Overview no. 63 Non-equilibrium grain boundary segregation of boron in austenitic stainless steel-II. Fine scale segregation behaviour

A combination of TEM, FIM, AP and IAP has been used to study boron grain boundary segregation in austenitic stainless steels of the types 316L (with 40 ppm or with <1 ppm boron) and “Mo-free 316L” (23 ppm boron). High resolution segregation profiles were determined for cooling rates from 0.29 to 530°C/s for three starting temperatures: 800, 1075 and 1250°C. Boron grain boundary segregation was found after all heat treatments. The segregation behaviour was mainly of the nonequilibrium type after cooling from 1075 or 1250°C whereas equilibrium segregation dominated after rapid cooling from 800°C. The influence of the relative grain orientation on the amount of non-equilibrium segregation was small for general boundaries. However, no segregation was detected at coherent twin boundaries. The binding energy of boron to austenite grain boundaries was estimated at 0.65 ± 0.04 eV for both types of steels. The influence of the composition and boron content of the steels on the segregation behaviour is discussed and the experimental techniques used are presented.     

Mechanism of the formation of lamellar M23C6 at and near twin boundaries in austenitic stainless steels

Precipitation characteristics of M₂₃C₆ during aging at 800 °C were studied by transmission electron microscopes in two austenitic stainless steels, A (type AISI 316L) and B (type DIN 4981), which had been quenched in water after their solution treatment at high temperatures (1150 °C and 1275 °C, respectively). After precipitation at grain boundaries, M₂₃C₆ precipitated at incoherent and coherent boundaries of twins and inside austenite grains. Close to an incoherent twin boundary on either side of the boundary, M₂₃C₆ mostly grew as elongated plates, although elsewhere in austenite matrix these grew usually as equiaxed particles. The plates lied on planes parallel to the twinning plane and were aligned unidirectionally along the axis of intersection of the twinning plane and a {110} plane perpendicular to the twinning plane. On coherent twin boundaries, similar plates of M₂₃C₆ formed along with some equiaxed particles. Existing models for the mechanism of formation of these lamellar carbides fail to explain the observed features of these carbides formed around different twins. It is suggested that the residual stress localized in the vicinity of twin boundaries in quenched specimens influences the nucleation and growth of these carbides during aging, resulting in their specific morphology.     

Abnormal grain growth in pure materials

Thompson et al. [Acta metall.35, 887 (1987)] have proved that if a large grain of size R_A is introduced in a matrix which has Hillert's [Acta metall.13, 227 (1965)] quasi-stationary grain size distribution this large grain will decrease its size relative to the critical radius of the distribution, R_Cr, so that d(R_A/R_Cr)/dt < 0. In the present work the effect of having a matrix distribution different from Hillert's is examined. It is shown that in such case the behaviour of an abnormally large grain is more complex and either d(R_A/R_Cr)/dt < 0 or d(R_A/R_Cr)/dt > 0 are initially possible. The present analysis is applied to lognormal distributions with different values of side deviation.     

Mechanical twinning in E2412 electrical steel

The deformation of fine-grain E2412 steel containing 3.63% Si, with different grain size is considered. Twinning predominates in its deformation. The samples are extended on an Instron-5565 machine at strain rates \(\dot \varepsilon \) ≈ 0.002–0.660 s^–1 at 183–393 K. Samples of two types are investigated: with around 80% of the grains in the ranges 1.5–9.0 mm and 0.025–0.225 mm. (The mean grain sizes are d _me1 = 3.55 mm and d _me2 = 0.12 mm, respectively.) The relation between the number of twins and the types of steps formed on the strain and strain-rate diagrams in loading is established for the grains with d _me1 = 3.55 mm. At small loading rates, on account of the high growth rate of the twins, step formation on the strain diagrams is accompanied by marked decrease in the load. With increase in the loading rate, the spread of the load is reduced; Δσ changes sign at a strain rate \(\dot \varepsilon \) ≈ 0.04 s^–1. In fine-grain steel (d _me2 = 0.12 mm), there are no visible jumps in the load with the appearance of twins. In the fine-grain steel, the growth time of twins in the grain is small. On account of their high rate of development, the number of twins is also small. The distribution of twinned grains is plotted as a function of the grain size, at different temperatures and loading rates. The peak in the size distribution of the twinned grains is shifted to larger grains in comparison with the overall grain-size distribution of the polycrystal. There is some optimal grain size that facilitates twinning. As a rule, it is higher than the mean grain size determined from the initial grain-size distribution. The number of twins in a particular grain depends on the test temperature and the strain rate. There is a deformation temperature at which the number of twins in the grain is constant at the strain rates employed.     

Dissociation of asymmetric ∑9, ∑27a and ∑81d 〈110〉 tilt boundaries

∑9, ∑27a and ∑8ld〈110〉 tilt boundaries in annealed bulk polycrystals of a Cu + 6 at.% Si alloy are shown to dissociate, and in each case a new intermediate grain is formed which is separated from one of the original grains by a ∑3 coherent twin interface. For ∑9 dissociation, the other two interfaces bounding the intermediate grain are lower valued ∑ interfaces i.e. coherent and incoherent ∑3 boundaries; for ∑27a dissociation, the other interfaces bounding the intermediate grain are higher valued ∑ interfaces, i.e. ∑81d boundaries; and for ∑81d dissociation the other interfaces bounding the intermediate grain are also higher valued ∑ interfaces, i.e. ∑243a boundaries. A mechanism is proposed which describes the atom movements involved in the dissociation of a ∑9 boundary, and it is suggested that similar types of mechanism may account for the observed dissociation of ∑27a and ∑81d boundaries to higher valued ∑ boundaries. The primary GBD structure of a ∑243a〈110〉 tilt boundary is determined and shown to correspond to a 7.36 low angle tilt boundary.     

The mechanisms of improved creep strength in a new austenitic stainless steel

An addition of 40 ppm boron, 0.4 pct vanadium and 0.12 pct nitrogen to an austenitic stainless type steel AISI 316L (0.02 C, 18 Cr, 12 Ni, 2.7 Mo) has considerably improved the creep properties. The improved creep properties are due to a combination of the precipitation of fine stable vanadium nitrides on the dislocations and the precipitation of chromium carbides (M₂₃C₆) in the grain boundaries. The latter process is thought to be enhanced by the presence of boron and helps to improve the creep ductility. The precipitation of vanadium nitrides on the dislocations retard the creep rate. The nitrides retain their small size even after long creep testing times. A model is proposed to explain this behavior of the precipitated particles and their interactions with the dislocations.     

Overview no. 63 Non-equilibrium grain boundary segregation of boron in austenitic stainless steel—IV. Precipitation behaviour and distribution of elements at grain boundaries

The distribution of elements and the precipitation behaviour at grain boundaries have been studied in boron containing AISI 316L and “Mo-free AISI 316L” type austenitic stainless steels. A combination of microanalytical techniques was used to study the boundary regions after cooling at 0.29–530°C/s from 800, 1075 or 1250°C. Tetragonal M₂B, M₅B₃ and M₃B₂, all rich in Fe, Cr and Mo, precipitated in the “high B” (40 ppm) AISI 316L steel whereas orthorhombic M₂B, rich in Cr and Fe, was found in the “Mo-free steel” with 23 ppm B. In the “high B steel” a thin (<2 nm), continuous layer, containing B, Cr, Mo and Fe and having a stoichiometry of typically M₉B, formed at boundaries after cooling at intermediate cooling rates. For both types of steels a boundary zone was found, after all heat treatments, with a composition differing significantly from the bulk composition. The differences were most marked after cooling at intermediate cooling rates. In both types of steel boundary depletion of Cr and enrichment of B and C occurred. It was found that non-equilibrium grain boundary segregation of boron can affect the precipitation behaviour by making the boundary composition enter a new phase field. “Non-equilibrium phases” might also form. The synergistic effect of B and Mo on the boundary composition and precipitation behaviour, and the observed indications of C non-equilibrium segregation are discussed.     

On the kinetics of the spreading of extrinsic grain boundary dislocations

The kinetics of the disappearance (spreading) of extrinsic grain boundary dislocations (EBGDs) at random grain boundaries has been investigated in a low-carbon austenitic stainless steel (0.01C-19Cr-18Ni) by measuring in TEM their disappearance times at the temperature range 360 to 450 °C. Experimental results fit most satisfactorily to a kinetic equation of the general form:t _d =AT _d exp(ΔH _b /RT _d ) derived by Johannesson and Thölen10 and Lojkowski and Grabskiⁱⁱ¹² assuming either a dissociation into discrete structural grain boundary dislocation or the core spreading, respectively. Additional spreading experiments performed on commercial 316 austenitic stainless steel also support this general equation. Calculations employing experimental results indicate that the spreading time of a parallel component of the EGBD's Burgers vector is negligible and does not play any role during the process. Johannesson and Thölen¹⁰ and Lojkowski and Grabskiⁱⁱ¹² models are practically indistinguishable assuming that in the J & T model an EGBD dissociates into an array of grain boundary dislocations with the magnitude of the Burgers vectorb _b ? 0.07 nm. The TEM photographic documentation of the spreading clearly indicates that some grain boundaries exhibiting fine lines (not Moiré fringes) behave during spreading as random (general) boundaries. On the other hand, some grain boundaries free of any fine lines exhibit very high spreading temperature indicating a special behavior.     

Interactions between lattice dislocations and grain boundaries in Ni3Al investigated by means of in situ TEM and computer modelling experiments

The interaction between lattice dislocations and grain boundaries in Ni₃Al has been investigated by means of in situ TEM deformation experiments. The interaction between screw dislocations and a coherent twin boundary could be analyzed in detail. The interaction mechanism found experimentally was compared to the results of a computer modelling study. In the computer modelling study, many-body potentials representing Ni₃Al were used. The results of the in situ straining indicate that 〈110〉 screw dislocations impinging on a Σ = 3 coherent twin boundary that have a Burgers vector that is parallel to the grain boundary plane can be transmitted to the symmetric slip plane in the other grain under influence of an applied stress. A one-to-one comparison with the results of a computer modelling study of exactly the same system in Ni₃Al can be made and the experiment agrees with the simulations. Also, observations were made of superlattice intrinsic stacking faults (SISF) that were formed as a result of the interaction between gliding dislocations and the dislocations of a low angle grain boundary (cell wall). The creation of jogs in the line of the gliding dislocation may be the cause of the SISF formation.     

The Role of Grain Orientation and Grain Boundary Characteristics in the Mechanical Twinning Formation in a High Manganese Twinning-Induced Plasticity Steel

In the current study, the dependence of mechanical twinning on grain orientation and grain boundary characteristics was investigated using quasi in-situ tensile testing. The grains of three main orientations (i.e., 〈111〉, 〈110〉, and 〈100〉 parallel to the tensile axis (TA)) and certain characteristics of grain boundaries (i.e., the misorientation angle and the inclination angle between the grain boundary plane normal and the TA) were examined. Among the different orientations, 〈111〉 and 〈100〉 were the most and the least favored orientations for the formation of mechanical twins, respectively. The 〈110〉 orientation was intermediate for twinning. The annealing twin boundaries appeared to be the most favorable grain boundaries for the nucleation of mechanical twinning. No dependence was found for the inclination angle of annealing twin boundaries, but the orientation of grains on either side of the annealing twin boundary exhibited a pronounced effect on the propensity for mechanical twinning. Annealing twin boundaries adjacent to high Taylor factor grains exhibited a pronounced tendency for twinning regardless of their inclination angle. In general, grain orientation has a significant influence on twinning on a specific grain boundary.     

The mechanisms of improved creep strength in a new austenitic stainless steel

An addition of 40 ppm boron, 0.4 pct vanadium and 0.12 pct nitrogen to an austenitic stainless type steel AISI 316L (0.02 C, 18 Cr, 12 Ni, 2.7 Mo) has considerably improved the creep properties. The improved creep properties are due to a combination of the precipitation of fine stable vanadium nitrides on the dislocations and the precipitation of chromium carbides (M₂₃C₆) in the grain boundaries. The latter process is thought to be enhanced by the presence of boron and helps to improve the creep ductility. The precipitation of vanadium nitrides on the dislocations retard the creep rate. The nitrides retain their small size even after long creep testing times. A model is proposed to explain this behavior of the precipitated particles and their interactions with the dislocations. Formerly with Uddeholm AB, Hagfors, Sweden     

The effect of extrinsic grain boundary dislocations on the grain size strengthening in polycrystals

The grain size dependence of flow stress at room temperature in the regime of small strains (0–2%) has been examined in as-annealed and 2% pre-strained and subsequently annealed specimens of 316L stainless steel. Grain sizes in the range of 3.4–22.4 μm were considered. The stress-strain curves exhibit linear hardening characteristic beyond 0.2% plastic strain. The analysis of the Hall-Petch parameters show a linear increase in σ₀(ϵ) and a linear decrease in K(ϵ) with strain in the as-annealed specimens. The increase in σ₀(ϵ) has been associated with both, the work-hardening processes in the grain interior and the long range stress field of extrinsic grain boundary dislocations (EGBDs). The EGBDs also act as sites of stress concentration thereby making it easier to generate dislocations in the vicinity of grain boundaries. Therefore, K(ϵ) which is function of the stress required to generate dislocations decreases with increasing strain. The observed drop in flow stress as a result of annealing of pre-stained specimens at 800°C has been related with the annihilation of dislocations at and in the vicinity of grain boundaries. Annealing at 550°C (this temperature is sufficient for the delocalization of the cores of EGBDs) does not have any significant effect on the density of dislocation at and in the vicinity of grain boundaries. Therefore, no significant change in the flow stress was obsereved.     

In-situ straining of Ni3Al in a transmission electron microscope

TEM in-situ straining experiments have been performed on thin foils of Ni₃Al and Ni₃Al containing 750 ppm (0.35 at.%) boron. During the straining, gliding APB-coupled dislocations were observed to leave superlattice intrinsic stacking faults (S-ISF's) as debris in their wake confirming a mechanism for S-ISF formation first suggested by Pak et al. [Scripta metall.19, 1081 (1976)]. The APB-coupled dislocations (i) pile-up at grain boundaries, (ii) become extrinsic grain boundary dislocations and (iii) initiate slip in adjacent grains. The characteristics of the crack propagation process indicated that plastic flow, albeit localised, precedes fracture.     

Influence of rolling temperature on the austenite structure and properties of low-carbon microalloyed steel

In laboratory conditions, the influence of the temperature interval in the finishing stage of controlled rolling on the structure of hot-deformed austenite, the final microstructure, and the properties of low-carbon microalloyed steel is studied. Experiments show that reducing that temperature interval within the range T _nr-A _r3 reduces the grain size in the direction transverse to rolling and increases the number of deformational twins within the hot-deformed austenite grains. The effect is to improve the grain size and uniformity of the ferrite-bainite microstructure and hence to improve the mechanical properties of the thick sheet produced.     

Radiation-induced precipitation at grain boundaries in Ni-8 at.% Si

High-resolution and quantitative energy dispersive X-ray spectroscopy (EDX) and transmission electron microscopy (TEM) were used to study radiation-induced precipitation at grain boundaries. A Ni-8 at.% Si alloy which is in the solid solution region of the phase diagram, was irradiated by either protons or α particles to 0.1–0.3 dpa at 750 K. Radiation-induced layers of γ′(Ni₃Si) precipitates were produced on all grain boundaries and even on some coherent twin boundaries. The layers formed asymmetrically only on one side of the grain boundaries with the exception of the coherent twin boundaries on which precipitation was asymmetric. By using helium pre-implantation at 975 K to monitor the original grain boundary locations it was revealed that grain boundary migration was associated with the radiation induced precipitation.     

The Effect of shock hardening on the impact resistance of low-carbon steel

Mild steel plates were deformed by explosive loading prior to the testing of notched specimens in slow and impact bending and of plain specimens in uniaxial tension. The shock loading produced a significant increase (up to 25 ksi) in tensile yield strength σ _Y . The increase resulted from both a high dislocation density and a high density of mechanical twins that were introduced by shock loading. At the same time, the slow-bend and impact transition temperatures were unaffected by the shock hardening. Elastic-plastic analysis showed that the shock loading results in an increase inσ _? ^* , the critical tensile stress required to initiate cleavage below the notch root. It is this increase inσ _? ^* that is responsible for preventing a rise in the brittleness transition temperature in shock-hardened plates, despite the increase inσ _Y . The increase inσ _? ^* was attributed to the increase in twin density, hence an increase in the number of twin boundaries. Although grain boundaries have a primary role in preventing the initiation or propagation of microcracks, extensive testing and analysis demonstrated that the twin boundaries also serve as barriers in this regard.     

Role of Boundary Strengthening on Prevention of Type IV Failure in High Cr Ferritic Heat-Resistant Steels

Microstructure evolution of newly developed 9Cr-3W-3Co-V, Nb steel with boron addition (B steel) has been analyzed during HAZ thermal cycle at the peak temperature of around A_c3 (A_c3 HAZ) and post-weld heat treatment (PWHT) to elucidate the prevention mechanism of type IV failure by boron addition. It was found that enhancement of the boundary strengthening by precipitates is the main reason for prevention of type IV failure by boron addition. In B steel HAZ, original austenite is reconstituted through martensitic α to γ reverse transformation during the heating and original martensite is reconstituted through martensitic transformation during cooling of the A_c3 HAZ thermal cycle. This process allows M₂₃C₆ carbides to precipitate at the prior austenite grain (PAG) and block boundaries during PWHT even if the chemical segregation of carbide forming elements exists. The effect of boundary strengthening on the creep property has also been investigated. Microstructure evolution during creep was compared among Gr.92 with different A_c3 HAZ microstructures prepared by three kinds of heat treatments and B steel. The results revealed that both the boundary length and kernel average misorientation value decreased in all samples during creep. However, this process occurred very rapidly in A_c3 HAZ simulated Gr.92, whereas it was significantly retarded in the other samples with sufficient boundary strengthening by precipitates. This result confirms that the precipitates formed at PAG and block boundaries play the most important role to stabilize the microstructure of A_c3 HAZ simulated samples during creep and prolong the creep life.     

Theory of modelling the isothermal austenite grain growth in a Si‐Mn TRIP steel

The development of a model to predict the isothermal austenite grain growth during soaking of a low carbon Si‐Mn TRIP steel is described. After reviewing the existing models for isothermal grain growth, a general model dⁿ=d₀ⁿ + K₁t exp(K₂/T) was selected and a procedure was delineated to calculate the values of the different constants of the equation starting with the real three‐dimensional austenite grain size. This paper also deals with an improved etching technique to reveal the austenite grain boundaries.     

A rationalisation of fatigue thresholds in pearlitic steels using a theoretical model

From a detailed re-examination of results in the literature, the effects of microstructure sizes, namely interlamellar spacing, pearlitic colony size and the prior austentitic grain size on the thresholds for fatigue crack growth (ΔK_th) and crack closure (K_{cl, th}) have been illustrated. It is shown that while interlamellar spacing explicitly controls yield strength, a similar effect on ΔK_th cannot be expected. On the other hand, the pearlitic colony size is shown to strongly influence ΔK_th and K_{cl, th} through the deflection and retardation of cracks at colony boundaries. Consequently, an increase in ΔK_th and K_{cl, th} with colony size has been found. The development of a theoretical model to illustrate the effects of colony size, shear flow stress in the slip band and macroscopic yield strength on K_{cl, th} and ΔK_th is presented. the model assumes colony boundaries as potential sites for slip band pile-up formation and subsequent crack deflection finally leading to zig-zag crack growth. Using the concepts of roughness induced crack closure, the magnitude of K_{cl, th} is quantified as a function of colony size. In deriving the model, the flow stress in the slip band has been considered to represent the work hardened state in pearlite. Comparison of the theoretically predicted trend with the experimental data demonstrates very good agreement. Further, the intrinsic or closure free component of the fatigue threshold, ΔK_{eff, th} is found to be insensitive to colony size and interlamellar spacing. Using a criterion for intrinsic fatigue threshold which considers the attainment of a critical fracture stress over a characteristic distance corresponding to interlamellar spacing, ΔK_th values at high R values can be estimated with reasonable accuracy. The magnitude of ΔK_th as a function of colony size is then obtained by summing up the average value of experimentally obtained ΔK_{eff, th} values and the predicted K_{cl, th} values as a function of colony size. Again, very good agreement of the theoretically predicted ΔK_th values with those experimentally obtained has been demonstrated.     

The microstructural path of grain-boundary-nucleated phase transformations

Microstructural path functions, S_v(t) = f{X_v(t)}, where X_v is the volume fraction transformed and S_v is the interfacial area per unit volume (separating transformed from untransformed material), are derived for transformations in which the product phase nucleates either at parent phase grain boundaries, grain edges or grain corners and grows at a constant rate. The treatment considers that either there is a fixed number of pre-existing nuclei, or nucleation is at a constant rate, or some combination of these two means of nucleation are operative. The rate equations for X_v are deduced using existing formalisms and by using the same basic premises to treat the problem of impingement, the S_v rate equations are derived for the first time. The microstructural path functions are obtained by eliminating time from these rate equations. An example of the usefulness of the microstructural path function to model recrystallization in an aluminum-gold alloy is presented.