Intragranular ferrite nucleation on MX carbonitrides and dislocations

The competing mechanisms of ferrite nucleation on (Ti,V)(C,N) MX carbonitrides and dislocations, as well as their dependence on deformation, are investigated experimentally by thermo-mechanical treatments and via simulation. The impact of recrystallization and the resulting austenite grain size on intragranular ferrite nucleation is evaluated. The austenite-to-ferrite transformation temperatures, affected by different microstructures, are examined by isothermal uniaxial single-hit compression tests on a dilatometer DIL 805 and compared to thermal treatments without deformation. Different resulting microstructures are analyzed by using optical light microscopy. The experimental data is used for the validation of thermo-kinetic simulations with the mean-field modeling software MatCalc using the implemented models for the austenite grain evolution, the on-particle nucleation of ferrite on the surface of MX particles, the dislocation density evolution, and recrystallization.


Introduction
Microstructure refinement is induced by intragranular ferrite (IGF) formation, thus realizing optimized mechanical properties of microalloyed steel [1].In contrast to allotriomorphic ferrite (GBF), which forms at austenite grain boundaries, IGF precipitates at nucleation sites within austenite grains [2].There are two types of IGF, which differ in their morphology: the polygonal-shaped idiomorphic ferrite (PF) and the acicular ferrite (AF) with a needle-shaped appearance [3].The preference for AF nucleation rises with a bigger size of the potential nucleation site and lower transformation temperatures [4,5].The intragranular precipitation of both ferrite variants is affected by the interplay of the prior austenite grain size (PAGS), the cooling rate, the steel composition, and the associated varying available nucleation sites [6].For high amounts of IGF, austenite grains have to be quite large (with an optimum of ~150 µm for AF) to ensure the presence of sufficient nucleation sites within the grains [7].The requirements on the cooling rate to room temperature depends on the previously applied thermomechanical treatments (tmt), e.g., austenitization and deformation.It should be chosen fast enough to impede the formation of pearlite and grain boundary ferrite [8].Microalloying of low carbon steels with Ti, Nb, or V is necessary for ferrite formation on MX (M=Ti,Nb,V; X=C,N) precipitates [1].Besides the nucleation of MX phases on primary precipitates [9], they are known to form on dislocations [10].
Zajac et al. [11,12] have shown that vanadium carbonitrides represent the most effective nucleation site for IGF.Ferrite additionally nucleates on MX, which form on primary MnS during thermal treatments without deformation [9,13].Deformation during thermal treatments generally initiates strain-induced precipitation of MX particles, which can act as a formation site for IGF [14][15][16].Increased stress also elevates the number of dislocations serving as potential nucleation sites [17][18][19].However, to date, the competition between these two mechanisms has not been researched systematically.
Previously published work by the authors [20] revealed that when computationally simulating thermal treatments without deformation, the formation of intragranular ferrite can be correctly described by the particle-stimulated nucleation of ferrite on MX carbonitrides without the effect of dislocations considered.In this study, heat treatments and compression tests are performed to investigate the tendency for the nucleation of ferrite on MX particles (MXF), as compared to the impact of dislocations on intragranular ferrite formation, which is denoted DF in the following.This will give insight into the competing precipitation sites for IGF, and will serve as validation base for simulative prediction.

Material and methods
Within the present work, microalloyed steel with low carbon content is investigated.The chemical composition of the examined steed is given in Table 1.

Experimental
The thermo-mechanical experiments are performed on a dilatometer DIL 805.They are schematically illustrated in Figure 1.S-type thermocouples are welded to the specimens under protective argon atmosphere, in order to control the temperature during the tests.Molybdenum plates are applied to both ends of the samples before clamping them inside a copper coil.The experiments are performed under a vacuum atmosphere, and cooling is conducted using argon.The cylindrical samples with 5 mm diameter and 10 mm length are solution annealed at 1553 K for 10 min and subsequently cooled down with 20 Ks −1 to 1153 K.At this temperature, they are either kept for 125 s/1600 s, without deformation, or held for 25 s/1500 s with additional deformation via uniaxial isothermal single-hit compression.The uniaxial deformation comprises compression with 0.01 s −1 to a strain of 1.It is therefore assured that the samples remain the same time at 1153 K, regardless of whether the samples get compressed or not.The two holding times at 1153 K are selected to investigate the impact of precipitates on the microstructure evolution.The dilatometer needs at least 25 s to reach 1153 K without fluctuations.This time is, therefore, the shortest time to be chosen for the state where hardly any particles are present.Contrary, the longer holding time of 1600 s at 1153 K results in higher amounts of precipitates.
Subsequently, the samples are cooled down with 7 Ks −1 to room temperature.Comparative preliminary tests with varying cooling rates (water quench, 70 Ks −1 , 35 Ks −1 , 7 Ks −1 , 0.7 Ks −1 ) from 1153 K to room temperature on a Gleeble ® 3800 assured that the cooling rate of 7 Ks −1 delivers the highest amounts of IGF.The microstructure of the specimens after cooling to room temperature is analyzed using optical light microscopy (LOM) using a Zeiss AxioImager.The sample preparation consists of cutting, hot embedding, grinding, polishing, and 5% nital etching to examine the final microstructure.To allow for the examination of prior austenite grain size, the PAGS is evaluated on the water-quenched samples from the preliminary tests [20].The procedure is based on DIN EN ISO 643 [21] using the etching methods "Bechet Beaujard" (100 ml picric acid, 1.6 g CuCl2, and 0.5 ml aegon) and "Görens" (40 cm³ distilled water, 60 cm³ ethanol, and 20 g FeCl3) [22].
As the formation of IGF depends on the austenite grain size, the austenite grain growth must be computationally assessed.It is calculated using a model developed by Rath and Kozeschnik [26].The on-particle nucleation model [20] is used for the nucleation of carbonitrides on primary phases and the formation of ferrite on the surface of MX particles.Here, the phase boundary between the new phase and the pre-existing particle serves as nucleation site.The formation process is controlled by the diffusion rate of atoms within the matrix.A description of the strain-dependent dislocation density evolution is necessary for the deformation-induced nucleation of MX carbonitrides [27] and ferrite on dislocations.The model in MatCalc is based on the findings of Kocks and Mecking [28,29] and consists of one term that describes the generation of dislocations (term A), a second one that represents dynamic recovery (B), and a third one for static recovery (C) [30].The authors discussed the physics-based determination of these parameters in previously published work [31].Not only is the dislocation density evolution needed for the calculation of potential nucleation sites, but also, the dislocation density evolution plays a decisive role for recrystallization.This work uses the model by Buken et al. [32][33][34][35][36] for nucleation, growth, and coarsening of recrystallized grains.
The model parameters and their value settings for the computation of grain size, MX precipitation, and ferrite formation during the studied tmt, which is equal to the experimental treatment of Figure 1, are discussed by the authors in [20].

Results and discussion
3.1.Microstructure analysis 3.1.1.Microstructures after treatments without deformation.Figure 2 shows that the intermediate holding time at 1153 K hardly impacts the non-deformed final microstructure at room temperature.A ferritic pearlitic structure is achieved for both cases, 125 s and 1600 s of holding.Former austenite grains can be clearly distinguished due to the presence of grain boundary ferrite.Examination of the final PAGS revealed that the austenite grain size reaches 142 µm after 125 s holding and 156 µm after 1600 s holding.The deviation of 14 µm lies within the statistical error range.Therefore, there is no indication of coarsening of the austenite grains during isothermal holding.Within the prior austenite grains, acicular ferrite is formed.Intragranular-formed polygonal-shaped idiomorphic ferrite is not visible.It is assumed that the size of the MX particles present is big enough for AF to suppress the formation of PF since the tendency for acicular ferrite formation is increased with a bigger size of the potential nucleation site.In the close-up of the microstructure images (Figure 2 (c) and (d)), intragranular nucleation of ferrite on pre-existing particles is marked by arrows.Compared to the non-compressed samples, ferrite predominates, and the amount of pearlite decreases.The final PAGS of 16 µm at the experiment with 25 s holding and 21 µm with prolonged holding before the compression indicate that coarsening is insignificant.The prior austenite grains reach lower values due to recrystallization when comparing the deformed samples to the non-deformed ones.Within the final microstructure of the compressed specimens, the distinction between grain boundary ferrite and IGF, especially polygonal-shaped idiomorphic ferrite, is hardly possible.Areas of clearly distinguishable sole acicular ferrite are present; they are assumed to be located within subgrains or at subgrain boundaries.

Comparison of simulation results and experiments 3.2.1. Prior austenite grain size evolution.
The experimental PAGS is compared to the results from the simulation in Figure 4 (a).During the simulation of the annealing treatment at 1553 K, the austenite grains grow from an initial grain size of 135 µm to 153 µm.Prolonged holding at the test temperature of 1153 K does not impact the austenite grain size.The final PAGS (156 µm) of the non-deformed samples from the metallographic examination correlates well with the simulation result.The simulation of the thermal treatment without compression indicates that the equilibrium dislocation density of 10 11 remains unchanged during the whole process, Figure 4 (b).As a result, no recrystallization of the austenite grains occurs.In contrast, the additional deformation at 1153 K raises the strain-dependent dislocation density, leading to recrystallization of the austenite grains.From 153 µm, before the deformation, the grains reach a size of 23 µm after compression.This value coincides with the recrystallized austenite grains' experimental final grain size (21 µm).

Ferrite evolution.
The evolution of microalloy carbonitride and ferrite precipitation during the thermo-mechanical treatments is simulated.As vanadium carbonitrides are expected to be the most effective MX nucleation sites for IGF, the V(C,N) precipitates are analyzed in detail, Figure 5. Here, the term V(C,N) refers to the sum of all vanadium carbonitride phases formed independently if they nucleated on dislocations or on the surface of pre-existing MnS.
During the thermal treatment demonstrated in Figure 5 (a) as well as the compression test (b), precipitation of V(C,N) starts during the first quench.Holding at 1153 K increases the formed particles' size and phase fraction while the number density remains unchanged.Additional compression at 1153 K significantly increases the V(C,N) number density during the second quench.Deformation induces the formation of dislocations acting as nucleation sites for MX phases during cooling to room temperature.The mean radius decreases, and the phase fraction rises.Within the computational analysis of the ferrite formation, we divide IGF into two phases: ferrite nucleating on the surface of pre-existing MX particles (MXF) and ferrite, which forms on dislocations (DF).The evolution of grain boundary ferrite GBF, MXF, and DF during the thermo-mechanical treatment is illustrated in Figure 6.When comparing the final ferrite phase fraction of the pure heat treatments, Figure 6 (a), to the compression tests, Figure 6 (b), the GBF drops from 10% to 1%, MXF decreases from 25% to 3%, and DF increases from 17% to 85%, respectively.Therefore, the total phase fraction of intragranularly formed ferrite rises from 43% to 88%.These results correlate well with the microscopic images of nondeformed and compressed samples Figure 2 and Figure 3, respectively.
As deformation is applied, the number of dislocations increases, and potential nucleation sites for ferrite within the austenite grains thus escalate.This correlates well with the final microstructure after compression, Figure 3.The strain-induced formation of dislocations enhances the precipitation of DF during cooling, while the tendency of ferrite to form on MX phases is negligible.The high amount of dislocations acting as potential formation sites for ferrite outnumber the carbonitride phases available for on-particle nucleation.When only a heat treatment is applied and there is no deformation, the probability for IGF to form on the surface of particles is higher.In contrast, ferrite prefers to nucleate on the numerous dislocations at tests with deformation.

Validation of simulated MXF and DF.
Plotting the dilatation signal against temperature [37], the austenite-to-ferrite (γ→α) transformation temperatures during the individual treatments are assessed.It is examined whether the experimental transformation temperature corresponds to the formation of MXF or DF. Figure 7 displays the experimentally determined and simulated γ→α transformation temperatures.Figure 7 (a) shows the results of the heat treatments without compression of the samples.The results from the experiments correlate well with the simulated formation temperature of ferrite nucleating on the surface of MX precipitates with a maximum deviation of only 7 K.In contrast, the γ→α temperature of ferrite forming on dislocations would be higher.Notably, during holding at 1153 K, coarsening of the precipitates (as discussed in Figure 5 (a)) occurs, and therefore more preferable nucleation sites for MXF are available.As a consequence, the experimental as well as the simulated transition temperature after prolonged holding rises slightly.Simulation supports the interpretation that the dilatometer signal refers to the nucleation of MXF (with 25 % phase fraction), as it represents the predominating nucleation mechanism compared to DF (with 17 % phase fraction).
Compression elevates the austenite-to-ferrite transformation temperature from dilatometer tests to >1000 K, independently of varying intermediate holding times, Figure 7 (b).As the γ→α temperature for MXF differs from the experimental values >40 K, the results from the dilatometer tests are assumed to correspond to the formation of DF.This indicates that deformation boosts the number of dislocations, acting as nucleation sites, to such an extent that the prolonged holding and the associated coarsening of carbonitrides have no impact on the ferrite formation.

Summary and conclusion
Heat treatments and compression tests with varied intermediate holding at the test temperature of 1153 K were performed on microalloyed steel to examine the nucleation tendency of intragranular ferrite to form on MX carbonitrides or on dislocations and to illuminate the role of deformation.Microstructure analysis showed that prolonged holding hardly impacts the final microstructure of the samples for both non-deformed and deformed treatments.Acicular ferrite prevails at the treatments without compression.Deformation raises the amount of precipitated IGF to higher values.Simulation results of thermal treatments without deformation agree with the experimental data that ferrite nucleation on MX particles controls the IGF formation.Analysis of the microstructures showed that higher amounts of PF, in comparison to the predominating acicular ferrite of the non-deformation tests, are achieved.The austenite grain evolution from experiments and simulation agree that deformation induces recrystallization, leading to the reduction of PAGS by one order of magnitude.The increased dislocation density results in higher amounts of DF in comparison to MXF.The thermomechanical treatments involving compression showed that dislocations are obviously increased to such a high level that ferrite nucleation on dislocations predominates.Accordingly, the γ→α temperature from dilatometer tests matches the simulated DF formation temperature.

3. 1 . 2 .
Microstructures after deformation treatments.The light microscopy images of the samples with deformation show that the resulting microstructures after different holding times are the same, Figure3.

Figure 3 .
Figure 3.Light microscopy images of deformed dilatometer samples after cooling with 7 Ks −1 to room temperature.(a), (c) 25 s holding before deformation at 1153 K; (b), (d) 1500 s holding before deformation at 1153 K.

Figure 4 .
Figure 4. Evolution of parameters during thermal treatments (no def.) and compression tests (def.) with 1600 s holding at 1153 K. Comparison of experimental (Exp.) and computational (Sim.)austenite grain evolution (a); Dislocation density evolution and recrystallized fraction of austenite grains from simulations (b).

Figure 5 .
Figure 5. Phase fraction, mean radius, and number density of V(C,N) phases during simulation of tmt with 1600 s holding at 1153 K; (a) no deformation; (b) with deformation.Within the computational analysis of the ferrite formation, we divide IGF into two phases: ferrite nucleating on the surface of pre-existing MX particles (MXF) and ferrite, which forms on dislocations (DF).The evolution of grain boundary ferrite GBF, MXF, and DF during the thermo-mechanical treatment is illustrated in Figure6.

Figure 6 .
Figure 6.phase fraction, mean radius, and number density of ferrite phases during simulation of tmt with 1600 s holding at 1153 K; (a) no deformation; (b) with deformation.When comparing the final ferrite phase fraction of the pure heat treatments, Figure6(a), to the compression tests, Figure6(b), the GBF drops from 10% to 1%, MXF decreases from 25% to 3%, and DF increases from 17% to 85%, respectively.Therefore, the total phase fraction of intragranularly formed ferrite rises from 43% to 88%.These results correlate well with the microscopic images of nondeformed and compressed samples Figure2and Figure3, respectively.As deformation is applied, the number of dislocations increases, and potential nucleation sites for ferrite within the austenite grains thus escalate.This correlates well with the final microstructure after compression, Figure3.The strain-induced formation of dislocations enhances the precipitation of DF during cooling, while the tendency of ferrite to form on MX phases is negligible.The high amount of dislocations acting as potential formation sites for ferrite outnumber the carbonitride phases available for on-particle nucleation.When only a heat treatment is applied and there is no deformation, the probability for IGF to form on the surface of particles is higher.In contrast, ferrite prefers to nucleate on the numerous dislocations at tests with deformation.

Figure 7 .
Figure 7.Comparison of austenite-to-ferrite transformation temperature during thermal treatments and compression tests, determined by simulation and dilatometer experiments; (a) no deformation; (b) with deformation.Figure7(a) shows the results of the heat treatments without compression of the samples.The results from the experiments correlate well with the simulated formation temperature of ferrite nucleating on the surface of MX precipitates with a maximum deviation of only 7 K.In contrast, the γ→α temperature of ferrite forming on dislocations would be higher.Notably, during holding at 1153 K, coarsening of the precipitates (as discussed in Figure5 (a)) occurs, and therefore more preferable nucleation sites for MXF are available.As a consequence, the experimental as well as the simulated transition temperature after prolonged holding rises slightly.Simulation supports the interpretation that the dilatometer signal refers to the nucleation of MXF (with 25 % phase fraction), as it represents the predominating nucleation mechanism compared to DF (with 17 % phase fraction).Compression elevates the austenite-to-ferrite transformation temperature from dilatometer tests to >1000 K, independently of varying intermediate holding times, Figure7 (b).As the γ→α temperature for MXF differs from the experimental values >40 K, the results from the dilatometer tests are assumed to correspond to the formation of DF.This indicates that deformation boosts the number of dislocations, acting as nucleation sites, to such an extent that the prolonged holding and the associated coarsening of carbonitrides have no impact on the ferrite formation.