The role of Cr and Mo alloying element additions on... kinetics and effects of Upper Bainite formation in quench

The role of Cr and Mo alloying element additions on... kinetics and effects of Upper Bainite formation in quench
The role of Cr and Mo alloying element additions on the
kinetics and effects of Upper Bainite formation in quench
and tempered plate steels
By
Lindsay Josephine Leach
Supervised by
Prof W.E. Stumpf and Dr C.W. Siyasiya
Submitted in partial fulfilment of the requirements for the degree of
Master of Science (Applied Science) Metallurgical Engineering
In the
Department of Materials Science and Metallurgical Engineering,
Faculty of Engineering, Built Environment and Information Technology,
University of Pretoria
South Africa
February 2013
© University of Pretoria
ACKNOWLEDGEMENTS
I would firstly like to thank God for the being an unfailing source of strength.
I extend my sincere gratitude to my supervisors Professor Stumpf and Dr Siyasiya for their
insightful guidance and encouragement.
I thank my parents Joseph and Angeline Leach for the continuous love and support they give
me.
The help of Carel Coetzee, Dr Kevin Banks and the staff of the Industrial Minerals and
Materials Research institute in training of the SEM, Gleeble and B ̈ hr dilatometer is greatly
appreciated.
I wish to thank Louise Ackerman, Sarah Havenga and Elsie Snyman-Ferreira for their
cheerful support in administrative matters.
The provision of financial support and provision of materials by ArcelorMittal South Africa
is acknowledged.
I wish to thank Professor Robert Knutsen of The Centre for Materials Engineering at the
University of Cape Town for assistance with their Instrumented Charpy testing unit.
I wish to thank Dr N. van der Berg of the Department of Physics, University of Pretoria for
his discussions and help with shadowing of carbon extraction replica samples and Chris van
der Merwe of the Laboratory for Microscopy and Microanalysis for his assistance with TEM
microscopy.
Finally I extend my gratitude to my good friends Asimenya Kapito and Given Maruma whose
companionship made the duration of this study a pleasant and memorable one.
i
© University of Pretoria
THE ROLE OF Cr AND Mo ALLOYING ELEMENT ADDITIONS ON
THE KINETICS AND EFFECTS OF UPPER BAINITE FORMATION IN
QUENCH AND TEMPERED PLATE STEELS
Supervisor: Professor WE Stumpf
Co-Supervisor: Doctor CW Siyasiya
Department: Materials Science and Metallurgical Engineering
Degree: Master of Science (Metallurgy)
ABSTRACT
The aim of the work presented was to investigate the effects of upper bainite on impact
toughness in quench and tempered low alloy plate steels. The experimental research included
construction of CCT diagrams by dilatometry, verification of phases by optical microscopy
(OM), Vickers hardness, scanning electron microscopy (SEM), transmission electron
microscopy (TEM) on precipitates extracted by carbon replica and by electrolytic means and
finally impact testing of Charpy specimens with mixed bainite:martensite microstructures.
Bainite was formed in High Chromium Low Molybdenum (HCrLMo) and in High
Molybdenum Low Chromium (HMoLCr) steel samples by isothermal annealing within the
bainite C-curve of the respective CCT diagrams. The isothermal kinetics of the upper bainite
transformation was modelled with the Johnson Mehl Avrami Kolmogorov (JMAK) model.
Avrami exponents of 1.4 and 1.3 were obtained for the HCrLMo and HMoLCr steels
respectively which indicated linear growth with a considerable lengthening rate of laths and
negligible thickening.
The measurably slower growth kinetics in the HMoLCr steel as observed in the JMAK model
and the higher hardenability with reference to its CCT diagram, suggested a strong Mo
alloying element effect. The stronger effect of Mo compared to Cr was attributed to a solute
drag like effect.
The effect of upper bainite in a tempered martensitic matrix was investigated for the
following amounts of bainite; 0%, 10%, 25%, 60%, 75%, 90% and 100%. The impact
toughness of the mixed bainite:martensite samples was evaluated against the toughness of
100% bainite and 100% martensite. It was demonstrated that upper bainite reduces the total
ii
© University of Pretoria
absorbed impact energy by an adverse effect on crack nucleation energy and crack
propagation energy.
Keywords: bainite, bainite:martensite structures, toughness, instrumented impact testing,
dilatometry, isothermal transformation.
Publications and Conference Presentations
1. L.J. Leach, C.W. Siyasiya, W.E. Stumpf: Effect of dual phase microstructure on the
toughness of a Cr-Mo low alloy plate steel – Presented at the Ferrous and Base Metals
Development Network Conference, Magaliesburg, South Africa, 15-17 October 2012
– Published conference proceedings: Symposium series S73, Paper F09.
2. L.J. Leach, C.W. Siyasiya, W.E. Stumpf: Effect of dual phase microstructure on the
toughness of a Cr-Mo low alloy plate steel – Accepted for publication in the Journal
of the South African Institute of Mining and Metallurgy
3. L.J. Leach, C.W. Siyasiya, W.E. Stumpf: Effect of the alloying elements Cr and Mo
on isothermal transformation kinetics in two plate steels – In preparation.
iii
© University of Pretoria
Table of Contents
Acknowledgements……………………………………………………………………………i
Abstract………………………………………...……………………………………………..ii
Table of Contents……………………………………..……………………………………..iv
Table of Figures………………………..….………………………………………………..viii
List of Tables……………………………...…………………………………………………xv
Table of Abbreviations……………………...……………………………..………………xvi
_Toc348086409
1 Chapter 1…………………………………………………………………………………...1
1.1 Introduction…………………………………………………………………………...1
1.2 Problem statement…..………………………………………………………………...2
2 Chapter 2: Literature Survey……………………………………………………………..3
2.1 Alloying elements and the austenite to bainite transformation……..………………...3
2.2 Kinetics……...………………………………………………………………………...9
2.2.1
Nucleation and growth transformations………………………………………..9
2.2.2
The Johnson Mel Avrami equation…………………………………………...10
2.3 Bainite formation models…...……………………………………………………….15
2.3.1
General features………………………………………………………………15
2.3.1.1 Lower bainite………………………………………………………………...17.
2.3.1.2 Upper bainite………………………………………………………………...19.
2.3.2
The Displacive model…………………………………………………………21
2.3.2.1 Surface relief………………………………………………………………...21.
iv
© University of Pretoria
2.3.3
The Diffusional model………………………………………………………..24
2.3.3.1 The Kinetic definition……………………………………………………….24
2.3.4
The Stasis……………………………………………………………………..25
2.3.4.1 The influence of To………………………………………………………….25
2.3.4.2 The Solute Drag-Like Effect………………………………………………...27
2.4 Mechanical properties……………………………………………………… ……….31
2.4.1
Ductile and Brittle fracture……………………………………………………31
2.4.2
Effect of precipitates on fracture……………………………………………...32
2.4.3
Effect of a bainitic microstructure on fracture………………………………..33
2.4.4
Hardness of bainitic microstructures………………………………………….36
2.4.5
Effect of bainite on toughness………………………………………………...37
3 Chapter 3: Experimental procedures…………………………………………………...41
3.1 Metallography……………………………………………………...……...…….…..41
3.1.1
Secondary Electron Microscopy……………………………………………...41
3.1.2
Transmission Electron Microscopy…………………………………………..42
3.1.3
Carbon extraction replica sample preparation………………………………..42
3.1.4
Electrolytic extraction………………………………………………………..42
3.1.5
Etchants………………………………………………………………………44
3.2 Thermal analysis…………………………………………………………………….44
3.2.1
Dilatometry…………………………………………………………………...44
3.2.2
Gleeble………………………………………………………………………..48
3.3 Mechanical testing….……………………………………………………………….48
3.3.1
Hardness……………………………………………………………………...48
v
© University of Pretoria
3.3.2
Instrumented Charpy impact………………………………………………….48
4 Chapter 4: Results………………………………………………………………………..51
4.1 Microstructures of as the received material...……………………………………….51
4.1.1
The microstructure of the HCrLMo steel……………………………………..52
4.1.2
The microstructure of the C-Mn steel………………………………………...53
4.2 Hardness Profiles…………………………………………………................……….56
4.2.1
Hardness profiles of the HCrLMo steel………………………………………57
4.2.2
Hardness profiles of the C-Mn steel………………………………………….59
4.3 Tempering characteristics…………..………………….……………………………60
4.4 Continuous Cooling Transformation Diagrams…………………………………......62
4.4.1
Preliminary tests………………………………………………………………62
4.4.1.1 Determination of austenitising temperatures………………………...………61
4.4.1.2 Carbide and nitride dissolution………………………………………….......65
4.4.1.3 Tests for Ac1 and Ac3 temperatures…………………………………………..67
4.4.2
Partial CCT diagrams…………………………………………………………73
4.5 Isothermal transformations……………………………………….………………….76
4.5.1
Preliminary tests………………………………………………………………76
4.5.2
Measurement of the volume fraction of bainite………………………………78
4.6 Johnson Mehl Avrami Kolmogorov kinetics………………………….………….....80
4.6.1
Sigmoidal curves and Avrami exponents……………………………………..80
4.6.2
Depression of Ms temperatures……………………………………………….81
4.6.3
Electrolytic extraction of precipitates…………………………………………83
4.7 Instrumented Charpy impact tests………………………………………………….. .90
vi
© University of Pretoria
4.7.1
Test parameters………………………………………………………………..90
4.7.1.1 Effect of tempering on hardness……………………………………………..89
4.7.1.2 Effect of notch position relative to the rolling direction………………...…..92
4.7.1.3 Distribution of bainite in martensite…………………………………………93
4.7.1.4 Temperature gradients……………………………………………………….95
4.7.2
Effect of the amount of bainite on the impact energy…………………..…….97
4.7.3
Crack initiation and propagation energies…………………………………….99
4.7.3.1 The HMoLCr steel…………………………………………………………100
4.7.3.2 The HCrLMo steel…………………………………………………………103
4.7.4
Fractography of the Charpy fracture surfaces………………………………105
4.7.5
Shear fracture measurements………………………………………………..107
5 Chapter 5: Discussion…………………………………………………………………..111
5.1 Tempering characteristics of the C-Mn and Cr-Mo steels………………................111
5.2 Effect of the steel compositions on the partial CCT diagrams……..…………....…113
5.3 JMAK kinetics of the isothermal transformations….……………………………...115
5.4 Effect of alloying elements on growth rates..……….……………………………..117
5.5 Effect of isothermal holding on Ms temperatures…..……………………………...121
5.6 Impact toughness…...……………………………………………………………...124
5.6.1
Effect of bainite on the total absorbed energy………………………………124
5.6.2
Crack initiation and propagation energies…………………………………..125
5.6.2.1 Comparison of crack initiation energies………………………………...…124
5.6.2.2 Comparison of crack propagation energies……………………………...…124
5.6.2.3 Relation of microstructure to energies………………………………..……126
vii
© University of Pretoria
6 Chapter 6: Conclusions…………………………………………………………………130
7 References………………………………………………………………………………..131
viii
© University of Pretoria
Table of Figures
Figure 2.1.1 Effect of Mo additions to a Low Alloy steel with (a) 0.03wt% Nb (b) 0.03wt%
Nb-0.3wt% Mo and (c) 0.03wt% Nb-0.6wt% Mo [3]. .............................................................. 5
Figure 2.1.2 The effects of (a) Mn and (b) Cr on the Bs temperature as determined by
Artificial Neural Network (ANN) modelling [14]. .................................................................... 8
Figure 2.2.1 Temperature dependence of the Avrami exponent n (a) in Fe-Ni-Si-C, Fe-Mn-SiC and 300M steels [19] and (b) in a 0.66%C steel [16]. .......................................................... 12
Figure 2.2.2 Correlation of Avrami exponents with temperature and transformation products
[16]. .......................................................................................................................................... 14
Figure 2.2.3 Average Avrami exponent n of 2 is shown for both upper and lower bainite in a
0.31C-1.22Mn-0.25Si-0.14Cr-0.10Ni steel (in wt%) [18]....................................................... 14
Figure 2.3.1 The mechanism of formation of upper and lower bainite [23]............................ 16
Figure 2.3.2 SEM of a Medium Carbon steel isothermally transformed for 900s at 450⁰C [18]
UB is upper bainite and LB is lower bainite.. .......................................................................... 17
Figure 2.3.3 The transition temperature between upper and lower bainite as a function of
carbon content [25]. ................................................................................................................. 17
Figure 2.3.4 The decrease of ferrite plate thickness with decreasing isothermal treatment
temperature of a medium carbon steel [18]. ............................................................................ 18
Figure 2.3.5 SEM micrograph of lower bainite in a 2358 steel austempered at 260 for 100
minutes [26]. ............................................................................................................................ 19
Figure 2.3.6 TEM micrograph of lower bainite showing the parallel intralath carbides within
ferrite laths [27]........................................................................................................................ 19
Figure 2.3.7 Formation of upper bainite by repeated nucleation of sheaves [28] ................... 20
Figure 2.3.8 SEM micrograph of upper bainite in a 0.31C-1.22Mn-0.14Cr-0.25Si steel
isothermally transformed at 525 for 900s [18]. .................................................................... 20
Figure 2.3.9 Diagram of austenite and ferrite free energies and To curve on the phase
diagram. ................................................................................................................................... 22
ix
© University of Pretoria
Figure 2.3.10 (a) Surface relief of bainite for a 0.33C-0.74Mn-3.47Ni (wt%) steel treated at
574 for 22hr. Magnification 300x [12]. (b) An atomic force microscope scan across a
bainite sub-unit with surface relief [28]. .................................................................................. 23
Figure 2.3.11 The lever rule as applied to the To line to find the maximum volume fraction of
bainite that can form at a specific temperature under diffusionless conditions. ...................... 26
Figure 2.3.12 Phase diagram showing calculated Ae3, To and To’ curves and experimental
carbon concentration of residual austenite after isothermal bainite formation in a 15H2VT
steel [28]................................................................................................................................... 27
Figure 2.3.13 Phase diagram showing calculated Ae3, To and To’ curves and experimental
carbon concentration of residual austenite after isothermal bainite formation in an Fe-Cr-Si-C
steel [39]................................................................................................................................... 27
Figure 2.3.14 Isothermal section of a Fe-C-Mo phase diagram where solute drag results in
lower volume fractions of ferrite with the lever rule applied to the para-equilibrium (dashed)
boundary.[36]. .......................................................................................................................... 29
Figure 2.3.15 Schematic illustrations of reaction kinetic behaviour below the Bs and the
sequence of transformation [47]. SN = sympathetic nucleation and SDLE = solute drag-like
effect. ....................................................................................................................................... 31
Figure 2.4.1 Schematic illustration of the difference in upper shelf energy of lower and upper
bainite....................................................................................................................................... 32
Figure 2.4.2 Microcrack formation by dislocation pile-up. ..................................................... 33
Figure 2.4.3 SEM micrograph of crack deflection at (a) high angle boundaries and (b) low
angle boundaries in lower bainitic microstructure of a 4150 steel transformed at 375 [53].
.................................................................................................................................................. 34
Figure 2.4.4 SEM micrograph of cleavage cracks in upper bainite in a steel transformed at
450 [53]. ............................................................................................................................... 34
Figure 2.4.5 Cleavage crack deflection in a fragmented austenite grain at (a) bainite packet
boundaries and (b) at bainite + martensite packet boundaries. ................................................ 35
Figure 2.4.6 Diagram showing fragmentation of a prior austenite grain due to formation of
bainite sheaves (2) within austenite (1) where growth starts at the austenite grain boundary
(3). ............................................................................................................................................ 36
Figure 2.4.7 Hardness decrease with isothermal transformation temperature increase [18]. .. 37
x
© University of Pretoria
Figure 2.4.8 Fracture surfaces of a 26Kh1MFA steel consisting of (a) martensite and lower
bainite and (b) martensite and upper bainite [1]. ..................................................................... 38
Figure 2.4.9 Cleavage fracture of an AISI 4340 steel isothermally transformed to upper
bainite at 430 [57]. ............................................................................................................... 39
Figure 2.4.10 SEM micrograph of Carbide cracking and debonding in upper bainitic
microstructure of a AISI 4150 steel isothermally transformed at 450 for 24hrs [53]. ......... 39
Figure 3.1.1 Circuit used for electrolytic extraction. ............................................................... 43
Figure 3.1.2 Tinted ethanol in the beaker after centrifuging and centrifuged sample with
extracted precipitate collected at the bottom. .......................................................................... 44
Figure 3.2.1 The chamber of a Bӓhr DIL 805 dilatometer. The schematic diagram shows the
coils surrounding a test specimen. ........................................................................................... 45
Figure 3.2.2 Diagram showing contraction during transformation of austenite to ferrite. ...... 46
Figure 4.1.1 Optical images of the HCrLMo steel: 0.17C-1.076Mn-0.73Cr-0.23Mo-0.002B
steel taken through the mid-thickness at (a) near the top surface (b) in the centre and (c) near
the bottom surface. ................................................................................................................... 52
Figure 4.1.2 (a) SEM micrograph of the HCrLMo steel and (b) TEM micrograph of a carbon
extraction replica taken from the HCrLMo steel. .................................................................... 53
Figure 4.1.3 Optical micrographs of the C-Mn steel taken through the mid-thickness (a) near
the top surface (b) in the centre and (c) near the bottom surface of the steel plate. ................ 54
Figure 4.1.4 SEM micrographs of the C-Mn steel taken (a) in the centre and (b) near the
surface of the plate. .................................................................................................................. 55
Figure 4.1.5 TEM micrographs of carbon extraction replicas taken from the C-Mn steel (a)
shadowed with Au-Pd and (b) unshadowed replica. Note the lath with carbides of size less
than 1μm. ................................................................................................................................. 56
Figure 4.2.1 Schematic illustration of lines along which hardness measurements were taken
on the C-Mn and HCrLMo steel plates. RD is the rolling direction. T1 is a path along the
centre of the width of the plate and T2 is a path along the centre of the thickness of the plate.
.................................................................................................................................................. 56
xi
© University of Pretoria
Figure 4.2.2 (a) Hardness profiles as measured along the centre thickness through the sample
width and (b) measured along the centre width of the HCrLMo steel plate. Macro Vickers
load: 10kg................................................................................................................................. 57
Figure 4.2.3 (a) Hardness profiles measured along the centre thickness and (b) along the
centre width of the C-Mn steel plate. Macro Vickers load: 10kg ............................................ 59
Figure 4.3.1 Tempering curves for (a) the C-Mn steel and (b) the HCrLMo steel. The samples
were tempered for 30mins at temperature after soaking at 900 and water quenching. Macro
Vickers hardness load: 10 kg with 5 readings per data point. ................................................. 61
Figure 4.4.1 Dilatometric signal obtained for the C-Mn steel sample which was heated at
0.5 /s to 1100 before slow cooling. The critical temperatures found were Ac1 = 722.3
and Ac3 = 865.3 . ................................................................................................................... 63
Figure 4.4.2 Dilatometric signal for the HCrLMo steel sample heated at 1.5 /s to 1200
before slow cooling. Homogeneous austenite was obtained at 931.9ᵒC, where the expansion
became linear. .......................................................................................................................... 63
Figure 4.4.3 Superimposed dilatometric signals of austenitisation of HCrLMo steel samples
at 955 and 935 and cooled at 10 /s. ............................................................................... 65
Figure 4.4.4 Superimposed dilatometric signals of HCrLMo steel samples austenitised at
900 and 935 respectively and cooled at 18 /s................................................................. 65
Figure 4.4.5 Thermocalc graphs of temperature versus AlN content for (a) the HCrLMo steel
and (b) C-Mn steel. NMP = mole fraction. .............................................................................. 67
Figure 4.4.6 Microstructure of a HCrLMo steel sample austenitised at 955 and soaked for
20 minutes before cooling at 60 /s. The spots arrowed were found to be etch pits and not
second phases. .......................................................................................................................... 68
Figure 4.4.7 (a) Dilatometric signal of the HCrLMo steel using a 0.22 /s cooling rate after
austenitising at 900 . The Ar3 and Ar1 temperatures recorded were 621 and 452
respectively.(b) Optical micrograph showing the microstructure obtained. ............................ 69
Figure 4.4.8 (a) The graph and (b) microstructure of the decomposition product in the steel
HCrLMo after slow cooling at 0.07 /s. The product is a mixture of ferrite (F), pearlite (P)
and bainite (B).......................................................................................................................... 70
Figure 4.4.9 Graph of a sample of the steel HCrLMo with a 0.074ᵒC/s cooling rate showing
two transformations at Ac3 = 813.5 , Ac1 = 656.9 , Bs = 561.8 and Bf = 398.5 . .......... 71
xii
© University of Pretoria
Figure 4.4.10 Microstructure of sample of the steel HCrLMo slow heated at 0.246 /s and
cooled at a cooling rate of 0.074 /s........................................................................................ 72
Figure 4.4.11 Graph of a sample of the steel HCrLMo with a cooling rate of 0.028 /s and
where the critical transformation temperatures were found at Ar3 = 829.7 and Ar1 =
682.4 . .................................................................................................................................... 73
Figure 4.4.12 Partial CCT diagrams for (a) the HCrLMo steel (b) the HMoLCr steel and (c)
the C-Mn steel. F represents ferrite, P pearlite and B bainite. ................................................. 75
Figure 4.5.1 Double quench on steel HCrLMo, cooling rate interrupted at 443 , Bs = 520 .
.................................................................................................................................................. 77
Figure 4.5.2 Optical micrographs of the mixed upper bainite-martensite microstructures in
steel HCrLMo obtained after continuous cooling at 5 /s and 60 /s after (a) a Nital etch and
(b) a 10% SMB etch. B is bainite and M is martensite. ........................................................... 78
Figure 4.5.3 Dilatometric signal of isothermal transformation on a sample of the steel
HCrLMo austenitised at 900 then cooled at 8
to 484
and held for 20 s to form
67.8% bainite before cooling at 30
to form martensite. The strains measured at e1, e2 and
ex were used to calculate the bainite volume fractions. ........................................................... 79
Figure 4.6.1 Volume fractions of bainite plotted as a function of time at isothermal
temperatures for the two steels HCrLMo and HMoLCr. The time taken for 50% to form is
shown as t50. ............................................................................................................................. 81
Figure 4.6.2 Graph of Ms temperatures with isothermal holding time for both steels. ............ 82
Figure 4.6.3 Graph of the effect of increasing volume fractions of bainite on the measured Ms
temperatures. ............................................................................................................................ 83
Figure 4.6.4 TEM images taken from carbon extraction replicas of HCrLMo steel with a
100% bainitic structure. ........................................................................................................... 84
Figure 4.6.5 EDS spectra of (a) AlN and (b) TiN precipitates. ............................................... 85
Figure 4.6.6 (a) TEM image of TiN particle. EDS maps of theTiN particle showing the
distributions of (b) Ti, (c) N and (d) C. The spectrum of the particle is shown in (e). ............ 87
Figure 4.6.7 (a) TEM image of a Mo rich particle and EDS maps showing the distributions of
(b) Mo (c) Fe (d) C (e) Cr and (f) is the spectrum of the particle. ........................................... 88
xiii
© University of Pretoria
Figure 4.6.8 (a) TEM image of a Cr rich particle and EDS maps showing the distributions of
(b) C (c) Cr (d) Mo and (e) Fe. The spectrum of the particle is shown in (f). ......................... 90
Figure 4.7.1 Sigmoidal plots and extrapolations used for heat treatment durations (a) in the
HMoLCr and (b) HCrLMo steel Charpy specimens................................................................ 91
Figure 4.7.2 Change in hardness with tempering parameter (x10 3) for bainitic steel. T is the
absolute temperature and t is time in hours. The dotted line represents the tempering
parameter used in tempering the martensite in both of the steels HCrLMo and HMoLCr at
500 for 30 minutes to lower its hardness to that of the upper bainite. ................................. 92
Figure 4.7.3 Hardness profiles in the HMoLCr steel across a Charpy sample taken (a) parallel
to the Charpy V-notch and (b) transverse to the notch. ........................................................... 94
Figure 4.7.4 Hardness profiles measured (a) parallel to the Charpy V-notch and (b) transverse
to the V-notch across a HMoLCr steel Charpy specimen heat treated to form 50% upper
bainite....................................................................................................................................... 96
Figure 4.7.5 (a) Time-Temperature profile on Gleeble of the heat treatment to produce bainite
(b) a magnified view of the quenching process. PTemp is the programme temperature and
TC3 the control temperature. ................................................................................................... 97
Figure 4.7.6 Graphs of the total energy absorbed as a function of the amount of bainite in the
Charpy samples of (a) steel HMoLCr and (b) steel HCrLMo. ................................................ 98
Figure 4.7.7 Graphical output from instrumented impact tests showing different regions of
fracture on load and energy curves plotted as a function of time.Taken from a HMoLCr steel
sample with 50% bainite. ......................................................................................................... 99
Figure 4.7.8 Characterisation of Charpy fracture surface in a HMoLCr steel sample consisting
of 75% bainite. I is the fracture initiation region, II is the britte propagation region, III is a
shear lip and IV is the final fracture....................................................................................... 100
Figure 4.7.9 Plots of (a) crack initiation and (b) crack propagation energies in the HMoLCr
steel Charpy samples. The respective energies were measured as a fraction of the total
energy. .................................................................................................................................... 102
Figure 4.7.10 Optical images of fracture surfaces of HMoLCr samples containing (a) 100%
M, (b) 10% B, (c) 25% B, (d) 90% B and (e) 100% B. M = martensite and B = bainite Note
the reduced ductile appearance from 10 to 100% bainite. ..................................................... 103
Figure 4.7.11 Plots of the fractions of (a) the crack initiation energy and (b) the crack
propagation energy of the HCrLMo steel. ............................................................................. 105
xiv
© University of Pretoria
Figure 4.7.12 SEM micrographs of the fracture initiation region in (a) a HCrLMo steel
Charpy sample with 75% bainite and (b) in a HMoLCr steel Charpy sample with 10% bainite.
Note the particles within ductile dimples, most likely MnS inclusions. ................................ 106
Figure 4.7.13 SEM micrograph taken in a HMoLCr steel Charpy sample containing10%
bainite at (a) the transition region between crack initiation and brittle fracture region (b) a
region of brittle fracture with numerous cleavage facets. ...................................................... 106
Figure 4.7.14. SEM micrograph of tear ridges within the brittle fracture region of a HCrLMo
steel Charpy sample with 75% bainite. Note the secondary cracks (arrowed) that developed
amongst the cleavage facets. .................................................................................................. 107
Figure 4.7.15 Lateral expansions measured on (a) the HMoLCr steel Charpy samples and (b)
on the HCrLMo Charpy samples. .......................................................................................... 108
Figure 4.7.16 Shear fracture measured on fractured Charpy samples of (a) the HMoLCr and
(b) HCrLMo steels as a function of the % bainite. ................................................................ 110
Figure 5.2.1 Effect of boron on transformation [27] . ........................................................... 113
Figure 5.3.1 Plots of lnt versus lnln(1/1-Vf) for (a) the HMoLCr steel and (b) the HCrLMo
steel isothermally transformed at 498 and 484 respectively. .......................................... 116
Figure 5.5.1 Graph of the Ms depression with carbon content as modelled according to the
equation by Krauss [38]. ........................................................................................................ 122
Figure 5.5.2 T-zero temperature plotted against carbon content (weight fraction) for (a) steel
HMoLCr and (b) steel HCrLMo showing the T-zero temperatures To and the maximum
carbon content wTo at the isothermal treatment temperatures, as estimated by Thermocalc.
................................................................................................................................................ 123
Figure 5.6.1 Plots of absolute crack initiation energies of steels HCrLMo and HMoLCr as a
function of the amount of bainite in the Charpy specimens .................................................. 125
Figure 5.6.2 Plots of crack propagation energies of steels HCrLMo and HMoLCr as a
function of the amount of bainite in the Charpy specimens. ................................................. 126
Figure 5.6.3 SEM micrograph of bainite in the HCrLMo steel with 90% bainite ................. 127
Figure 5.6.4 SEM micrograph of bainite in the HMoLCr steel. Sample contains 56% bainite.
................................................................................................................................................ 127
xv
© University of Pretoria
List of Tables
Table 2.1 Avrami exponents in relation to nucleation and growth .......................................... 11
Table 2.2 n values obtained for upper and lower bainite in different alloy steels [21]. .......... 14
Table 2.3 Comparison of the transformation characteristics of bainite αb with martensite α‟
and Widmanst ̈ tten ferrite αw [28]. ........................................................................................ 21
Table 4.1 Chemical compositions of the steel plates. .............................................................. 51
Table 4.2 Experimental and calculated Ac3 temperatures. ....................................................... 64
Table 4.3 Transformation temperatures obtained after austenitising HCrLMo steel samples at
900 and 930 and cooling at different rates........................................................................ 66
Table 4.4 HMoLCr steel micro-Vickers hardness with cooling rate. Load = 300 gf. ............. 76
Table 4.5 HCrLMo steel micro-Vickers hardness with cooling rate. Load = 100 gf. ............. 76
List of Abbreviations
AQ
As quenched
B
Bainite
xvi
© University of Pretoria
BCC
Body Centred Cubic
BE
Back Scattered Electrons
CCT
Continuous Cooling Transformation
EDS
Energy Dispersive Spectroscopy
F
Ferrite
FCC
Face Centred Cubic
HCrLMo
High Chromium Low Molybdenum
HMoLCr
High Molybdenum Low Chromium
HSLA
High Strength Low Alloy
JMAK
Johnson Mehl Avrami Kolmogorov
M
Martensite
NMP
Mole fraction
NPLE
No Partition Local Equilibrium
OM
Optical Microscope
P
Pearlite
Q&P
Quench and Partitioned
RD
Rolling direction
SDLE
Solute Drag Like Effect
SE
Secondary Electrons
SEM
Scanning Electron Microscope
SN
Sympathetic Nucleation
TEM
Transmission Electron Microscope
TTT
Time Temperature Transformation
xvii
© University of Pretoria
UTS
Ultimate Tensile Strength
YS
Yield Strength
xviii
© University of Pretoria
1
1.1
Chapter 1
Introduction
Structural steels are produced on an industrial scale as plate steels which are hot rolled in a
mill and then quenched and tempered to achieve their final strength. Due to their use, it is
imperative that structural steels are manufactured to standards conforming to the desired
qualities.
Firstly, good reliability is necessary as the steel undergoes fabrication processes involving
bending, shearing, grinding, welding, drilling and punching without any detriment to its
properties. Such properties include sufficient ductility, strength and notch toughness.
Weldability is improved by limiting the addition of carbon to relatively low values, hence
low carbon steels serve this purpose well, although comparable properties are obtained with
High Strength Low Alloy steels (at a higher cost) and higher carbon mild steels. Alloy
additions to the steels improve hardenability. In so doing, the quench severity may be reduced
thus averting quench cracks. This is achieved by the addition of amounts of Manganese
greater than 1% and smaller amounts of the strong carbide formers Chromium and
Molybdenum as well as Boron, the latter to enhance the hardenability even more. Residual
Sulphur reacts with the Mn and during hot rolling, and MnS stringers may develop. These
compounds often result in non-isotropy of mechanical properties and so-called “shape
control” additives such as Ca and Ce are often used to ensure rounded non-metallic
inclusions.
Despite rigorous efforts to ensure good quality steel products, it may so happen that in one
aspect or another, the product falls short of a certain requirement. When martensitic
hardenability is insufficient in the centre of a steel plate it tends to have a different structure
to the martensite that forms on the surface. Lower cooling rates within the centre result in
greater retention of heat and the possibility of formation of a higher temperature
transformation product is likely. This compromises the integrity of the material, firstly by the
inhomogeneity of the material and secondly, by the type of additional transformation product
that forms and the collective effect this imposes on the mechanical property requirements of
1
© University of Pretoria
the material. The stress systems to which the steel is exposed mostly require that strength and
toughness are optimal and the steel must possess a good measure of resistance to failure.
Therefore, a sufficient amount of ductility is desired in these steels as a signal of imminent
collapse and a guarded measure against brittle fracture resistance. In this regard,
microstructural properties exert a significant influence. The presence of inclusions, their
morphology and distribution is also an important factor.
It is imperative that such plate steels manufactured for structural purposes possess reliability
and integrity taking into consideration the factors that may influence or compromise the
performance of the final product.
1.2
Problem statement
Plate steels with insufficient hardenability may consist of a combination of two phases,
namely lath martensite and bainite. In many studies, dual phase microstructures of martensite
and lower bainite are shown to have improved mechanical properties including strength and
toughness [1, 2]. However, it is doubtful that the same can be said for dual phase upper
bainite and martensite. This being so by the inherent structural difference that exists between
upper and lower bainite. Therefore, this study endeavours to determine the quantitative effect
of upper bainite in quench and tempered martensitic steel with regards to impact energy,
thereby establishing the microstructural tolerances for quench and tempered plate steels.
2
© University of Pretoria
2
Chapter 2: Literature survey
There has been extensive research on the models of bainite formation and the factors that
influence the transformation. This chapter is an overview of the literature on bainite
characteristics and mechanical properties typical of steels with bainitic microstructures.
2.1
Alloying elements and the austenite to bainite transformation
Continuous Cooling and Isothermal Transformation diagrams are a powerful metallurgical
tool as they are an accurate means of predicting microstructure-property relationships in
steels, properties which are modifiable in accordance with the combination of alloying
elements present. Equilibrium solubility temperatures/phases are altered such that the critical
temperatures, i.e Ac1 (equilibrium temperature at which austenite forms during heating), Ac3
(temperature at which transformation of ferrite to austenite is complete), Ms (martensite start)
and Bs (bainite start) temperatures are readily altered by these elements. This is proven in the
quantitative empirical relationships between critical temperatures and alloying element
combinations that have been established. Moreover, equilibrium phase regions in steels are
enlarged or reduced by the relative amounts of alloying elements that are either ferrite
formers or austenite formers.
Ferrite formers such as Cr, Si and Mo stabilise the ferrite regions and the austenite formers
Ni, Mn and C stabilise the austenite regions. In addition to this, Ni, Cr and Mo are known to
exert a significantly positive effect on the depth of hardenability. When added in sufficient
quantities, V and Mo will form coherent carbides that are beneficial for strength while
incoherent Cr carbides are beneficial for improved wear-resistance. However, small additions
of Mo and particularly Boron for instance, retard the austenite to ferrite and pearlite
transformations thus allowing bainite to form more readily, see Figure 2.1.1. Additions of 0.3
and 0.6wt% Mo to a High Strength Low Alloy (HSLA) steel increase the amount of bainite
that forms.
3
© University of Pretoria
4
© University of Pretoria
Figure 2.1.1 Effect of Mo additions to a Low Alloy steel with (a) 0.03wt% Nb (b) 0.03wt%
Nb-0.3wt% Mo and (c) 0.03wt% Nb-0.6wt% Mo [3].
Small amounts of boron remarkably increase the steel‟s hardenability, thus reducing the
amount of expensive alloying element additions required, i.e. Cr, Mo or Ni. The hardenability
effect with additions of between 10 and 30 ppm of boron is caused by the segregation of
solute boron to austenite grain boundaries where it reduces the interfacial energy between
austenite grains and thus slows the nucleation of pro-eutectoid ferrite and pearlite on these
grains. Formation of Fe23(C,B)6 or M23(C,B)6 by excessive amounts of boron (generally 50
ppm and higher) at the austenite grain boundaries severely deteriorates hardenability because
the boro-carbide is a nucleation site for ferrite [4].
As mentioned above, the main mechanism responsible for boron hardenability is that it
reduces preferential sites for ferrite nucleation by reduction of the grain boundary energy, i.e.
a thermodynamic effect. However, other mechanisms proposed for the hardenability effect
[4] are that;
1) The self-diffusion coefficient of iron at grain boundaries is reduced and thus the ferrite
nucleation rate is also reduced, i.e. a kinetic effect;
2) The very act of boron segregation to austenite grain boundaries eliminates preferential
sites for ferrite nucleation (thermodynamic effect) and
5
© University of Pretoria
3) Fine borides that are coherent with the matrix form along the boundaries, eliminating
ferrite nucleation between the boride and matrix, i.e. a structural thermodynamic effect.
During quenching, quenched-in vacancies allow for the formation and diffusion of the
vacancy-boron complex into the grain boundary and boron is thus prevented from returning
to the grain interior [5].
Boron segregation can also lead to the formation of BN, Fe2B and oxides. Low alloy steels
usually contain small amounts of boron; 0.001 – 0.003wt.% [6]. Boron in such quantities
slows the austenite decomposition which allows for lower quenching rates.
In the intermediate temperature range where bainite forms, the austenite grain boundary is
thus stabilised by boron. This causes a lowering of the upper limiting temperature of the Ccurve for bainite, i.e. the Bs [7]. In addition to decreasing the Bs temperature, there is an
increase in the proportion of bainite obtainable and a refinement of the lath size with boron
additions [4].
A long holding time causes boron segregation and subsequent precipitation of borides at
austenite grain boundaries. Such an excessive amount of boron drastically decreases the
steel‟s hardenability. Hardenability initially increases with increasing boron content up to an
optimum boron level after which no improvement and even a decline in its effectiveness is
found [5]. In combination with Mo, the optimum amount of boron increases as the addition of
Mo retards the precipitation of the boro-carbides M23(C,B)6 [8]. In the execution of an
experiment to determine austenitising temperature effects on boron segregation in alloys
soaked at 900
boron led to no observable precipitation while those austenitised at 1100
had boro-carbides on their grain boundaries [5]. Therefore, an optimal boron concentration is
generally considered to be between 10 and 30 ppm to ensure enhanced hardenability.
Cementite formation in bainite is delayed by high Al or Si content, usually in amounts
exceeding 1wt% for Si. Consequently, the resulting microstructure is described as “carbidefree” bainite [9-12]. It is interesting that a study on a no Si-containing alloy showed that the
bainitic ferrite subunits grow at a considerably higher rate with regards to plate lengthening
than bainitic ferrite in which interlath carbide formation is supressed by the addition of Si
6
© University of Pretoria
[13]. The reason proposed is that the carbides are additional sinks for carbon in close
proximity to the bainitic ferrite and thus increase the driving force for ferrite growth. It is
thought after the first ferrite plate forms, further development is triggered by the carbide,
which in turn triggers nucleation of another ferrite lath and so forth. When measured against
Widmanstätten ferrite, which is considered the carbide-free analogue to upper bainite because
of the orientation relationship of the ferrite plates with the matrix, the latter has even higher
growth rates. It would, therefore, seem that the difference arises from the formation of the
carbides, which seem to accelerate the growth rate of bainite [12].
(a)
7
© University of Pretoria
(b)
Figure 2.1.2 The effects of (a) Mn and (b) Cr on the Bs temperature as determined by
Artificial Neural Network (ANN) modelling [14].
Mn decreases the free energy difference between ferrite and austenite as it is a strong
austenite stabiliser and the incubation period is then prolonged by the decrease in Gibbs free
energy. The effect of Mn on the Bs is shown in Figure 2.1.2 (a).
Cr is highly effective in reducing the Bs (Figure 2.1.2 (b)) and increasing the incubation
period of the bainite reaction. However, Mn is more effective in shifting the bainite region
towards longer times in the CCT diagram.
Si and Al function by removing any oxygen present in the steel. In addition to this, Si has
also been shown to suppress the formation of cementite during the austenite to bainite
transformation resulting in a carbide-free bainite [15]. Si suppresses the Bs slightly and
decreases the overall reaction kinetics of bainite formation.
Another beneficial quality of Al is the formation of its nitride, AlN. When it is formed, this
compound binds the free nitrogen and thus reduces the free nitrogen content of the steel
which would otherwise be detrimental to ductility and toughness because of the
embrittlement effect of free nitrogen in a ferritic lattice.
8
© University of Pretoria
Titanium forms a stable nitride. When used in combination with Boron, the steel‟s
hardenability is further improved by alleviating Boron-Nitride formation and thus promoting
the existence of elemental Boron.
2.2
Kinetics
2.2.1 Nucleation and growth transformations
In a reaction where nucleation occurs homogeneously, the volume transformed from (V) to
a new phase (Vβ) after a time interval t is proportional to the untransformed volume and
can be expressed as a first order rate process:
eq 2.1
where k is the rate constant.
A particle will originate at a time η, known as the incubation period, before which there is no
detectable growth. Thus at t > η the volume of new phase
where
η
is given by
τ
= Y1Y2Y3 (t – τ)
represent principle growth velocities in three mutually perpendicular directions
[16]. At the early stages of transformation, there is no interference between the growth fronts
of the new β phase. Thus at an interval between the incubation t = τ and t = τ + dτ the
volume fraction of transformation product becomes:
dVe =
τ Vo
ν
I dτ
eq 2.2
where Ve is the extended volume, Vo the untransformed volume and νI is the nucleation
rate per unit volume. Extended space is defined as regions where particles can grow through
each other and where particles can nucleate in areas which are already transformed [17]. The
new phase in the untransformed regions is given by dV = (1 -
τ
ν
I dτ
Integrating the above expression and setting
leads to:
9
© University of Pretoria
dVe which becomes:
∫
∫
where η is a shape factor. Assuming that νI is constant, the general expression obtained is:
V = 1 – exp (-ktn)
eq 2.3
The general equation 2.3, known as the Johnson Mehl Avrami Kolmogorov equation, can be
applied to most S-shaped transformation curves. It is based on the supposition that the
growing regions do not interfere with each other and case of impingement between growing
regions is dealt with by the concept of extended volume fraction proposed by Avrami [16].
2.2.2 The Johnson Mehl Avrami Kolmogorov equation
In the JMAK equation (eq 2.3) k is a temperature dependent constant given by:
k= η
ν
I
eq 2.4
where R is the gas constant in kJ/mol.K and T is the absolute temperature and Q is the
activation energy. The time exponent n for bainite is a dimensionless quantity generally
having values ranging from 1 to 4 for most transformations [16].
Equation 2.3 models the time dependence of an isothermal phase transformation and at
different isothermal transformation temperatures, the activation energy Q for the mechanism
of the phase formation can be derived from equation 2.4.
The physical interpretation of the exponent n (which is derived from the slope of the graph
of lnln(1/(1-Vf)) against lnt ) is that it accounts for the dimensionality of the growth. Pearlite,
for instance has been found to have an n-value of 4 [17], which represents constant nucleation
rate and three dimensional growth of pearlite at grain corners. Bainite on the other hand has
been found to have n values of roughly 2 [18], which is indicative of a constant nucleation
rate and linear growth. The lengthening of bainite plates is faster relative to the thickening of
the plates which is said to occur after growth has reached completion, hence growth is
regarded as one dimensional.
10
© University of Pretoria
Possible values of the constant n are given in the table below. It is often used as a criterion
of the mechanisms that accompany a transformation. A change in the value of n in a given
transformation is indicative of a change in mechanism though it cannot be used to predict the
exact nature of the transformation mechanism.
Table 2.1 Avrami exponents in relation to nucleation and growth
Conditions
All shapes growing from small dimensions, increasing nucleation rate
All shapes growing from small dimensions, constant nucleation rate
All shapes growing from small dimensions, decreasing nucleation rate
All shapes growing from small dimensions, zero nucleation rate
Growth of particles of appreciable initial volume
n
>2
2
1 -2
1
1-1
Needles and plates of finite long dimensions, small in comparison with their separation
1
Thickening of long cylinders (needles) (e.g. after complete edge impingement)
1
Thickening of very large plates (e.g. after complete edge impingement)
Precipitation on dislocations (very early stages)
For transformations occurring in a thin sheet, growth along the plane of thickness is over
relatively fast. The reaction is then considered to be two dimensional and the growth rate
previously expressed in terms of the three mutually perpendicular planes is now confined to
two planes. That is to say, the growth rate is proportional to ηY1 Y2 (t - η)2. Values of the
exponent n in this case are limited to between 2 and 3. Similarly, in a thin wire, growth is
confined to a direction along the length of the wire and is thus one-dimensional. The value of
n is now between 1 and 2.
The experimental value of n obtained is partly influenced by the method used. A technique
which records information in one dimension, such as length change, measures the
transformation kinetics in that direction alone and the resulting n is between 1 and 2 as for a
one-dimensional transformation. Maintaining three-dimensional results would require
11
© University of Pretoria
sampling in all 3 directions [16]. Although n is regarded as a constant, some studies have
shown a temperature dependence of n [16, 19] as shown in Figure 2.2.1.
(a)
(b)
Figure 2.2.1 Temperature dependence of the Avrami exponent n (a) in Fe-Ni-Si-C, Fe-Mn-SiC and 300M steels [19] and (b) in a 0.66%C steel [16].
The assumptions on which the JMAK equation is based are that of random nucleation,
constant nucleation rate, three dimensional growth and constant growth rate. The value of n
is thus a function of the nucleation rate and the type of site at which nucleation occurs. The
increasing n-values shown in Figure 2.2.1 by Fang et al [16] were found to be due to an
increase in the number of nucleation sites.
When the formation of bainite is modelled with the JMAK equation, generally an exponent of
1 or 2 is found. Fang et al [16] used n as a transformation sensitive parameter in a 0.69wt%
C steel to determine the temperature limits of a transformation product and obtained separate
C-curves for upper and lower bainite but with some overlap. However, there is not always a
transition in the value of the exponent n as the transformation temperature changes from
upper bainite to lower bainite [18] as was found for the medium carbon steel (see Figure
2.2.3) studied by Caballero et al [18].
When modified to suit non-isothermal conditions, it is shown that the Avrami exponent is
still a useful parameter in a qualitative assessment of the kinetics. Gupta et al [20] showed
that each cooling rate yields a specific exponent and that its change with a decrease in cooling
12
© University of Pretoria
rates is related to the morphological changes of the transformation product. A cooling rate of
105 /min gave an exponent of 1, indicating one dimensional growth. The corresponding
microstructure was one of bainitic ferrite plates with interlath films of retained austenite. At
much lower cooling rates an Avrami exponent of 3 was obtained. The corresponding bainite
microstructure was coarser than that at higher cooling rates, indicating significant thickening
of the bainite laths.
13
© University of Pretoria
Figure 2.2.2 Correlation of Avrami exponents with temperature and transformation products
[16].
Figure 2.2.3 Average Avrami exponent n of 2 is shown for both upper and lower bainite in a
0.31C-1.22Mn-0.25Si-0.14Cr-0.10Ni steel (in wt%) [18].
In the work done by Caballero et al, the Avrami exponents obtained for upper and lower
bainite structures in a medium carbon steel were found to be close to 2 [18]. They indicated
that the bainitic reaction occurred by linear growth of ferrite plates nucleated at a constant
rate on the austenite grain surfaces. Thickening of the plates was negligible relative to their
lengthening. The two bainite morphologies were not distinguishable by any significant
difference between their n values [18].
In another study [21] it was found that lower and upper bainite possessed different n values
with the former generally possessing higher n values than the latter. The observation was
made on six different alloy steels, shown in Table 2.2. The variation in Avrami exponents
was interpreted as being the result of different transformation mechanisms.
Table 2.2 n values obtained for upper and lower bainite in different alloy steels [21].
n
Structures
Granular structure
Upper bainite
Lower bainite
15CrMnMoV 18Cr2Ni4W
3.8
2.0
3.5
1.8
1.7
0.6
Steels
30CrMnSiNi2
--3.5
2.8
14
© University of Pretoria
40CrMnSiMoV
--2.0
0.6
GCr15 9CrSi
----6.4
3.3
4.2
2.4
2.3
Bainite formation models
2.3.1 General features
Bainite is referred to generally as a non-lamellar aggregate of ferrite and carbide to
distinguish it from pearlite. Crystallographic mechanisms and features of bainitic and
Widmanstätten ferrite are similar according to Bhadeshia et al with the exception being that
the latter is formed with carbon diffusion as the rate limiting factor and the former without
carbon diffusion [12]. Due to the marked similarities that exist between bainite and
martensite and Widmanstätten ferrite, various postulates have been brought forward on the
mechanisms of bainite formation.
Figure 2.3.1 is an illustration of a widely accepted mechanism for bainite formation. In high
silicon steels, carbide precipitation is suppressed or even eliminated, consequently, the
transformation mechanism is believed to involve only the nucleation of ferrite sheaves [22,
23]. This is utilised commercially in the manufacture of steels with a high level of resistance
to cleavage fracture and void formation.
15
© University of Pretoria
Figure 2.3.1 The mechanism of formation of upper and lower bainite [23].
According to some authors, the transformation mechanisms for upper and lower bainite are
not fundamentally dissimilar. Bainitic ferrite grows with a supersaturation of carbon in both
cases but it is in the partitioning of the surplus carbon from the ferrite into the residual
austenite or its precipitation within the ferrite where the difference arises. The prevalence of
either process will determine the type of bainite obtained. Moreover, the faster the
partitioning of carbon into the untransformed austenite, the less likely is precipitation within
the ferrite and the more likely structure is upper bainite. Hence at temperatures in the region
of the lower bainite start temperature, it is possible to obtain both upper and lower bainite in a
microstructure as in Figure 2.3.2 below [18].
16
© University of Pretoria
1μm
Figure 2.3.2 SEM of a Medium Carbon steel isothermally transformed for 900s at 450⁰C
[18]. UB is upper bainite and LB is lower bainite.
At the transition temperature between upper and lower bainite, which is believed to be at
350
[11, 24], the kinetics of the bainite reaction changes. It is thus possible to form both
upper and lower bainite at temperatures in the region of the transition temperature.
The transition temperature from upper to lower bainite is not a fixed value as it varies with
the carbon content. As can be seen from Figure 2.3.3 the transition temperature initially
increases to a maximum before decreasing to a stable value.
Figure 2.3.3 The transition temperature between upper and lower bainite as a function of
carbon content [25].
The width of a bainitic ferrite plate increases with the isothermal transformation temperature
as shown in Figure 2.3.4. At higher isothermal transformation temperatures, the yield strength
of the austenite decreases and therefore there is plastic relaxation in the austenite bordering
17
© University of Pretoria
the newly formed bainite plates, allowing the plates to widen more. Dislocations are thus
induced in the austenite and they pose a resistance to the advance of the bainite:austenite
interface [18].
Figure 2.3.4 The decrease of ferrite plate thickness with decreasing isothermal treatment
temperature of a medium carbon steel [18].
2.3.1.1 Lower bainite
The two main morphological variants of bainite (upper bainite and lower bainite) are
distinguishable by the precipitation and distribution of the iron carbide. Lower temperatures
of formation lead to a finer carbide dispersion within the ferrite laths. Lower bainite, as
shown in the SEM micrographs of Figures 2.3.5 and 2.3.6 below, has plate-like carbides
aligned at an angle of approximately 55 to the long axis of the ferrite lath due to their
orientation relationship with specific crystallographic ferrite planes. The carbide may be ɛcarbide (Fe2.4C) or cementite and precipitates within the ferrite lath and thus precipitation is
from ferrite rather than austenite. It is presumed that this is a result of the lower carbon
diffusion rates at the reduced temperatures at which lower bainite forms, whereby rejection of
carbon atoms into the adjacent austenite is delayed.
18
© University of Pretoria
1μm
Figure 2.3.5 SEM micrograph of lower bainite in a 2358 steel austempered at 260
minutes [26].
for 100
Figure 2.3.6 TEM micrograph of lower bainite showing the parallel intralath carbides within
ferrite laths [27].
19
© University of Pretoria
2.3.1.2 Upper bainite
Upper bainite is distinguishable from lower bainite by the distribution of carbides that lie
along and between the ferrite laths. The cementite precipitates from the carbon enriched
austenite between the ferrite laths [22, 27]. Lower bainite is predominantly plate-like whereas
upper bainite forms elongated laths. The laths grow from austenite grain boundaries in a
Kurdjumov Sachs orientation relationship [28]. Each lath grows by repeated nucleation of
ferrite subunits which grow to a limited size and collectively form a sheaf, as Figure 2.3.7
shows.
Figure 2.3.7 Formation of upper bainite by repeated nucleation of sheaves [28]
1μm
Figure 2.3.8 SEM micrograph of upper bainite in a 0.31C-1.22Mn-0.14Cr-0.25Si steel
isothermally transformed at 525 for 900s [18].
Whilst the ferrite plates of lower bainite have a high aspect ratio, increased thickness of the
plates and thus smaller aspect ratios are characteristic of upper bainite [29]. The SEM
20
© University of Pretoria
micrograph of Figure 2.3.8 shows upper bainite in a 0.31C-1.22Mn-0.14Cr-0.25Si steel.
Higher micro-hardnesses (295-315 HV) were found for a lower bainite structure which
consists of small sheaves of straight laths with thicknesses less than 1µm. Coarser bundles of
lower hardnesses (240-265 HV) and wider laths on the other hand, characterised upper
bainite [30].
2.3.2 The Displacive model
In the context of a displacive transformation, bainitic ferrite forms by a shear mechanism and
the subsequent precipitation of carbides is a secondary reaction which occurs by diffusion of
carbon into the residual austenite [19, 28, 31]. In accordance with shear or martensitic
transformation characteristics, the bainitic transformation is diffusionless and occurs by the
coordinated movement of atoms [28]. The transformation shares some characteristics of the
martensite formation.
Displacive transformations, such as is proposed for the bainite reaction, involve the
simultaneous shear transformation of a substitutional lattice and diffusion of interstitial
species, which in this case is carbon. The main features between bainite and martensite are
compared in Table 2.3.
Table 2.3 Comparison of the transformation characteristics of bainite αb with martensite α’
and Widmanst ̈ tten ferrite αw [28].
Nucleation and growth reaction
Plate shape
Diffusionless nucleation
Carbon diffusion during
nucleation
Substitutional diffusion during
nucleation
Confined to austenite grains
Large shear
Invariant plane strain shape
deformation
Diffusionless growth
Carbon diffusion during growth
Substitutional diffusion during
growth
Glissile interface
21
© University of Pretoria
Consistency of a comment with the transformation concerned is indicated by ( ),
inconsistency by
.
The growth is initially diffusionless [22] and the upper limiting temperature at which bainite
forms is To, which is defined as the temperature at which austenite and ferrite of the same
composition have equal free energies. Figure 2.3.9 shows a schematic illustration of the To
curve on a phase diagram. The stored energy of 400 J/mol due to the invariant plane strain
change in shape during the formation of bainitic ferrite is accounted for in the To’ curve. The
To curve on a temperature/carbon plot has a negative slope, showing that austenite can
accommodate more carbon at lower temperatures. According to Bhadeshia et al [27] there is
no diffusion of substitutional solute elements and their effect is only a thermodynamic shift in
the To. The majority of alloying elements decrease the carbon concentration at the To curve
except for Si, Al and Co, which increase the carbon concentration [23]. Alloying elements
shift the To curve thus the maximum possible carbon concentration in retained austenite can
be optimised.
Figure 2.3.9 Diagram of austenite and ferrite free energies and To curve on the phase
diagram.
2.3.2.1 Surface Relief
The formation of both upper and lower bainite is accompanied by a martensite-like surface
relief typical of invariant plane strain deformation. It is thus thought that the surface relief of
22
© University of Pretoria
bainite is evidence of a displacive or shear-driven transformation mechanism akin to a
martensitic transformation. According to the displacive mechanism, bainite is a
transformation product resulting from cooperative movement of atoms which produces
growth by shear. The austenite:ferrite interface moves without any thermal activation and the
reaction is initially diffusionless and involves high growth rates given the absence of a
temperature dependence [32]. Figure 2.3.10 shows the surface relief that develops when
bainite forms.
(a)
(b)
Figure 2.3.10 (a) Surface relief of bainite for a 0.33C-0.74Mn-3.47Ni (wt%) steel treated at
574 for 22hr. Magnification 300x [12]. (b) An atomic force microscope scan across a
bainite sub-unit with surface relief [28].
Unlike martensite, bainite forms at higher temperatures where the austenite cannot elastically
accommodate the shape deformation and thus in regions adjacent to bainite, it experiences
plastic deformation [28]. The plastic deformation restricts the growth of each bainite plate
thus each plate grows to a limited size (much smaller than the austenite grain size) before
successive nucleation of new plates, giving rise to the sheaf morphology (see Figure 2.3.7).
23
© University of Pretoria
The observation by Ko et al [25] and Bhadeshia [33] of surface relief on bainite as with
martensite brought about the conclusion that both structures form or nucleate by lattice shear,
however, with the bainite growth rate then governed by carbon diffusion. Due to the lower
growth rate of bainite relative to martensite, Ko and Cottrell proposed that bainite growth is
controlled by carbon diffusion. The surface relief was considered vital evidence in support of
the displacive theory and for growth that occurs by migration of a glissile ferrite:austenite
interface where the bainite that forms inherits the carbon content of the austenite from which
it forms [12].
2.3.3 The Diffusional model
The diffusional or reconstructive model ascribes the nucleation and growth of bainite to a
carbon diffusion controlled process [12]. The diffusion controlled transformation of austenite
to bainite is said to occur by a diffusional ledge movement mechanism [28, 34]. Diffusional
growth of bainite, according to Hultgren [35], occurs under full local equilibrium along the
Ae3 phase boundary by short-range diffusion at ledges in the austenite:ferrite interface. Upper
bainite grows by movement of austenite:ferrite interfaces of low coherency and lower bainite
by the movement of more coherent interfaces which require little self-diffusion [24]. Within
the diffusional model, as with the displacive model, martensite, Widmanst ̈ tten ferrite and
bainite are a continuous series of transformation products that result from increased carbon
trapping as the transformation temperature is reduced [36].
At lower transformation temperatures, substitutional atoms diffuse much slower than
interstitial atoms and therefore no diffusion of substitutional elements is observed. Hultgren
first found that there was no partitioning of alloying elements between ferrite and carbide
during the formation of bainite. It is generally agreed that bainite forms under paraequilibrium conditions whereby only carbon partitions between austenite and ferrite. The
upper limiting temperature for bainite forming in the reconstructive model is given by the Bs
rather than the To temperature. Hillert examined the relationship between the Bs and To as a
function of carbon concentration and found that their respective slopes differed and that they
intersected at some point [35]. Bainite formed at carbon concentrations greater than the upper
24
© University of Pretoria
limit xTo imposed by the To line. In a similar experiment, Agren et al [12] also found that
bainite forms at carbon contents greater than given by the To curve (see Figure 2.3.13).
2.3.3.1 The Kinetic definition
The Bs is the temperature above which it is not possible for bainite to form. It lies
considerably below the eutectoid temperature and is the upper limit to the kinetically
determined bainite C-curve [25], [37]. The C-curve identifying bainite is intermediate with
respect to the temperature ranges for martensite and polygonal ferrite/pearlite. The C-curve
on the CCT represents the nucleation and early stage growth [14].
The addition of alloying elements has a significant effect on the Bs. Several authors [38] have
established equations of the Bs temperature as influenced by the addition of alloying elements
to a plain carbon steel. According to phase transformation theory, the formation of each phase
is reflected by a C-curve in a CCT diagram. However, both upper and lower bainite are
contained within the same C-curve of a CCT diagram.
2.3.4 The Stasis
The formation of a maximum amount of ferrite that is significantly less than the amount
indicated by the lever rule as applied to the para-equilibrium Ae3 line or γ/γ+α phase
boundary, is referred to as the incomplete transformation or the transformation stasis [28]. As
the Bs is approached, it is found that in some steels the bainite transformation ceases
momentarily and at times for lengthy periods before it resumes and achieves the volume
fraction predicted by the lever rule. The incompleteness of the bainite transformation is
inevitably linked to the mechanism of transformation and is thus accounted for in both
displacive and diffusional models.
25
© University of Pretoria
2.3.4.1 The Influence of To
Figure 2.3.11 The lever rule as applied to the To line to find the maximum volume fraction of
bainite that can form at a specific temperature under diffusionless conditions.
Figure 2.3.11 is an illustration of the lever rule applied to the To on a phase diagram. The To
curve typically lies between the Ae1 and Ae3 temperatures, therefore applying the lever rule to
the Ae3 curve where equilibrium conditions are obtained, gives the amount of bainite that
ought to form at a specific temperature. The reaction is incomplete in terms of its failure to
reach completion with respect to the equilibrium amounts predicted by applying the lever rule
to the Ae3 curve [27].
With reference to the displacive model, the incomplete transformation may be explained by
considering the progressive accumulation of carbon from bainitic ferrite. For instance,
consider the initial diffusionless formation of a plate of bainite. Due to the low solubility of
carbon in ferrite, excess carbon from the newly formed bainitic ferrite plate is rejected into
the surrounding austenite. The next plate of bainitic ferrite that grows does so from carbon
enriched austenite. The process is repeated during the growth of bainite so that each
successive plate grows from a consistently higher carbon enriched austenite such that the
process ceases when the carbon concentration in the austenite reaches the To curve [18, 35].
Thus the reaction is incomplete as the equilibrium composition as given by the Ae3 curve is
not attained [28]. To illustrate the relation of the incomplete reaction to the To curve, carbon
concentrations of residual austenite in isothermally transformed steels were plotted on a
phase diagram together with the Ae3, To and To’ curves, which were determined by
26
© University of Pretoria
thermodynamic assessment. The results are given in Figures 2.3.12 and 2.3.13. The authors
found that the reaction stopped when the carbon concentration in the residual austenite was
closer to the To curves than to the Ae3.
Figure 2.3.12 Phase diagram showing calculated Ae3, To and To’ curves and experimental
carbon concentration of residual austenite after isothermal bainite formation in a 15H2VT
steel [28].
Figure 2.3.13 Phase diagram showing calculated Ae3, To and To’ curves and experimental
carbon concentration of residual austenite after isothermal bainite formation in an Fe-Cr-SiC steel [39].
It can be seen from Figures 2.3.12 and 2.3.13 that for a given carbon concentration, a larger
undercooling below To allows a larger quantity of bainite to form. Likewise, with lower
undercoolings at transformation temperatures approaching To, smaller amounts of bainite are
27
© University of Pretoria
formed [27]. Diffusionless growth of bainite, therefore, requires that the carbon concentration
in the austenite is below that given by the To curve.
2.3.4.2
The Solute Drag-Like Effect
The framework for a solute drag-like effect is created in the context of a diffusional
mechanism of bainite formation, particularly at higher temperatures as in upper bainite. The
solute drag-like effect accounts for retarded transformation of austenite to bainite by the
involvement of substitutional solute atoms during the transformation. There are various solute
drag-like models which illustrate the capacity of involvement of substitutional solute atoms:
Hultgren first observed that there is no partitioning of alloying elements between ferrite and
bainitic carbide. This “no partitioning”, in the view of Hillert, could also be explained by a no
partitioning local equilibrium (NPLE) [35]. The concept incorporates the participation of
alloying elements during the formation of bainite by considering an alloying element
accumulation adjacent to the moving interface which Hillert termed a „spike‟. If the spike is
of a thickness below atomic dimensions, it is more likely to move into the moving
austenite:ferrite interface and effects of such an alloy element aggregate within the interface
would be similar to the solute drag effect originally proposed by Cahn [40].
Another explanation for the incomplete reaction bay in the TTT diagram and for the overall
reduced kinetics was established by Aaronson et al [41] whose work on the diffusional
reaction of pro-eutectoid ferrite in Fe-C-X alloys (where X is a substitutional alloying
element which significantly reduces the activity of carbon in austenite) showed that
nonequilibrium absorption of the element X to austenite:ferrite grain boundaries occurs.
Since growth occurs under para-equilibrium conditions, accumulation of substitutional solute
atoms in the austenite:ferrite boundaries occurs when the moving boundary retains some
solute atoms in excess of their average concentration in the alloy [42-44]. In alloys where X
is an element that reduces the activity of carbon [41], the carbon concentration gradient in the
austenite ahead of the growth front is reduced. Stasis will occur when the reduction in carbon
activity at the α:γ grain boundaries leads to a carbon activity which is the same as that further
from the boundary such that there is no gradient of the carbon activity in the austenite. The
28
© University of Pretoria
diminished carbon concentration gradient, which drives the growth of ferrite, slows the ferrite
growth kinetics [45].
One quantitative explanation for the stasis based on the solute drag effect, is that involving
ortho- and para-equilibrium austenite/ferrite phase boundaries on an Fe-C-Mo phase diagram
[36]. Under para-equlibrium, the fraction transformed is given by:
f = (c2 - c0)/(c2 - cα)
eq 2.5
and under local equilibrium it is given by
f = (c3 - c0)/(c3 - cα)
eq 2.6
where c0, cα , c2 and c3 are the carbon concentrations in the bulk ferrite, at the paraequilibrium γ/α+γ and local equilibrium at the γ/α+γ phase boundaries respectively. Solute
drag forces the boundaries to the left in Figure 2.3.14 and smaller transformed fractions are
obtained.
Figure 2.3.14 Isothermal section of a Fe-C-Mo phase diagram where solute drag results in
lower volume fractions of ferrite with the lever rule applied to the para-equilibrium (dashed)
boundary.[36].
The rate of formation of bainite is limited, though not necessarily controlled by diffusion of
carbon, and also by the diffusion of substitutional elements and interfacial reactions. In Fe-C
steels, the absence of substitutional alloying elements and thus without their slower diffusion
29
© University of Pretoria
rate, allows higher reaction rates. The alloying elements are, therefore, responsible for the
slower kinetics in alloy steels [46].
The decomposition of austenite to bainite tends to be more complete as the transformation
temperature decreases and the stasis is thus more likely to occur in the upper bainite range
[24].
The incomplete transformation in steels, in addition to being temperature dependent, is also
dependent on the chemical composition. In some steels where the stasis is not observed, it is
as a result of the overlap of the bainite and pearlite C-curves which interfere with the reaction
at higher temperatures. Where carbide forming elements capable of separating the C-curves
are present, the stasis is more observable. The addition of Mo to steel produces a bay in the
TTT diagram as well as the incomplete transformation of austenite to bainite. Figure 2.3.15
below shows sigmoidal curves that vary in the extent to which the formation of bainite is
stalled. Curve IV is subdivided into 3 sections with the middle stage having a slope of zero
where df/d(log t) = 0. At this point f, which is the volume fraction of austenite transformed,
is less than that allowed by the lever rule. As the temperature and compositions are varied
from the conditions under which stasis occurs, the curve changes from III to curve II, where a
non-zero gradient is found at the middle stage and eventually to curve I where no stasis is
observed. In the type I curve, JMAK type kinetics is consistent throughout the
transformation. It was found for this steel that the stasis will develop only when specific CMo ratios are exceeded [42].
30
© University of Pretoria
Figure 2.3.15 Schematic illustrations of reaction kinetic behaviour below the Bs and the
sequence of transformation [47]. SN = sympathetic nucleation and SDLE = solute drag-like
effect.
2.4
Mechanical properties
2.4.1 Ductile and brittle fracture
The fracture of a metal occurs in either a ductile or brittle manner. Ductile fractures are
characterised by formation and coalescence of microvoids while brittle fractures occur by
cleavage in a transgranular or by intergranular manner.
The shape and orientation of ductile dimples are indicative of the manner of loading
experienced during fracture [48]. Homogeneous plastic strain in the loading direction
produces equiaxed dimples. A non-uniform force instead produces dimples elongated in the
direction of crack extension. Microvoids that initiate the formation of dimples tend to
nucleate at regions of local strain discontinuity such as inclusions and grain boundaries.
Brittle fracture is characterised by cleavage fracture. This is a low energy type of fracture that
propagates intragranularly along cleavage planes. River patterns or chevron markings are a
key feature of brittle crack propagation and can aid in locating the origin of a crack by tracing
back the direction of crack propagation. Cleavage occurs on specific, low-index
crystallographic planes and it is the mismatch of these planes across grain or subgrain
boundaries that produce cleavage steps and ultimately, river patterns. In this way, obstruction
31
© University of Pretoria
to advancement of a cleavage crack is created. Misorientation at grain boundaries also causes
tearing in the vicinity of the grain boundary together with localized deformation [49].
2.4.2 Effect of precipitates on fracture
In microstructures containing a hard or brittle second phase, embrittlement is likely to occur
and often leads to brittle fracture. The type of bainite that forms in a steel affects its fracture
resistance. Lower bainite generally shows improved toughness, ductility and strength in
comparison to upper bainite [50]. Steels containing upper bainite as a secondary
decomposition product are susceptible to brittle fracture. In such instances, the Ductile to
Brittle transition temperature is raised and the Upper Shelf Energy is lowered as shown in
Figure 2.4.1. The influence of precipitate morphology is thus illustrated schematically for the
case of upper and lower bainite.
Figure 2.4.1 Schematic illustration of the difference in upper shelf energy of lower and upper
bainite.
The simplest mechanism for nucleation of a microcrack as suggested by Stroh [49] and as
illustrated in Figure 2.4.2, is the result of a stress concentration produced at the tip of a
dislocation pile-up when the material is subjected to stress. In dual phase steels containing
precipitates or carbides, the pile up may terminate at a carbide interface. Under such
circumstances, the stress generated at the tip of the dislocation pile-up may lead to cracking
of the carbide or decohesion at the carbide:matrix interface. This is especially likely if the
particle is of a brittle nature. The effect of brittle precipitates is incorporated in the Smith
model [51]:
32
© University of Pretoria
1/2
eq 2.6
where ζf is the fracture stress, γf is the effective surface energy of the interface between the
grain and precipitate, Co is the thickness of the precipitate, v is Poisson‟s ratio and E is
Young‟s Modulus. A dislocation pile-up will cause a grain boundary precipitate to fracture if
the stress exceeds a certain value which is in part dependent on the precipitate morphology by
the term Co. From this model, it is evident that coarser precipitates will lower the fracture
stress (ζf). McMahon and Cohen found that the incidence of a microcrack forming increased
with increasing thickness of carbides and decreasing test temperature [85].
Brittle grain boundary films not only facilitate crack nucleation, they also reduce energy for
crack propagation, thus cracks grow easily [51].
Figure 2.4.2 Microcrack formation by dislocation pile-up.
2.4.3 Effect of a bainitic microstructure on fracture
A bainite packet is defined as a group of parallel laths that subdivide an austenite grain. The
packets are separated by high angle boundaries. The packet is further divided into blocks by
smaller lath bundles. Laths are separated by small angle boundaries. Consequently, a crack is
only slightly deviated at lath boundaries, as shown in Figure 2.4.3, while higher angle
boundaries such as exist at extremities of a grain are capable of significantly changing
direction of microcrack propagation and even arresting its motion. Boundaries are considered
33
© University of Pretoria
to be highly angular in the order of: prior austenite grain boundaries, block and packet
boundaries [52].
(a)
(b)
Figure 2.4.3 SEM micrograph of crack deflection at (a) high angle boundaries and (b) low
angle boundaries in lower bainitic microstructure of a 4150 steel transformed at 375 [53].
It is worth noting, that in upper bainitic structures, the crack passes or proceeds seemingly
oblivious of the carbides present and smooth cleavage cracks are found, such as in Figure
2.4.4 below.
Figure 2.4.4 SEM micrograph of cleavage cracks in upper bainite in a steel transformed at
450
[53].
34
© University of Pretoria
Zhi-jun et al [52] considered the effect of packet and block boundaries on crack propagation.
In body-centred-cubic structured low alloy steel cracks extend along the [100] cleavage
planes, as shown by the analysis that was based on the [100] angle between adjacent blocks
within a packet. They found that the angle of the [100] planes within the same block was
similar to that of adjacent blocks in different packets. Thus having shown the similar effects
of blocks and packets on the hindrance of crack propagation, it was concluded that the block
is the main substructure controlling toughness rather than the packet. The large angle
transition of a crack as it encounters a high angle boundary is the underlying factor behind the
linear proportional relationship between high angle boundary length and toughness.
Figure 2.4.5 Cleavage crack deflection in a fragmented austenite grain at (a) bainite packet
boundaries and (b) at bainite + martensite packet boundaries.
When subdivision of an austenite grain by martensite and bainite packets occurs, a finer facet
size is obtained as shown in Figure 2.4.5 (b). In upper bainitic structures, it is often found that
the packet size is smaller than the cleavage facet size due to the deflections at the colony
boundaries Figure 2.4.5 (a). As illustrated in Figure 2.4.5 above, deviation of a crack is
promoted by fragmentation of a prior austenite grain as illustrated in Figure 2.4.6. Several
35
© University of Pretoria
studies show that the aforementioned subdivision tends to have an effect analogous to that of
grain size refinement on improved impact toughness [1, 52, 54, 55].
Figure 2.4.6 Diagram showing fragmentation of a prior austenite grain due to formation of
bainite sheaves (2) within austenite (1) where growth starts at the austenite grain boundary
(3).
2.4.4 Hardness of bainitic microstructures
The main factors responsible for the hardness of bainitic microstructures are the plate
thickness and the location of the cementite particles. Lower transformation temperatures lead
to harder microstructures. The reason is for this is the decrease in lath width with lower
transformation temperatures and the density increase of carbides within the bainitic ferrite
plates [18]. Figure 2.4.7 shows the change in hardness as the isothermal transformation
temperature increases.
36
© University of Pretoria
Figure 2.4.7 Hardness decrease with isothermal transformation temperature increase [18].
Plastic relaxation in the austenite at the higher isothermal transformation temperatures
generates dislocations which resist the advance of the austenite:ferrite interface. Since the
lengthening rate is higher than the thickening rate, the bainitic ferrite plate thickness would
thus increase with higher transformation temperatures. The result is that with lower
temperature bainite, higher hardness‟s are attained as the plate width is smaller at these
temperatures.
2.4.5 Effect of bainite on toughness
The positive influence of bainite has been associated with the fragmentation of the prior
austenite grains as shown in Figure 2.4.5 above. The crack intersects the laths without much
deviation and its direction of propagation is altered only as it passes from one packet to
another. Retained austenite may hinder the progress of a crack. Fractographic analysis of
upper bainite-martensite fracture surfaces in Figure 2.4.8, revealed coarser fractures in a
cleavage and detachment mode. Plastic deformation at the tip of a crack formed in upper
bainite is minimal and the fracture mechanism becomes cleavage and less plastic deformation
ensues. In microstructures consisting of lower bainite:martensite, the existence of cleavage
fracture was limited to regions consisting of martensite only.
37
© University of Pretoria
Figure 2.4.8 Fracture surfaces of a 26Kh1MFA steel consisting of (a) martensite and lower
bainite and (b) martensite and upper bainite [1].
The decrease in impact toughness for the structure in Figure 2.4.8 (b) is attributed to
heterogeneous distribution of carbides of a coarse morphology. In such cases, crack
deflection only occurred on intersection with high-angle martensite packet boundaries.
The good impact toughness of lower bainite is due to the high density of high-angle
boundaries presented by the morphological packets and blocks that make up its
microstructure. A propagating cleavage crack encountering such a block boundary will
change its plane of propagation in accordance with the crystallography ahead of the
boundary. In order for effective deflection to be attained, the crystallographic misorientation
between boundaries ought to be greater than 15 . Ferrite plates aligned at 15 and lower
consist of crystallographic packets [56].
The subdivision of a prior austenite grain by laths and packets of martensite and bainite has a
significant effect by the crack propagation energy absorption that occurs at the boundaries of
these subunits. In one study [2] it was shown that the size of packets and laths was smallest in
a mixed bainite-martensite microstructure. The packet size was 28µm and the lath size 2µm.
The martensitic microstructure had larger packet and lath sizes of 47 and 4µm respectively
and the largest substructure sizes measured were for the bainitic structures of 60µm and 9µm.
Instrumented impact tests, which measure the crack initiation and propagation energies
separately, show very low fracture energy for upper bainite structures [57]. A rapid decrease
38
© University of Pretoria
of crack propagation energy was found for fracture in upper bainite microstructures in
Figures 2.4.9 and 2.4.10. Fracture surfaces observed consisted entirely of cleavage facets.
Figure 2.4.9 Cleavage fracture of an AISI 4340 steel isothermally transformed to upper
bainite at 430 [57].
Figure 2.4.10 SEM micrograph of Carbide cracking and debonding in upper bainitic
microstructure of a AISI 4150 steel isothermally transformed at 450 for 24hrs [53].
Steel alloyed with 2.3wt% Si to produce carbide-free upper bainite and subjected to low
temperature tempering at 180
showed a fracture surface consisting of an even distribution
of fine dimples [58]. These initiation points were considered to have been local regions of
high residual stresses produced during the formation of martensite. Tempering at 250
39
© University of Pretoria
eliminated these initiation points by reducing the micro-stresses in the martensite and large
dimples were then found [58].
According to the above authors, structures consisting of 10 - 30% bainite in martensite,
formed by isothermal transformation at a temperature above the Ms followed by quenching to
martensite, had the same strength as martensite although they showed no improvement of
toughness or ductility. An increase of the bainite content to 36% resulted in a quasi-cleavage
brittle fracture surface. It was suggested that the micro-ductility decreased due to the higher
levels of carbon enrichment of austenite which then produced a harder and more brittle
martensite.
When the isothermal treatment temperature was increased from 300
to 400 , the YS/UTS
ratio was increased together with a simultaneous decrease in Charpy impact energy. This can
be attributed to the finer and more homogeneous distributions of carbides, a higher
dislocation density and higher bainitic ferrite aspect ratios found in the lower bainite formed
at 300
[29].
40
© University of Pretoria
3
3.1
Chapter 3: Experimental procedures
Metallography
3.1.1 Secondary Electron Microscopy
Scanning Electron Microscopy (SEM) is a microstructural characterisation technique that
utilises the interaction of primary electrons with a sample surface. It is useful in obtaining
higher lateral resolution and depth of field images than can be obtained with the conventional
optical microscope. Incident electrons that leave the surface of a sample, known as Back
Scattered Electrons (BE), are quantified by the backscattering coefficient (η), which is a
function of the atomic number. Due to this dependence, BE mode is usually used for
identifying compositional contrasts on a sample. Regions of the sample surface with higher
atomic number appear brighter than areas of low atomic number. The contrast differences
become more visible with lower atomic numbers. BE SEM is thus a means of qualitatively
identifying microstructural constituents.
Secondary Electrons (SE) are produced by the interaction of the incident beam electrons with
the loosely bound electrons on the sample surface. SE mode is useful for topographic contrast
as the secondary electrons to incident electrons ratio (δ) is more sensitive to the tilt angle (i.e.
angle between the incident beam and sample surface) than to the atomic number. However,
the backscattering coefficient also shows a strong dependence on the tilt angle and BE is thus
also used to image surface morphology [59].
Samples mounted in bakelite were prepared for SEM examination by grinding and polishing.
Silicon carbide paper of 180, 220, 340, 400, 600 and 800 grit was used to grind samples in
this sequence with a continuous flow of water. Samples were then polished to a mirror finish
using 3µ and 1µ diamond suspension paste or alumina suspension mixture. The polished
samples were then etched in a 2% Nital solution.
41
© University of Pretoria
3.1.2 Transmission Electron Microscopy
The Transmission Electron Microscope (TEM) is an instrument used most effectively in high
resolution studies of the metallic microstructures. Electrons produced as the source of
illumination are transmitted through the sample, which must be thin enough to allow the
passage of electrons. Electrons that strike the thin film can either be diffracted by the atomic
planes or pass through as transmitted electrons. Interactions of the sample with the electron
beam that produce diffracted electrons results in images with internal structural detail of very
high resolution of the order of nanometres. The diffracted beams produce dark field images
while transmitted electrons produce bright field images.
TEM was used to examine precipitates extracted from samples by the two methods given
below in subsections 3.1.3 and 3.1.4.
3.1.3 Carbon extraction replica sample preparation
The procedure for extraction replica sample preparation was as follows:
Mounted samples were etched for approximately 60 seconds using 2% Nital before coating
with Au-Pd at an angle small enough to cast a shadow on a surface feature. A coat of carbon
was then laid on vertically above the sample surface. The replicas were removed by
immersion in 5% Nital for a period of time sufficient to detach the coating from the sample
surface. They were floated off in demineralised water and carefully captured on TEM Copper
grids. In a final step, the grid was dried on filter paper.
3.1.4 Electrolytic extraction
Samples of dimensions 15x10x10 mm were heat treated to form a completely upper bainite
microstructure. The bainitic carbide was extracted from the samples electrolytically as
follows:
A 35% HCl was diluted in de-ionised water to obtain a solution of 10% HCl, which was the
electrolyte. An insulated copper wire was attached to the sample with conductive epoxy,
which was set in a furnace at 100 . The connection was then mounted in cold curing resin to
42
© University of Pretoria
protect it from being corroded. The cathode consisted of a stainless steel electrode. The
copper wire connected to the sample formed the anode. The electrode potential of Fe in the
reaction Fe2++ e ↔ Fe is ≈ -0.447V thus a potential of 0.50V was maintained via a
potentiostat connected to the circuit. Figure 3.1.1 shows the circuit setup used.
The residue collected was rinsed in ethanol by centrifuging repeatedly to remove the
electrolyte. Centrifuging was done at a speed of 18000 rpm for a minimum of 10 times. The
rinse was to remove all traces of electrolyte and as Figure 3.1.2 shows, the ethanol turned
yellow after the first few rinses. The electrolyte was completely washed off the residue when
the ethanol returned colourless after centrifuging.
To examine the extracted precipitates in the TEM, a drop containing residue suspended in
solution with ethanol was dispersed on a thin carbon substrate on Cu grids and allowed to
dry.
Figure 3.1.1 Circuit used for electrolytic extraction.
43
© University of Pretoria
Figure 3.1.2 Tinted ethanol in the beaker after centrifuging and centrifuged sample with
extracted precipitate collected at the bottom.
3.1.5 Etchants
Samples mounted for optical microscopic examination were etched by swabbing a 2% Nital
solution prepared from Nitric acid (2%) and Ethanol (98%).
A 10% Sodium Metabisulphite colour etchant was used to distinguish between bainite and
martensite. The solution consisted of approximately 10g Na2S2O5 dissolved in 100ml distilled
water. Specimens ground and polished to mirror finish were immersed face up in the etchant
followed by gentle agitation till there was visible darkening of the polished surface. The
samples were viewed under polarised light to enhance colouration in which bainite is tinted
blue and martensite brown [60].
3.2
Thermal analysis
3.2.1 Dilatometry
An induction type dilatometer heats a sample suspended inside an induction-heating coil by
passage of a current through the coil. The electrically conductive sample is heated in an
induction coil in a vacuum chamber at pressure 0.05Pa. Figure 3.2.1 shows the chamber of
the dilatometer used. A reduction in the current flowing through the coil as well as an influx
of inert gas (Helium or Nitrogen) controls the cooling. Valves inject blasts of He or Nitrogen
gas during quenching at maximum rates of 100Ks-1 when in deformation mode. Higher
44
© University of Pretoria
cooling and heating rates of up to 2500Ks-1 and 4000Ks-1 respectively are achieved in quench
mode, depending on the pressure with which the gas is released.
Figure 3.2.1 The chamber of a Bӓhr DIL 805 dilatometer. The schematic diagram shows the
coils surrounding a test specimen.
Temperature changes of the sample are measured by a thermocouple that is spot-welded to
the sample. Usually an S-type thermocouple was used.
45
© University of Pretoria
The use of a dilatometer is based on the dimensional change that occurs on heating and
cooling of steels. The dimensional change occurs as a result of thermal strain or
crystallographic change. Thermal strain is a result of temperature change and is usually
linear. The change in strain of the sample is measured by push-rod apparatus located at the
longitudinal ends of the specimen. During crystallographic phase change, the change in strain
is detectable due to the difference in density of the body centred cubic (BCC) and face
centred cubic (FCC) unit cells. BCC ferritic structures are stable up to the Ae3 temperature
which for pure iron is Ac1 = 910 . Above the upper critical temperature, FCC austenite is
stable. On a heating cycle, the non-linear volume change associated with the BCC to FCC
transformation is read off a strain-temperature plot. Figure 3.2.2 shows one such plot. A
phase transformation is thus distinguishable by the deviation from linear thermal strain. The
start and finish of formation of microstructures is read off at the point of deviation from linear
thermal strain on the strain-temperature plot obtained upon cooling. Continuous cooling
transformation diagrams can be obtained by monitoring the transformation temperatures and
times at various cooling rates [61].
Figure 3.2.2 Diagram showing contraction during transformation of austenite to ferrite.
Solid cylindrical rods of 5mm diameter and 10mm length were machined from the steel
plates in the rolling direction. An S-type thermocouple was welded to the surface of the
longitudinal centre of the sample. Samples were heated in vacuum to the austenitising
46
© University of Pretoria
temperatures of at a rate of 1.5 /s then soaked for 15minutes. During cooling to room
temperature at various rates from the soaking temperature, critical points for the Partial CCT
diagrams were determined. A Bӓhr DIL805 dilatometer was used for the thermal heat
treatments. To avoid thermal cycling effects, no sample was reused.
Heat treatments were designed to produce partial CCT diagrams by varying the linear cooling
rates as shown in the schematic figure below. The Ac1 and Ac3 temperatures were obtained by
using low equilibrium cooling rates of 0.03 /s.
Heat treatments to obtain microstructures consisting of bainite and martensite were obtained
by isothermally holding the samples for various times. The schematic figure below shows the
temperature cycle used;
Solution treatment
Temperature
900⁰
( )
C
Cool to bainite C-curve
Bainite Bs boundary
1.5⁰ C/
s
Bainite formation
He quench to
martensite
Time /s
Figure 3.5 Schematic diagram of isothermal heat treatments.
The cooling rates to the bainite C-curve were designed to exclude incubation time from the
holding time at the isothermal transformation temperatures. i.e. after cooling the isothermal
treatment entered the bainite nose immediately.
47
© University of Pretoria
3.2.2 Gleeble
The temperature cycles of Figure 3.5 were simulated on a Gleeble 1500D TM
thermomechanical simulator in which full size Charpy samples were heat treated to produce
various proportions of bainite and martensite. The method was preferable to salt bath heat
treatment as specific rates of heating and cooling could not be maintained in the latter and
difficulty was encountered in controlling the microstructures produced in the specimens.
However, in the Gleeble it is possible that a temperature gradient occurs between the surface
and centre of a sample. This is due to direct electrical heating of the sample where the current
concentrates more on the surface of the sample.
Standard Charpy samples of dimension 10x10x55 mm machined in the rolling direction of
the low alloy steel plates were used. A K-type thermocouple welded to the centre of the
sample controlled the temperature during heat treatment. Samples were notched after heat
treatment. A total of 21 Charpy samples per steel were heat treated with three samples
assigned to any given heat treatment.
3.3
Mechanical testing
3.3.1 Hardness
A quick and simple way of differentiating between phases is by the micro-hardness of a
phase. This is useful for instance, when combinations of phases are obtained in samples
continuously cooled from the austenitising temperature. A Micro-Vickers hardness tester was
used for measuring the hardness of finer features and a Macro-Vickers hardness testing unit
was used to measure the hardness profiles of the steel plates. A dwell time of 10s was used
throughout all the tests and five readings were taken for every data point.
3.3.2 Instrumented Charpy impact
The impact toughness of dual phase bainite-martensite Charpy samples was measured by
impact tests conducted with an instrumented Instron Dynatup 9210 impact tester. Figure 3.6
shows the drop tower of the instrumented impact tester that was used. To ensure sufficient
thrust to fracture the steel samples, a mass of 19.54kg was suspended at a height of 1.65m
48
© University of Pretoria
above the position of the Charpy sample. The velocity of the tup was ≈ 3.9m/s, in accordance
with standardised procedures which requires velocity in the range of 3 – 6m/s. The tests were
conducted at ambient temperature. The sensor situated in the tup unit relays the information
to a data acquisition system which records the variation of the force with time/displacements
and integrates this information to energy-displacement numerical data on an Excel spread
sheet.
Instrumented impact testing gives detailed information about the fracture history of a
specimen. The total energy absorbed during the fracture of a Charpy specimen is distributed
between elastic and plastic deformation as well as ductile and brittle propagation. Thus the
sequence of events leading to failure of a sample can be analysed both visually and with loaddisplacement or load-time data.
49
© University of Pretoria
Figure 3.6 Drop tower of the instrumented impact tester utilised, Instron Dynatup 9210.
50
© University of Pretoria
4
4.1
Chapter 4: Results
Microstructures of as the received material
The chemical compositions of the steels used are given in Table 4.1. The steels are
commercial grade low alloy plate steels taken from normal production by ArcelorMittal
South Africa (Vanderbijlpark plant) in which they are hot rolled and quench and tempered.
Table 4.1 Chemical compositions of the steel plates in wt%.
C
Mn
Si
Ni
Cr
Mo
Ti
Al
N
B
HC6
0.112
1.351
0.465
0.037
0.291
0.014
0.023
0.032
0.01
0.00243
HC7-03
0.171
1.076
0.295
0.024
0.728
0.226
0.002
0.098
0.01
0.0035
HC7-05
0.17
1.10
0.31
0.04
0.10
0.54
0.005
0.07
0.01
0.002
The HC7-05 steel will be referred to as the HMoLCr steel and the HC7-03 as the HCrLMo
steel due to the relative Cr to Mo ratios in these two steels. The HC6 steel will be referred to
as the C-Mn steel. The HCrLMo and C-Mn steel plates of thicknesses 41 and 32 mm
respectively were first examined in the as-received condition which reportedly had undergone
a quench and tempering treatment for the former low alloy steel and as-quenched for the
latter C-Mn steel. Microstructural analyses were carried out on the HCrLMo and C-Mn steels
by means of optical microscopy, scanning electron microscopy (SEM) and transmission
electron microscopy (TEM). The micrographs were taken along the centre thickness of the
plates. Note that the HMoLCr steel samples were received at a later stage of the study and
therefore the initial analysis up till section 4.3 was conducted on the HCrLMo and C-Mn
steels.
51
© University of Pretoria
4.1.1 The microstructure of the HCrLMo steel
50µm
(a)
bainite
20µm
(b)
(c)
50µm
Figure 4.1.1 Optical images of the HCrLMo steel: 0.17C-1.076Mn-0.73Cr-0.23Mo-0.002B
steel taken through the mid-thickness at (a) near the top surface (b) in the centre and (c) near
the bottom surface.
The optical micrographs show the presence of martensite at the upper and lower surfaces
(Figure 4.1.1 (a) and (c)) of the plate and some bainite at the centre (Figure 4.1.1(b)).
52
© University of Pretoria
AlN
(a)
(b)
15μm
1μm
Figure 4.1.2 (a) SEM micrograph of the HCrLMo steel and (b) TEM micrograph of a carbon
extraction replica taken from the HCrLMo steel.
The SEM micrograph in Figure 4.1.2 (a) shows grains of lath martensite. The large
precipitates in the TEM micrograph in Figure 4.1.2 (b) are AlN particles. The smaller
particles that seem to have formed with a specific crystallographic orientation are most likely
to be the iron carbides formed in lower bainite.
4.1.2 The microstructure of the C-Mn steel
(a)
20μm
53
© University of Pretoria
(b)
20μm
(c)
20μm
Figure 4.1.3 Optical micrographs of the C-Mn steel taken through the mid-thickness (a) near
the top surface (b) in the centre and (c) near the bottom surface of the steel plate.
The optical micrographs taken near the top and bottom surfaces of the C-Mn steel plate
showed a martensitic structure. The wider ferrite laths in the microstructure at the centre of
the plate shown in the micrograph of Figure 4.1.3 (b) suggest that the structure may be
bainite. The SEM micrograph of this lath structure in the centre of the plate is shown in
Figure 4.1.3 (a).
54
© University of Pretoria
(a)
(b)
Figure 4.1.4 SEM micrographs of the C-Mn steel taken (a) in the centre and (b) near the
surface of the plate.
The higher resolution SEM images of the C-Mn plate confirmed the morphology of the
different microstructures that were present at the surface and centre of the plate. According to
B ́ ranger et al. [62], in steels with carbon levels less than 0.2wt% the volume fraction of
cementite decreases and the shapes of the ferrite laths and lath bundles become more irregular
and less distinct. Further evidence of the possibility that the microstructure at the centre of the
plate is bainite was found in the carbon extraction replica in Figure 4.1.5 (b). The lath in the
microstructure has a fine distribution of small carbides which are at a fixed orientation, as is
typical of lower bainite.
55
© University of Pretoria
TiN
AlN
lath
(a)
(b)
1μm
2μm
Figure 4.1.5 TEM micrographs of carbon extraction replicas taken from the C-Mn steel (a)
shadowed with Au-Pd and (b) unshadowed replica. Note the lath with carbides of size less
than 1μm.
Many of the precipitates in the carbon extraction replicas were either TiN or AlN as
determined by EDS analysis.
4.2
Hardness Profiles
T2
RD
T1
Figure 4.2.1 Schematic illustration of lines along which hardness measurements were taken
on the C-Mn and HCrLMo steel plates. RD is the rolling direction. T1 is a path along the
centre of the width of the plate and T2 is a path along the centre of the thickness of the plate.
Macro Vickers hardness measurements were determined directly on the surfaces of the asreceived steel plates, which were ground to a smooth finish. The hardness indentations were
made along paths T1 (width profile) and T2 (thickness profile) on a plane transverse to the
rolling direction on the received samples. The test load used was 10kg.
56
© University of Pretoria
4.2.1 Hardness profiles of the HCrLMo steel plate near the surface
Hardness (Hv10)
HCrLMo steel: sample width profile
450
400
350
300
250
200
150
100
50
0
plate
Right
Left
0
10
20
30
40
50
60
70
80
Distance (mm)
(a)
HCrLMo steel: sample thickness profile
400
Hardness (Hv10)
350
300
250
200
150
plate
100
50
Top
0
0
Bottom
10
20
30
40
50
Distance (mm)
(b)
Figure 4.2.2 (a) Hardness profiles as measured along the centre thickness through the
sample width and (b) measured along the centre width of the HCrLMo steel plate. Macro
Vickers load: 10kg
Figure 4.2.2 (b) shows the hardness profile across the thickness of the HCrLMo steel plate.
The lowest hardness is at the top and bottom edges of the plate and at the centre. It rises at
57
© University of Pretoria
about a quarter and three quarter way through the thickness by 31.8HV from the top surface
of the plate and by 40HV from the bottom surface of the plate. The centre hardness was
309.9HV, which agreed at the point where the two measurement paths crossed.
The width profile in Figure 4.2.2 (a) increased linearly from the left surface to the right
surface of the plate. Note that the edge‟s hardness‟s were taken as close to the periphery as
was possible, i.e. ~ 2.5 mm from the edge without distortion of the indentation. The width
referred to above is that of the received samples rather than the actual plate width produced
by ArcelorMittal, which is much larger.
The dip in hardness at the plate centre can be correlated with the micrographs of Figure 4.1.1
(b) which present martensite together with bainite at this position. Bainite is softer than
martensite and thus the plate‟s hardness is lower where some bainite is found. The
micrographs of samples taken close to the top and bottom surfaces of the plate in Figure 4.2.2
(a) and (c) contained martensite and thus had a higher hardness. The presence of bainite in
the steel plate resulted in a non-uniform microstructure which formed during cooling as a
result of slower heat transfer from the steel plate during quenching. The steel plate during
production is quenched in a quench brake press by sprays of water on both the top and
bottom surfaces, it thus experiences the highest heat transfer at regions close to these
surfaces. Slower heat transfer occurs towards the centre and consequently the slower cooling
rate results in formation of bainite rather than martensite. The areas at half thickness (see
Figure 4.2.2 (a)) are equidistant from the quenched surface and are therefore assumed to
experience the same rate of heat transfer, hence no significant variation in width hardness
profile was found.
58
© University of Pretoria
4.2.2 Hardness profiles of the C-Mn steel
Hardness (Hv10)
C-Mn steel: sample width profile
400
350
300
250
200
150
100
50
0
plate
Right
Left
0
10
20
30
40
50
60
70
80
90
Distance (mm)
(a)
C-Mn steel: sample thickness profile
400
Hardness (Hv10)
350
300
250
200
150
plate
100
50
Bottom
Top
0
0
5
10
15
20
25
30
35
Distance (mm)
(b)
Figure 4.2.3 (a) Hardness profiles measured along the centre thickness and (b) along the
centre width of the C-Mn steel plate. Macro Vickers load: 10kg
The hardness profile in Figure 4.2.3 (b) showed a decrease in hardness towards the centre.
The hardness dropped by 75.5HV from the top surface and by 63.4HV from the bottom
surface of the plate to a hardness of 271.3HV at the centre. On the contrary, the width
59
© University of Pretoria
hardness profile in Figure 4.2.3 (a) had the lowest hardness at the edges and peaked towards
the centre with hardness in the region of 277 - 286HV.
As was described in section 4.2.1 for the HCrLMo steel, during quenching the gradient in
cooling rate from the plate‟s surface to centre leads to the formation of different phases which
results in a variation of hardness. Since the quench is only directed at the wide top and
bottom surfaces and not at the two side edges, the thickness face experiences the least heat
transfer and the width hardness profile in Figure 4.2.3 (a) testifies to this. The lowest
hardness is found at the width extremities.
The C-Mn steel plate was received in the as-quenched condition and was untempered. By
correlating the hardness measurements taken on the C-Mn steel plate with the amount of
martensite corresponding to specific carbon contents [63] the amount of martensite contained
in the plate can thus be estimated. The hardness data comparison shows that the centre of the
plate most likely contains less than 50% martensite while the edges have 95 - 99%
martensite.
Hardenability calculations were done for both the HCrLMo and C-Mn steels using a quench
factor H = 1 and ASTM 7 grain size and alloying element multiplying factors DX were
obtained from standard tables ASTM A255-02 [63]. From this the Grossmann equivalent
thickness for the C-Mn steel plate was found to be 45mm and that of the HCrLMo was
92.5mm. The actual plate thicknesses were 32mm and 41mm and complete hardening was
therefore expected. However, the hardness profiles above show that the hardenability was
compromised, most likely through inadequate quenching.
4.3
Tempering Characteristics
The tempering curves were plotted from an average of at least five Vickers hardness
measurements. The macro-Vickers hardness tests were made on tempered samples using a
test load of 10kg. Samples from the HCrLMo and C-Mn steel were austenitised at 900
then quenched in water. Tempering was done in increments of 100
100
and
from a temperature of
to 600 . After tempering, the samples were air cooled and the tempering curves were
plotted as Vickers hardness against tempering temperature and are presented in Figures 4.3.1
and 4.3.2.
60
© University of Pretoria
C-Mn steel tempering curve
400
AQ = 357.50HV
Hardness (HV)
350
300
250
200
150
AQ: As Quenched
Hardness
100
50
0
0
100
200
300
400
500
600
700
Temperature (ᵒC)
(a)
Hardness (HV)
HCrLMo steel tempering curve
500
450
400
350
300
250
200
150
100
50
0
AQ = 443.24HV
AQ: As Quenched
Hardness
0
100
200
300
400
500
600
700
Temperature (ᵒC)
(b)
Figure 4.3.1 Tempering curves for (a) the C-Mn steel and (b) the HCrLMo steel. The samples
were tempered for 30mins at temperature after soaking at 900 and water quenching.
Macro Vickers hardness load: 10 kg with 5 readings per data point.
The C-Mn steel tempering curve in Figure 4.3.1 (a) showed a continual decrease in hardness
as the tempering temperature was increased from 100
61
© University of Pretoria
to 600 . The first tempering
treatment at 100
produced only a slight reduction in hardness while the largest hardness
decrease (72.1HV) was found after tempering at 500 .
The highest hardness for both steels was in the as-quenched state. For the HCrLMo steel, a
hardness reduction of 49HV from the as-quenched hardness (443HV) occurred on tempering
at 100 . Thereafter an increase in hardness by 19.4HV was found on tempering at 200 .
This resulted in a slight peak in the tempering curve of Figure 4.3.1 (b). The results are
discussed in section 5.1.
4.4
Continuous Cooling Transformation Diagrams
4.4.1 Preliminary tests
4.4.1.1 Determination of Austenitising Temperatures.
In order to establish the temperature at which homogeneous austenite is formed, dilatometric
samples machined from the C-Mn and HCrLMo steels were heated to 1100
and 1200
respectively in a Bähr dilatometer. The high temperatures were used so that the specimen
completely transforms to austenite and the linear region above the Ac3 temperature in which
the samples have completely transformed to austenite, could be identified. Figures 4.4.1 and
4.4.2 show the strain – temperature graphs obtained for the C-Mn and HCrLMo steels
respectively.
62
© University of Pretoria
0.5⁰C/s heat to 1100⁰C
140
120
Ac3
Dilatation ΔL/L
100
80
60
40
20
0
-20 0
200
400
-40
600
800
1000
1200
Temperature (⁰C)
Figure 4.4.1 Dilatometric signal obtained for the C-Mn steel sample which was heated at
0.5 /s to 1100 before slow cooling. The critical temperatures found were Ac1 = 722.3
and Ac3 = 865.3 .
180
1.5⁰C/s heat to 1200⁰C
160
Ac3
140
Dilatation ΔL/L
120
100
80
60
40
20
0
-20
0
200
400
600
800
Temperature/ ⁰C
1000
1200
1400
Figure 4.4.2 Dilatometric signal for the HCrLMo steel sample heated at 1.5 /s to 1200
before slow cooling. Homogeneous austenite was obtained at 931.9ᵒC, where the expansion
became linear.
The experimental Ac3 temperature obtained for the C-Mn steel was 865.3
HCrLMo was 931.9 .
63
© University of Pretoria
and that for the
In addition to the experimentally obtained temperatures, the Ac3 temperatures were modelled
according to the Kasatkin [38] equation which is given by:
Ac3 = 912 – 370C – 27.4Mn +27.3Si – 6.35Cr – 32.7Ni + 95.2V + 190Ti + 72.0 Al + 64.5Nb
+ 5.57W + 332S +276P + 485N – 900B + 16.2C Mn + 32.3C Si + 15.4C Cr + 48.0C Ni +
4.32Si Cr – 17.3 Si Mo – 18.6 Si Ni + 4.80Mn Ni + 40.5 Mo V + 174C2 + 2.46Mn2 – 6.86 Si2
+0.322Cr2 + 9.90Mo2 + 1.24Ni2 – 60.2V2
The calculated and experimental tempertures are given in Table 4.2 below for the two steels.
Table 4.2 Experimental and calculated Ac3 temperatures.
Calculated Ac3(oC)
Steel
Experimental Ac3 (oC)
C-Mn
843.7
865.3
HCrLMo
847.2
931.9
The experimentally obtained Ac3 temperature for the HCrLMo steel was 82
higher than the
calculated value, while that for the C-Mn was higher by 22 . Typically, austenitising
temperatures are taken 50
above the Ac3 temperature. Therefore a suitable austenitising
temperature determined for the C-Mn steel was 916 , an approximate 50
above the
experimental Ac3 temperature. The HCrLMo steel was subjected to further scrutiny since
austenitising 50
above the experimental Ac3 seemed rather high and was likely to result in
grain growth.
Two samples from the HCrLMo steel were thus heated to 935
and 955
respectively
before cooling in the dilatometer. The change in Ms temperature was considered due to the
impact of complete carbide dissolution on transformation temperatures. A fully austenitised
material with all of its alloying elements in solid solution will give a lower Ms temperature on
quenching and the latter can then be used as an indication of the effectiveness of the solution
treatment.
64
© University of Pretoria
120
100
100
80
80
60
60
40
40
20
20
0
0
Dilatation (soaked at 955oC)
120
-20
Dilatation (soaked at 935oC)
955oC
935oC
-20
0
200
400
600
800
1000
o
Temperature( C)
Figure 4.4.3 Superimposed dilatometric signals of austenitisation of HCrLMo steel samples
at 955 and 935 and cooled at 10 /s.
Figure 4.4.3 shows dilatometric signals of samples of the HCrLMo steel austenitised at 955
and 935
respectively. No difference was found in the Ms temperature of these two samples.
This indicated that soaking of the HCrLMo steel samples at 955
did not induce any further
dissolution of carbides than soaking them at 935 .
120
100
100
80
80
60
60
40
40
20
20
0
0
-20
o
120
Dilatation (soaked at 900 C)
Dilatation (soaked at 935oC)
935oC
900oC
-20
0
200
400
600
800
1000
o
Temperature ( C)
Figure 4.4.4 Superimposed dilatometric signals of HCrLMo steel samples austenitised at
900 and 935 respectively and cooled at 18 /s.
The change in Ms temperature was also considered between samples austenitised at 900
and 935 . The Ms obtained for the sample austenitised at 900
65
© University of Pretoria
was surprisingly lower than
that for the sample soaked at 930
(undissolved carbides at 900
should lead to a higher
Ms). However, the difference was small.
Next, the effect of austenitising temperature on grain growth was considered. The grain sizes
of samples solution treated at 900
and 935
respectively were measured on optical
micrographs using the linear intercept method. At 935
the mean linear intercept was
26.7µm and 20.8µm at 900 . Thus a rise in austenitisng temperature by 35
resulted in an
increase in grain size by 5.9 microns. It is known that austenite grain size can affect
hardenability and thus cause a shift in the CCT diagram.
However, it was found that austenitising the HCrLMo samples at 930
did not shift the CCT
diagram significantly if compared to a soaking temperature of 900 . Table 4.3 shows the
transformation temperatures obtained after austenitising HCrLMo samples at 900
930
and
respectively and cooling at the same rates. The transformation temperatures from 2 to
18 /s are the temperatures at which bainite began to form whilst those at 30 /s represented
the Ms. The relatively small difference between the transformation temperatures implied that
CCT curves constructed from the two austenitising temperatures would be similar.
Table 4.3 Transformation temperatures obtained after austenitising HCrLMo steel samples at
900 and 930 and cooling at different rates.
Cooling
rate
(oC/s)
2
5
10
18
30
Transformation
Transformation
Absolute
temperature
temperature
difference
after soaking at after soaking at (oC)
930oC
900oC
558.0
572.1
14.1
519.8
517.0
2.8
476.0
472.3
3.7
441.6
439.0
2.6
434.8
421.0
13.8
4.4.1.2 Carbide and nitride dissolution
Thermocalc simulations of the dissolution temperature of AlN show that at all the
austenitising temperatures used for the HCrLMo steel these precipitates are not dissolved.
However, some amount of dissolution for the C-Mn steel is expected in the region of 900
although complete dissolution of AlN is only attained at about 1000 .
66
© University of Pretoria
(a)
(b)
Figure 4.4.5 Thermocalc graphs of temperature versus AlN content for (a) the HCrLMo steel
and (b) C-Mn steel. NMP = mole fraction.
A high austenitising temperature of 1000
at which complete dissolution of AlN would be
achieved according to Figure 4.4.5, also promotes excessive grain growth. This was therefore
ruled out as a possible austenitising temperature.
67
© University of Pretoria
Optical micrographs of the samples which were austenitised at 930 , 935
and 955
had
dark spots remaining in the microstructures that at first appeared to be carbides. The
micrograph of such a sample which was austenitised at 955
and quenched to form
martensite at 60 /s is shown in Figure 4.4.6.
Figure 4.4.6 Microstructure of a HCrLMo steel sample austenitised at 955 and soaked for
20 minutes before cooling at 60 /s. The spots arrowed were found to be etch pits and not
second phases.
The suspicion that the spots are carbides was verified by SEM-EDS examination of the
sample. The dark spots had the same approximate composition as the matrix and when the
sample was viewed in backscatter mode, no compositional contrast was observed. Therefore
what appeared to be carbides or nitrides appear may have been to be etch pits. Therefore, it
was ascertained that carbides were in solution at the solution treatment temperature of 955 .
4.4.1.3 Tests for Ac1 and Ac3 temperatures
In this section, attempts at finding the equilibrium Ac1 and Ac3 temperatures of the HCrLMo
steel were made by using slow cooling rates after austenitising.
The first trial run was done by heating a dilatometer sample to 900
followed by cooling
over the duration of 1 hour at a rate of 0.22 /s to about 100 . The temperatures were
68
© University of Pretoria
measured from dilatometric signals by reading off the points of inflection on cooling in
Figure 4.4.8.
1hr cool at 0.22⁰C/s
120
Dilatation ΔL/L
100
80
60
40
20
0
-20
0
200
400
600
Temperature (oC)
800
1000
(a)
Ferrite
Pearlite
100μm
(b)
Figure 4.4.7 (a) Dilatometric signal of the HCrLMo steel using a 0.22 /s cooling rate after
austenitising at 900 . The Ar3 and Ar1 temperatures recorded were 621 and 452
respectively.(b) Optical micrograph showing the microstructure obtained.
The Ar3 and Ar1 temperatures of 621
and 452
respectively were lower than expected.
Bands of ferrite and pearlite were obtained with possibly some martensite or bainite in the
grey areas of the micrograph in Figure 4.4.7. To obtain equilibrium temperatures, a lower
69
© University of Pretoria
cooling rate of 0.07 /s was then used. The heating rate used for both the 0.22 /s and the
0.07 /s cooling rates was 1.5 /s.
120
3hr cool at 0.07⁰C/s
100
Dilatation ΔL/L
80
60
40
20
0
0
200
-20
400
600
800
1000
Temperature (oC)
(a)
BB
F
P
100μm
(b)
Figure 4.4.8 (a) The graph and (b) microstructure of the decomposition product in the steel
HCrLMo after slow cooling at 0.07 /s. The product is a mixture of ferrite (F), pearlite (P)
and bainite (B).
The temperatures of the transformation obtained in this case were Ar3 = 553
and Ar1 =
404 . The result was still lower than was previously obtained and it seems that the slower
degree of cooling did not result in the equilibrium conditions as was expected. Before the
70
© University of Pretoria
expansion at 553 , a slight deviation from linearity on cooling was observed at 836
as
indicated by the arrow in the graph in Figure 4.4.8. The corresponding micrograph shows
three distinct phases. In accordance with this, it was deduced that the ferrite formed at the
higher temperature of 836 , although no contraction was detected at this point and that
553
should not be regarded as the Ar3 temperature. It also explains why a lower „Ar3‟ was
found at this cooling rate than at the higher cooling rate of 0.22 /s.
The sample heating rate was reconsidered since an equilibrium heating is more likely to
produce a more homogeneous austenite and the slow cooling that follows will thus give
transformation temperatures within a more acceptable range. A sample of the steel HCrLMo
was thus heated at 0.246 /s to 900
then cooled at the rate previously employed of
0.07 /s. The result is given below.
100
0.07⁰C/s cool after heating at
0.246oC/s
Dilatation ΔL/L
80
60
40
20
0
0
200
-20
400
600
800
1000
Temperature (⁰C)
Figure 4.4.9 Graph of a sample of the steel HCrLMo with a 0.074ᵒC/s cooling rate showing
two transformations at Ar3 = 813.5 , Ar1 = 656.9 , Bs = 561.8 and Bf = 398.5 .
The Ar3 temperature of 813.5
obtained in this instance was much higher than previously
found. Two distinct transformations were observed on the dilatometric signal in Figure 4.4.9.
The first transformation as indicated by the inflection at the higher transformation
temperature of 813.5
was a ferrite transformation. The corresponding micrograph in Figure
4.4.10 confirmed this. The second inflection at the lower transformation temperature of
561.8
was where bainite began to form from the remaining austenite. These results confirm
71
© University of Pretoria
the importance of slow heating in addition to slow cooling when searching for equilibrium
critical temperatures.
F
P
B
50μm
Figure 4.4.70 Microstructure of sample of the steel HCrLMo slow heated at 0.246 /s and
cooled at a cooling rate of 0.074 /s.
Since there was a second transformation after the pearlite-ferrite formed, this showed that not
all the austenite transformed to pearlite, as would be expected in an equilibrium cool.
Therefore a final attempt at obtaining the Ar1 and Ar3 temperatures approaching Ac1 and Ac3
equilibrium temperatures respectively was made by cooling at a rate of 0.028 /s. Figure
4.4.11 shows the dilatometric signal with the transformation temperatures obtained.
72
© University of Pretoria
8hr cool at 0.028⁰C/s
120
Dilatation ΔL/L
100
80
60
40
20
0
-20 0
200
400
600
Temperature( oC)
800
1000
Figure 4.4.11 Graph of a sample of the steel HCrLMo with a cooling rate of 0.02 /s and
where the critical transformation temperatures were found at Ar3 = 829.7 and Ar1 =
682.4 .
The Ar temperatures found after cooling a sample at 0.028 /s, which was the lowest cooling
rate used, were found to be closest to equilibrium and thus the final critical temperatures were
taken as Ac3 = 829.7
and Ac1 = 682.4 .
Note that the dilatometer cannot distinguish between pro-eutectoid ferrite and pearlite
formation hence a single expansion is recorded where the ferrite forms below the Ac3
temperature [64].
4.4.2 Partial CCT diagrams
The partial CCT diagrams of the HCrLMo and HMoLCr steels were constructed by
austenitising dilatometer samples at 900
and 930
respectively, followed by continuous
cooling at various linear cooling rates to room temperature. The same followed for the C-Mn
steel which was austenitised at 916 . Figure 4.4.12 shows the CCT diagrams that were
constructed from the transformation temperatures given in the dilatometry curves. The
cooling rates plotted on a logarithmic scale show a slight curvature not visible on the figure.
73
© University of Pretoria
o
Cooling rates C/s
2
HCrLMo steel
5
8
10
14
18
24
26.14
30
900
700
o
Temperature ( C)
800
600
500
B
400
300
200
100
1
10
100
log time (s)
(a)
o
Cooling rate C/s
900
HMoLCr steel
30
14
5
3.5
2
0.7
800
o
Temperature ( C)
700
600
500
B
400
300
200
100
10
100
log time (s)
(b)
74
© University of Pretoria
1000
1000
o
Cooling rates ( C/s)
900
C-Mn steel
F+P
700
o
Temperature ( C)
800
600
500
B
------ 2
7
10
20
24
28
33
40
400
300
200
100
1
10
100
log time (s)
(c)Figure 4.4.12 Partial CCT diagrams for (a) the HCrLMo steel (b) the HMoLCr steel and
(c) the C-Mn steel. F represents ferrite, P pearlite and B bainite.
The Ms temperatures of the HCrLMo steel were found from an average of four cooling rates
and were 423 , 432 , 430
and 421
average Ms of 426.5 .
A critical cooling rate of 5 /s to form only martensite was found for the HMoLCr steel
while that for the HCrLMo steel of 30 /s was found. This reflects the higher hardenability
imparted by the high Mo content in the steel HMoLCr if compared to that of a high Cr
content in the steel HCrLMo. The change in hardness with cooling rate is given in Table 4.4.
Since the partial CCT consisted of only a bainite nose, the decreasing hardness with lower
cooling rates was due to the bainitic ferrite which is a phase that is much softer than
martensite. Cooling rates of 14 /s and 5 /s for the HMoLCr steel produced martensite. A
larger quantity of bainite was produced when a sample was cooled at 0.7 /s, this gave a
lower hardness. The same followed for the HCrLMo steel, where the slowest cool of 2 /s
passed through a wider region of the bainite nose and resulted in more bainite in the
microstructure.
75
© University of Pretoria
Table 4.4 HMoLCr steel micro-Vickers hardness with cooling rate. Load = 300 gf.
Cooling rate /ᵒCs-1
14
5
2
0.7
Hardness/HV
391.3
377.6
313.4
283.7
Table 4.5 HCrLMo steel micro-Vickers hardness with cooling rate. Load = 100 gf.
Cooling rate/ᵒCs-1
100
30
26
24
14
10
8
5
2
4.5
Hardness/HV
381.9
402.7
409.7
387.8
391.5
376.0
360.8
335.7
272.4
Isothermal transformations
4.5.1 Preliminary tests
A preliminary test was conducted on a HCrLMo steel dilatometer sample to show the
feasibility of producing a bainite-martensite structure. In this trial run the sample was heated
to 900
and solution treated at this temperature for 15 minutes. The sample was then cooled
to 443
at 5 /s and subsequently quenched in He at 60 /s. The cooling rate was chosen to
enter the bainite nose without any isothermal incubation time.
When analysed individually, the 5 /s cool is slow enough to allow bainite to form and the
60
is a fast quench that results in martensite forming from any remaining austenite.
Therefore, cooling a sample at the two cooling rates combined in the sequence given was
expected to result in a microstructure containing bainite and martensite. The dilatometric
curve in Figure 4.5.1 was used to verify this.
76
© University of Pretoria
5⁰C/s and 60⁰C/s cool
120
Dilatation ΔL/L
100
80
60
40
Ms'
20
0
-20 0
200
400
600
800
1000
Bs
-40
Temperature ( ⁰C)
Figure 4.5.1 Double quench on steel HCrLMo, cooling rate interrupted at 443 , Bs = 520 .
Notice that the two transformations occurred as expected. The temperature of the first
transformation corresponded to the Bs temperature found on quenching at 5 /s. The arrow at
the point labelled Ms’ indicates the start of the second transformation at which martensite
forms from the untransformed austenite. Microstructural evidence in Figure 4.5.2 also
showed that the two transformation products were indeed bainite and martensite.
M
M
B
(a)
20μm
77
© University of Pretoria
B
M
M
B
(b)
50μm
Figure 4.5.2 Optical micrographs of the mixed upper bainite-martensite microstructures in
steel HCrLMo obtained after continuous cooling at 5 /s and 60 /s after (a) a Nital etch
and (b) a 10% SMB etch. B is bainite and M is martensite.
The sample was first etched in a 2% Nital solution to reveal the microstructure. The same
sample was polished and then etched in a 10% Sodium Metabisulphite (SMB) solution. This
solution is a colour etchant that gives bainite a blue appearance and martensite a brown
appearance when viewed in an optical microscope under polarised light.
4.5.2 Measurement of the volume fraction of bainite.
Isothermal heat treatments to form mixed structures of bainite and martensite were conducted
and the volume fraction of bainite was calculated from the expansion line of the dilatometric
signal. Figure 4.5.3 shows the dilatometric curve when dual phase bainite/martensite is
obtained on cooling.
78
© University of Pretoria
Figure 4.5.3 Dilatometric signal of isothermal transformation on a sample of the steel
to 484 and held for 20 s to form
HCrLMo austenitised at 900 then cooled at 8
67.8% bainite before cooling at 30
to form martensite. The strains measured at e1, e2
and ex were used to calculate the bainite volume fractions.
The magnitude of expansion of the dilatometric signal at the isothermal holding temperature
was a measure of the amount of bainite that formed and was taken as a fraction of the
maximum possible expansion of the sample at this temperature. This was done in accordance
with the standard ASTM A1033-10. The formula for calculating the volume fraction of
bainite is essentially derived from the lever rule:
Volume fraction =
……………………………………..………………………eq 4.1
Where ei is the dilatation ΔL/L for:
i=2 at the maximum possible expansion at isothermal treatment temperature,
i=1 at the extrapolated cooling curve and
i=x at the point where the transformation of austenite to bainite is terminated by the quench.
79
© University of Pretoria
4.6
Johnson Mehl Avrami Kolmogorov (JMAK) kinetics
4.6.1 Sigmoidal curves and Avrami Exponents
The Johnson Mehl Avrami approach was applied to the isothermal transformation of
austenite to bainite in the HCrLMo and HMoLCr steels. Upper bainite was formed in the
steels by isothermal heat treatment at 484
for the HCrLMo steel samples and at 498
for
the HMoLCr steel samples. The temperatures used resulted in similar estimated
undercoolings of 132
and 134
below the Bs temperatures. Similar undercoolings below
the Bs were used to ensure that the estimated driving forces for the bainite transformation
were approximately similar by assuming an equivalent enthalpy of formation in the two
steels. The driving force to form bainite was calculated in Thermo-Calc from the difference
in free energies of the system at the austenitising temperatures and at the To temperature. The
difference in free energy for the HCrLMo steel was -13409.1J/mol and for the HMoLCr steel
-15274J/mol. The driving forces for the formation of bainite in the two steels differed by
about 14% and were, therefore, relatively similar.
The volume fractions obtained after varying the isothermal holding time were plotted as a
function of isothermal transformation time.
80
© University of Pretoria
1.0
Volume fraction bainite
0.8
HCrLMo
0.6
HMoLCr
0.4
0.2
t50
0.0
0
t50
50
100
150
200
250
Time (s)
Figure 4.6.1 Volume fractions of bainite plotted as a function of time at isothermal
temperatures for the two steels HCrLMo and HMoLCr. The time taken for 50% to form is
shown as t50.
The sigmoidal curves for the two steels give some indication of the growth kinetics under the
conditions employed. Figure 4.6.1 shows that the HMoLCr steel took a significantly longer
time to transform to a fully bainitic structure than the HCrLMo steel. The transformation was
halfway complete in 47 seconds in the HMoLCr steel whilst it took a mere 14 seconds in the
HCrLMo steel. This shows that the nucleation and growth rate of upper bainite was
significantly higher in the HCrLMo steel than in the HMoLCr steel. The results are further
discussed in section 5.4.
4.6.2 Depression of Ms temperatures
Using the above dilatometric curves, the progressive lowering of the Ms temperature of the
remaining austenite could be followed as the volume fraction of bainite increased.
81
© University of Pretoria
440
HCrLMo
435
430
HMoLCr
o
Ms temperature ( C)
425
420
415
410
405
400
395
390
385
0
20
40
60
80
100
120
140
160
180
Isothermal holding time (s)
Figure 4.6.2 Graph of Ms temperatures with isothermal holding time for both steels.
A decrease in the Ms temperatures as increasing amounts of bainite were formed was evident
in both steels. Nash et al observed the same reduction of Ms in a 0.58%C – 0.82%Mn –
0.04%Si – 1.9%Ni – 0.55%Cr – 0.77%Mo – 0.028%Al and 0.82%C – 0.86%Mn – 0.04%Si –
1.9%Ni – 0.55%Cr – 0.77%Mo – 0.031%Al (in wt%) steels. The Ms is dependent on carbon
content and thus a decrease in Ms for both steels suggests that the carbon content in the
untransformed austenite increased as more bainite formed [26, 65], a known feature of the
bainite transformation as carbon is rejected from the ferritic bainite into the austenite.
82
© University of Pretoria
440
HCrLMo
HMoLCr
430
420
o
Ms ( C)
410
400
390
380
370
0
10
20
30
40
50
60
70
80
90
100
Amount of bainite (%)
Figure 4.6.3 Graph of the effect of increasing volume fractions of bainite on the measured Ms
temperatures.
The data in Figure 4.6.3 shows the effect of increasing amounts of bainite on the Ms
temperatures of the untransformed austenite. There is an overlap of the Ms temperatures in
the HCrLMo and HMoLCr steels which decrease as more bainite is formed. The dependence
of the Ms temperatures on the volume fraction of bainite rather than the chemical
composition, already points to a volume-driven effect, whereby the accumulation of rejected
carbon in the untransformed austenite is the same for both steels in spite of large differences
in the rate of transformation.
4.6.3 Electrolytic extraction of precipitates
The differing rates of decrease in the Ms temperature as can be seen from the graph of Figure
4.6.2, suggests that the rate of carbon rejection into the untransformed austenite was higher in
the HCrLMo steel samples than in the HMoLCr samples. Together with the higher rates of
formation of bainite (see Figure 4.6.1.) in the HCrLMo steel, this implies that bainite
formation in the HMoLCr steel is slower due to the influence of the alloying elements Cr and
83
© University of Pretoria
Mo on the reaction. One possible cause of slower growth kinetics is the formation of
cementite containing Cr or Mo (M3C) rather than pure Fe3C.
To determine if the upper bainite carbide is cementite containing some alloying elements,
carbon extraction replicas were made from the HMoLCr steel sample consisting of a
completely upper bainitic microstructure. The experiment was unsuccessful as AlN and TiN
precipitates rather than cementite were extracted using this technique. The failure to extract
the cementite using carbon extraction replicas was likely the result of insufficient etching
depth. Figure 4.6.4 shows TEM images of the extracted particles.
100nm
500nm
Figure
4.6.4 TEM images taken from carbon extraction replicas of HCrLMo steel with a 100%
bainitic structure.
(a)
84
© University of Pretoria
(b)
Figure 4.6.5 EDS spectra of (a) AlN and (b) TiN precipitates.
The high levels of copper detected in the spectra of Figure 4.6.5 are due to the copper TEM
grids on which the specimens were mounted while the high carbon levels were detected from
the carbon film which held the precipitates.
An alternative method was then used to extract larger volumes of carbide particles by
electrolytic extraction. As before, a sample in which 100% upper bainite was formed was
used. The precipitates were examined in a TEM in scanning mode. Images of particles were
taken at randomly selected areas where their corresponding element maps and EDS spectra
gave their chemical composition. Many of the larger precipitates extracted were TiN and
AlN.
The TEM micrograph in Figure 4.6.6 shows a blocky precipitate which was likely a Titanium
nitride particle. EDS maps of the precipitate had a high concentration of Titanium and
Nitrogen. In addition to this, the spectrum affirmed the high Titanium and Nitrogen levels
shown in the maps.
85
© University of Pretoria
(a)
(b)
(c)
(d)
(d)
86
© University of Pretoria
(e)
Figure 4.6.6 (a) TEM image of TiN particle. EDS maps of theTiN particle showing the
distributions of (b) Ti, (c) N and (d) C. The spectrum of the particle is shown in (e).
(a)
(a)
(b)
(c)
1μm
87
© University of Pretoria
1μm
(d)
1μm
(e)
1μm
(f)
Figure 4.6.7 (a) TEM image of a Mo rich particle and EDS maps showing the distributions of
(b) Mo (c) Fe (d) C (e) Cr and (f) is the spectrum of the particle.
The TEM micrograph in Figure 4.6.7 of round particles had a considerable concentration of
Mo and Fe. The Mo and Fe EDS maps distinctively delineate the particles and their peaks on
the spectrum are prominent. Minute amounts of Cr were also detected in these smaller
particles as shown on the EDS map and the spectrum in Figure 4.6.7. The chemical
compositions suggest that the particles are iron carbides with some Mo, most likely with a
M3C composition.
88
© University of Pretoria
(a)
(b)
1μm
(c)
1μm
(d)
1μm
(e)
1μm
89
© University of Pretoria
(f)
Figure 4.6.8 (a) TEM image of a Cr rich particle and EDS maps showing the distributions of
(b) C (c) Cr (d) Mo and (e) Fe. The spectrum of the particle is shown in (f).
The particle analysed in Figure 4.6.8 shows the presence of Cr, Mo and Fe. The evidence
from the EDS maps and spectra of the extracted particles suggests that the cementite
contained some Cr and Mo elements.
4.7
Instrumented Charpy impact tests
4.7.1 Test parameters
4.7.1.1 Effect of tempering on hardness
Specimens cut from the HCrLMo and HMoLCr steel plates were machined into Charpy
samples of dimension 10x10x55mm. The samples were heat treated in a thermomechanical
simulator Gleeble 1500DTM to produce various amounts of bainite in a martensitic matrix.
The HCrLMo samples were cooled from the austenitising temperature at a rate of 8 /s to
484 , and the HMoLCr samples were cooled at 2 /s to 498 . The cooling rates were
chosen to enter the bainite C-curve without any isothermal incubation time. After machining
the Charpy groove, the samples were impact tested in an instrumented Instron Dynatup 9210.
The isothermal holding times to produce various amounts of bainite were found from the
sigmoidal curves. Figure 4.7.1 shows the extrapolations from the selected amounts of bainite.
The selected amounts were 10%, 25%, 60%, 75% and 90% bainite. Specimens with 100%
90
© University of Pretoria
martensite were produced by fast quenching from a muffle furnace into water and those with
100% bainite were produced in the muffle furnace after 20 minutes of isothermal holding.
HMoLCr sigmoidal curve
Volume fraction bainite
1.0
t = 100.13, Vf = 0.90
0.8
t = 70.43, Vf = 0.75
0.6
t = 54.27, Vf = 0.60
0.4
t = 25.88, Vf = 0.25
0.2
t = 11.03, Vf = 0.10
0.0
0
50
100
150
200
250
Time(s)
(a)
HCrLMo Sigmoidal curve
Volume fraction bainite
1.0
t = 33.74, Vf = 0.90
0.8
t = 22.9, Vf = 0.75
0.6
t = 17.09, Vf = 0.60
0.4
t = 7.41, Vf = 0.25
0.2
t = 3.97, Vf = 0.10
0.0
0
10
20
30
40
50
Time (s)
(b)
Figure 4.7.1 Sigmoidal plots and extrapolations used for heat treatment durations (a) in the
HMoLCr and (b) HCrLMo steel Charpy specimens.
To ensure that the effect of bainite on toughness was isolated from that of the martensite
which is brittle in the untempered state, the micro-hardness of the two phases was equalised
91
© University of Pretoria
by tempering. The hardness of untempered martensite was 419.3HV and that of the
untempered bainite was 275.8HV. Tempering the mixed martensite-bainite samples at 500
for 30 minutes reduced the hardness of martensite to a level sufficient to eliminate the
brittleness of martensite that in the untempered state would have obscured the effect of
bainite. The bainite remained unaffected by tempering under these conditions. According to
Bhadeshia [27], bainite has a greater degree of stability over a longer tempering period during
tempering than does martensite. Figure 4.7.2 shows the effect of tempering on the Vickers
hardness in a 0.14 wt% C steel containing bainite.
Figure 4.7.2 Change in hardness with tempering parameter (x10 3) for bainitic steel. T is the
absolute temperature and t is time in hours. The dotted line represents the tempering
parameter used in tempering the martensite in both of the steels HCrLMo and HMoLCr at
500 for 30 minutes to lower its hardness to that of the upper bainite [27].
When bainite forms much of the carbon is precipitated as cementite, there is therefore little
carbon in solid solution in the ferrite, as opposed to quenched martensite. At the relatively
high temperatures at which upper bainite forms, the microstructure undergoes recovery to a
greater extent than does autotempered martensite. Tempered bainite therefore undergoes little
recovery and negligible change in the morphology and density of the carbides. In contrast to
this, the carbon in solid solution in martensite precipitates as fine carbides and the strength
92
© University of Pretoria
decreases. The coarsening rate of these very fine carbides after nucleation is thus higher in
martensite than the already coarse cementite in upper bainite. The carbides in upper bainite in
the untempered state are already coarse due to the inherent tempering that leads to growth
during its formation. The tempering parameter calculated for the steels in this study which
have a carbon content of 0.17 wt% was found to be 14 912 for 484
15 188 for 498
(757K) tempering and
(771K) tempering for a duration of 30 minutes. These values, with
reference to Figure 4.7.1.2, show no change in the hardness of bainite when it is as-quenched
versus when it is tempered. Therefore, tempering upper bainite did not produce any change in
its toughness.
On the contrary, there is a definite increase in the toughness of martensite after it is tempered.
An impact test on tempered and untempered Charpy samples of martensite in the steel
HCrLMo was conducted. The Charpy impact energy of the tempered martensite sample was
69.7J and that of the untempered martensite sample was 28.3J of energy. Thus tempering
mixed upper bainite-martensite samples isolated the effect of bainite on toughness by
eliminating the brittleness of the martensite.
4.7.1.2 Effect of the notch position relative to the rolling direction
As the steel plates from which the Charpy samples were machined were hot rolled during
their production, it was necessary to examine the effect on the impact toughness of the
position of the V-notch relative to the rolling direction. This would serve to eliminate the
effect of banding in the microstructure on toughness.
Two Charpy samples were machined from the HMoLCr steel plate in the rolling direction.
One sample was notched on a plane transverse to the rolling direction, i.e. in the so-called LT orientation and the other was notched on an orthogonal plane also transverse to the rolling
direction or in the L-S orientation (see Figure 4.2.1). The specimens were then fractured and
the impact energies obtained were 71.6 J and 67.7 J respectively. The difference of 3.9 J was
not significant and it was thus concluded that the effect of microstructural banding was not
significant in this steel if at all present. In addition to this, the hardness profiles taken along
the two transverse directions were relatively constant, as Figure 4.7.3 below shows.
93
© University of Pretoria
400
350
Hardness (HV10)
300
250
200
150
100
notch
50
HMoLCr steel
0
0
2
4
6
8
10
Distance (mm)
(a)
400
350
Hardness (HV10)
300
250
200
150
100
HMoLCr steel
50
0
0
2
4
6
8
10
Distance (mm)
(b)
Figure 4.7.3 Hardness profiles in the HMoLCr steel across a Charpy sample taken (a)
parallel to the Charpy V-notch and (b) transverse to the notch.
4.7.1.3
Distribution of bainite in martensite
The heat treatment of the Charpy samples was conducted in a Gleeble 1500DTM
thermomechanical simulator which heats samples by direct electric current. The thermal
94
© University of Pretoria
cycles used to obtain the various amounts of bainite in the Charpy samples were based on
measurements from a Bӓhr dilatometer which uses induction heating, where the small sample
size of 5 mm diameter enables the thermal cycle to be followed or adhered to with a high
degree of response precision. During direct electrical heating of the Charpy samples where
the electric current tends to concentrate more on the surface, the sample is heated from the
surface inwards and a temperature gradient may result between the surface and the centre of
the sample within the plane on which the thermocouple is attached. The thermal gradient may
be tolerable for some purposes, however for the purpose of this experimental work where the
amount of bainite transformed had strict time-temperature dependence, it was important to
understand the effects of any likely thermal gradient. This would ensure that a homogeneous
distribution of bainite in martensite in the plane below the Charpy groove is achieved. To
obtain some idea of the distribution of these phases, the hardness profiles were taken across a
Charpy specimen of the steel HMoLCr in which 50% bainite had been formed. The hardness
profiles were taken along orthogonal, intersecting paths and the results are shown in Figure
4.7.4.
Some periodic scatter was observed, which was expected from hardness indentations on the
softer bainite which would give lower readings and higher readings on the harder martensite.
Apart from the scatter, no definite trend was observed. It was concluded that the bainite was
reasonably evenly distributed in the martensitic matrix.
350
300
Hardness (HV)
250
200
150
100
V-notch
50
HMoLCr steel
0
0
2
4
6
Distance (mm)
95
© University of Pretoria
8
10
(a)
400
350
Hardness (HV)
300
250
200
150
100
HMoLCr steel
50
0
0
2
4
6
8
10
Distance (mm)
(b)
Figure 4.7.4 Hardness profiles measured (a) parallel to the Charpy V-notch and (b)
transverse to the V-notch across a HMoLCr steel Charpy specimen heat treated to form 50%
upper bainite.
4.7.1.4 Temperature gradients
PTemp
TC3
1000
Solution treatment
Cooling to enter the
bainite nose directly
o
Temperature ( C)
800
600
Bainite formation
400
He quench
200
0
0
500
1000
1500
Time (s)
(a)
96
© University of Pretoria
2000
PTemp
TC3
600
400
o
Temperature ( C)
500
300
200
100
1700
1800
1900
2000
Time (s)
(b)
Figure 4.7.5 (a) Time-Temperature profile on Gleeble of the heat treatment to produce
bainite (b) a magnified view of the quenching process. PTemp is the programme temperature
and TC3 the control temperature.
Figure 4.7.5 (a) shows the time – temperature profile used on the Charpy samples. There was
a discrepancy between the temperature of the thermocouple to which the sample was welded
(TC3) and the program temperature (PTemp) but only during the final quench. A magnified
section of the He quench is shown in Figure 4.7.5 (b) which shows that the final quench
temperature of 150
was reached at an average cooling rate of about 9 /s. Since the Ms
temperatures are significantly higher than the quench temperature, it can be ascertained that
the martensite volume fraction attained was within the expected range.
4.7.2 Effect of the amount of bainite on the impact energy
The total Charpy impact energy was plotted as a function of the amount of bainite in Figure
4.7.6.
97
© University of Pretoria
160
150
140
130
120
Total energy (J)
110
100
90
80
70
60
50
40
30
HMoLCr steel
20
10
0
0
20
40
60
80
100
Amount of bainite(%)
(a)
150
140
130
120
Total energy (J)
110
100
90
80
70
60
50
40
HCrLMo steel
30
20
10
0
0
20
40
60
80
100
Amount of bainite(%)
(b)
Figure 4.7.6 Graphs of the total energy absorbed as a function of the amount of bainite in the
Charpy samples of (a) steel HMoLCr and (b) steel HCrLMo.
The trend from Figure 4.7.6 (a) for the HMoLCr steel shows that the total absorbed impact
energy already decreased when only 10% bainite was present in the microstructure but there
98
© University of Pretoria
after in the range of 10 - 80% bainite, the impact energy seems to level out until a second
decrease is found at 100% bainite.
The HCrLMo steel, on the other hand, showed no significant variation in the absorbed energy
throughout the entire range of mixed structures, although the fully bainitic samples had the
lowest total absorbed energy.
4.7.3 Crack initiation and propagation energies
The total energy absorbed in an impact fracture consists of the energy required to initiate a
crack and the energy to propagate the crack. From the instrumented Charpy impact data the
initiation and propagation energies were obtained from the individual load-deflection curves
as typically shown in Figure 4.7.7.
40
80
60
20
40
10
Average energy (J)
Average load (kN)
30
20
(I) (II) (III)
0
0
0
1
2
3
4
5
6
Time (ms)
Figure 4.7.7 Graphical output from instrumented impact tests showing different regions of
fracture on load and energy curves plotted as a function of time.Taken from a HMoLCr steel
sample with 50% bainite.
The arrows on the peak load and the corresponding energy are indicative of the energy
expenditure in initiating a crack, since beyond this point, a decrease in load occurs as the
energy is dissipated in crack movement. The regions (I), (II) and (III), are demarcated by the
two vertical lines. The load sustained up to the intersection of the first vertical line
99
© University of Pretoria
demarcates region (I). This load is the maximum sustainable load before plastic deformation
begins at the root of the notch. Region (I) is therefore the area under the load-time curve
representing the crack initiation energy. Plastic deformation at the notch root mentioned
above is represented in region (II). The area under the remaining part of the curve, region
(III), is the energy expended on rapid crack growth where the unstable crack propagation
begins at the point where the load markedly drops. The energy given by regions (II) and (III)
are collectively taken as representing the crack propagation energy for each sample, which
includes both ductile and brittle crack growth. Note that the maximum rate of energy
absorption during the impact test occurs in regions (I) and (II).The regions of the Charpy
sample that can be associated with crack initiation and propagation fracture regions are
shown in Figure 4.7.8 below for the steel HCrLMo.
I
II
IV
III
Figure 4.7.8 Characterisation of Charpy fracture surface in a HMoLCr steel sample
consisting of 75% bainite. I is the fracture initiation region, II is the britte propagation
region, III is a shear lip and IV is the final fracture.
The morphology of the fracture initiation region I is plastic deformation characterised by
ductile tearing. The initiation region is formed by shearing along slip lines. It is followed
immediately by ductile crack propagation. At the rapid load drop, brittle fracture begins and
is visible as the relatively flat central region on a fracture surface.
100
© University of Pretoria
4.7.3.1 The HMoLCr steel
Following the method to find the crack initiation energy in Figure 4.7.7 the graphs of crack
initiation and propagation energy were plotted as a function of % bainite and are shown in
Figure 4.7.9 for the HMoLCr steel.
Crack Initiation
1.0
0.9
Initiation energy fraction
0.8
0.7
0.6
0.5
0.4
0.3
0.2
HMoLCr steel
0.1
0.0
0
20
40
60
80
Amount of bainite (%)
(a)
101
© University of Pretoria
100
Crack propagation
1.0
0.9
Propagation energy fraction
0.8
0.7
0.6
0.5
0.4
0.3
0.2
HMoLCr steel
0.1
0.0
0
20
40
60
80
100
Amount of bainite (%)
(b)
Figure 4.7.9 Plots of (a) crack initiation and (b) crack propagation energies in the HMoLCr
steel Charpy samples. The respective energies were measured as a fraction of the total
energy.
The crack initiation energy graph in Figure 4.7.9 shows a significant drop in crack initiation
energy between 100% martensite specimens and 10% bainite specimens. The decrease is to
less than 0.5 of the total absorbed energy and precedes an increase in crack initiation energy
as the amount of bainite increases. It is evident that the effect of low volume fractions of
bainite in the HMoLCr steel is to increase its susceptibility to brittle cracking as little energy
is required to initiate a crack. This finding is the reason for the decrease in total absorbed
energy in Figure 4.7.6 between 0% bainite (or 100% martensite) and 10% bainite.
The crack propagation energy decreased with an increase in the amount of bainite. This is a
result of an increase in the amount of the brittle carbide phase. The images in Figure 4.7.10
show the fracture surfaces of samples containing various amounts of bainite.
102
© University of Pretoria
(a) 100%M
(b) 10% B
(c) 25%B
(d) 90%B
(e) 100%B
Figure 4.7.10 Optical images of fracture surfaces of HMoLCr samples containing (a) 100%
M, (b) 10% B, (c) 25% B, (d) 90% B and (e) 100% B. M = martensite and B = bainite Note
the reduced ductile appearance from 10 to 100% bainite.
It can be seen from the optical images of the fracture surfaces, that there is a marked
transition from predominantly ductile fracture at the 10% B sample surface to a largely brittle
fracture surface at 100%B. This is consistent with the decreasing crack propagation energy
with increasing amounts of bainite.
The total absorbed energy plot shows a horizontal midsection between 10% and 80% bainite.
Consider the crack initiation energy fraction plot in Figure 4.7.9 where the energy fractions of
the 10%, 25% and 75% bainite specimens are all roughly 0.3. Because of the relation of the
total absorbed energy to the crack initiation and propagation energies described earlier, the
103
© University of Pretoria
corresponding crack propagation energy fraction for these specimens are all approximately
0.7. This consistency resulted in the plateau of the total absorbed energy.
4.7.3.2 The HCrLMo steel
The samples from the HCrLMo steel, on the other hand, had the highest crack initiation
energies in the microstructures containing small amounts of bainite and a trend in crack
propagation energy that increased with increasing amounts of bainite. Figure 4.7.11 shows
the respective initiation and propagation energies for the HCrLMo steel samples.
Crack Initiation
1.0
0.9
Initiation energy fraction
0.8
0.7
0.6
0.5
0.4
0.3
HCrLMo steel
0.2
0.1
0.0
0
20
40
60
80
Amount of bainite (%)
(a)
104
© University of Pretoria
100
Crack Propagation
1.0
0.9
Propagation energy fraction
0.8
0.7
0.6
0.5
0.4
HCrLMo steel
0.3
0.2
0.1
0.0
0
20
40
60
80
100
Amount of bainite (%)
(b)
Figure 4.7.11 Plots of the fractions of (a) the crack initiation energy and (b) the crack
propagation energy of the HCrLMo steel.
The crack initiation energies in the HCrLMo contribute a significant amount to the total
absorbed energy as they lie within 95 to 75% of the total energy. This means that a large
amount of ductility predominates the HCrLMo specimens and corresponds with the high
amounts of total absorbed energy shown in Figure 4.7.6.
4.7.4 Fractography of the Charpy fracture surfaces
The SEM micrographs in Figure 4.7.12 below were taken within the crack initiation region at
the ductile/plastic deformation zones where shear fracture produced elongated dimples.
The density of dimples present on a fracture surface is a function of the number of nucleation
sites available. Ductile dimples occur by a microvoid coalescence mechanism and are
nucleated at any point of strain discontinuity such as at inclusions, which is evident in the
inclusions found in many dimples. Other possible points of nucleation are grain or subgrain
boundaries and second phase particles.
105
© University of Pretoria
MnS
(a)
(b)
Figure 4.7.12 SEM micrographs of the fracture initiation region in (a) a HCrLMo steel
Charpy sample with 75% bainite and (b) in a HMoLCr steel Charpy sample with 10%
bainite. Note the particles within ductile dimples, most likely MnS inclusions.
(a)
(b)
Figure 4.7.13 SEM micrograph taken in a HMoLCr steel Charpy sample containing10%
bainite at (a) the transition region between crack initiation and brittle fracture region (b) a
region of brittle fracture with numerous cleavage facets.
The mode of fracture changed from ductile fracture in the crack initiation region to brittle
fracture in the rapid crack propagation region. Cleavage facets were found in these brittle
crack propagation regions.
106
© University of Pretoria
Figure 4.7.14. SEM micrograph of tear ridges within the brittle fracture region of a HCrLMo
steel Charpy sample with 75% bainite. Note the secondary cracks (arrowed) that developed
amongst the cleavage facets.
Tear ridges were found within the brittle fracture region as shown in Figure 4.7.14. Tear
ridging is a sign of local energy absorption during cleavage crack propagation. The
connection of these cleavage facets by tear ridges is a dominant energy absorption
mechanism during cleavage crack propagation [66]. The occurrence of localized features
exhibiting both cleavage and plastic deformation is known as quasi-cleavage [67] and is said
to be caused by tri-axial stress states, material embrittlement and or impact loading within the
DBTT region. However, the occurrence of localised plastic deformation on a largely cleavage
fractured surface is thought to be due to ductile tearing of regions that were not broken or
opened by the cleavage crack path.
The occurrences of secondary cracks shown by the arrow in Figure 4.7.14 during crack
propagation in the area of the main cleavage crack propagation indicate areas of local energy
absorption.
4.7.5 Shear fracture measurements
The lateral expansion was measured as a function of the amount of bainite. Figure 4.7.15
shows the graphs of the lateral expansion of Charpy samples of the HCrLMo and HMoLCr
steels.
107
© University of Pretoria
1.3
1.2
1.1
Lateral expansion (mm)
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
HMoLCr steel
0.1
0.0
0
20
40
60
80
100
Amount of bainite (%)
(a)
1.5
1.4
1.3
Lateral expansion (mm)
1.2
1.1
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
HCrLMo steel
0.1
0.0
0
20
40
60
80
100
Amount of bainite (%)
(b)
Figure 4.7.15 Lateral expansions measured on (a) the HMoLCr steel Charpy samples and (b)
on the HCrLMo Charpy samples.
The HCrLMo steel showed an increase in the lateral expansion at 10% bainite, indicating an
increased amount of ductile deformation and thereafter remained approximately constant.
108
© University of Pretoria
The HMoLCr steel‟s lateral expansion also increased up to 50% bainite but thereafter
decreased to its lowest value at 100% bainite.
The amount of ductile fracture was estimated from digital images of the fracture surfaces.
The shear fracture was distinguished from the brittle fracture region by the large flat area in
the centre of the specimen. The amount of shear fracture was calculated as a fraction of the
total fracture area. Figure 4.7.16 shows the graphs of shear fracture measured on Charpy
Shear fracture
samples of the two steels.
1.00
0.95
0.90
0.85
0.80
0.75
0.70
0.65
0.60
0.55
0.50
0.45
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
HMoLCr steel
0
20
40
60
80
Amount of bainite (%)
(a)
109
© University of Pretoria
100
Shear fracture
1.00
0.95
0.90
0.85
0.80
0.75
0.70
0.65
0.60
0.55
0.50
0.45
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
HCrLMo steel
0
20
40
60
80
100
Amount of bainite (%)
(b)
Figure 4.7.16 Shear fracture measured on fractured Charpy samples of (a) the HMoLCr and
(b) HCrLMo steels as a function of the % bainite.
The shear fracture plot of steel HMoLCr confirms the decrease in ductility with increasing
amounts of bainite that is evident in the lateral expansion plot. In steel HCrLMo, the shear
fracture decreased up to 25% bainite and remained fairly constant thereafter. According to the
shear fracture plot, therefore, smaller amounts of bainite are more critical to the toughness of
tempered martensite in the HCrLMo steel if compared to larger amounts of bainite.
110
© University of Pretoria
5
Chapter 5: Discussion
In this chapter, the effect of Cr and Mo on the bainite transformation is discussed by
examining the impact of these elements on hardenability, the growth rate and on the Ms
temperatures. Finally, the impact energy of martensite-bainite Charpy samples is examined
together with the effect of various amounts of bainite on the crack initiation and crack
propagation energies.
5.1
Tempering characteristics of the C-Mn and HCrLMo steels
Both the C-Mn and HCrLMo steels have relatively low carbon contents of 0.11wt% and
0.17wt% respectively and therefore their Ms temperatures (which are 441.1
steel and 419.9
for the C-Mn
for the HCrLMo steel) are significantly above room temperature. It is
expected that autotempering of the martensite may have occurred during quenching.
Therefore, prior to tempering the steel specimens, the first stages of tempering of martensite
had already occurred through autotempering. The carbon atoms segregated to dislocations
and the matrix was relieved of strain. Decomposition of martensite to a less supersaturated
martensite and ε-carbide occurs up to 200 .
The hardness data in Figure 4.3.1 shows a hardness peak at 200
in the tempering curve of
the HCrLMo steel. The peak in hardness is ascribed to the formation of coherent epsilon
carbides that precipitate during tempering from the carbon-supersaturated martensite.
According to Leslie et al the precipitation of epsilon carbides that produces a coherency peak
is more pronounced in steels containing amounts of carbon in excess of 0.2wt% than in steels
containing lower carbon contents [68]. This is consistent with the absence of a hardness peak
in the C-Mn steel tempering curve, in which the carbon content was less than 0.2wt%.
It is known that the as-quenched hardness of martensite is dependent upon the carbon
concentration in the martensite. This relationship is reflected in the as-quenched hardness
values obtained which for the C-Mn steel was lower at 357.50HV compared to 443.2HV for
the HCrLMo steel.
111
© University of Pretoria
The secondary hardening peaks from coherent carbides such as M2X (in 12% Cr steels),
Mo2C and V4C3 [69] that are typical in steels with alloying elements Cr, V, Mo, were not
observed in this study as the quantity of these alloying elements in the steels were present in
too small concentrations.
112
© University of Pretoria
5.2
Effect of the steel compositions on the partial CCT diagrams
The partial Continuous Cooling Transformation diagrams which were obtained for the
HCrLMo and HMoLCr steels consisted of only the bainite nose which was the area of
interest. The absence of the ferrite nose in both CCT diagrams over the range of cooling rates
used is attributed to the effect of boron. Boron segregates to austenite grain boundaries and
reduces the grain boundary energy thus rendering the grain boundaries less effective as
potential nucleation sites for ferrite. As little as 0.002wt% boron is typically sufficient to
achieve this effect. The steels from which the CCT‟s in Figure 4.4.13 were constructed
contain 0.0035 and 0.002wt% boron, both in sufficient quantities to achieve the hardenability
effect illustrated in Figure 5.2.1 below.
Figure 5.2.1 Effect of boron on transformation [27] .
In addition to the absence of the ferrite C-curve over the range of cooling rates used, the
bainite C-curve was shifted to longer times in the HMoLCr steel. A useful measure of the
position of the bainite C-curve with respect to time is the critical cooling rate for 100%
martensite to form upon the initial quench. From the CCT diagrams in Figure 4.4.13 (a) and
(b) the critical cooling rate for the formation of martensite from the initial quench was 30 /s
for the HCrLMo steel and only 5 /s for the HMoLCr steel. Despite a higher level of Cr in
the HCrLMo steel, the steel is significantly less hardenable than the HMoLCr steel.
113
© University of Pretoria
The effects of boron in combination with Cr or Mo was delineated in the work by Han et al
[70] who found that Mo was more effective than Cr in enhancing the hardenability of boron
added steels. The authors compared the hardenability of two low carbon steels of which one
was a Mo-B steel (Fe-0.1C-1.5Mn-0.2Mo-0.002B) and the other a Cr-B steel (Fe-0.1C1.5Mn-0.5Cr-0.002B) with a boron baseline steel of composition Fe-0.1C-1.5Mn-0.002B (in
wt%). After cooling at various cooling rates, the fastest cooling rate during which pearlite and
polygonal ferrite evolved was 3 /s in the boron steel, 1 /s in the Cr-B steel and 0.2 /s in
the Mo-B steel. It was thus demonstrated that Mo has a more substantial effect on
hardenability than Cr [70]. Han and co-workers ascribed the hardenability effect of Mo to the
deterioration of phase stability of M23(C,B)6 which reduces hardenability by providing
nucleation sites for ferrite.
The Wagner Interaction coefficient
was introduced as a measure of the strength of
interaction of an alloying element i with carbon [41, 71] and is given by the expression
where
activity
is the activity coefficient for carbon which is given by the carbon
divided by the carbon concentration:
.
Aaronson , Reynolds and Purdy [41] found that more negative values of
produced a
pronounced solute drag effect [41]. In the work done by Wada, the Wagner interaction
coefficient for carbon in Fe-C-Cr alloys with between 0.2 and 8wt% Cr, was found to be
[72] and in Fe-C-Mo alloys of Mo content between 0.2 and 3wt% Mo, the
interaction coefficient was
[73], showing that Mo has a stronger effect than Cr.
Therefore, the suppression of a ferrite C-curve in the partial CCT diagrams of both the
HCrLMo and HMoLCr steels in Figure 4.4.13 is attributed to the hardenability effect of
Boron. However, the difference observed in the bainite C-curves is ascribed to the stronger
effect of Mo by its synergistic effect in combination with Boron and its higher negative
Wagner interaction coefficient.
114
© University of Pretoria
5.3
JMAK kinetics of the isothermal transformations
The Johnson Mehl Avrami Kolmogorov model was used to fit the volume fraction – time
data. By taking the natural logarithm of the JMAK equation, the linear equation below is
obtained:
*
+
eq 5.1
where Vf is the volume fraction and k is a temperature dependent constant.
The Avrami exponent n was then calculated from the slope of the graphs of ln t against
+ in Figure 5.3.1.
2
1
lnln [1/(1-Vf)]
*
0
-1
HMoLCr steel
n = 1.3
ln k= -5.45
R-Square = 0.97503
-2
-3
2.0
2.5
3.0
3.5
4.0
4.5
lnt
(a)
115
© University of Pretoria
5.0
5.5
2
lnln [1/(1-Vf)]
1
0
-1
HCrLMo steel
n = 1.4
ln k = -4.12
R-Square = 0.97739
-2
1.0
1.5
2.0
2.5
3.0
3.5
4.0
lnt
(b)
Figure 5.3.1 Plots of lnt versus lnln(1/1-Vf) for (a) the HMoLCr steel and (b) the HCrLMo
steel isothermally transformed at 498 and 484 respectively.
The exponent n is, according to Christian [74] an indication of the dimensionality of the
growth. Typically, an exponent of 2 represents constant nucleation and linear growth.
Christian states that the general expression for volume transformed (equation 2.3) in the
Avrami theory of nucleation at preferred sites is valid if
dimensional growth and if
which leads to one-
leads to two-dimensional growth. One-dimensional
growth in the case of bainite refers to the lengthening rate of ferrite laths. The thickening of
bainitic ferrite laths is negligible and growth of the ferrite laths occurs mainly along the
length of the lath. This has been demonstrated, for instance, in the work done by Kostic,
Hawbolt and Brown [75] on volume diffusion controlled growth of bainite, which showed
that the rate of lengthening of bainite plates occurred at least two orders of magnitudes higher
than the rate of thickening of the plates. A typical aspect ratio of 4 was obtained for the
bainite formed in this study. The exponents of 1.3 and 1.4 obtained for the HMoLCr and
HCrLMo steels therefore indicate one dimensional growth.
116
© University of Pretoria
5.4
Effect of alloying elements on growth rates
The major question to be addressed here is that the rates of bainite formation during
isothermal transformation shown in Figure 4.6.1 differed with the HCrLMo steel
transforming significantly faster than the HMoLCr steel. Why does the addition of Mo in a
relatively small amount to these plate steels have such a marked effect on the bainite
formation if compared to Cr additions? Is this effect a thermodynamic or a kinetic one? As
both steels were isothermally transformed at equivalent estimated undercoolings, the
thermodynamic driving force for nucleation could be considered approximately equivalent.
The growth rate of bainite can be modelled according to the Trivedi model for diffusion
controlled lengthening of plates in which the growth rate v is given by the equation [17, 75]:
eq 5.2
where
is the carbon supersaturation, D is the average carbon diffusivity and ρ the radius
of the tip of the advancing interface.
The equation depicts a linear proportional relationship between the diffusivity of carbon and
the growth velocity
. This implies that factors affecting the diffusivity of carbon
consequently affect the growth rate directly.
The effect of various alloying elements on carbon diffusivity in austenite has been studied by
several researchers [73, 76-78]. It is well documented that the diffusivity of carbon in an
alloy steel is a function of its chemical composition. For instance manganese and chromium
tend to decrease the carbon diffusivity while carbon and nickel increase it [71]. It is thought
that elements which reduce the carbon activity such as Cr and Ni, attract C atoms, slowing
them down whilst elements that increase carbon activity do so by repelling C atoms. The
lowering of carbon activity and consequently diffusivity and mobility usually occurs where
carbide-forming elements are present. ̌ erm ́ k et al put forward the relationship between the
carbon tracer diffusivity and the carbon activity coefficient [76].
D = D(0) ( ⁰)
eq 5.4
117
© University of Pretoria
where D(0) is the diffusion coefficient in the unalloyed metal, γ is the carbon activity
coefficient in an alloyed metal and γ⁰ is the carbon activity coefficient in unalloyed steel.
From equation 5.4 it is evident that as γ γ⁰, D
D(0). The relationship was tested by
measuring the diffusivity of carbon in Fe-Ni-Cr alloys [76]. The carbon diffusion coefficient
was found to increase as γ approached γ⁰. This held for temperatures ranging from 570
to
850 . Diffusivity is related to carbon activity and chemical potential and it has been shown
that the effect of Mo and Cr is to decrease the thermodynamic activity of carbon in austenite
[79].
Based on the equation for carbon diffusivity (eq 5.5) in binary Fe-C alloys introduced by
̇ gren [80]; Seok-Jae Lee et al [78] derived an equation for the activity coefficient of carbon
in austenite (eq 5.6).
(
)
(
)
( (
)
)
eq 5.5
The equation expresses the activity coefficient fC in a multi-component alloy as a function of
the absolute temperature and a concentration parameter unique to each alloying element.
(
(
)
(
)
)
(
(
)
)
(
(
)
)
eq 5.6
In a separate experiment to verify carbon diffusivity at temperatures in the range 500
900
to
[81] the authors found good correlation between their experimental data at the lower
temperature of 500
with the expression proposed by ̇ gren. They concluded that the latter
can thus be used in the temperature range 500
to 900 . On this basis, the use of equation
5.6 in the upper bainitic temperature range is qualified. From this equation it can be seen that
at equivalent Cr and Mo amounts yCr = yMo and at the same temperature, Mo reduces the
activity coefficient of carbon slightly more than Cr.
118
© University of Pretoria
To verify the activity of carbon in the HCrLMo and HMoLCr steels during isothermal
conditions, the thermodynamic activities were simulated by Thermo-Calc software with data
from the TCFE7 database. The activities of carbon in fcc austenite and bcc ferrite in the
HMoLCr steel at 498
were found to be 0.21 and 3.41 respectively. The activity of an
element is related to the chemical potential
where
through the expression
is the chemical potential of component i in the standard state, which for carbon is
graphite. Using this expression, the difference in chemical potential between austenite and
ferrite in the HMoLCr steel calculated at the transformation temperature of 771K is 18368.76 J/mol. In the HCrLMo steel, the activities of carbon in fcc austenite and bcc ferrite
at 484
were 0.20 and 3.70 respectively and the difference in chemical potential at the
transformation temperature of 757K is -17825.56J/mol. The chemical potential of carbon in
the HMoLCr steel thus is lower than in the HCrLMo steel, implying the presence of a
reduced chemical potential gradient in the HMoLCr steel. The implication of a lower
chemical potential on the transformation may be considered according to Christian‟s [71]
treatment of diffusional transformations. The diffusion current or flux of a species i as
related to its chemical potential is given by:
potential,
is the mobility and
where
is the chemical
is the density of atoms. The lower chemical potential
gradient in steel HMoLCr causes a reduction in the thermodynamic driving force to form
upper bainite from austenite, which is in agreement with the experimental findings of a much
reduced growth rate in the HMoLCr steel.
It has been shown that Mo reduces the carbon diffusivity [70]. As carbon diffusion is central
to the formation of bainite, a reduction in the rate of carbon diffusion will retard the growth
of the bainite. The diffusivities of carbon in the HCrLMo and HMoLCr steels as estimated by
Dictra were found to be of the same order of magnitude and did not differ significantly
between the two steels. It therefore seems that the observed greater retarding effect of Mo on
upper bainite formation may not necessarily be carbon diffusion related if compared to that of
the higher Cr steel.
However, given that the undercoolings are equivalent, another variable that is likely to affect
the growth rates of the two steels is the nature of the substitutional alloying elements present.
Solute drag models assume solute accumulation at growth interfaces and short range
119
© University of Pretoria
diffusion of these solute atoms as they move with the interface. Substitutional solute atoms
can therefore cause a delay in the growth kinetics of the reaction. In the light of such a model,
substitutional atoms can be said to slow down or retard growth if they participate in the
growth of either bainitic ferrite or cementite. In verifying the involvement of substitutional
atoms, the chemical composition of iron carbides extracted from upper bainite samples by
electrolytic means were analysed and Mo and Cr peaks were observed in the spectra of
numerous carbide particles. As these are strong carbide formers, the results suggest that the
cementite in the upper bainite in these steels was alloyed iron carbide M3C rather than iron
carbide Fe3C. The diffusivity of Cr and Mo in ferrite in the HCrLMo and HMoLCr steels at
the respective transformation temperatures was predicted to be of the order 10-22 m2/s and 1023
m2/s in the respective steels. These values are many orders of magnitude lower than the
carbon diffusivities. It can therefore be speculated that the formation of alloy carbides rather
than pure cementite in bainite slows the growth rate of bainite in both alloys by the
involvement of alloying atoms because of the significantly lower diffusion rate for the
substitutional alloying elements than for carbon. It can also be seen that the slower diffusivity
of Mo (10-23 m2/s) during the formation of M3C is a likely cause of the low growth rates.
However, according to equation 5.2 it is not only diffusion that affects the growth rate but
also the radius of the advancing interface. Figures 5.6.3 and 5.6.4 show the microstructures of
the steels. The bainite laths in the steel HCrLMo were smaller or narrower than the laths in
steel HMoLCr where widening or merging of the laths had occurred. The difference in lath
size suggests that the radii of the tip of the advancing interface ρ was also a variable
contributing to the difference in growth rate. By virtue of the inverse relationship between υ
and ρ, thicker bainite laths in steel HMoLCr imply a slower growth rate than the thinner
bainite laths in steel HCrLMo.
The difference in carbon activity obtained for the two steels implied that a lower chemical
potential gradient at the growth front in the HMoLCr steel could have caused slower bainite
growth. Furthermore, the observed participation of Cr and Mo in the formation of bainitic
carbide may also have been a rate limiting factor and it is likely that a solute drag like effect,
which accounts for the involvement of alloying elements, was operative during the growth of
bainite in the HMoLCr steel.
120
© University of Pretoria
5.5
Effect of isothermal holding on Ms temperatures
The Ms temperature of the remaining austenite after bainite formation was found to decrease
when measured as a function of the isothermal holding time in Figure 4.6.2. Since the Ms
temperature is sensitive to the amount of carbon in the austenite from which it forms, its
decrease with longer holding times indicates that some carbon enrichment occurs in the
remaining austenite after bainite has formed. The carbon enrichment of austenite from bainite
has been used in quench and partitioned (Q&P) steels where carbide-free bainite is preferred.
The steels of which the Ms temperatures in Figure 4.6.2 were measured, contain alloyed
cementite as is the case when upper bainite forms. Although the formation of cementite in the
upper bainite consumed some of the carbon, the decreasing Ms testifies to a non-negligible
carbon enrichment in the remaining austenite. The possible carbon content in the austenite
was calculated at each Ms temperature according to the equation [38]:
Ms = 561 - 474C - 33Mn - 17Cr - 17Ni - 21Mo
eq 5.7
The experimental Ms temperatures are plotted in Figure 5.5.1 as a function of the carbon
content.
440
430
420
410
HCrLMo
o
Ms ( C)
HMoLCr
400
390
380
0.16
0.18
0.20
0.22
0.24
Carbon (wt%)
121
© University of Pretoria
0.26
0.28
Figure 5.5.1 Graph of the measured Ms depression with carbon content as modelled
according to the equation by Krauss [38].
The implication of the carbon enrichment predicted in Figure 5.5.1 as derived from equation
5.7, is that the austenite that forms martensite is more enriched in carbon with the progressive
formation of bainite. It follows that the martensite thus formed should be harder, yet hardness
tests on samples did not reveal an increase in as-quenched hardness. When plotted as a
function of time as in Figure 4.6.2 the Ms – time curves correspond with the volume fraction
– time sigmoidal curves in Figure 4.6.1. The HCrLMo steel which grows at a higher rate than
the HMoLCr steel also has a higher Ms – time gradient. However, the Ms plotted as a function
of the amount of bainite for the two steels shows no significant difference in the two steels.
Furthermore, the question arises whether the estimated carbon concentrations in the
untransformed austenite in Figure 5.5.1 as the percentage bainite increases, are realistic? The
To curves for the steels were modelled using Thermo-Calc and are shown in Figure 5.5.2
below. The plots show that the carbon enrichments predicted in Figure 5.5.1 are well below
the To carbon concentrations (wTo). The plots also show that the To temperature for bainite is
727
in steel HMoLCr and
717
in steel HCrLMo and therefore the undercoolings
below the To at the isothermal transformation temperatures used were 229
and 233
respectively which are practically the same. This supports the earlier objective that the
driving forces for bainite formation in both steels were practically the same in the selection of
the two respective transformation temperatures.
122
© University of Pretoria
To
Isothermal bainite at 771K
(a)
To
Isothermal bainite at 757K
(b)
Figure 5.5.2 T-zero temperature plotted against carbon content (weight fraction) for (a) steel
HMoLCr and (b) steel HCrLMo showing the T-zero temperatures To and the maximum
carbon content
at the isothermal treatment temperatures, as estimated by Thermocalc.
123
© University of Pretoria
5.6
Impact toughness
5.6.1 Effect of bainite on the total absorbed energy
As Figure 4.7.6 shows, both steels had the lowest total absorbed energy in specimens with
100% bainite and it confirmed the effective brittleness of upper bainite in contrast to
tempered martensite. However, the graphs did not show a consistent decrease in total
absorbed energy with an increasing amount of bainite. The total energy in the figure for the
HMoLCr steel initially decreased and reached a plateau between 10% and 80% bainite before
a further decrease at more than 90 % bainite leading up to 100% bainite whilst the decreasing
trend in the HCrLMo steel graph was of a more general nature with some considerable
amount of scatter. In the mixed bainite/martensite samples a crack can initiate in either
tempered martensite or bainite, and the distribution of bainite in the martensite matrix may
have been the underlying reason for the scatter. This especially applies to the distribution of
phases at the root of the notch. Studies on the fracture toughness of bainite-martensite steels
showed that the overall toughness depends on the phase in which the cleavage fracture is first
initiated. Some studies revealed that [82] cleavage will occur at a critical distance ahead of
the precrack tip, in this case the V-notch. The occurrence of cleavage crack commencing in
the bainite or martensite will depend on the probability of the occurrence of either phase at
this critical distance. This implies that the impact energy of mixed bainite-martensite samples
may approach the impact energy of 100% martensite or 100% bainite specimens where the
energies of the latter are upper and lower bounds or limits. Consequently, the toughness of
the mixed microstructures ought not be any higher than that of 100% martensite or lower than
that of 100% bainite.
Overally, the mixed microstructures in the HCrLMo steel exhibited a greater degree of
toughness as their absorbed energies were within a higher range if compared to that of the
steel HMoLCr.
124
© University of Pretoria
5.6.2 Crack initiation and propagation energies
5.6.2.1 Comparison of crack initiation energies
HMoLCr
150
HCrLMo
140
130
Crack initiation energies (J)
120
110
100
90
80
70
60
50
40
30
20
10
0
0
20
40
60
80
100
Amount of bainite (%)
Figure 5.6.1 Plots of absolute crack initiation energies of steels HCrLMo and HMoLCr as a
function of the amount of bainite in the Charpy specimens
Figure 5.6.1 shows how the absolute values of crack initiation energy vary with the amount
of bainite. From the figure, it can be seen that the crack initiation energy levels for the
HCrLMo steel are distinctly higher than those of the HMoLCr steel and in both steels there is
an initial decrease in the crack initiation energy from 100% martensite to 10% bainite. The
decreasing trend in the absolute value of crack initiation energy for the HCrLMo steel is
consistent with its corresponding trend of the initiation energy as a fraction of the total as
shown in Figure 4.7.11 and the crack initiation energy of steel HMoLCr is relatively
independent of the quantity of bainite, with the exception of outliers at 60% and 90% bainite.
5.6.2.2 Comparison of crack propagation energies
Figure 5.6.2 shows how the absolute crack propagation energy varies with the amount of
bainite for steels HMoLCr and HCrLMo.
125
© University of Pretoria
HMoLCr
90
HCrLMo
Crack propagation energy (J)
80
70
60
50
40
30
20
10
0
0
20
40
60
80
100
Amount of bainite (%)
Figure 5.6.2 Plots of crack propagation energies of steels HCrLMo and HMoLCr as a
function of the amount of bainite in the Charpy specimens.
The figure above shows a decrease of crack propagation energy in steel HMoLCr from
roughly 85J at 100% martensite to an average of 28J at 100% bainite. However, the HCrLMo
steel exhibited an increase in the crack propagation energy from the lowest energies at 100%
martensite to about 20J at 100% bainite. At higher amounts of bainite, the crack propagation
energies of the two steels overlap, showing that the effect of bainite on crack propagation
energy in the two steels is roughly the same, although that of the HMoLCr steel remains
within the higher range. The latter is consistent with the crack propagation energy fractions in
Figure 4.7.9 which are between 0.4 and 0.6 at 100% bainite whereas the propagation energy
fractions in steel HCrLMo (Figure 4.7.11) are
0.2.
126
© University of Pretoria
5.6.2.3 Relation of microstructure to energies
Figure 5.6.3 SEM micrograph of bainite in the HCrLMo steel with 90% bainite
Figure 5.6.4 SEM micrograph of bainite in the HMoLCr steel. Sample contains 56% bainite.
The respective microstructures of the upper bainite in the two steels are shown in the SEM
micrographs in Figures 5.6.3 and 5.6.4. In steel HCrLMo, the bainite appears to have a welldefined lath structure with longer carbide particles whereas the structure of the bainite in steel
127
© University of Pretoria
HMoLCr can be described as being similar to coalesced bainite, The term coalesced indicates
a merging of bainite plates [83] and the lack of a distinctly lath structure in the HMoLCr steel
suggests the possibility of coalesced bainite. The coalesced platelets with identical
crystallographic orientation result in regions devoid of crystallographic discontinuities such
as packet or block boundaries, thereby reducing the ability to deflect propagating cracks [83].
Therefore, the decrease of crack propagation energy in the HMoLCr steel shown in Figure
5.6.2 is attributed to this effect.
The high angle boundaries at block boundaries cause significant deviation of cracks.
Hindrance to the movement of a microcrack that has nucleated and is now propagating in the
matrix, will be most effective at high angle boundaries [84]. Generally a misorientation
greater than 30 between grains is likely to lead to crack arrest. Despite the relatively larger
appearance of the carbides in steel HCrLMo, the bainitic block substructure deviates
propagating cracks to an appreciable extent as is evident in the increasing crack propagation
energy in Figure 5.6.2. The same effect is depicted in the increasing trend of crack
propagation energy fractions in Figure 4.7.11. However, the low crack propagation energy
fractions of at most 0.2 are a consequence of the bainitic carbide.
The bainitic carbide can reduce the crack propagation energy as it is prone to cracking and
debonding with the matrix [84], however it also acts as a point of crack initiation as a result
of the stress concentration produced by its elongated shape. Thus the progressive reduction in
crack initiation energy in the HCrLMo steel is related to the increase in crack initiation sites
most of which are possibly due to cementite in bainite.
The Smith theory [51] for brittle fracture incorporates the effect of second phase particles on
grain boundaries during cracking. The model takes into account the thickness of brittle
second phase particles on grain boundaries or, as in this case, on interlath boundaries. The
condition for the particle cracking is given by the expression:
*
–
+
eq 5.5
128
© University of Pretoria
where
is the fracture stress, Co is the precipitate thickness, E is Young‟s Modulus,
is Poisson‟s ratio and
is the effective surface energy of the interface between the carbide
and ferrite. The cracking of a large carbide particle is usually continued in the adjacent matrix
[30]. The coarse cementite in upper bainite „influences‟ the impediment of dislocations by the
lath boundary [54].
Thus the elongated bainitic carbides in the HCrLMo steel samples appear to have been
responsible for the reduction in the crack initiation energy, however, the lath structure of
bainite in this steel was an effective barrier to the propagation of cracks. As a result, although
the crack initiation energy trend decreased with increasing amounts of bainite, their absolute
values were much higher than those of the HMoLCr steel samples. In addition to this, the
beneficial effects of the subgrain structure showed in the increasing crack propagation energy
trend of steel HCrLMo. A large proportion of the total absorbed impact energy in the
HCrLMo steel (75-95%) consisted of the crack initiation energy. The absence of a distinct
lath bainite structure in steel HMoLCr led to lower crack initiation energies and reduced
resistance to crack propagation.
129
© University of Pretoria
6
Chapter 6: Conclusions
Two low alloy steels containing different Chromium and Molybdenum ratios were
isothermally treated to form upper bainite. The growth and transformation kinetics of the
steels were examined in the light of the effects of these alloying elements and it may be
concluded that:

Mo had a stronger effect than Cr on the hardenability as well as on the measured
upper bainite formation with a markedly lower growth rate of bainite in the HMoLCr
steel than in the HCrLMo steel.

The undercoolings below the To temperatures obtained from Thermo-Calc predictions
were approximately similar at 229

and 233
for the two steels.
It was confirmed by TEM-EDS of extracted carbides from the upper bainite, that the
carbide composition was not pure cementite but rather a (Fe,Cr,Mo)3C or M3C.
Substitutional diffusion or participation in the bainite transformation, therefore, has
taken place in these steels transformed under these conditions;

The lower growth rate of bainite in the HMoLCr steel compared to the HCrLMo steel
may be attributed to the formation of M3C bainitic carbide rather than pure cementitie
Fe3C. This is possible by the diffusional involvement of Cr and Mo during the
formation of M3C where diffusivity of Cr and Mo in the HMoLCr steel were found to
be an order of magnitude lower than in the HCrLMo steel.

The differences in carbon activities of 0.17 in steel HCrLMo and 0.31 in steel
HMoLCr suggest the occurrence of a solute drag like effect during the growth of
bainite in the HMoLCr steel.

The progressive formation of larger amounts of bainite during isothermal
transformation caused the untransformed austenite to become more enriched in
carbon, thereby lowering the measured Ms temperature of this remaining austenite
progressively as the bainite transformation proceeded. The carbon enrichments were
within the limits set by the predicted To temperature.
130
© University of Pretoria
The Charpy impact energy of steels HCrLMo and HMoLCr heat treated to form 0%, 10%,
25%, 60%, 75%, 90% and 100% upper bainite was measured in an instrumented impact
tester. The following conclusions were reached:

Upper bainite in the steels HCrLMo and HMoLCr reduced the total absorbed energy
with the lowest impact energy obtained in 100% bainite samples.

The absolute crack initiation energies in the HCrLMo steel were higher than those of
the HMoLCr steel.

The plot of crack initiation energy as a fraction of the total energy in the HCrLMo
steel decreased to
0.8 at 100% bainite indicating that most of the total absorbed
impact energy was expended on crack initiation.

The absolute values of crack propagation energy decreased in the HMoLCr steel and
increased in the HCrLMo steel with % bainite formed. In the latter, crack propagation
energy increased to only 20% of the total absorbed energy.

Microstructural examination of the bainite formed in the HCrLMo steel revealed a
structure consisting of well-defined bainite packets and blocks that developed from
the fragmentation of the prior austenite grains during the growth of bainite. The block
structure forms an effective barrier to cracks and therefore resulted in the high crack
initiation energies and the increasing trend of crack propagation energy in steel
HCrLMo. The low values of crack propagation energy in this steel are attributed to
the embrittling effect of bainitic carbides.

The decreasing crack propagation energies in the HMoLCr steel are attributed to the
absence of the distinct subgrain structure that was evident in the HCrLMo steel.

Crack propagation in the HCrLMo steel consumed the least amount of energy and is
therefore more critical to impact toughness than crack initiation.

Despite the reduction in toughness, the overall Charpy impact energy for both
HCrLMo and HMoLCr steel‟s mixed microstructures was within acceptable
toughness ranges.
131
© University of Pretoria
7
References
1.
S.V. Belikov, K.I. Sergeeva, O.Yu. Kornienko, I.N. Ashikhmina, A.I. Stepanov,
Special features of the formation of structure and properties of steels with
heterogeneous bainite-martensite structure for oil and gas pipelines. Metal Science
and Heat Treatment, 2011. 52: p. 581-587.
2.
A. Gulyaev, Y.S. Golovanenko, and V. Zikeev, Effect of the quantity of
nonmartensitic transformation products on the resistance to fracture of quenched and
tempered structural steels. Metal Science and Heat Treatment, 1978. 20 (7): p. 584594.
3.
W. Lee, S. Hong, C. Park and S. Park, Carbide precipitation and high-temperature
strength of hot-rolled high-strength low-Alloy Steels containing Nb and Mo.
Metallurgical and Materials Transactions A, 2002. 33: p. 1689-1698.
4.
X.L. He, X.M. Wang, Effect of boron addition on structure and properties of low
carbon bainitic steels. ISIJ International, 2002. 42 (S38-S46).
5.
B. Hwang, D.W. Su, S.J Kim, Austenitising temperature and hardenability of lowcarbon boron steels. Scripta Materialia, 2011. 64.
6.
W. Garlipp, M. Cilense, S.I. Novaes Gomes, Austenite decomposition of C-Mn steel
containing boron by continuous cooling. Journal of Materials Processing Technology,
2001. 114: p. 71-74.
7.
Y. Jung, K. Nakai, H. Ohtsubo, Y. Ohmori, Isothermal transformation behaviour of
extremely low carbon 3%Mn - 1%Cr steels. ISIJ International, 1994. 34 (1): p. 43-50.
8.
H. Asahi, Effects of Mo addition and austenitizing temperature on hardenability of
low alloy B-added steels. ISIJ International, 2002. 42 (10): p. 1150-1155.
9.
A. Clarke, J.G. Speer, M.K. Miller, R.E. Hackenberg, D.V. Edmonds, D.K. Matlock,
F.C. Rizzo, K.D. Clark, E. De Moor, Carbon partitioning to austenite from martensite
or bainite during the quench and partition (Q&P) process: A critical assessment.
Acta Materialia, 2008. 56 (1): p. 16-22.
10.
F.G. Caballero, M.K. Miller, A.J. Clark, C. Gracia-Mateo, Examination of carbon
partitioning into austenite during tempering of bainite. Scripta Materialia, 2010. 63
(4): p. 442-445.
11.
J.G. Speer, D.V Edmonds, F.C. Rizzo, D.K. Matlock, Partitioning of carbon from
supersaturated plates of ferrite with application to steel processing and fundamentals
132
© University of Pretoria
of the bainite transformation. Current Opinion in Solid State and Materials Science,
2004. 8 (3): p. 219-237.
12.
A. Borgenstam, M. Hillert, J. Agren, Metallographic evidence of carbon diffusion in
the growth of bainite. Acta Materialia, 2009. 57: p. 3242-3252.
13.
D. Quidort, Y. Brechet, Isothermal growth kinetics of bainite in 0.5% C steels. Acta
Materialia, 2001. 49 (20): p. 4161-4170.
14.
J. Wang, P.J. Van Der Wolk, S. Van Der Zwaag, On the influence of alloying
elements on the bainite reaction in low alloy steel during continuous cooling. Journal
of Materials Science 2000. 35: p. 4393-4404.
15.
D.H. Kim, J.G. Speer, H.S. Kim, B.C. De Cooman, Observation of an isothermal
transformation during quenching and partitioning processing. Metallurgical and
Materials Transactions A, 2009. 40 (9): p. 2048-2060.
16.
L. Fang, W.E. Wood, D.G. Atterdge, Identification and range quantification of steel
transformation products by transformation kinetics. Metallurgical and Materials
Transactions A, 1997. 28 A: p. 5-14.
17.
D. Quidort, Y.J.M. Brechet, A Model of isothermal and non-isothermal
transformation kinetics of bainite in 0.5% C steels. ISIJ International, 2002. 42 (9): p.
1010-1017.
18.
F.G. Caballero, M.J. Santofimia, C. Gracia-Mateo, C. Gracia de Andres, Time temperature - transformation diagram within the bainitic temperature range in a
medium carbon steel. Materials Transactions, 2004. 45 (12): p. 3272-3281.
19.
H.K.D.H. Bhadeshia, Bainite: Overall transformation kinetics, Journal DePhysique
Colloques 1982. p. C4-443 - C4-448.
20.
C. Gupta, G.K. Dey, J.K. Chakravartty, D. Srivastav, S. Banerjee, A study of the
bainite transformation in a new CrMoV steel under continuous cooling conditions.
Scripta Materialia, 2005. 53: p. 559-564.
21.
M.K. Kang, D.M. Chen, S.P. Yang, G.L. Hu, The time-temperature-transformation
diagram within the medium temperature range in some alloy steels. Metallurgical
Transactions A, 1992. 23 A: p. 785-795.
22.
H.K.D.H. Bhadeshia, D. Edmonds, The bainite transformation in a silicon steel.
Metallurgical and Materials Transactions A, 1979. 10 (7): p. 895-907.
23.
M. Takahashi, Recent progress: kinetics of the bainite transformation in steels.
Current Opinion in Solid State and Materials Science, 2004. 8 (3): p. 213-217.
24.
S.A. Sajjadi, Z.M. Zebarjad, Isothermal transformation of austenite to bainite in high
carbon steels. Journal of Materials Processing Technology, 2007. 189: p. 107-113.
133
© University of Pretoria
25.
R.F. Hehemann, K.R. Kinsman, H.I. Aaronson, A Debate on the Bainite Reaction.
Metallurgical Transactions, 1972. 3: p. 1077-1094.
26.
S. Chupatanakul, P. Nash, D. Chen, Carbon partitioning during bainite
transformation in 4317 type steels. Metals and Materials International, 2006. 12 (6):
p. 453-458.
27.
H. Bhadeshia, Bainite in Steels, 2001. London, Institute of Materials.
28.
Z. Ławrynowicz, A. Barbacki, Features of bainite transformation in steels. Advances
in Materials Science, 2002. 2 (1): p. 5-32.
29.
R. Bakhtiari, A. Ekrami, The effect of bainite morphology on the mechanical
properties of a high bainite dual phase (HBDP) steel. Materials Science and
Engineering A, 2009. 252: p. 159-165.
30.
A. Vashurin, E.S. Palatnikova, E.M. Ushakova, A.P. Chuchvaga , Effect of structure
on bainite embrittlement of steel 30Kh3MF. Metal Science and Heat Treatment, 1971.
13 (11): p. 963-965.
31.
A. Ali, H. Bhadeshia, Growth rate data on bainite in alloy steels. Materials Science
and Technology, 1989. 5 (4): p. 398-402.
32.
M. Hillert, The Nature of Bainite. ISIJ International 1995. 35 (9): p. 1134-1140.
33.
H. Bhadeshia, D. Edmonds, The mechanism of bainite formation in steels. Acta
Metallurgica, 1980. 28 (9): p. 1265-1273.
34.
J.M. Prado, Bainitc transformations of steels. Journal of Materials Science Letters,
1986. 5: p. 1075-1076.
35.
M. Hillert, Diffusion in Growth of Bainite. Metallurgical and Materials Transactions
A, 1994. 25: p. 1957-1966.
36.
M. Enomoto, Partition of carbon and alloying elements during the growth of ferrous
bainite. Scripta Materialia, 2002. 47: p. 145-149.
37.
S.W. Thompson, D.J. Colvin, G. Krauss, Continuous cooling transformations and
microstructures in a low-carbon high-strength low-alloy plate steel. Metallurgical and
Materials Transactions A, 1990. 21 A: p. 1493-1507.
38.
A.A. Gorni, Steel forming and heat-treating hand book, 2004, Sao Vicente, Brazil.
39.
Z. Ławrynowicz, Influence of cementite precipitation on the extent of bainite reaction
in Fe-Cr-Si-C steel. Journal of Polish CIMAC Selected full texts, 2011. 6 (3): p. 137144.
134
© University of Pretoria
40.
J.W. Cahn, The impurity-drag effect in grain boundary motion. Acta Metallurgica,
1962. 10 (9): p. 789-798.
41.
H. Aaronson, W.T. Reynolds, G. Purdy, Coupled-solute drag effects on ferrite
formation in Fe-C-X systems. Metallurgical and Materials Transactions A, 2004. 35
(4): p. 1187-1210.
42.
G. Shiflet, H. Aaronson, Growth and overall transformation kinetics above the bay
temperature in Fe-C-Mo alloys. Metallurgical and Materials Transactions A, 1990. 21
(6): p. 1413-1432.
43.
H. Goldenstein, H. Aaronson, Overall reaction kinetics and morphology of austenite
decomposition between the upper nose and the Ms of a hypoeutectoid Fe-C-Cr alloy.
Metallurgical and Materials Transactions A, 1990. 21 (6): p. 1465-1478.
44.
H.I. Aaronson, W.T. Reynolds Jr, G.J. Shiflet, G. Spanos, Bainite viewed three
different ways. Metallurgical Transactions A 1990. 21 p. 1343-1380.
45.
E. Humphreys et al., Molybdenum accumulation at ferrite: austenite interfaces during
isothermal transformation of an Fe-0.24 pct C-0.93 pct Mo alloy. Metallurgical and
Materials Transactions A, 2004. 35 (4): p. 1223-1235.
46.
G.J. Shiflet, R.E. Hackenberg, Partitioning and the growth of bainite. Scripta
Materialia, 2002. 47: p. 163-167.
47.
W.T. Reynolds Jr, H.I. Aaronson, G. Spanos, A summary of the present diffusionist
views on bainite. Materials Transactions, JIM (Japan), 1991. 32 (8): p. 737-746.
48.
K. Mills, J.R. Davis, ASM Handbook, Volume 12 - Fractography, ASM International.
p. 12-14.
49.
R.J. Bishop, R.E. Smallman, Strengthing and Toughening. Modern Physical
Metallurgy and Materials Engineering, 1999. p. 288-293.
50.
A. Barbacki, The role of bainite in shaping mechanical properties of steel. Journal of
Materials Processing Technology, 1995. 53: p. 57-63.
51.
E. Smith, The nucleation and growth of cleavage microcracks in mild steel.
Proceedings of the conference on physical basis of yield and fracture, Institute of
Physics and Physical Society, London, 1966: p 36.
52.
L. Zhi-jun, S. Jun-chang, S. Hang, D. Yue-hua, Y. Cai-fu, Z. Xing, Effect of
substructure on toughness of lath martensite/bainite mixed structure in low carbon
steels. Journal of Iron and Steel Research International, 2010. 17 (11): p. 40-48.
53.
D.R. Johnson, W.T. Becker, Toughness of Tempered Upper and Lower Bainitic
Microstructures in a 4150 Steel. Journal of Materials Engineering and Perfomance,
1993. 2 (2): p. 255-263.
135
© University of Pretoria
54.
D.V. Edmonds, R.C. Cochrane, Structure-property relationships in bainitic steels.
Metallurgical Transactions A, 1990. 21: p. 1527-1540.
55.
Y. Tomita, K. Okabayashi, Mechanical properties of 0.40 pct C - Ni - Cr - Mo High
strength steel having a mixed structure of martensite and bainite. Metallurgical
Transactions A, 1985. 16 A: p. 73-82.
56.
F. Caballero, et al., The microstructure of continuously cooled tough bainitic steels.
2nd Int. conference super high strength steels, 2010.
57.
G. Krauss, Steels - Processing, structure and performance, ASM International. 2005.
p. 87-98. Online version available at www.knovel.org
58.
N.G. Pokrovskaya, L.N. Belyakov, I.P. Zhegina, E.Y. Grigor'eva, Influence of
structure on the crack resistance of a high-strength structural steel. Metal Science
and Heat Treatment, 1997. 39 (9-10): p. 415-420
59.
G.F. Vander Voort, Metallography and Microstructures. ASM Handbook, Volume
09, ASM International. p. 355-361.
60.
Standard Practice for Microetching Metals and Alloys. ASTM E 407 - 99, 1999.
www.astm.org
61.
Standard practice for quantitative measurement and reporting of hypoeutectoid
carbon and low alloy steel phase transformation. ASTM A1033 - 10, 2010.
www.astm.org.
62.
G. B ́ ranger, G. Henry, G. Sanz, 1996. The book of steel. Andover, UK. Intercept.
p.134.
63.
Standard test method for determining hardnenability of steel. ASTM A255 - 02, 2002.
www.astm.org.
64.
T. Kop, J. Sietsma, S. Van Der Zwaag, Dilatometric analysis of phase
transformations in hypo-eutectoid steels. Journal of Materials Science, 2001. 36 (2):
p. 519-526.
65.
S. Chupatanakul, P. Nash, Dilatometric measurement of carbon enrichment in
austenite during bainite transformation. Journal of Materials Science, 2006. 41 (15):
p. 4965-4969.
66.
P. Hausild, I. Nedbal, C. Berdin, C. Prioul, The Influence of ductile tearing on
fracture energy in the ductile-to-brittle transition temperature range. Materials
Science and Engineering A, 2002. 335: p. 164-174.
67.
K. Mills, J.R. Davis, Fractography, ASM Handbook, Volume 12, ASM International.
136
© University of Pretoria
68.
W.C. Leslie, 1981. The physical metallurgy of steels. Auckland, New York. McGrawHill: p. 396.
69.
J.H. Woodhead, A.G. Quarrell, Role of carbides in low-alloy creep-resisting steels. J.
Iron Steel Inst., 1965. 203 (6): p. 605-19.
70.
F. Han et al., Effect of molybdenum and chromium on hardenability of low-carbon
boron-added steels. Metals and Materials International, 2008. 14 (6): p. 667-672.
71.
J. Kucera, K. Stransky, J. Dojiva, Diffusion interaction coefficients
and
thermodynamic interaction coefficients
of carbon in alloyed austenitic steels.
Mater. Sci. Eng. A, 1990 (1): p. 75-82.
72.
H. Wada, Thermodynamics of the Fe-Cr-C system at 985 K. Metallurgical and
Materials Transactions A, 1985. 16 (8): p. 1479-1490.
73.
H. Wada, Thermodynamics of the Fe-Mo-C System at 985K. Metallurgical
Transactions A, 1986. 17 A: p. 391-398.
74.
J. Christian, The theory of transformation in metals and alloys, Part 1. Equilibrium
and General Kinetic Theory, 1975: p. 542.
75.
M. Kostic, E. Hawbolt, and L. Brown, Growth kinetics of bainite plates and
widmanst ̈ tten needles in the Β′ phase of a Ag-45at. pct Cd alloy. Metallurgical and
Materials Transactions A, 1976. 7 (11): p. 1643-1653.
76.
J. Cermak, J. Sopousek, K. Stransky, Relation between carbon tracer diffusivity and
carbon activity coefficient in austenitic Ni-Fe-Cr alloys. Scr. Metall. Mater., 1993. 29
(12): p. 1581-1585.
77.
T. Wada, H. Wada, J.F. Elliot, J. Chipman, Activity of carbon and solubility of
carbides in the fcc Fe-Mo-C, Fe-Cr-C and Fe-V-C alloys. Metallurgical Transactions,
1972. 3: p. 2865-2872.
78.
S.J. Lee, D.K. Matlock, and C.J. Van Tyne, Carbon diffusivity in multi-component
austenite. Scripta Materialia, 2011. 64 (9): p. 805-808.
79.
J.A. Lobo, G.H. Geiger, Thermodynamics of carbon in austenite and Fe-Mo austenite.
Metallurgical Transactions A, 1976. 7 A: p. 1359-1364.
80.
J. Ågren, Computer simulations of the austenite/ferrite diffusional transformations in
low alloyed steels. Acta Metallurgica, 1982. 30 (4): p. 841-851.
81.
P. Thibaux, A. Metenier, C. Xhoffer, Carbon diffusion measurement in austenite in
the temperature range 500
to 900 . Metallurgical and Materials Transactions A,
2007. 38 (6): p. 1169-1176.
137
© University of Pretoria
82.
X.Z. Zhang, J.F. Knott, Cleavage fracture in bainitic and martensitic microstructures.
Acta Metallurgica, 1999. 47 (12): p. 3483-3495.
83.
F. Caballero et al., Toughness deterioration in advanced high strength bainitic steels.
Materials Science and Engineering A, 2009. 525 (1): p. 87-95.
84.
L. Lan, C.Qui, D. Zhao, X. Gao, L. Du, Microstructural characteristics and
toughness of the simulated coarse grained heat affected zone of high strength low
larbon bainitic steel. Materials Science and Engineering A, 2011. 529: p. 192-200.
85
C.J. McMahon, M. Cohen, Initiation of cleavage in polycrystalline iron. Acta
Metallurgica, 1965. 13: p. 591-604
138
© University of Pretoria
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement