 850˚C, and 950˚C which had an initial condition of mill-annealed samples. The optimized value for interfacial energy is 0.01 mJ m⁄ 2 and nucleation sites is 8.68 x 1031 m-3, the details of which is described in Section 4.5.2 Precipitation calculations – Thermo-Calc PRISMA software.
3.5 Results of first objective
3.5.1 Temperature Modelling
The simulated temperature distribution profile is shown in Figure 18, where the temperatures were calculated based on the temperature input given from the TC measurements and the boundary conditions. An area below the fusion zone (along the z-axis or the symmetry line, marked by a-a) was considered for analyzing the microstructural changes (moving from the fusion zone to the base of the sample).
Figure 18 Temperature profile for 30 minutes sample simulated by COMSOL software z
y
Fusion Zone
23
The values for the simulated temperatures with respect to the temperature profile from Figure 18 across the section a-a, for times of 30 minutes and 4 hours, is shown in Figure 19. The difference between 30 minutes and 4 hours sample was due to the difference in the fusion zone geometry, which caused different temperature distribution in a-a line. The change in the fusion zone geometry needs to be investigated further in the future studies.
Figure 19 Distance vs. Temperature graph for the different heat-treatment times across a-a section. Different gradients for 30 minutes and 4 hours times are due to the different fusion zone geometries.
3.5.2 Initial conditions for the samples
The samples have two different initial conditions, as received and solutionized, which have hardness of 336 HV and 185 HV respectively. The initial microstructure for both conditions is shown in Figure 20. In the as received samples, the carbides are seen along the grain boundary and as stringers. In the solutionized samples, only the MC carbides are seen, while the grain is clean.
Figure 20 (a) Solutionized sample with the all the phases removed except MC carbides. (b) As received sample showing MC carbides, M6C carbides (marked with the blue ellipse), γ' precipitates (the yellow- brown tinge of the image is due to the
etching response from the γ' precipitates) and the grain-boundary carbides (marked with the white ellipse).
200 400 600 800 1000 1200 1400 0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5 5,0 5,5 T em per atu re (˚ C )
Distance from fusion zone (mm)
30 mins 4 hours
Base of the sample
24
3.5.3 Optical Microscopy
The cross section of samples after arc heat treatment is shown in Figure 21 and Figure 22. The stereo micrographs in Figure 21 clearly shows the fusion zone shape and MC carbide bands in different locations. This zone was in the liquid state during arc heat treatment. The optical micrographs in Figure 22 show different etching response for different samples. This shows the formation different microstructure at different locations heat treated at different temperatures.
Figure 21 Stereo microscope images at 10x mag. showing the cross section of sample and the area of dissolution of MC carbides (marked by the red eclipses) in as-received arc-heat treated samples. The sample of 1.5 minutes shows the start of the dissolution of the phases below the fusion zone. The samples of 30 minutes and 4 hours show the area for dissolution of
phases increasing with time.
Figure 22 Evaluation of γ́ band with increasing times for both initial conditions. The as received samples are etched for the
𝛾′ band and the base metal while the solutionized samples are etched for only the 𝛾′ precipitates.
As described, in both conditions, MC carbides, remaining from previous processing such as solidification and forging, were present. After arc heat treatment, it was observed that some area under the fusion zone did not show MC carbides anymore, as shown for the samples with as-received initial condition in Figure 21. The MC carbides dissolution area increased with increasing arc heat treatment time.
25
similar etching response which is an indication of presence of 𝛾′ in both regions, later confirmed by the hardness mapping of the 𝛾′ band and the base metal (336 HV). It was observed that the width of 𝛾′ band increased with the increase in heat treatment time for both initial conditions.
Figure 23 (a) depicts the area showing the change in the microstructure. Different areas between the white area and 𝛾′ band are observed in detail. As may be seen, the micrograph shows where the MC carbides could not be seen, grain boundary carbides formed, and 𝛾′ precipitated. Figure 23 (b) shows the region where on the top no secondary phases are present, but below a line, fine MC carbides are still visible. This boundary moved towards lower temperatures with increasing heat treatment time. Below the area of the dissolution of phases, an area of grain boundary carbide precipitation is seen, as shown in Figure 23 (c). In this region, grain boundary carbides, as well as MC carbides, were observed. Below this region of MC carbides is a band of differently etched 𝛾′precipitate region and the change in the microstructure is as shown in Figure 23 (d).
Figure 23 (a) Image to depict the change in the microstructure in the different regions of the arc heat treated sample showing
26
3.5.4 Scanning Electron Microscopy
The SEM micrograph of the fusion zone boundary is shown in Figure 24. On the top side of the fusion zone a dendritic microstructure and inter-dendritic phases are shown. On the bottom of the micrographs, precipitation free zone is clear, where all phases dissolve as discussed in the previous section.
Figure 24 SEM micrograph showing fusion zone with dendritic structure and MC carbides formed due to segregation in the fusion zone (shown in the insert at higher magnification) along the boundary
The evaluation of 𝛾′ precipitate size was performed using a FEG SEM for the solutionized samples, as shown in Figure 25.The temperature indicated in the figure are based on the distance measured from the fusion zone and compared with the modelled temperature values. The precipitate sizes decrease from the start of gamma prime band towards the lower temperatures, while the volume fraction of the precipitates was observed to be increasing with distance away from the fusion zone. As may be seen, for 1.5 minutes all locations and 30 minutes and 4 hours sample at lower temperature, the 𝛾′ is not clear enough for measurement of the sizes. In all temperatures and times, the 𝛾′shape is spherical, which is
27
Figure 25 Evolution of microstructure for γ́ precipitates within the 𝛾′ band for the solutionized samples.
𝛾′ sizes are presented in Table 8. As may be seen, at around 1000 ˚C, the 𝛾′ size reached to ~24 nm after 30 minutes and to ~45 nm after 4 hours. However, at 870˚C, this size is only ~10 nm after 30 minutes and ~23 nm after 4 hours.
Table 8 Sizes with respect to the different times and temperatures for the solutionized samples.
Times Temperatures 1.5 minutes 30 minutes (Average Std Dev: 1.97) 4 hours (Average Std Dev: 7.14) ̴ 990-1000˚C Not Resolvable ~24 nm ~45 nm ̴ 970-980˚C Not Resolvable ~21 nm ~43 nm ̴ 950-960˚C Not Resolvable ~20 nm ~35 nm ̴ 930-940˚C Not Resolvable ~19 nm ~28 nm ̴ 900-920˚C Not Resolvable ~14 nm ~26 nm
̴ 885-895˚C Not Resolvable ~13 nm Not Resolvable
̴ 860-870˚C Not Resolvable ~10 nm ~23 nm
̴ 840-850˚C Not Resolvable Not Resolvable ~21 nm
28
Figure 26 Grain boundary carbides and precipitates along a grain boundary. (a) High temperature grain boundary showing only grain boundary carbide with discrete morphology. (b) Grain boundary within the start of the γ' band showing a low fraction of γ' precipitates within the grain and a depleted region near the grain boundary. (c) & (d) Grain boundary at a lower temperature show a higher fraction of precipitates in the grain but still reduced fraction near the grain boundary. (e)
& (f) show a similar trend for the grain boundaries at lower temperatures with higher fraction of precipitates in the grains and along the grain boundaries as well.
3.5.5 Microhardness
29
Figure 27 Micro-hardness map depicting the evaluation of hardness for the γ́ band for all the cases. The top hardness maps show arc heat treated as received condition, which have higher initial hardness than the solutionized samples. Arc heat
treatment increased hardness within the 𝛾′ band showed by different etching response in both sets of samples.
3.5.6 Thermodynamic calculations
3.5.6.1 Equilibrium calculations
Figure 28 shows an equilibrium diagram for Haynes 282 calculated from JMatPro version 6.2.1, where the temperature ranges of interest are marked as Region 1 for temperatures between 700-1000˚C to study the 𝛾′ precipitation, Region 2 marks temperatures between 1050-1120˚C to study the precipitation and dissolution of grain-boundary carbides and Region 3 covers the 1200 -1300˚C temperature range for the dissolution of MC carbides. As may be seen, in the equilibrium phase diagram, MC carbides are stable at high temperature
Figure 28 Equilibrium Diagram for Haynes 282 from JMatPro software. The different regions marked on the equilibrium diagram represent the different phases of interest for the study, region 1 for the precipitation of 𝛾′, region 2 for grain
boundary carbides precipitation and region 3 for the precipitation of MC carbides.
30
3.5.6.2 TC PRISMA Calculations
The 𝛾′ precipitation calculations were performed for the solutionized state using TC PRISMA. The temperatures considered for the calculations were based on the values commonly observed for the two times (30 minutes and 4 hours). Particle sizes were calculated for the temperatures of 760˚C, 820˚C, 950 ˚C, and 980˚C, as shown in Figure 29. The temperatures of 760˚C and 820˚C are chosen to compare the simulated value against the values observed by Haas et. al. [49] and the other temperature values (950˚C and 980˚C) are chosen to compare the experimentally obtained values in this work. As can be seen, at higher temperatures the 𝛾′ precipitates are larger.
Figure 29 Simulated precipitate sizes at 760˚C, 820˚C, 950˚C and 980˚C. The values in the table shows the calculated sizes for 30 minutes and 4 hours. All the simulations were carried out taking solutionized initial condition.
3.6 Discussions about the first objective
3.6.1 Microstructural map for arc heat treated sample
31
Figure 30 Comparison between etched sample and temperature plot for arc-heat treatment.
3.6.2 Region of dissolution of phases
In the samples with the solutionized (all phases removed except MC carbides) initial condition, MC carbides were dissolved within the dissolution area, while for the samples with as received initial condition, all the phases (MC carbides, grain boundary carbides and 𝛾′) were dissolved. This area showed similar low hardness values in both as received and solutionized samples. This region was found to be thermodynamically not stable for 𝛾′ and grain boundary carbides as shown in Figure 28 (regions 1 &2). MC carbides, however, were expected to be stable at high temperature in the equilibrium diagram (as shown in Figure 28 region 3), but the results showed that they are not stable. Further study at high temperatures (> 1300˚C) is needed to verify the observations on the dissolutions of MC carbides and possible improvement is thermodynamic databases must be applied. The increase in the area of dissolution of phases with time could be the result of chemical homogenization [51-53]. The dissolution occurs faster at higher temperatures, as the diffusion is faster. At lower temperature, it takes much longer time for dissolution of phases.
3.6.3
𝜸′ precipitation zone
The radii of γ′after 30 mins and 4 hrs measured in this study were compared with the predicted values
32
Figure 31 Size comparison for γ' precipitates between the experimentally obtained values and the simulated values. The values for 820˚C and 760˚C are compared between the simulated (Thermo-Calc) and experimental values (S Haas et. al. and
Joseph [4].) from literature. The values for 980˚C and 950˚C are compared between the simulated (Thermo-Calc) and experimental values (from this work).
From Figure 31, it was observed that the 𝛾′ sizes observed in SEM and the TC PRISMA predicted values were in close agreement with each other. For 980˚C, the experimental and calculated values are ~21 nm and ~23 nm for 30 minutes and ~43 nm and ~46 nm after 4 hours. For 950˚C, for 980˚C, the experimental and calculated values are ~20 nm and ~19 nm after 30 minutes and ~35 nm and ~38 nm after 4 hours.
Haas et. al. [49] solutionized a Haynes 282 sheet sample at 1120˚C - 30 minutes and aged at the temperatures of 820˚C and 760˚C. The radii values were of the order of ̴ 12 nm for 820˚C and ̴ 6 nm 760˚C. In the present study, the radii for 867˚C is about 10 nm, which is lower than Hass et al.’s work, at a temperature of 820 ˚C.
The Thermo-Calc simulations 760˚C for 30 minutes yielded ~3 nm, which were under-predicted compared to Haas et. al. [49]. The experimental values which were reported on a mill-annealed sheet material for the temperature of 950˚C by Joseph et al [4] were also larger than the simulated values. Therefore, observation showed that thermodynamic calculation fits very well with the result in this study for the solutionized initial condition, but they have underestimation compared to other results. The possible reason for these observations is as follow:
• The material form: the thermomechanical history before the experiment can significantly influence the results. For instance, in mill annealed condition, it is expected to have some 𝛾′, which could result in larger values.
33
resulting in larger 𝛾′ sizes in those studies. Arc heat treatment, in contrast, provided rapid heating and cooling and represented the results closer to the selected temperatures.
3.6.4 Time – Temperature Diagrams
Based on all results obtained from the previous sections, Temperature-Hardness (TTH) and Time-Temperature-Precipitate (TTP) plots are generated.
3.6.4.1 Time – Temperature – Hardness (TTH) diagram
The TTH diagram, shown in the Figure 32, was determined by taking 250HV, 300 HV and 350 HV as the reference values, which were identified on the samples from the microhardness maps. The distance from the fusion zone on the hardness maps are compared with the COMSOL temperature predictions. Based on the COMSOL predictions, the temperature ranges are identified for the various reference hardness values.
Figure 32 Plot for TTH for Haynes 282 for the hardness values of 250 HV (Blue), 300 HV (Green), 350 HV (Orange). The plot shown in Figure 32, for the case of 1.5 minutes, the curve for 250 HV covers a larger range of temperatures as compared to the other two curves. This could be due to the slight variation of hardness in a-a section. As 250 HV hardness is value closer to base metal hardness, it is expected to more variation. As may be seen in Figure 27, some location in the base metal of solutionized sample showed these values. The area covered by the curve for 250 HV is larger than the other reference values which also can be seen in the work done by Fahrmann et. al. [28]. In addition, it should be noted that the obtained temperature ranges are based on modelled values for temperature gradients from COMSOL simulations. For the case of 4 hours, the mapping for range of temperatures was slightly uncertain due to the influence of the shape of the fusion zone (shown in Figure 19) on the temperature prediction.
For the higher reference values of 300 HV and 350 HV, with increasing the hardness value, γ′ led to a
34
3.6.4.2 Time – Temperature – Precipitation (TTP) diagram
TTP diagram, Figure 33, shows the temperatures at which the 𝛾′ precipitates were observed. The temperatures were correlated with distance of each characterized spot from the model. For the case of 1.5 minutes, the 𝛾′ precipitates were observed but the size could not be measured due to resolution limit of the SEM itself. As may be seen, compared to TTH diagram, it shows different ranges. The challenge in this case is that the smaller 𝛾′ precipitates were not seen under SEM, but they increased hardness values significantly. Therefore, hardness mapping can represent the transformation better than SEM analysis in the present study.
Figure 33 Plot for TTP for Haynes 282
Figure 34 depicts the measured sizes for the particles which were resolvable at each point evaluated across the axis a-a. As can be seen, however the presence of 𝛾′ precipitates were clear in some location/temperature, but it was impossible to measure the size.
35
3.7 Conclusions for the first objective
The evolution of microstructure and hardness of Haynes 282 was investigated using a novel arc heat treatment technique. Two different initial conditions and arc heat treatment times of 1.5 minutes, 30 minutes and 4 hours was examined. The experiments were complemented with temperature modelling, thermodynamic calculations, and precipitation simulations.
1. The technique consisted of applying a stationary TIG arc on stationary water-cooled discs, providing a wide range of temperatures (liquidus to ambient temperature). The steady state temperature gradient was quickly achieved after igniting the arc. For each condition, only three samples were produced, which significantly decreased the research time compared to the studies using furnace heat treatment.
2. A graded microstructure covering wide range carbide and 𝛾′ sizes and morphologies was produced. The arc heat treated samples showed different regions for microstructural development based on the temperatures in the samples: fusion zone, dissolution of secondary phases area, grain boundary carbide area, and gamma prime precipitation area.
3. The MC carbides were dissolved at high temperatures irrespective of the initial microstructure condition. Minimum hardness was obtained in this temperature region.
4. The 𝛾′ precipitation was observed to start in the temperature range of 990-1000˚C. The 𝛾′ particle sizes for the temperature ranges increased with increasing heat treatment times. This was also reflected by increasing the microhardness for all the samples in this region.
5. The TC PRISMA simulated values for the temperatures of 980˚C and 950˚C were found to be in good agreement with the experimentally obtained values obtained in this study.
