IRON-CARBON BASED ALLOYS
IRON-CARBON BASED ALLOYS
Proefschrift
ter verkrijging van de graad van doctor aan de Technische Universiteit Delft
op gezag van de Rector Magnificus prof.dr.ir. J.T. Fokkema, voorzitter van het College voor Promoties,
in het openbaar te verdedigen op maandag 26 april om 13.00 uur
door
Paola Valentina MORRA
Dottore in Chimica, Università degli Studi di Torino (Italië) Geboren te Asti (Italië)
Samenstelling promotiecommissie:
Rector Magnificus voorzitter
Prof. dr. ir. S. Radelaar Technische Universiteit Delft, promotor
Prof. dr. R.Boom Technische Universiteit Delft
Prof. dr. ir. J.Th.M. de Hosson Rijksuniversiteit Groningen Prof. S. Ioannides Imperial College, London, UK
Dr. A.J. Böttger Technische Universiteit Delft
Dr. J.T. Slycke SKF Engineering Research Centre,
Nieuwegein
Prof. Dr. I.M. Richardson Technische Universiteit Delft, reservelid
Dr. A.J. Böttger heeft als begeleidster in belangrijke mate aan het totstandkomen van het proefschrift bijgedragen.
This research was carried out under project number MS 97003 in the framework of the Strategic Research Programme of the Netherlands Institute for Metals Research (NIMR) in The Netherlands.
1. GENERAL INTRODUCTION 1
1.1 Introduction ... 2
1.2 Martensitic transformation and structure of iron-based martensite... 3
1.3 Tempering behaviour of iron-based martensite... 4
1.4 Influence of an external stress on the tempering and creep behaviour of martensitic steels ... 7
1.5 Outline of this thesis ... 8
References ... 10
2. THE INFLUENCE OF AN EXTERNAL LOAD ON THE TEMPERING OF IRON-BASED MARTENSITES 11 2.1 Introduction ... 12
2.2 Experimental procedures ... 13
2.2.1 Specimen preparation ... 13
2.2.2 Kinetic analysis... 16
2.2.2.1 Differential Scanning Calorimetry ... 16
2.2.2.2 Dilatometry... 17
2.2.3 X-ray diffraction ... 18
2.2.4 Electron Probe MicroAnalysis... 18
2.3 Results and discussion... 19
2.3.1 As-quenched condition: composition and partitioning of alloying elements... 19
2.3.2 Influence of alloying elements and stress on phase Transformations... 23
2.3.2.1 Segregation and clustering of carbon atoms... 26
2.3.2.2 Precipitation of the η transition carbide... 29
2.3.2.3 Decomposition of retained austenite ... 31
2.3.2.4 Precipitation of cementite... 33
2.4 Conclusions ... 36
THE FIRST STAGE OF TEMPERING 39 3.1 Introduction ... 40 3.2 Experimental procedures ... 41 3.2.1 Sample preparation ... 41 3.2.2 Loading technique ... 43 3.2.3 X-ray diffraction ... 43
3.2.4 Scanning Electron Microscopy... 50
3.2.5 Transmission Electron Microscopy ... 46
3.3 Results ... 47
3.3.1 Microstructural analysis ... 47
3.3.1.1 X-ray diffraction ... 47
3.3.1.2 Scanning Electron Microscopy... 51
3.3.1.3 Transmission Electron Microscopy ... 52
3.3.2 Dimensional changes: dilatometry ... 58
3.4 Discussion... 61
3.4.1 Effect of the temperature ... 66
3.4.2 Effect of the stress ... 70
3.4.3 Material related effect... 70
3.5 Conclusions ... 71
References ... 72
4. THE COARSENING-INDUCED PLASTICITY MODEL 73 4.1 Introduction ... 74
4.2 The coarsening-induced plasticity model ... 75
4.3 Conclusions ... 82
COARSENING-INDUCED PLASTICITY MODEL 85
5.1 Introduction ... 86
5.2 Experimental procedures ... 87
5.2.1 Sample preparation ... 87
5.2.2 Transmission Electron Microscopy ... 87
5.2.3 Loading technique ... 88
5.2.4 X-ray diffraction ... 90
5.3 Results ... 91
5.3.1 Microstructure of as tempered material... 91
5.3.1.1 X-ray diffraction ... 91
5.3.1.2 Transmission Electron Microscopy ... 91
5.3.2 Dimensional changes: dilatometry ... 98
5.4 Discussion... 100 5.5 Conclusions ... 106 References ... 107 SUMMARY 108 SAMENVATTING 112 DANKWOORD 117 CURRICULUM VITAE 118
1.1. INTRODUCTION.
The term “martensite” was first proposed in 1895, in honour of the German metallurgist Adolf Martens, to designate the “constituent or structure” found in hardened steel. Martens performed a microstructural analysis of different steels and found that the hardest steel had a regular crystalline structure1.
Later it became clear that the hardness of quenched steel is a property of a plate- or needle-like constituent formed upon quenching from the austenitic field with a rate fast enough to hinder the eutectoidal diffusion controlled decomposition processes (see Fig. 1.1) and the term martensite was used for the constituent itself.
α γ α + cementite 540 580 620 660 700 740 780 820 860 900 C 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Wt %C
Fig. 1.1. Schematic representation of the Fe-rich side of the Fe-C phase diagram.
In the 1920’s it was demonstrated2 (see Fig. 1.2) that martensite derives from a particular type of transformation, a so-called “diffusionless” transformation, completely different from the other phase transformations occurring by diffusion of interstitial or substitutional atoms. Furthermore, this “martensitic” type of transformation has been found in a number of alloy systems, including non-ferrous alloys, and undoubtedly
general terms for any alloy, “martensitic transformation” and “martensite” to designate the process and the product respectively.
Fig. 1.2. Schematic representation of the γ → α’ correspondence according to Bain2.
Iron atoms are indicated by white spheres while possible interstitial sites for carbon atoms are indicated by grey spheres.
1.2. MARTENSITIC TRANSFORMATION AND STRUCTURE OF
IRON-BASED MARTENSITE.
The cooling rate necessary for the martensitic transformation in steels to occur is such that the majority of the carbon atoms in solution in the face centred cubic austenite (γ) phase remain in solution in the ferrite (α) phase. Steel martensite is therefore a supersaturated solid solution of interstitial carbon atoms in a body centred tetragonal metal lattice consisting of iron and substitutional alloying elements. The interstitials tend to occupy dominantly one type of octahedral interstices1, thereby causing a tetrahedral distortion of the metal lattice.
Upon cooling the first martensite plates form at the Ms (martensite start, see Fig. 1.3)
temperature. In low carbon steels Ms is about 773 K, but increasing the carbon content
progressively decreases the Ms temperature. Also substitutional alloying elements such
as Cr, Si, and Ni influence the Ms value3. The Mf (martensite finish, see Fig. 1.3)
increase in the amount of austenite transformed. In practice, in particular in the case of high carbon steel, some retained austenite is left below Mf. The retention of austenite is
attributed to the high stresses between the last martensite plates that form, which tend to suppress growth or thickening of the existing plates.
T
Log time
γ γ+α α+Fe3c
Eutectoid temperature
Fig. 1.3. Schematic representation of the TTT diagram of an eutectoid steel.
1.3. TEMPERING BEHAVIOUR OF IRON-BASED MARTENSITE.
The as-prepared martensite is very hard and brittle because of the strain induced by the interstitial atoms, which distort the metal lattice; therefore most technological steels have to be heat treated after quenching in order to improve ductility and toughness, and in some cases also strength. Furthermore, since the dimensional stability of these steels is very important for their industrial applications, phase transformation-induced size changes occurring at temperatures lower than the in-use temperature of the specific component must be allowed to occur before finishing grinding the single parts.
A thorough understanding of the phase transformations occurring upon tempering of steel martensites becomes thus of fundamental importance in order to be
Mf
required by a specific application. From the results reported in literature, the tempering process of steel martensites can be divided in five steps.
I) The pre-precipitation stage, occurring in the temperature range between room temperature and 373 K, consists in the segregation of carbon atoms to lattice defects and grain boundaries and in the subsequent clustering of the carbon atoms4. This stage is associated with a small volume decrease of the material.
II) The first stage of tempering, occurring in the temperature range between 348 and 373 K, is associated with the precipitation of a transition carbide (first discovered by Jack5 and referred to as ε carbide and later as η carbide by Nagakura6) uniformly
throughout the martensite phase. These precipitates enhance the strength of the material, resulting in the so-called “primary hardening”7, but tend to coarsen with increasing temperature, thereby leading to a strength decrease of the material.
The mechanism of the carbide nucleation is still not well established, but it has been proposed as a possible mechanism that the displacement of the iron atoms produced by interstitial carbon atoms favour the nucleation. Different morphologies, such as platelike8 or rodlike9, have been reported for this carbide, suggesting that alloy composition could have an important influence on the carbide morphology. A decrease of specific length occurs in the temperature range concerned.
III) The second stage of tempering, occurring in the temperature range between 473 and 573 K, is associated with the transformation of retained austenite (γR) in ferrite (α) and
cementite (θ, an orthorhombic M3C carbide). The retained austenite is the fraction of
austenite that has not transformed into martensite after the austenising treatment and the quenching and its quantity can vary depending on the carbon content and on the preparation procedure of the steel. In the case of low-alloy commercial steels (C wt % < 0.2) with high MS temperatures the decomposition of γR has a very important role in the
tempering process because the retained austenite is not stabilised; in medium-carbon martensites γR forms thin interlath films10 (associated with phenomena as the tempered
martensite embrittlement); in steels forming plate martensite (with a C weight % higher than 0.6) γR is trapped in small volumes between martensite plates, the decomposition
often begins at the martensite/austenite interface6. In high-alloy steels γR may be
stabilised so that no observable decomposition occurs during tempering treatments. If internal macrostresses are low the correlation between the amount of the transformed retained austenite and the dimensional change is linear11. The dimensional change, an expansion for carbon contents lower than 2 wt%, depends strongly on the austenite carbon content.
IV) The third stage of tempering, occurring in the temperature range between 523 and 623 K, is associated with the precipitation of cementite (M3C, θ carbide, orthorhombic),
which results in a length reduction. At the same time as the cementite precipitation the carbides formed in the first stage dissolve and supply carbon atoms for the growing θ particles. The cementite initially adopts a platelike morphology; prolonging the tempering leads to coarsening and spheroidisation of cementite particles12. This phenomenon, called Ostwald ripening13, consists in the shrinking and disappearing of the smaller particles and in the consequent growth at their expense of the larger particles. The theory of this process has been developed by Lifshitz and Slyozov14; normally the larger particles situated on interlath boundaries grow at the expense of the intralath particles, giving as a result a final dispersion of cementite particles only on interlath boundaries.
The coarsening of the particles causes a decrease of hardness during tempering of martensite, but alloying additions resulted to have a strong retarding effect on the rate of tempering, even if the addition does not lead to the formation of special carbides13. This effect is very complicated and has not been treated analytically; in order to describe it a model based on the assumption of a local equilibrium at the cementite/ferrite interfaces has been proposed. From this assumption it resulted that the rate of a moving interface in most cases is controlled either by the diffusion of carbon or of a substitutional alloying element15.
V) The fourth stage of tempering is associated with the dissolution of cementite particles that are replaced by more stable alloy carbides; this occurs only in the presence of strong carbide formers such as chromium, molybdenum, tungsten, and vanadium.
alloy carbides are much more finely dispersed than the cementite they replace and, as a consequence, hardness may increase during tempering at these temperatures, resulting in the so-called “secondary hardening”7. Typical carbides that form are: Cr7C3, Mo2C,
W2C, and VC.
1.4. INFLUENCE OF AN EXTERNAL STRESS ON THE
TEMPERING AND CREEP BEHAVIOUR OF MARTENSITIC
STEELS.
Due to the complex microstructure of martensitic steels and to the overlapping sequence of tempering steps, little work has been performed in order to study the influence of an external applied stress on isochronal tempering of martensite. The few data found in literature16 point to the fact that an applied stress does not have an influence on the start temperature and on the kinetics of the precipitation of the transition carbides, whereas it does have an influence (strongly dependent on the type of applied stress) on the kinetics of decomposition of retained austenite, the tempering stage associated with the largest length change in high carbon steels.
More work has been performed in order to study the thermo-mechanical behaviour of heat-treated martensitic steels. In these materials some of the transformations occurring upon tempering have been allowed to occur before the external stress was applied. The elevated temperature creep deformation of engineering materials has received significant attention for many years17,18,19. Unlike the high temperature case, creep deformation at low homologous temperatures (T/Tm < 0.3),
“low temperature creep”, has been given less attention because the creep deformation in this case is small and therefore difficult to detect. Due to the increasing need for high precision components and to the improved measuring accuracy this research field gained interest recently. In particular research was carried out in order to identify the mechanisms leading to the low temperature creep and to be able to predict the thermo-mechanical behaviour of different materials.
Up to now two different approaches have been used in order to describe the low temperature creep of engineering steels: i) the dislocation glide model20 and ii) the transformation-induced plasticity (TRIP) model21.
In the first approach the creep deformation is attributed to dislocation glide. A material stressed at a constant temperature above a certain threshold stress will have a number of dislocations, which are mobile. These mobile dislocations will become immobilised by interaction with other dislocations or obstacles such as precipitates during creep, leading to strain hardening since the stress field around the dislocation clusters will effectively prevent further dislocation glide.
The transformation-induced plasticity model attributes the plastic flow to variations of the phase proportions. Generally two contributing mechanisms are considered:
The Greenwood-Johnson22 mechanism is based on the idea that the
transformation from one phase to another one generates, as a consequence of the volume difference between the two phases, microscopic plasticity in the weaker one, which permits macroscopic plastic flow in the presence of an external load, even if this load in absence of internal stresses would be insufficient to induce plasticity.
The Magee23 mechanism (at first derived for the martensitic transformation and subsequently applied to other transformations with a shear component24,25, e.g. the bainitic transformation) argues that if a martensitic transformation takes place under external loading, formation of martensite plates with a preferred orientation is enhanced, which affects the overall shape of the body.
1.5. OUTLINE OF THIS THESIS.
In this work the attention has been focussed on a martensitic ball bearing steel, alloy SAE 52100, and on two different test materials, a pure Fe-Cr-C and a pure Fe-C alloys with the same chromium and carbon content as the commercial material.
Chapter 2 describes the sequence of transformations occurring upon tempering of these martensitic materials in the temperature range between room temperature and
the activation energies of the different steps and the underlying mechanism are discussed. The influence of an external applied stress on the different tempering stages is investigated and explained in terms of the effect of the load on the rate-determining step of each process.
Chapter 3 focuses on the low-temperature creep behaviour of the commercial ball bearing material and the Fe-Cr-C alloy tempered up to the completion of the first stage of tempering. The materials in this annealing condition consist of a very complex mixture of phases (i.e. ferrite, retained austenite, transition carbides and primary cementite) and from X-ray diffraction data it is clear that retained austenite partially decomposes upon thermo-mechanical treating. A modification of the dislocation glide model taking into account the influence of the phase transformations on the mobile dislocation density is proposed.
In Chapter 4 a low-temperature creep model based on the combination of the Greenwood-Johnson transformation-induced plasticity model with the Ostwald ripening of precipitates is derived. This model, called the coarsening-induced plasticity model, allows coupling the length changes occurring upon coarsening to the dissolution of the precipitates present in the material.
Finally, in Chapter 5 the coarsening-induced plasticity model is successfully applied to the thermo-mechanical behaviour of alloy SAE 52100 tempered up to the completion of the second stage of tempering. In this annealing condition, the material consists of ferrite, transition carbides and primary and secondary cementite. Evidence of growth of the particles located at grain boundaries is found by means of transmission electron microscopy, and the volume transfer caused by the diffusion of carbon atoms from dissolving to coarsening carbides is linked to the low-temperature creep behaviour of the material investigated.
REFERENCES.
1 Z. Nishiyama, “Martensitic Transformation”, Academic Press, New York, 1978, p.3. 2 E.C. Bain, Trans. AIME, 70 (1924) 25.
3 J. Wang, P.J. van der Wolk, S. van der Zwaag, Mat. Trans., JIM, 41 (2000) 761.
4 L. Cheng, C.M. Brakman, B.M. Korevaar, E.J. Mittemeijer, Met. Trans. A, 19A (1988) 2415. 5 G. Krauss, “Principle of heat treatment of steels”, American Society for Metals, Metals Park, Ohio 44073, 1980.
6 S. Nagakura, Y. Hirotsu, M. Kusunoki, T. Suzuki, Y. Nakamura, Met. Trans. A, 14 (1983) 1025. 7 R.W. Honeycombe, “Structure and strength of alloy steels”, Climax Molybdenum CO. LTD., London, 1974.
8 M.G.H. Wells, Acta Met., 12 (1964) 389.
9 Y. Tanaka, K. Shimizu, Trans. JIM, 22 (1981) 779.
10 M. Sarikaya, A.K. Jhingan, G. Thomas, Met. Trans. A, 14 (1983) 1121.
11 F. Hengerer, W. Nierlich, J. Volkmuth, H. Nützel, Ball Bearing Journal, vol. 231, p.26. 12 E.D. Hyam, J. Nutting, JISI, 184 (1956) 148.
13 S. Björklund, L.F. Donaghey, M. Hillert, Acta Met., 20 (1972) 867. 14 I.M. Lifshitz, V.V. Slyozov, J. Phys. Chem. Solids, 19 (1961) 35. 15 M. Hillert, Proc. Int. Conf. Sc. and Techn. Iron and Steel, Tokyo, 1970.
16 R. Millot, P. Archimbault, E. Gautier, J.P. Houin, A. Badard, J. Bellus, C. Hunter, Y. Desalos, F. Rucksthul, Proc. 3rd European Mechanics of Materials Conf., 1998.
17 D.R. Hayhurst, J. Mech. Phys. Solids, 20 (1972) 381.
18 L. Finnie, W.R. Feller, “Creep of engineering materials”, London, MaGraw-Hill Book Company, 1959. 19 L. Bendersky, A. Rosen, A.K. Mukherjee, Int. Met. Rev., 130 (1985) 1.
20 F.R.N. Nabarro, Mat. Sc. Eng. A 309 (2001) 227.
21 J.B. Leblond, J. Devaux, J.C. Devaux, Int. Journ. Plast., 5 (1989), 551. 22 G.W.Greenwood, R.H. Johnson, Proc. Roy. Soc., A283 (1965) 403.
23 C.L. Magee, “Transformation Kinetics, Microplasticity and Aging of Martensite in Fe-31Ni”, PhD Thesis, Carnegie Institute of Technology, Pittsburgh, PA.
24 H.K.D.H. Bhadeshia, S.A. David, J.M. Vitek, R.W. Reed, Mat. Sc. Techn., 7 (1991) 686. 25 G.I. Rees, P.H. Shipway, Mat. Sc. Eng. A, 223 (1997) 168.
THE INFLUENCE OF AN EXTERNAL LOAD ON THE
TEMPERING OF IRON-BASED MARTENSITES
2.1. INTRODUCTION.
Iron-based martensitic (α’) alloys are a class of materials widely used in applications because of their excellent mechanical properties. They are obtained by quenching the appropriate iron-based alloy from the austenitic phase field and by subsequently tempering the hard but brittle material thus obtained in order to adjust the mechanical properties and to bring about a microstructural stabilisation.
There are two main parameters that can be varied to optimise the mechanical properties of precision components meant for determinate working conditions: the composition and the tempering treatment. Obviously in order to choose the appropriate parameter values a detailed knowledge of the processes occurring upon tempering these materials and of the effect of alloying elements on these processes is required.
The as-quenched martensitic interstitial carbon steels consists of three phases: martensite (a supersaturated solid solution of interstitial carbon atoms in a metal lattice1), retained austenite (γR) and primary carbides (cementite θ, an orthorhombic
M3C carbide) that remained partly undissolved after the austenitisation treatment and
the quench. During tempering the following processes2,3 occur: the segregation and clustering of interstitial carbon atoms, the precipitation of a transition carbide, the decomposition of retained austenite and the conversion of the transition carbide into the more stable cementite. These phase transformations can partially or completely overlap depending on the steel composition and microstructure.
Since the iron-based martensitic parts are mostly used under stress a thorough understanding of the effect of an applied stress on the phase transformations occurring, and therefore of the dimensional stability of the material4, becomes very important.
In this chapter the tempering processes of martensites obtained from cast material SAE 52100, and the test cast materials Fe-Cr-C and Fe-C with about the same C and Cr contents as the commercial alloy are discussed. The Fe-C test material was chosen in order to compare the results with literature data, whereas the Fe-Cr-C test material was selected in order to investigate the influence of Cr and of the minor alloying elements on the transformation kinetics. The effect of alloying elements on the decomposition behaviour of these martensitic alloys is investigated in the temperature
scanning calorimetry (DSC) and dilatometry. The combination of these techniques is essential here because some of the phase transformations (segregation and clustering of carbon atoms) show a significant heat release but almost no length change whereas others such as the precipitation of cementite are attended with a pronounced length change and a small heat release. In order to investigate the effect of the stress on the phase transformations occurring upon tempering of the commercial materials, dilatometer experiments under compressive stress (25 to 200 MPa) have been performed in the temperature range from RT to 723 K. Furthermore the use of dilatometry gives directly information about the dimensional stability of the investigated materials. Activation energies have been obtained by performing a Kissinger-like analysis5 (see
section 2.2.2) and were used to infer the rate-determining step for the different stages of decomposition. Structural and compositional analysis of the martensitic specimens has been performed by means of X-ray diffraction (XRD) and electron-probe microanalysis, respectively.
2.2. EXPERIMENTAL PROCEDURES.
2.2.1. SPECIMEN PREPARATION.
Three different cast materials of various overall composition, denoted by alloys Fe-C, Fe-Cr-C and SAE 52100 respectively, were investigated. The compositions were obtained by IR-detection according to ASTM E1019 for carbon and sulphur and by optical emission according to ASTM E415 for all the other alloying elements (see table 2.1). After casting the alloys have been processed in a hot iso-static press at a temperature of 1423 K and a pressure of 120 MPa during two hours, followed by slow cooling, allowing some homogenisation and the elimination of any porosity. Finally a soft annealing treatment, consisting of austenitising at 1093 K for 1 hour followed by (i) slow cooling (10 K/hour) down to 963 K and (ii) air cooling down to room temperature, was performed.
For the DSC and XRD experiments small disc-shaped specimens (diameter of about 5 mm and thickness of about 0.2-0.3 mm) have been cut from the cast pieces. These discs have been austenitised in a vertical tube furnace at 1133 K for 14 minutes under a H2
flow and then quenched in brine (10 wt % NaCl in water) at room temperature (BQ) in order to obtain the martensitic structure. To avoid decarburisation and oxidation during the austenitising treatment the specimens were covered with Condursal 0090 paint. For the preparation of martensitic specimens for the dilatometer experiments see section 2.2.2.2.
Tab le 2.1 . Co mpo sition s o f th e cas t allo ys in ves tig ated in th e pres en t w ork. Cr at % (wt%) C a t% (wt%) Mn at % (wt%) Si at % (wt%) Ni at % (w t %) Cu at % (wt%) Mo at % (w t%) Al at % (w t%) S a t% (wt%) P at% (wt% ) A lloy F e-C 0.02 (0.02) 4. 35 (0.97) 0.09 (0 .09) 0.02 (0.01) 0. 03 (0.03) 0.006 (0.007) 0. 005 (0. 009) 0.010 (0.005) 0.013 (0.00 8) 0.010 (0.006) A lloy Fe -C r-C 1.46 (1.39) 4. 39 ( 0.98 ) 0.09 (0.09) 0. 02 (0.01) 0. 03 (0.03) 0.007 (0.008) 0. 006 (< 0.01) 0.016 (0.008) 0.013 (0.00 8) 0.010 (0.006) A lloy SA E 52100 1.43 (1.36) 4. 51 (1.01) 0.31 (0.32) 0.48 (0.25) 0. 15 (0.16) 0.10 (0.12) 0.02 (0.04) 0.0026 (0.0013) 0. 033 (0.020) 0.023 (0.01 3)
2.2.2. KINETIC ANALYSIS.
The values of activation energy for the phase transformation occurring during tempering of the martensitic alloys have been determined using a Kissinger-like method5 based on the equation: constant ln ' 2 ' = + f f RT E T ϕ
where R is the gas constant, Tf’ is the transformation temperature corresponding to a
certain fraction (f’) of phase transformed and ϕ is the heating rate in K/min. The peak maximum has been chosen in most of the cases as Tf’; when it was not possible to
distinguish a distinct peak maximum (DSC measurements pertaining to the segregation of carbon atoms) because of the overlap with another phase transformation, Tf’ was
chosen as the point where the slope of the measured curve (i.e. the measured heat flow) diverges 2 % from the tangent of the rising part of the peak3. The activation energy was
calculated by taking the slope of the straight line obtained by plotting ϕ 2 ' f T ln against ' f RT 1
. The standard deviation of this slope was used as estimation for the error in the
activation energy.
2.2.2.1 DIFFERENTIAL SCANNING CALORIMETRY.
Differential scanning calorimetry (DSC) was performed using a Perkin-Elmer DSC7 apparatus in a protective Ar atmosphere. Calibration was performed by measuring the well-established melting points of high purity indium and lead. The mass of the specimens used varied between 30 and 60 mg. The specimens were sealed in an aluminium pan and heated from 303 to 773 K with heating rates (ϕ) of 5 to 30 K/min and then cooled down to 303 K; an empty pan was used as a reference. The baseline was determined by reheating the specimens with the same heating rates.
differential scanning calorimeter. The activation energy for the formation of cementite could not be determined by DSC because of the small heat release involved.
2.2.2.2. DILATOMETRY.
Dilatometry was carried out using a Bähr 805 dilatometer. Specimens of alloys Fe-Cr-C (of length L = 5 mm and diameter φ = 5 mm) and SAE 52100 (L= 7 mm, φ = 4 mm) have been heated up to 1133 K in 5 minutes, held at that temperature for 14 minutes under vacuum (0.01 Pa) and then quenched to 303 K in He (denoted by QM, i.e. quench mode). This treatment led to a retained austenite content in the martensitic specimens of about 12-14 vol %; this amount allowed to investigate the decomposition of γR also by
means of dilatometry. The quenching rate thus obtained was however not high enough to allow the formation of martensite in alloy Fe-C. After this treatment specimens of alloy Fe-Cr-C were tempered up to 823 K with heating rates of 5 to 20 K/min and specimens of alloy SAE 52100 were tempered up to 823 K with heating rates of 5 to 30 K/min without an applied stress and under a compressive stress of 25, 50, 100, 150 and 200 MPa.
In order to reach a quenching rate high enough to allow the formation of martensite in alloy Fe-C specimens of alloys Fe-C (L= 5 mm, φ = 4 mm), Fe-Cr-C (L = 5 mm, φ = 5 mm) and SAE 52100 (L = 7 mm, φ = 4 mm) have been heated up to 1133 K in 5 minutes, held at that temperature for 14 min under vacuum (0.01 Pa), quenched to 183 K in He (denoted by DC, i.e. deep cooling mode). After this treatment the retained austenite content of the samples was reduced: this enabled to observe more accurately the transformation temperature of the precipitation of cementite (that is accompanied by a length decrease), which otherwise is hindered by the overlapping decomposition of γR
(that is accompanied by a length change of opposite sign i.e. a length increase). In order to determine the transformation temperatures pertaining to each phase transformation the data have been arithmetically averaged over time and subsequently differentiated with respect to the temperature. The temperatures corresponding to a maximal change in length are taken as the transformation temperatures, no further corrections were applied. The values of activation energy for the precipitation of the η carbide, the decomposition of retained austenite and the formation of cementite have been determined using
dilatometry. The other transformations are accompanied by a length change too small to be measured.
2.2.3. X-RAY DIFFRACTION.
X-ray diffractometry was performed at RT using a Bruker-AXS type D5005 diffractometer equipped with a Co tube and a graphite monochromator in the diffracted beam to select the Co Kα radiation. Measurements have been performed in the range
63-86 °2θ with a step size of 0.020 °2θ and a counting time per step of 60 seconds on brine-quenched alloys Fe-C, Fe-Cr-C and SAE 52100 in order to determine the lattice parameters of martensite (using the reflections {002} and {200} of martensite). The lattice parameters can be related to the carbon content of the martensite matrix1. Before being measured the samples have been polished in order to remove the oxide layer on the surface; an electrolytical polishing method has been chosen because mechanical thinning would have caused the decomposition of retained austenite. The electrolytical polishing leads to some roughening of the sample surface, which could cause an error in the determination of the lattice parameter. This effect of displacement has been taken into account by applying a thin layer of a reference powder (silicon: NB6 Si 640a) to the surface of the sample. The Si reflections have been used to correct the 2θ values pertaining to the martensite reflections.
2.2.4. ELECTRON PROBE MICRO-ANALYSIS.
In order to investigate the partition of the alloying elements between the primary carbides and the matrix (martensite/austenite) in the as-quenched condition, the martensitic alloys Fe-Cr-C and SAE 52100 have been investigated by means of electron probe micro-analysis (EPMA).
The measurements were performed with a JEOL JXA 8900R microprobe using an electron beam with an energy of 10 keV and a current of 20 nA. The composition of the measured points was determined using the X-ray intensities for the constituent elements. All X-ray intensities of the main alloying elements (i.e. C and Cr in the case of alloy
Fe-except that of Fe, which was calculated by difference. The surface of the specimen was decontaminated using an air-jet prior to each measurement.
The samples (prepared as described in section 2.2.2.2, QM) were polished with 0.25 µm diamond paste and a 0.05 µm alumina suspension. In order to reveal the microstructure of finely dispersed carbides (approx. 1 µm) the specimens were etched for 5 seconds in 1% Nital solution.
For both materials (alloy Fe-Cr-C and SAE 52100) the composition of points at a distance of about 0.5 µm along two straight lines crossing a primary carbide has been measured. The points with a carbon content lower than 1 wt% have been considered to be part of the martensitic / austenitic matrix and the points with a carbon content higher than 5 wt% have been considered as carbides.
2.3. RESULTS AND DISCUSSION.
2.3.1. AS-QUENCHED CONDITION: COMPOSITION AND PARTITIONING OF ALLOYING ELEMENTS.
In order to study the influence of the alloying elements on the phase transformations occurring upon tempering of iron-based martensites, a detailed knowledge of the actual composition of the phases taking part to the transformations is necessary. In what follows the results of the analyses aimed to investigate the partition of the alloying elements between the martensitic matrix and the primary carbides will be given.
The average values of the composition of the matrix and the carbides of alloys Fe-Cr-C and SAE 52100 as obtained by EPMA are shown in table 2.2. Although quantitatively the results differ somewhat from the overall compositions of the alloys, as established by IR-detection and optical emission (see table 2.1), clear trends in the distribution of the elements are observed. Part of the carbon is contained in the primary carbides; as a consequence the carbon content of the matrix is lower than the overall carbon content. The content of elements such as Cr, Mn and Mo (see Fig. 2.1) increases with increasing carbon content, suggesting that these elements tend to concentrate in the primary carbides. The Si, Cu and Ni contents (see Fig. 2.2) decrease with increasing carbon content, pointing to the fact that these elements are mostly contained in the matrix.
Table 2.2. Average va lues of the com position of the ma trix and t he c arbide s of all oys F e-Cr-C a nd S A E 52100 as obta ine d by EPMA. Cr at% (w t %) C at% (wt %) Mn at% (wt %) Si at% (w t %) Ni at% (w t% ) C u at% (w t% ) Mo at% (w t% ) Alloy F e-Cr-C matrix 1.28 (1.22) 4.22 (0.94) - --Alloy F e-Cr-C carbides 5.58 (6.30) 23.33 (6.18) - --Alloy SAE 52100 matrix 1.16 (1.10) 3.69 (0.82) 0.203 (0.206) 0.562 (0.292) 0.226 (0.246) 0.128 (0.151) 0.015 (0.026) Alloy SAE 52100 carbides 7.00 (7.74) 20.94 (5.43) 0.524 (0.622) 0.115 (0.070) 0.077 (0.098) 0.012 (0.017) 0.057 (0.119)
8 6 4 2 0 Cr wt% 7 6 5 4 3 2 1 C wt% SAE 52100 Fe-Cr-C 0.8 0.6 0.4 0.2 0.0 5 4 3 2 1 C wt% Mn wt% Mo wt%
Fig. 2.1. Correlation between the carbon content and a) the Cr content of martensitic alloys SAE 52100 and Fe-Cr-C and b) the Mn and Mo contents of martensitic alloy SAE 52100.
a)
0.3 0.2 0.1 0.0 Si wt% 5 4 3 2 1 C wt% 0.5 0.4 0.3 0.2 0.1 0.0 5 4 3 2 1 C wt% Ni wt% Cu wt%
Fig.2.2. Correlation between the carbon content and a) the Si content and b) the Ni and Cu contents of martensitic alloy SAE 52100.
The carbon content of the matrix of the alloys Fe-C, Fe-Cr-C and SAE 52100 has been determined also by means of X-ray diffraction. To this end the dependency of the lattice parameters of martensite on the carbon content given in Ref. 6 has been used. It was
a)
(Fe,Cr)3C carbides has been estimated on the basis of the carbon content of the matrix
and the overall carbon content of the steels. A volume fraction of retained austenite of 14% (as obtained by X-ray diffraction, see Chapter 3) is taken in the calculation. The resulting carbon content of the matrix and the volume fraction of primary carbides are shown in table 2.3.
Clearly alloying elements such as Cr stabilise the primary carbides leading to volume fractions of about 4-5 vol% whereas less than 1 vol% of primary carbides is present in the unalloyed Fe-C specimen.
Table 2.3.
C at % (wt. % ) matrix vol% carbides θ
Alloy Fe-C 4.16 (0.93) 0.6
Alloy Fe-Cr-C 3.26 (0.73) 3.8
Alloy SAE 52100 3.12 (0.70) 4.7
2.3.2. INFLUENCE OF ALLOYING ELEMENTS AND STRESS ON PHASE TRANSFORMATIONS.
Analysis of enthalpy and length changes was used to investigate the influence of alloying elements and the presence of a compressive stress on the tempering process. The heat evolutions and changes in specific length induced in the investigated alloys by non-isothermal annealing employing a heating rate of 10 K/min are shown in Figs. 2.3 (DSC, BQ), 2.4 (dilatometry, QM) and 2.5 (dilatometry, DC). The effect of stress on the specific length changes is reported in Fig. 2.6.
70 60 50 40 30 20 10 0 Heat flow · 10 3 (W/g) 700 600 500 400 Temperature (K) alloy Fe-C alloy Fe-Cr-C alloy SAE 52100 segregation and clustering of C atoms precipitation of ε/η decomposition of γR precipitation of θ
Figure 2.3. DSC curves of three brine quenched martensitic specimens on isochronal (φ = 10 K/min) annealing. 20 15 10 5 0
Derivative of length change · 10
6 (1/K) 700 600 500 400 Temperature (K) alloy Fe-Cr-C
alloy SAE 52100 precipitation of ε/η decomposition of γR
precipitation of θ
Fig. 2.4. Derivative of length change on isochronal (φ = 10 K/min) annealing of two martensitic specimens in the QM (quenching mode).
-20 -15 -10 -5 0 5 10 15
Derivative of length change · 10
6 (1/ K) 700 600 500 400 Temperature (K) alloy Fe-C alloy Fe-Cr-C alloy SAE 52100 precipitation of ε/η decomposition of γR precipitation of θ
Fig. 2.5. Derivative of length change on isochronal (φ = 10 K/min annealing of two martensitic specimens in the DC (deep cooling mode).
-20 -10 0 10 20
Derivative of length change · 10
6 (1/ K) 700 600 500 400 Temperature (K) 0 MPa 25 MPa 50 MPa 100 MPa 150 MPa 200 MPa precipitation of ε/η decomposition of γR precipitation of θ
Fig. 2.6. Effect of an applied uniaxial compressive stress on the derivative of length change on isochronal (φ = 20 K/min) annealing of martensitic alloy SAE 52100.
2.3.2.1. SEGREGATION AND CLUSTERING OF CARBON ATOMS.
This so-called pre-precipitation stage starts already during the quenching process and proceeds at room temperature. It consists in the segregation of carbon atoms to lattice defects and grain boundaries and in the subsequent formation of carbon clusters in the iron matrix3. This tempering stage does not show a distinct length change, but is characterised by a significant heat release (see first peak in Fig. 2.3). The heat effect due to these transformations overlaps with the heat effect of the subsequent transformation (the precipitation of the η transition carbide) and therefore it is not possible to discern a distinct peak maximum in the heat flow. The activation energy was determined as indicated in section 2.2.2. According to high resolution transmission electron microscopy results for Fe-C 7 the segregation of carbon atoms to dislocations and grain
boundaries is followed by the development of carbon enrichments of variable carbon content that become a modulated, periodic structure upon decomposition. This modulated structure disappears after tempering above 373 K.
The values of the activation energies for the segregation and clustering, given in table 2.4, are within the range 81-94 kJ/mol, which is in good agreement with the values found in literature for Fe-C martensite 8 i.e. 83 kJ/mol for segregation; 79 kJ/mol for clustering. The diffusion of carbon in martensite (assuming the activation energy value for this process to be almost the same as for the diffusion of carbon in ferrite9, i.e. 80 kJ/mol) is considered as the rate-determining step for this pre-precipitation stage. The data for pre-precipitation suggest a slight increase of the activation energy for diffusion of carbon due to the presence of alloying elements.
The influence of stress on the activation energy of this process could not be investigated because of the absence of a distinct length change associated to the segregation and clustering; anyway no influence of stress on the mobility of interstitial carbon atoms in α-Fe is expected16.
Table 2.4. A ctivation ener gy values for te mp erin g p ro ces se s in F e-C b as ed marten si tes de rive d from non isot he rm al -a nne al in g e xperi m ent s. DSC Allo y F e-C (kJ/ mol ) DSC All oy Fe-Cr-C (kJ /m ol ) DSC All oy SAE 52100 (kJ/mol) Dil atometry Al lo y F e-C (kJ/ mol) DC D ilatometry Al lo y F e-C r-C (kJ/ mol ) DC / Q M D ilatometry A lloy S A E 52100 (kJ/mol ) DC / QM Se gr eg at io n and c luste ri ng of C a toms 81 ± 5 89 ± 1 6 94 ± 1 2 _ _ / _ _ / _ Pr ecipitation of η tr an si tio n carbide 118 ± 5 114 ± 2 111 ± 7 135 ± 3 102 ± 34 / 115 ± 11 115 ± 9/ 113 ± 2 D eco mp os itio n of re ta in ed au st enite ( γ)R 135 ± 2 141 ± 5 144 ± 2 _ _ / 156 ± 2 _ / 140 ± 3 Pr ecipitation of cemen tite (θ ) I _ _ _ 195 ± 2 168 ± 7 / 1 64 ± 8 170 ± 6 / 163 ± 33
T ab le 2. 5. A ctiv atio n e nerg y v alu es fo r tempe rin g p ro cesse s i n al lo y SAE 5 210 0 un der an appl ie d co m pr essi ve st re ss. D ilato metry A llo y S A E 5 21 00 (kJ/mo l) 0 MPa 25 M P a 50 MPa 100 MPa 150 MPa 20 0 M P a P recip itatio n of η tran sitio n ca rb ide 11 3 ± 2 121 ± 2 12 5 ± 8 10 7 ± 6 115 ± 5 113 ± 5 Deco m po si tion of reta ined au sten ite ( γR ) 14 0 ± 3 143 ± 7 14 6 ± 5 13 4 ± 8 119 ± 4 104 ± 11 P rec ipita tion of cemen tite ( θ) I 163 ± 33 13 8 ± 8 171 ± 2 151 ± 6 15 2 ± 1 5 79 ± 11 P rec ipita tion of ce men tite ( θ) I I _ _ _ 300 ± 14 38 3 ± 2 2 420 ± 33
2.3.2.2. PRECIPITATION OF THE η TRANSITION CARBIDE.
The precipitation of a transition carbide, first discovered by Jack10 and referred to as ε carbide and later as η carbide by Nagakura11 and Han7, occurs in the temperature range
of about 353-453 K. The precipitation is involving a pronounced heat production and a significant length decrease. The second heat release in Fig. 2.3 (differential scanning calorimetry) and the first maximum in the derivative of the length change in Figs. 2.4 and 2.5, (dilatometry) have been associated3 with the precipitation of the transition carbide.
The activation energies obtained without an applied stress, given in table 2.4, are within the range 110-120 kJ/mol (DSC) and 100-115 kJ/mol (dilatometry). The differences in the values of the activation energy obtained by DSC and those obtained by dilatometry could be caused by the different quenching procedures used (see sections 2.2.1 and 2.2.2.2.) which results in differences in microstructure a.o. internal stresses, dislocations densities. The values for the activation energies are compatible with those found in literature12 i.e. about 113 kJ/mol, obtained from dilatometer experiments13 for high purity Fe-C alloys and two commercial steels varying in carbon content from 0.6 to 1.4 wt%.
The activation energies observed are in between those for diffusion of C in α-Fe and those for diffusion of iron along dislocations (i.e. pipe diffusion of Fe, 152 kJ/mol14). Dislocations are expected to occur upon η precipitation in order to accommodate the specific volume misfit between the transition carbide and the matrix.
The activation energies observed are about the same within the experimental errors for all three alloys. On the basis of the transition carbide/matrix misfit and the composition of the alloys, however, it is expected that the activation energy of Fe-C is somewhat higher than that of the other two alloys (Fe-Cr-C and SAE 52100) as follows. The misfits are about -0.44%, -0.36% and -0.34% for alloy Fe-C, Fe-Cr-C and SAE 52100, respectively; these differences are mainly due to the differences in the amount of dissolved carbon in the matrix (see Table 2.3.). This suggests a somewhat higher amount of misfit dislocations formed in alloy Fe-C than in alloy Fe-Cr-C and SAE 52100.
For what concerns the influence of alloying elements on the activation energy it has been reported (on the basis of resistivity measurements) that carbide forming alloying
elements reduce the activation energy of the order of 5 to 10 kJ/mol per wt% alloying element 15 whereas elements like Ni and Si raise the activation energy. On the basis of the overall compositions the activation energies for alloy Fe-Cr-C and SAE 52100 are expected to be not more than 10 kJ/mol lower than the one for alloy Fe-C. From the discussion above it can be concluded that the expected effect of composition (substitutional alloying elements and carbon) on the activation energies is of the same order of magnitude as the experimental error implying that the current results are not conclusive in this respect.
In Fig. 2.7 the activation energy for alloy SAE 52100 is given as a function of applied stress. The activation energy does not seem to be affected by an external stress. If indeed diffusion processes related to growth are determining the kinetics of the precipitation of the transition carbide no influence of stress is expected because (i) strong effect of stress on dislocation core diffusion (pipe diffusion) seems not very likely, and (ii) on the basis of experimental data of activation volumes for carbon diffusion 16,17 the influence of (hydrostatic) pressures up to 200 MPa is estimated to be at most ~ 0.1 kJ/mol. 140 130 120 110 100
Activation energy (kJ/mol)
-200 -150 -100 -50 0
Stress (MPa)
precipitation transition carbide SAE 52100
Fig. 2.7. The activation energies pertaining to the precipitation of the transition carbide as a function of the applied external load for alloy SAE 52100. The application of an
The transformation temperature, extrapolated to zero scanning rate, for the precipitation of the transition carbide is the same for the two Cr containing materials (about 402 K for alloy Fe-Cr-C and SAE 52100) and somewhat lower for the Fe-C alloy (i.e. 397 K). This suggests that chromium retards the precipitation of the transition carbide. No influence of the presence of silicon is observed in this stage of tempering.
2.3.2.3. DECOMPOSITION OF RETAINED AUSTENITE.
The decomposition of retained austenite3 into a microstructure comprising that of lower bainite (ferrite and cementite), taking place in the temperature range of about 473-623 K, causes a large heat production and a pronounced length increase. The decomposition of retained austenite and the formation of cementite (length decrease) overlap completely (see Fig. 2.3). The activation energy values obtained for the decomposition of retained austenite without stress are in the range 135-156 kJ/mol and agree well (in particular the DSC data) with the values of the activation energy as calculated on the basis of the carbon content dependent diffusion of carbon in austenite18,19 : 138 kJ/mol for alloy Fe-C, 140 kJ/mol for alloy Fe-Cr-C and 141 kJ/mol for alloy SAE 52100). This indicates that carbon diffusion in austenite is the rate-determining step for the phase transformation. No significant influence of the alloying elements on the activation energy is observed. This is in contrast to the expected influence of Si, element that concentrates in the matrix (martensite/austenite) and therefore suppresses carbide precipitation20, or the increase of activation energy for carbon diffusion due to Cr (for Fe-0.8 wt%C-1.41 wt%Cr an activation energy of 164 kJ/mol is reported21 as compared to 130 kJ/mol for Fe-0.8 wt%). This hints to a paraequilibrium formation of cementite for which the concentration of the alloying elements is the same as that of the parent (austenite) matrix. The carbide precipitates observed in SAE 52100 specimens heat treated for 4 hours at 523 K (see Chapter 3) were established to be of the same composition of the matrix thereby confirming the presence of paraequilibrium.
The differences of the phase transformation temperatures for alloys Fe-C, Fe-Cr-C and SAE 52100 (occurring at 544, 547 and 564 K respectively for alloys Fe-C, Fe-Cr-C and SAE 52100, ϕ= 10 K/min) is ascribed to the effect of alloying elements. Since alloying elements such as Cr22 and Mo23 are concentrated in the undissolved (primary) carbides
the actual concentration in the matrix is lower than the overall composition. Unfortunately the compositions as obtained by EPMA are not sufficiently accurate to be used as an estimate for the composition of the matrix as the overall content obtained by EPMA differs from the actual overall content by about ± 20%. The alloys Fe-C, Fe-Cr-C and SAE 52100 contain (overall contents) in total 0.12, 0.12 and 0.60 wt% austenite stabilising24 elements i.e. manganese, nickel and copper, respectively. The differences in the austenite stabilising elements contents between alloy Fe-C, Fe-Cr-C and alloy SAE 52100, and the presence of silicon in the SAE 52100 alloy can be adduced as an explanation for the difference between the transformation temperatures of alloys Fe-C and Fe-Cr-C, which are about the same, and that of the commercial material, which is significantly higher. The presence of silicon in the SAE 52100 alloy does not seem to suppress the carbide precipitation. This is in agreement with literature as reported in a study of bainite forming in Fe-Ni-Si-C for Si contents below 0.5 at% 25.
150 140 130 120 110 100 90
Activation energy (kJ/mol)
-200 -150 -100 -50 0
Stress(MPa) decomposition retained austenite
SAE 52100
Fig. 2.8. The activation energies pertaining to the decomposition of retained austenite as a function of the applied external load for alloy SAE 52100. The application of an external compressive load higher than 50 MPa during the decomposition of retained
The applied stress influences both the transformation temperature and the activation energy of this stage of tempering (see Fig. 2.8 and Table 2.5). As a consequence of the maximum applied compressive stress of 200 MPa the transformation temperature is shifted to about 30-40 K higher values, while the activation energy decreases about 40-50 kJ/mol (see Table 2.5). The increase in the transformation temperature can be understood by the mechanical energy (compression) that suppresses the transformation, which is accompanied by a volume increase. The fact that the measured activation energy significantly decreases for applied compressive stresses above 50 MPa can be due to a change in the process occurring caused by the increasing influence of the shear component of the transformation at higher stress values. Also, at the transformation temperatures involved in this stage of tempering and under an applied external load, other processes such as creep and transformation plasticity occur, which influence the value of the (effective) activation energy measured.
2.3.2.4. PRECIPITATION OF CEMENTITE.
This phase transformation, occurring upon tempering between about 523 and 723 K3 and involving the conversion of the transition carbides into the more stable cementite, is associated with a small change of enthalpy and with a significant length decrease and therefore only dilatometry could be applied to investigate the kinetics of this stage of tempering. Since the precipitation of cementite and the decomposition of retained austenite overlap, martensitic samples with a lower retained austenite volume fraction have been prepared in the deep cooling mode (see section 2.2.2.2.). Clearly, the alloying elements play an important role in this transformation (see Fig. 2.5): the precipitation temperature is about the same for alloy Fe-C and alloy SAE 52100 and it is shifted to lower values in the case of alloy Fe-Cr-C. Alloy Fe-Cr-C contains in fact as only alloying element Cr, which favours the precipitation of cementite, while alloy SAE 52100 contains both carbide stabilising elements (i.e. Cr, Mn, Ni, Mo) and the element Si that suppresses carbide precipitation. Also the shape of the (derivative of) length change depends on the alloy composition. In the case of alloy Fe-C the derivative of the length change with respect to the temperature is a narrow, symmetrical peak, in the case of alloy Fe-Cr-C the peak becomes broader but it is still rather symmetrical and for what concerns alloy SAE 52100 a large broadening and strong asymmetry are
observed. According to the data reported in literature26 for a medium carbon steel, nickel and manganese shift the temperature of the end of the process towards higher temperatures while other alloying elements such as silicon, even in small concentration, tend to increase the transformation temperature. Atom probe analysis of tempered Fe-1.84 at%C-3.84 at%Si-2.95 at%Mn27 (for 0.5 hr at 623 K) showed that the cementite initially formed is neither enriched with the carbide forming Mn nor depleted of Si, whereas the carbon content is 25 at%, as it is expected for cementite.
The values of activation energy determined are in the range 163-195 kJ/mol, in between the values indicated in literature for the pipe diffusion of iron14 (152 kJ/mol) and the volume diffusion of iron28 (251 kJ/mol). Therefore the activation energy obtained could be interpreted as an effective one combining these two mechanisms. Note that the pipe diffusion of iron atoms (and most probably also of the substitutional alloying elements), that may occur along dislocations that form to accommodate the volume misfit caused by the phase transformation, may be less dominant for the precipitation of cementite than in the case of the precipitation of the transition carbide. The precipitation of cementite occurs in fact at higher temperatures, where the volume diffusion becomes relatively more important.
The application of a stress during tempering clearly influences the transformation: the minimum in the derivative of the length change with respect to the temperature associated with this transformation becomes broader and, starting from a stress of -100 MPa, it splits in two distinct minima. The activation energy values and the transformation temperatures for the first one are very difficult to evaluate (in particular in the case of the measurement performed under a stress of –200 MPa) because of the overlap with the previous transformation, which increases with increasing stress. As for the decomposition of retained austenite the transformation to cementite contains a shear component29,30. This could decrease the activation energy because nucleation is enhanced in those regions favourably oriented with respect to the externally applied stress. In addition, since increasing deformation of the matrix occurs upon increasing applied stress, it is suggested that also the formation of dislocations enhances the kinetics of transformation by the larger contribution of pipe diffusion to the process. The second minimum shifts to higher temperatures with increasing stress and the
diffusion of Cr, Si, Mn and Ni31 in ferrite, which suggests the attribution of this signal to the slow redistribution of alloying elements in the carbides. The increase in activation energy with increasing stress is, within the accuracy of the parameter determination, linear (see Fig. 2.9 ), but the order of magnitude of the increase (about 2.5 10-1 kJ/(mol·MPa)) is larger than that expected by the theory of substitutional diffusion under a stress field32 (about 2.5 10-3 kJ/(mol·MPa)).
200 160 120 80
Activation energy (kJ/mol)
-200 -150 -100 -50 0 Stress (MPa) cementite formation θ1 a 450 400 350 300
Activation energy (kJ/mol)
-200 -150 -100 -50 0
Stress (MPa) cementite formation θ2 b
Fig. 2.9. The activation energies pertaining to the precipitation of cementite as a function of the applied external load for alloy SAE 52100. The application of an external compressive load during the precipitation of cementite results in the observation of two separate length changes each with a different activation energy as given in a) for the first process (denoted by θ1) occurring at the lowest temperature and b) for the process (denoted by θ2) occurring at higher temperature.
2.4. CONCLUSIONS.
The precipitation of the intermediate carbides and the precipitation of cementite (decomposition of retained austenite) in the early stages of tempering of the ternary alloys Fe-Cr-C and SAE 52100 occurs in para-equilibrium conditions i.e. the concentrations of the substitutional alloying elements in the product phase are inherited from the parent phase whereas the equilibrium concentrations for carbon are reached thereby reducing the free energy of the whole system. Equilibrium partitioning of the alloying elements occurs at higher temperatures.
An externally applied stress does not affect the kinetics of the segregation and clustering of carbon atoms and the precipitation of the intermediate η carbide. The kinetics of these processes is mainly governed by the diffusion of carbon atoms and some pipe diffusion in the case of the precipitation of the η carbide.
The kinetics of the decomposition of retained austenite is governed by carbon diffusion in austenite, which is not affected by the presence of alloying elements. The kinetics of and the formation of cementite are governed by pipe diffusion and volume diffusion of matrix atoms. The kinetics of the decomposition of retained austenite and the formation of cementite both are enhanced in the presence of an externally applied compressive stress. This is attributed to the effect of mechanical energy on, in particular, the shear component of the nucleation.
REFERENCES.
1 C.M. Wayman, Introduction to the Crystallography of Martensitic Transformations, Macmillan, 1964.
2 G.R. Speich, Metals Handbook, vol. 8, p. 202, ASM, 1973.
3
M.J. van Genderen, A. Böttger, Th.H. de Keijser, E.J. Mittemeijer, Met. Trans. A, 28A (1997) 545.
4 E.B. Mikus, T.J. Hughel, J.M. Gerty, A.C. Knudsen, Trans. ASM, 52 (1960) 307.
5 E.J. Mittemeijer, J. Mat. Science, 27 (1992) 3977.
6 L. Cheng, A. Böttger, Th.H. de Keijser and E.J. Mittemeijer, Scripta Met. et Mat., 24 (1990) 509.
7 K. Han, M.J. van Genderen, A. Böttger, H.W. Zandbergen and E.J. Mittemeijer, Phil. Mag. A, 81 (2001) 741.
8 L. Cheng, C.M. Brakman, B.M. Korevaar, E.J. Mittemeijer, Met. Trans. A, 19A (1988) 2415.
9 A.E. Lord, D.N. Beshers, Acta Met., 14 (1966) 1659.
10
G. Krauss, Principles of Heat Treatment of Steels, American Society for Metals, Metals Park, Ohio 44073, 1980.
11 S. Nagakura, Y. Hirotsu, M. Kusunoki, T. Suzuki and Y. Nakamura, Met. Trans. A, 14A (1983) 1025.
12 B.S. Lement and M. Cohen, Acta Met., 4 (1956) 469.
13 C.S. Roberts, B.L. Averbach and M. Cohen, Trans. ASM, 45 (1953) 576.
14 M. Cohen, Trans. JIM, 11 (1970) 145.
15 H.W. King and S.G. Glover, JISI, 196 (1960) 281.
16 J. Bass and D. Lazarus, J. Phys. Chem. Solids, 23 (1962) 1820.
17 M. Wuttig, J. Keiser, Phys. Rev. B, 3 (1971) 815.
18 J. Ågren, Scripta Met., 20 (1986) 1507.
19 C. Wells, W. Batz and R.F. Mehl: Trans. AIME, 188 (1950) 553.
20 D. Quidort and Y.J.M. Brechet, Acta Mat., 49 (2001) 4161.
21
M. Munirajulu, B.K. Dhindaw and A. Biswas, Scripta Mat., 37 (1997) 1693.
22 Z. Glowacki and A. Barbacki, JISI, 185 (1972) 724.
23 R.C. Thomson and M.K. Miller, Applied Surface Science, 87/88 (1995) 185.
24 E.C. Bain and H.W. Paxton, Alloying Elements in Steel, American Society for Metals, 1966, p.124-125.
25 D. Quidort and Y.J.M. Brechet, Acta Mat., 49 (2001) 4161.
26 A.S. Kenneford and T. Williams, JISI, 185 (1957) 467.
27 S.S. Babu, K.Hono, T. Sakurai, Appl. Surf. Sc. 67 (1993) 321.
28
F.S. Buffington, K. Hirano and M. Cohen, Acta Met. 9 (1961) 434.
29 J.W. Stewart, R.C. Thomson, H.K.D.H. Bhadeshia, J. Mat. Sc. 29 (1994) 6079. 30 J.R. Patel, M. Cohen, Acta. Met. 1 (1953) 581.
31 E. A. Brandes and G.B. Brook, Smithells Metals Reference Book, 1992, p. 13-70 – 13-97.
LOW TEMPERATURE CREEP BEHAVIOUR OF
MARTENSITIC IRON-BASED ALLOYS TEMPERED UP
TO THE COMPLETION OF THE FIRST STAGE OF
TEMPERING
3.1. INTRODUCTION.
Hardened martensitic steels containing retained austenite are widely used in industrial applications because of their outstanding mechanical properties. In particular the presence of retained austenite was shown to improve fatigue life and to benefit fracture toughness1. Since the annealing stabilisation treatment implemented in order to obtain the desired microstructure in this class of materials is performed at relatively low temperatures, creep deformation at homologous temperatures, T/Tm below 0.3, can
occur. This phenomenon, known as low-temperature creep2, has not received as much
attention as high temperature creep, probably because the time dependent deformations are small and therefore difficult to detect.
Nevertheless, the relatively small dimensional changes occurring at low temperatures can affect the working life of precision components such as ball bearings, gears and shafts and therefore it is very important to understand the mechanisms inducing microplastic deformation. In the last fifteen years, thanks to the improved measuring accuracy, more attention was paid to low temperature creep and the theories developed in order to account for it suggest that the driving force for creep is provided by thermally activated processes like dislocation glide3 or transformation induced plasticity4.
In this work the attention was focussed on the martensitic ball bearing steel SAE 52100 and on a ternary test Fe-Cr-C material with the same Cr and C contents as the commercial alloy tempered up to the completion of the first stage of tempering5. It is well known that, hardened, as-quenched martensitic (SAE 52100) steels comprises three phases, martensite, retained austenite, and undissolved carbides6. Whilst the latter phase is stable up to temperatures of approximately 990 K, the martensite and retained austenite are subject to phase transformations at relatively low temperatures7 (T < 473 K). The microstructure of SAE 52100 and Fe-Cr-C, when heat treated up to the completion of the first stage of tempering, is very complex and consists of cementite and other carbides in a ferritic/austenitic matrix. The low temperature creep behaviour of these alloys, investigated by means of dilatometry, was correlated to the structural changes studied by means of X-ray diffraction and transmission electron microscopy. In
austenite, and the effect of a broad stress range going from values lower than the yield strength of the weaker phase (austenite) to values higher than the yield strength of the majority phase (ferrite).
3.2. EXPERIMENTAL PROCEDURES.
3.2.1. SAMPLE PREPARATION.
All experiments were performed using (parts of) the dilatometer samples that were prepared as follows. Small cylinders with a diameter of about 4 mm and a length of about 7 mm were cut from a bar 2 m long and with a diameter of 20 mm of soft-annealed alloy SAE 52100 and from an ingot 20 cm long and with a diameter of 25 mm of soft annealed pure Fe-Cr-C test material. The composition of the alloys is given in Table 3.1.
The heat treatment necessary to obtain the desired mixture of phases and microstructure was performed by using a Bähr dilatometer type 805 in the “quenching mode”. Specimens were heated up for austenitisation to 1133 K in 5 minutes, held at that temperature for 14 minutes under vacuum (10-2 Pa), quenched to 303 K in He gas, heated up to 428 K with a heating rate of 10 K/min, held at that temperature for 1 hour and then cooled down again to 303 K in He gas in 2 minutes. This tempering procedure leads to a material with a very complex microstructure consisting of at least two different types of carbides in a ferritic/austenitic matrix (see also section 3.3.1).
tion of al loys F e- Cr-C and SA E 52100 a s obt ai ne d by IR-de te ct ion for C a nd S (a cc ording to AS TM E1019) emi ssion for a ll other e lem ent s (a cc ording to AS TM E415). . Cr at% (wt%) C at% (wt%) Mn at% (wt%) Si at% (w t%) Ni at% (wt %) Cu at% (w t%) Mo at% (w t%) A l at% (wt% ) S at% (wt % ) P at% (wt % ) 1.46 (1.39) 4.39 ( 0.98 ) 0.09 (0.09) 0.02 (0.01) 0.03 (0.03) 0.007 (0.008) 0.006 (< 0.01) 0.016 (0.008) 0.013 (0.008) 0.010 (0.006) 1.43 (1.36) 4.51 (1.01) 0.31 (0.32) 0.48 (0.25) 0.15 (0.16) 0.10 (0.12) 0.02 (0.04) 0.0026 (0.0013) 0.033 (0.020) 0.023 (0.013)
3.2.2. LOADING TECHNIQUE.
The thermo-mechanical behaviour of the commercial and the test materials was investigated at different temperatures and several stress values in order to analyse the effect of these two parameters on the dimensional changes occurring.
Isothermal long duration load experiments were performed on alloys SAE 52100 and Fe-Cr-C samples prepared as described in section 3.2.1 with a Bähr dilatometer type 805 in the “constant load” mode. An uniaxial compressive stress (σA) of 300 MPa was
applied for various periods of time ranging from 1 to 4 days to SAE 52100 samples at temperatures between 348 and 418 K and to alloy Fe-Cr-C samples at temperatures between 398 and 418 K.
In order to evaluate the effect of the applied load additional experiments were performed at 418 K: uniaxial compressive stresses of 100, 200, 400 and 500 MPa were applied to the SAE 52100 samples and uniaxial compressive stresses of 100 and 500 MPa were applied to alloy Fe-Cr-C samples for periods of time ranging from 1 to 4 days.
Because of the long duration of the experiments and the relatively small length changes occurring, a correction procedure was required to compensate for the variations in the output of the linear variable differential transformer (LVDT) caused by variations of its temperature (not more than 5 K). A linear correction function was derived, taking 408 K as a reference temperature of the LVDT, from a calibration measurement during three days using a Pt sample. All the data used in this work are the result of averaging at least three different measurements performed at the same temperature.
3.2.3. X-RAY DIFFRACTION.
X-ray diffraction (XRD) experiments were performed using the dilatometer specimens prepared as described in section 3.2.1 in order to determine the volume fraction of retained austenite (γR) and to investigate the possible decomposition occurring during
thermo-mechanical treatments (see Table 3.2). In particular the attention was focussed on the effect of a compressive stress on the phase transformation and therefore most of the XRD measurements on alloy SAE 52100 were carried out on both specimens annealed under load (-300 MPa) as well as on specimens annealed without load.
The measurements were performed at room temperature using a Bruker-AXS type D5005 diffractometer equipped with a Co tube and a graphite monochromator in the diffracted beam to select the Co Kα radiation. The diffracted intensity was recorded in
the range 57-130 °2θ with a step size of 0.040 °2θ and a counting time per step of 32 seconds. The samples were fixed to the middle of the holder with tape and their surfaces were levelled to the height of the holder. In order to take into account the error caused by the displacement of the specimen from the diffractometer axis a thin layer of silicon was applied to the surface of the sample. The Si standard reflections were used to correct the 2θ values pertaining to the ferrite and austenite reflections.
Since the amount of retained austenite can differ of about ± 1% from sample to sample most probably due to small differences (e.g. ± 1 s in a quenching time of about 30 s) in the quenching procedure it was not possible to follow the decomposition process as a function of time by simply measuring the amount of phase transformed in different samples subjected to the same thermo-mechanical treatment for different periods of time. Therefore just one specimen was used for each type of thermo-mechanical treatment. The specimens were investigated by means of XRD before the tempering treatment and after each one of the annealing treatments of a few hours performed, adding up to in total 144 hours. The thermo-mechanical treatments (reported in Table 3.2) were performed in a Bähr dilatometer type 805 in the “constant load” set-up.
The volume fraction of retained austenite was calculated following the well-established methodology outlined in Ref. 8 based on the direct comparison method, by using the integral intensities of the {200} and {211} reflections of α and the {220} and {311} reflections of γR.
